stuck in mind - diva portaloru.diva-portal.org/smash/get/diva2:445438/fulltext01.pdfthank you steven...

Stuck in Mind

This dissertation is dedicated to my parents,

and to Johan & Joar

Örebro Studies in Psychology 21

IDA FLINK

Stuck in Mind The role of Catastrophizing in Pain

© Ida Flink, 2011

Title: Stuck in Mind: The role of Catastrophizing in Pain.

Publisher: Örebro University 2011 www.publications.oru.se

[email protected]

Print: Örebro University, Repro 10/2011

ISSN 1651-1328 ISBN 978-91-7668-826-7

http://www.publications.oru.se/�

mailto:[email protected]�

Abstract Ida Flink (2011): Stuck in Mind: The role of Catastrophizing in Pain. Örebro Studies in Psychology 21, 71 pp. Pain catastrophizing emerges in the literature as one of the most important psycho-logical determinants of both pain itself and the negative outcomes commonly asso-ciated with it. However, despite decades of research confirming the impact of catas-trophizing, there are still areas that remain unexplored or in which the surface has only been scratched. The overall aim of this dissertation was to expand existing knowledge about catastrophizing and to advance the theoretical framework around the concept.

The role of catastrophizing was explored in three distinct areas: during pain in childbirth, in exposure treatment for back pain patients, and in a problem solving context. The findings from the three studies confirmed the vital role of catastrophiz-ing in these areas. Firstly, catastrophizing played a critical role in pain in childbirth; women who catastrophized reported labor pain as more intense and the subsequent recovery period as longer than women who did not catastrophize. Secondly, cata-strophizing was identified as a moderator of treatment effect in exposure in vivo for back pain patients with pain-related fear; patients who catastrophized were not helped by the exposure. Thirdly, catastrophizing played a role in a problem solving context; although this is in line with contemporary models such as the misdirected problem solving model, the results suggested a somewhat different pathway to this previous model. Taken together, these findings underscore the instrumental role of catastrophizing in diverse areas and imply a need for catastrophzing to be assessed and addressed in clinical contexts. In addition, the findings highlight a need for further development of the theoretical framework around catastrophizing as well as treatment interventions that directly target catastrophizing.

Based on these needs, a new model of catastrophizing was proposed – a model of catastrophizing from a process perspective. In this model, the proposed function of catastrophizing is to down-regulate negative affect, as a form of internal avoidance. The model is a complement to existing theoretical models and provides a frame-work for developing treatment interventions that directly target catastrophizing, for example by problem solving skills training. Successful interventions for people who catastrophize would lead to several gains – for the individual in less suffering and increased ability to handle pain problems, and for society as a whole in reduced costs for health care for these individuals.

Keywords: catastrophizing; pain; fear avoidance; problem solving.

Ida Flink, Akademin för juridik, psykologi och socialt arbete Örebro University, SE-701 82 Örebro, Sweden, [email protected]

Acknowledgements This dissertation would have been difficult, if not impossible, to produce without the support and encouragement of a number of people who I sin-cerely want to thank.

First and foremost, I want to thank Steven J Linton for being the best supervisor and mentor that I could have ever wished for. During the last year at the psychology program, I discovered that writing a thesis was the most enjoyable part of my whole education. By then I had decided to enter research “at some point throughout my career”. However, I did not expect it to be so soon. Thank you Steven for believing in me, encouraging me, and opening up a new world to me! And thank you for continually bring-ing me into projects that I would never ever think of on my own. You are a true source of inspiration. I have learnt a lot from you and I hope to con-tinue learning – about work and life.

I also sincerely want to thank my second supervisor, Maria Tillfors, for providing unlimited encouragement and support. You have helped me to “hålla styrfart framåt”, and not get lost on the track. You are always there whenever I need guidance, and you are a spiritizing source of energy.

My third mentor, and my informal supervisor, has been Katja Boersma. Thank you Katja for your generosity and honesty – you have been so im-portant for me during these years! I have learned a lot through our colla-borations and I always appreciate your feedback. I often get the feeling that you understand what I mean, even when I am not really sure of it myself. I truly enjoy working with you, as well as confabbing about life in general.

Another highly important person that I want to thank is Sofia Bergbom. You are a truly sincere colleague and friend. Without our discussions and without your emotional support, these years would have been so much harder - and much more boring. I am addicted to you!

I also genuinely want to thank Markus Jansson-Fröjmark, Thomas Overmeer, Shane MacDonald, Sara Larsson, and the other members in CHAMP. A research group is nothing without its members. I enjoy our discussions and collaborations and I am grateful for working in such an open-minded and inspiring group.

I would also like to thank the staff at the Psychology program, for dee-pening my interest in psychology, and for teaching me to be open-minded and curious: Bengt Eriksson, Anders Agrell, Håkan Stattin, and Britt Eriks-son-Helleryd, to mention only a few.

There are other colleagues that also have inspired me and that I would like to thank. Thank you Madelon Peters, for carefully reading my disser-

tation, and for providing helpful comments that indeed improved it. Thank you Michael Sullivan, for the amount of interesting research about the concept of catastrophizing and for collaborating with us in Study I. Thank you Magdalena Mroczek, for sharing the experience of writing a (damn good) thesis. And thank you to all other researchers that I have had the honor to meet, to discuss with, and to collaborate with during the years as a doctoral student.

I would also like to thank the participants in my studies. Without you, these studies would not have been realized, and your experiences are what matters.

Last but not least, I want to thank the most important persons in my life: my family. Tack mamma för att du är du och för du alltid finns där för mig. Du ÄR ”the Queen of Fuckin’ Everything”. Tack pappa för att du är, och för att du har lärt mig att älska skrivandet – genom det blir livet mer verkligt. Tack mina syskon för allt vi har delat och fortfarande delar.

Och så förstås ett varmt tack till min egna lilla familj. Ni är allt för mig! Tack Joar för att du kom till oss och lärde mig vad som verkligen betyder något. Du kommer alltid att vara nummer ett! Din energi, nyfikenhet och livsglädje skapar mening i mitt liv. Och slutligen ett stort och kärleksfullt tack till Johan, för att du finns vid min sida. Tack för värme, skratt, gråt, trygghet och äventyr – och tack för att du ser till så att jag inte blir ”Stuck in mind”.

Örebro, September 2011

List of publications This dissertation is based on the following studies, which will be referred to in the text by their Roman numerals:

I. Flink, I.K., Mroczek, M.Z., Sullivan, M.J.L., & Linton, S.J.

(2009). Pain in childbirth and postpartum recovery – The role of catastrophizing. European Journal of Pain, 13, 312-316.

II. Flink, I.K., Boersma, K, & Linton, S.J. (2010). Catastrophizing moderates the effect of exposure in vivo for back pain patients with pain-related fear. European Journal of Pain, 14, 887-892.

III. Flink, I.K., Boersma, K., MacDonald, S., & Linton, S.J. (2011). Understanding catastrophizing from a misdirected problem solv-ing perspective [Electronic version]. British Journal of Health Psychology. DOI: 10.1111/j.2044-8287.2011.02044.x

The studies have been reprinted with permission from the copyright holders.

Table of contents

INTRODUCTION ................................................................................... 13 Overview .................................................................................................. 13 Definitions ............................................................................................... 14

Pain ...................................................................................................... 14 Pain catastrophizing ............................................................................. 15

Historical overview .................................................................................. 16 Assessment of catastrophizing .................................................................. 16 Relation to pain and overt behavior ......................................................... 18 Consequences of catastrophizing .............................................................. 18 Relation to other concepts ....................................................................... 19 Theoretical models ................................................................................... 20

Schema-activation model ..................................................................... 20 Fear-avoidance models ......................................................................... 21 Appraisal model ................................................................................... 23 Communal coping model ..................................................................... 23 Misdirected problem solving model ..................................................... 24 Catastrophizing from a process perspective ......................................... 25

Catastrophizing in treatment .................................................................... 26 Summary .................................................................................................. 28 Aim of the dissertation ............................................................................. 29

Specific aims and research questions .................................................... 29

SHORT DESCRIPTION OF THE STUDIES ........................................... 30 STUDY I .................................................................................................. 30

Introduction ......................................................................................... 30 Aim ...................................................................................................... 30 Overview of the design ......................................................................... 30 Participants .......................................................................................... 30 Measurements ...................................................................................... 31 Pain catastrophizing ............................................................................. 31 Pain ...................................................................................................... 31 Physical recovery in Activities of Daily Living (ADL) ......................... 31 Statistical analysis ................................................................................ 31 Results ................................................................................................. 31 Pain ...................................................................................................... 32 Physical recovery in ADL .................................................................... 32 Conclusions ......................................................................................... 32

STUDY II ................................................................................................. 34 Introduction ......................................................................................... 34

Aim ...................................................................................................... 34 Overview of the design ......................................................................... 34 Participants .......................................................................................... 35 Measurements ...................................................................................... 35 Background and pain ........................................................................... 35 Disability ............................................................................................. 35 Anxiety and depression ........................................................................ 35 Pain catastrophizing ............................................................................. 35 Fear-avoidance beliefs .......................................................................... 36 Statistical analysis ................................................................................ 36 Results ................................................................................................. 36 Conclusions .......................................................................................... 40

STUDY III ................................................................................................ 42 Introduction ......................................................................................... 42 Aim ...................................................................................................... 42 Overview of the design ......................................................................... 43 Participants .......................................................................................... 43 Measurements ...................................................................................... 43 Background, pain, and sick leave ......................................................... 43 Catastrophizing .................................................................................... 43 Biomedical problem framing ................................................................ 43 Medically oriented problem solving behavior ....................................... 43 Function ............................................................................................... 44 Statistical analysis ................................................................................ 44 Results ................................................................................................. 45 Conclusions .......................................................................................... 47

GENERAL DISCUSSION ........................................................................ 48 How are these findings related to theoretical frameworks around catastrophizing? ....................................................................................... 49 Clinical implications and directions for future research ........................... 55 Methodological limitations ...................................................................... 57 Summary and concluding remarks ........................................................... 58 Conclusions .............................................................................................. 59

REFERENCES ......................................................................................... 61

IDA FLINK Stuck in Mind I 13

Introduction

Overview As long as human beings have existed, different types of bodily pain have caused a huge amount of suffering all over the world. Over time, our view of pain has gone through remarkable changes. In the 17th century, Des-cartes achieved a major revolution by describing the human body as a ma-chine, and proposed that pain was a direct function of the degree of tissue damage (Descartes, 1989). According to this view, the experience of pain was equivalent with peripheral injury and there was no room for psycho-logical contributions such as interpretations or the individual’s ability to handle the situation. Although Descartes’s theory dominated for centuries, research from the last decades has resulted in a total shift away from this mechanistic view of pain. It is now well known and accepted that psycho-logical factors are important determinants for how people experience and handle pain.

Pain catastrophizing emerges in the literature as one of the most impor-tant psychological determinants of pain and pain-related disability (Keefe, Rumble, Scipio, Giordano, & Perri, 2004; Severeijns, Vlaeyen, van den Hout, & Weber, 2001; Sullivan, et al., 2002; Vlaeyen & Linton, 2000). Broadly speaking, catastrophizing refers to an exaggerated negative pattern of thinking that some people report when confronting pain. “I keep think-ing about how much it hurts”, “It is terrible and I think it’s never going to get any better” and “I wonder whether something serious may happen” are some examples of catastrophic thoughts (Sullivan, Bishop, & Pivik, 1995). Catastrophizing has repeatedly been associated with heightened pain and disability across different populations with both acute and chronic pain (for a review, see Sullivan, et al., 2001). However, despite years of research about catastrophizing and its negative impact on pain, there are still areas where the role of catastrophizing remains unexplored or where the surface has only been scratched.

The overall aim of this dissertation is to expand our knowledge about pain catastrophizing and to advance the theoretical framework around the concept. The role of catastrophizing will be explored in three new areas. Firstly, pain in childbirth is a unique, acute pain situation where the role of catastrophizing has not been studied. While childbirth induces pain that is often very intense, the situation is also unique because the pain is generally not a signal of harm and brings manifest positive associations, and it is not known what impact catastrophizing might have in this context. Secondly, there are some gaps in our knowledge about how catastrophizing influ-

14 I IDA FLINK Stuck in Mind

ences treatment. More specifically, it is not known how high levels of catastrophizing are related to the effect of exposure in vivo, a novel treat-ment approach in the pain field which was developed specifically for back pain patients with the high levels of pain-related fear which are often linked to high levels of catastrophizing. The third and final area concerns the role of catastrophizing in a problem solving context. In contemporary models, catastrophizing is linked to how pain patients frame and deal with their problem in terms of coping or problem solving. However, there is sparse empirical research supporting these links, which are of interest for the theoretical framework around the concept.

The introductory section will cover definitions and an historical over-view of the concept. A review of the literature about catastrophizing will then be presented, focusing on how catastrophizing is assessed and its rela-tion to pain, overt behavior and other concepts such as depressed mood and pain-related fear. Subsequently, theoretical models that have emerged to explain the concept will be summarized. Thereafter, research about the impact of catastrophizing in treatment will be reviewed in brief. Finally the aims and research questions will be presented.

Definitions

Pain This dissertation deals with pain of two types: pain in childbirth and long-lasting spinal pain. Pain is usually defined as “an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage” (IASP, 1994). Already in the defini-tion, three features that are relevant for this dissertation appear. First, pain is a subjective experience. Second, unpleasant emotions are involved. Third, it is associated with damage or potential damage which means that it involves an interpretation of the experience. This definition makes clear that psychological features today are viewed as integrated parts of the pain experience.

One categorization of pain is based on the temporal aspect. Pain that re-solves quickly is generally referred to as acute, and pain that lasts for a longer period is called chronic or persistent. The distinction between acute and persistent pain is traditionally based on an arbitrary interval of time from onset, usually 3 or 6 months (Turk & Okifuji, 2001). Sometimes the terms subacute and subchronic are used to describe the period of transition from acute to persistent pain (Main, Sullivan, & Watson, 2007). An alter-native definition of persistent pain which does not depend on a set time


interval is "pain that extends beyond the expected period of healing" (Turk & Okifuji, 2001). According to these definitions, Study I in this disserta-tion will focus on acute pain whereas Study II and III concerns persistent pain.

Study I deals with pain in childbirth. This refers to pain during labor, which begins with contractions and ends with the birth of a baby (Mel-zack, 1993). The pain is usually very intense although there is large vari-ability among women; women who are giving birth for the first time gen-erally report the pain as more intense (Ibid.). Since earlier labor experiences might influence pain ratings as well as the expectations of childbirth, only women who were giving birth for the first time were included in study I.

Study II and III deal with persistent musculoskeletal pain in the spinal region and the terms back pain, spinal pain and musculoskeletal pain will be used interchangeably. In most cases of spinal pain (90-95%), the pain is non-specific, which means that there is no known underlying pathology such as degenerative conditions, inflammatory conditions or infective causes (Krismer & Van Tulder, 2007). Typical signs or symptoms that may indicate underlying pathology are commonly referred to as red flags (Waddell, 2004), and may indicate that further medical examination is needed. In Study II and III, people with manifest red flags were excluded.

Pain catastrophizing A commonly used definition of pain catastrophizing is “an exaggerated negative mental set brought to bear during actual or anticipated pain ex-perience” (Sullivan, et al., 2001). This is a broad definition which basically remarks that catastrophizing involves (1) a fixed pattern of thinking (“mental set”) which is (2) disproportionate in regard to the circumstances (“exaggerated”), (3) focused on unpleasant aspects (“negative”) and (4) activated when the individual confronts imaginary or actual pain (“brought to bear”).

There are some remarks to make regarding this definition. Firstly, it does not include any standpoint about what type of phenomena catastro-phizing really is; if it for instance should be viewed as a thought process, an interaction between thoughts, feelings and overt behavior, or if it is more related to what some researchers call a cognitive schema, which refers to a mental structure which organizes our knowledge and determines how we interpret and process information (e.g., Cohen & Ebbesen, 1979). There is no general consensus around what type of phenomena catastrophizing is, which presumably explains why the definition fully omits it. Secondly, the definition does not explain whether the fixed pattern of thinking, the emo-tional component or the exaggeration of the experience is the primary


component. Thirdly, the definition does not express how catastrophizing is related to other constructs such as pain-related fear or depressed mood. Taken together, even though the above definition of catastrophizing is widely used, it does not provide a comprehensive theoretical and concep-tual framework for catastrophizing.

Historical overview The term catastrophizing has been used in the psychological literature for decades. The term was introduced as early as 1962 by Albert Ellis, the founder of rational-emotional therapy (Ellis, 1962), and was later adapted by Aaron Beck (Beck, 1976). In this early work, the term catastrophizing was used to describe a maladaptive cognitive style among patients with depressive and anxiety disorders and was considered as a tendency to mag-nify or exaggerate possible negative aspects of future events.

In the pain field, Chaves and Brown observed similar tendencies to mag-nify or exaggerate the negative aspects of pain in patients who went through a painful dental procedure (Chaves & Brown, 1987). In this study, catastrophizing was framed as a spontaneous cognitive strategy to mini-mize stress or pain, the most frequently reported by the patients. Spanos and colleagues (1979) continued to view catastrophizing as a cognitive strategy, but emphasized the impact of pain-related worry, perceived in-ability to cope and difficulties in diverting attention away from pain (Spanos, Radtke-Bodorik, Ferguson, & Jones, 1979). Rosentiel and Keefe also underscored the inability to cope and linked this to feelings of help-lessness when elaborating on the concept (Rosenstiel & Keefe, 1983). Taken together, the term catastrophizing has been used in the pain field for several years and researchers have differed in what aspects they view as primary. Despite notable differences in the conceptualization, recurrent features involve attention to negative aspects, persistent negative thinking, and perceived inability to handle the situation.

Assessment of catastrophizing Catastrophizing is usually assessed by self-report inventories. The Coping Strategies Questionnaire (CSQ) (Rosenstiel & Keefe, 1983) was an early attempt to capture catastrophizing by framing it as one of seven coping strategies used by people with back pain problems. The Catastrophizing subscale of the CSQ consists of six items reflecting helplessness and nega-tive self-statements which might arise when confronting pain (e.g. “I worry all the time whether it will end”, “It’s awful and it overwhelms me”). To cover more dimensions of the construct, the six items from the CSQ were


complemented by other examples of catastrophic ideation when Sullivan and his colleagues developed the Pain Catastrophizing Scale (PCS) a decade later (Sullivan, et al., 1995).

The PCS consists of 13 statements reflecting thoughts and feelings that might arise when people confront pain. Respondents rate on a 5-point scale the extent to which they agree with the statement when experiencing pain. Factor analytic work on the PCS yielded three separate second-order factors, namely magnification (e.g. “I wonder whether something serious may happen”), rumination (e.g. “I can’t seem to keep it out of my mind”) and helplessness (e.g. “There is nothing I can do to reduce the intensity of the pain”) (Ibid.). This factor structure has been replicated and the scale validated in pain-free populations (e.g., Osman, et al., 2000; Osman, et al., 1997), in people with persistent pain (Osman, et al., 2000), in different age-groups (Lu, Tsao, Myers, Kim, & Zeltzer, 2007), and in diverse cul-tures and languages (e.g., Lu, et al., 2007; Meyer, Sprott, & Mannion, 2008; Miró, Nieto, & Huguet, 2008; Yap, et al., 2008). The PCS is hence a robust and widely used tool for assessing self-reported pain catastrophiz-ing.

Self-report inventories like the CSQ or the PCS do not, however, assess overt behavior and might well be influenced by several sources of bias like the momentary emotional state of the respondent or individual differences in interpretation of the questions (for a discussion about bias, see e.g., At-kinson, Zibin, & Chuang, 1997). A critical review of pain catastrophizing recently pointed out that there is a need for a tool to systematically assess overt behavioral dimensions of the catastrophizing construct (Quartana, Campbell, & Edwards, 2009).

The CSQ and the PCS are generally used to assess catastrophizing as a stable tendency across situations. There is a continuing debate about whether catastrophizing really is a dispositional, trait-like tendency or if it might fluctuate across situations (see e.g., Turner & Aaron, 2001). On the one hand, ratings of catastrophizing have shown a fair amount of stability over time (Keefe, Brown, Wallston, & Caldwell, 1989; Sullivan, et al., 1995). Furthermore, catastrophizing has shown high correlation with measures of relatively stable, dispositional concepts like neuroticism (e.g., Drossman, et al., 2000). On the other hand, intensive cognitive-behavioral therapy (CBT) has led to notable reductions in catastrophizing (for a re-view, see Lohnberg, 2007), which indicates that contextual factors might well influence it. Moreover, it has been argued that situational demands are needed to activate the assumed dispositional components of catastro-phizing (e.g., Campbell, et al., 2010). There are indeed recent studies that have assessed catastrophizing as a state-like, situational tendency by adapt-


ing the PCS for patients to fill out in direct connection to a pain-stimuli (Campbell, et al., 2010; Dixon, Thorn, & Ward, 2004; Edwards, Camp-bell, & Fillingim, 2005; Edwards, Smith, Stonerock, & Haythornthwaite, 2006). According to the debate about dispositional versus situational catas-trophizing, a feasible interpretation based on existing knowledge is that everyone has more or less of this tendency, but that contextual factors influence whether the catastrophizing becomes salient.

Relation to pain and overt behavior One of the most robust findings around catastrophizing is the link to rat-ings of pain intensity. This relation has been observed in a wide range of situations involving both acute and persistent pain, and in healthy indi-viduals in addition to patients who suffer from chronic conditions (for a review, see Sullivan, et al., 2001). Although there is large variability be-tween studies, catastrophizing accounts for up to 31% of the variance in pain ratings, and years of research have consistently confirmed the link between catastrophizing and pain (Ibid.).

Catastrophizing is also related to how people behave when confronting pain. For example, catastrophizing has been associated with so-called pain behavior (Nicassio, Schoenfeld-Smith, Radojevic, & Schuman, 1995). This refers to motor and verbal responses that some people express when con-fronting pain, e.g. rubbing the pain area, frowning, or complaining about how much it hurts. Catastrophizing has also been linked to a heightened use of health care (e.g., Gil, et al., 1993) and medication (Bedard, Reid, McGrath, & Chambers, 1997). These findings indicate two things. First, people who catastrophize when experiencing pain seem to have an urgent need of expressing it to others. Second, they engage in activities they think might help to solve the pain problem such as seeking medical care. Thus, catastrophizing is closely linked to overt behavior which might be used to cope with a difficult situation.

Consequences of catastrophizing Catastrophizing has been linked to a number of negative consequences, cross-sectionally as well as prospectively (for reviews, see Sullivan, et al., 2001; Keefe, et al., 2004). For example, catastrophizing has repeatedly been associated with higher levels of disability. The link between catastro-phizing and heightened disability has been found in patients with acute pain (e.g., Swinkels-Meewisse, Roelofs, Oostendorp, Verbeek, & Vlaeyen, 2006) as well as in patients with chronic conditions (e.g., Peters, Vlaeyen, & Weber, 2005; Severeijns, Vlaeyen, van den Hout, & Weber, 2001a;


Turner, Jensen, Warms, & Cardenas, 2002) . To summarize, there is con-sistent evidence that people who catastrophize report the pain as being more intense, express more worries and suffer more adverse consequences of it than people who do not catastrophize.

However, the role of catastrophizing has mainly been studied in persis-tent and pathological pain situations and less is known about its impact in situations with more positive associations such as childbirth. Childbirth infers a remarkably intense pain. Although there are large individual differ-ences, 60 % of women who are giving birth for the first time rate their pain as severe or extremely severe (Melzack, 1993). The anticipation of this intense pain, which has been compared to the amputation of a finger without analgesia (Ibid.), might well be associated with psychological reac-tions such as fear and catastrophizing. On the other hand, childbirth is a unique experience and differs considerably from other situations involving pain. Labor pain is not a signal of harm but is rather an indication that the labor is proceeding. Furthermore, labor is usually restricted to a short pe-riod of time, and brings consequences that are generally viewed in a posi-tive light: namely the birth of a child. Therefore, it is not self-evident which role catastrophizing might play in this context.

Relation to other concepts Catastrophizing is closely related to negative emotionality. It has repeat-edly been associated with broad negative emotional responses such as de-pressed mood and anxiety (for a review, see Keogh & Asmundson, 2004) and there is a debate about whether catastrophizing really is a unique con-struct separate from these more general emotional constructs. Measures of catastrophizing overlap to a large extent with measures of negative emo-tionality and there is evidence that after controlling for the broad meas-ures, catastrophizing contributes minimally to the prediction of pain (Hirsh, George, Riley, & Robinson, 2007). There is, however, conflicting evidence indicating that catastrophizing is indeed a unique construct (Sulli-van, et al., 2001). For example, a recent clinical study showed that some pain patients presented either catastrophizing or depressed mood, whereas others presented both (Linton, et al., 2011). These findings indicate that it is relevant to separate catastrophizing from other reflections of negative emotionality in a clinical context.

Catastrophizing is also associated with specific negative emotional re-sponses to pain such as pain-related fear. In the fear-avoidance model of pain (Vlaeyen & Linton, 2000), catastrophizing is postulated as a precur-sor of pain-related fear and subsequent avoidance. There is some evidence supporting this sequential order of the model (Cook, Brawer, & Vowles,


2006; Leeuw, et al., 2007), but there are also contradictory findings (Wideman, Adams, & Sullivan, 2009). One interpretation of the conflict-ing results is that it is difficult to separate these constructs because of the large overlap between them.

Conceptually, there are recognized differences between catastrophizing and fear. Fear is commonly referred to as a basic survival mechanism oc-curring in response to a specific stimulus, such as pain or the threat of dan-ger (Ohman, 2000). According to this definition, fear does not involve the fixed pattern of thinking which is salient in catastrophizing. Pain catastro-phizing has broadly been referred to as “the cognitive element” of the fear network (Leeuw, et al., 2007). To narrow down catastrophizing to a purely cognitive element might, however, be an oversimplification since catastrophizing involves emotional aspects as well and is closely linked to overt behavior (e.g. pain behavior and health care seeking).

Taken together, catastrophizing is strongly related to negative emotion-ality in a broader sense, but also to more specific negative responses to pain, such as pain-related fear or anxiety although existing research indi-cates that catastrophizing is something over and above these other con-structs.

Theoretical models There is no clear consensus of how catastrophizing might be understood from a theoretical point of view. Even though most studies about catastro-phizing have focused more on its relation to pain and negative outcomes than on the theoretical understanding, there have been a few attempts to integrate catastrophizing in a theoretical framework. The main attempts will be summarized below.

Schema-activation model Before the term catastrophizing was introduced in the pain area, it was most frequently discussed in the cognitive literature about emotional disor-ders. Catastrophizing was proposed as a maladaptive pattern of thinking, commonly occurring among people with depressive and anxiety problems (Beck, 1976). Beck (1979) described how negative life events might activate depressive schemas and how such schemas might fuel different types of cognitive distortions whereof catastrophizing was one (Beck, 1979). From this perspective, catastrophizing was viewed as a pure cognitive concept which was a determinant for the development and perpetuation of emo-tional problems. Later on, the concept was transferred to the pain field and other theoretical frameworks emerged to explain specifically pain-related catastrophizing.

Injury

PAIN

Catastrophizing

Fear

Avoidance

Dysfunction

Depression

Vigilance and

tension

Normal fear

“Warning”

Confronting

Recovery


complicated than the original one and suggest in more elaborate detail how the different concepts such as anxiety, fear of pain and catastrophizing might be interrelated.

FIGURE 2. The fear-anxiety-avoidance model (Asmundson, et al., 2004) Note. This figure has been reproduced with the permission of Oxford University Press. The figure may not be reproduced for any other purpose without permission.

Since this model is intended to be more conceptually clear, it has been pro-posed as a sound basis for testing the relationship between variables (Boersma, 2005). One modification from the original model is that fear of pain has been separated from pain-related anxiety; with fear described as the response to a present threat whereas anxiety is the response to an an-ticipated threat (Asmundson, et al., 2004). Furthermore, the cues for anxi-ety are often vague or uncertain; in contrast to cues for fear that are more immediate (Barlow, 2004). Another modification is that pain beliefs, i.e. beliefs about pain being equivalent with harm, now have an articulated place in the model. This model infers that pain beliefs drive catastrophiz-ing. In other words, if patients hold strong beliefs that pain signifies some-thing is medically wrong, catastrophic thoughts are likely to emerge. How-ever, this assumption still lacks empirical support.

Injury or Organic

Pathology

PainPerception

PainCatastrophizing

NoCatastrophizing No Fear No Anxiety Recovery

Disuse/Deconditioning

FearOf Pain

DEFENSIVE

MOTIVATIONAU

TONO

MIC

AROUS

AL

THREATPERCEPTION

Pain-Related Anxiety

PREVENTATIVE

MOTIVATIONAU

TONO

MIC

AROUS

AL

HYPERVIGILANCE

Escape/DefensiveBehavior

Avoidance/Preventative

Behavior

No Escape/DefensiveBehavior

No Avoidance/Preventative

Behavior

PredisposingRisk Factors

PainBeliefs

Injury or Organic

Pathology

PainPerception

PainCatastrophizing

NoCatastrophizing No Fear No Anxiety Recovery

Disuse/Deconditioning

FearOf Pain

DEFENSIVE

MOTIVATIONAU

TONO

MIC

AROUS

AL

THREATPERCEPTION

FearOf Pain

DEFENSIVE

MOTIVATIONAU

TONO

MIC

AROUS

AL

THREATPERCEPTION


PREVENTATIVE

MOTIVATIONAU

TONO

MIC

AROUS

AL

HYPERVIGILANCE


PREVENTATIVE

MOTIVATIONAU

TONO

MIC

AROUS

AL

HYPERVIGILANCE

Escape/DefensiveBehavior

Avoidance/Preventative

Behavior

No Escape/DefensiveBehavior

No Avoidance/Preventative

Behavior

PredisposingRisk Factors

PainBeliefs


Appraisal model The appraisal model is another framework which has been used to explain pain catastrophizing (Severeijns, Vlaeyen, & van den Hout, 2006). This model is similar to the fear-anxiety-avoidance model in that it emphasizes the link between pain beliefs and catastrophizing. Based on the transac-tional model of stress and coping (Lazarus & Folkman, 1984), catastro-phizing is described as a result of underlying beliefs (i.e. pain beliefs), pri-mary appraisal (i.e. evaluation of the situation), and secondary appraisal (i.e. evaluation of ability to cope with the situation). To give an example, if a patient who suffers from low back pain holds beliefs that something is medically wrong in the back (pain beliefs), evaluates the situation as threat-ful (primary appraisal) and estimates that he or she cannot handle it (sec-ondary appraisal), catastrophic thoughts are likely to occur. The appraisal model positions catastrophizing in a coping framework stressing the cogni-tive aspects of the concept.

Communal coping model A model that also frames catastrophizing within a coping perspective, but from a different angle, is the communal coping model (CCM) (Sullivan, et al., 2001; Thorn, Ward, Sullivan, & Boothby, 2003). This model takes a step away from the cognitive conceptualization of catastrophizing by em-phasizing the importance of the social context. According to the CCM, people who catastrophize might exaggerate their expressions of pain as a way to maximize social support from people around them. Catastrophizing is here seen as a form of communication and support seeking. Even though a number of studies have provided preliminary support for the CCM in confirming the link between catastrophizing and the social environment (Giardino, Jensen, Turner, Ehde, & Cardenas, 2003; Keefe, et al., 2003; Lackner & Gurtman, 2004; Sullivan, Adams, & Sullivan, 2004), the model has also received substantial criticism (Severeijns, et al., 2006). One argu-ment is that the CCM focuses on the consequences of catastrophizing rather than on the origin. It is argued that it is feasible that people seek social support because they catastrophize, not that they catastrophize be-cause they elicit social support as the CCM implies. Furthermore, the CCM only focuses on one way to cope with pain and worrisome thoughts, namely through the support of others, although there may be additional ways of coping. Nevertheless, even though the CCM does not provide an absolute explanation of why people catastrophize, this theory has contrib-uted to an enhanced understanding of the context in which catastrophizing occur.


Misdirected problem solving model A recent model that has advanced the coping framework around catastro-phizing is the misdirected problem solving model (figure 3, Eccleston & Crombez, 2007). Coping is here conceptualized as an active attempt to solve a problem. This model builds on observations from the fear-avoidance model as well as the CCM and reframes them within a problem solving perspective. The misdirected problem solving model proposes that worry motivates pain patients to actively search for relief or a cure. These efforts, which are specifically directed at curing and getting rid of pain, have been described as assimilative coping (Crombez, Eccleston, van Hamme, & de Vlieger, 2008). In patients who frame the pain problem in purely biomedical terms (i.e. pain is equal to harm or injury), the attempts at problem solving might become “misdirected” if no medical solution exists, which is often the case in patients with long-term back pain. As the problem remains unsolved, the worry increases. When the worry progres-sively intensifies and the outcome is perceived as getting worse and worse, it has been defined as catastrophic worry (Davey & Levy, 1998), as cap-tured by the concept catastrophizing. The model illustrates how a perse-verance loop of catastrophic worry and misdirected problem solving devel-ops. In this model, catastrophizing is thus part of an unsuccessful problem solving strategy which involves repeated fruitless efforts at finding a cure for pain.

In relation to the above, earlier studies support an association between repeated efforts at finding a cure and higher levels of catastrophizing, greater disability and heightened attention to pain in patients with pain problems (Crombez, et al., 2008; de Vlieger, Bussche, Eccleston, & Crom-bez, 2006). However, neither the way in which problem framing is linked to catastrophizing and overt problem solving behavior or the sequential order of the concepts involved have been investigated. Indeed, earlier re-search has stressed the importance of specifically examining how catastro-phizing is related to overt problem solving behavior (Crombez, et al., 2008).

This model is similar to the fear-anxiety-avoidance model in emphasiz-ing the link between catastrophizing and the view patients have of their problem. In the fear-anxiety-avoidance model this view is called pain be-liefs, whereas in the misdirected problem solving model the term biomedi-cal problem framing is used. Both terms basically capture the same ten-dency - to strongly believe that pain is equal to harm or serious injury.


FIGURE 3. A stylized version of the misdirected problem solving model (Eccleston & Crombez, 2007) focusing on the perseverance loop. a Note. The original term worry is here replaced with the term catastrophic worry to underscore the intensity and the extreme nature of this tendency. b Note. This figure has been reproduced with permission of the International Asso-ciation for the Study of Pain® (IASP®). The figure may not be reproduced for any other purpose without permission.

Catastrophizing from a process perspective As opposed to some of the aforementioned models that argue around how different concepts are related to each other, an alternative perspective is to view catastrophizing as a process where thoughts, emotions and overt be-havior are intertwined. From a process perspective, catastrophizing might be conceptualized as a form of repetitive negative thinking, similar to worry or rumination. Rumination was recently defined as “passive focus on one’s symptoms of distress and the possible causes and consequences of these symptoms. The individual repeatedly goes over problems and his or her feelings about the problems, without moving into [constructive, auth. note] problem solving” (Nolem-Hoeksema, 2005). This definition might also serve for catastrophizing. In fact, one of the subscales in the PCS is labeled rumination, which might indicate that these processes indeed are

Pain

Hypervigilance

Catastrophic worry

Perseverance loop Biomedical problem frame

Problem unsolved

Problem solving behavior


interrelated, and may both be included in the overall term repetitive nega-tive thinking. It has been argued that repetitive negative thinking is an avoidant coping strategy (Stroebe, et al., 2007); the patient gets stuck in passive, abstract, catastrophic thinking, instead of really dealing with the problem by moving into constructive, concrete problem solving. From this perspective, focus is on the function of catastrophizing - to down-regulate negative emotions, the same function as overt behavioral avoidance. Ac-cording to this view, patients with high levels of pain-related fear and anxi-ety might engage in catastrophizing as a way of reducing the intensity of the aversive physiological and psychological aspects of the fear response. Outside of the pain field, catastrophizing has indeed been described as a perseverative iterative style which has been linked to worry and poor prob-lem-solving confidence (Davey & Levy, 1998). This perspective is in line with the misdirected problem solving model (Eccleston & Crombez, 2007) in that catastrophizing obstructs constructive problem solving. Taken together, the process perspective of catastrophizing provides an alternative way to look at catastrophizing as a form of covert avoidance and uncon-structive problem solving.

In sum, pain catastrophizing has been described within quite diverse theoretical frameworks. It is noteworthy that these frameworks do not always contradict each other and might also be seen as complementary. Moreover, there are some recurring aspects. One such aspect is the view of catastrophizing as a form of coping, or an attempt to handle a difficult situation. In fact, in the majority of the frameworks, catastrophizing is in one way or another related to coping. Another recurring aspect is the link between catastrophizing and beliefs the person holds about pain, or in other words problem framing. However, these models are to a large extent based on theoretical assumptions and empirical support is still needed to refine the theoretical framework around catastrophizing.

Catastrophizing in treatment As catastrophizing is an important factor in how people experience and handle pain problems, it might well influence how patients respond to treatment. In the treatment of chronic back pain, different approaches have been developed, and cognitive behavioral therapy (CBT) has made a note-worthy contribution to the field (for reviews, see e.g., Butler, Chapman, Forman, & Beck, 2006; McCracken & Turk, 2002; Eccleston, Williams, & Morley, 2009). CBT is generally quite effective for reducing the negative outcomes of a pain problem such as depressed mood, catastrophizing and disability. However, the effect sizes for CBT for pain are fairly modest and there are large individual differences in how well patients respond to


treatment (Eccleston, et al., 2009; Nicholas, 2008; van Tulder, et al., 2001; Vlaeyen & Morley, 2005). When scrutinizing why some patients are not helped by general CBT programs, catastrophizing has emerged as one im-portant factor. More specifically, high levels of catastrophizing have been associated with poor outcome (McCracken & Turk, 2002; Turner, Holtz-man, & Mancl, 2007). Thus, patients who catastrophize do not benefit to the same extent as others from CBT programs for pain, but these programs are usually not specifically developed to target catastrophizing.

Exposure in vivo is a novel treatment approach within CBT for pain that is specifically developed for patients with high levels of pain-related fear. Several studies have indeed shown that exposure is effective for reduc-ing fear, catastrophizing, disability and pain (Boersma, et al., 2004; de Jong, et al., 2005; Vlaeyen, de Jong, Geilen, Heuts, & van Breukelen, 2001; Leeuw, et al., 2008; Linton, et al., 2008; Woods & Asmundson, 2008). In addition, recent reviews have highlighted exposure as the treat-ment of choice for patients with high levels of fear and catastrophizing (Bailey, Carleton, Vlaeyen, & Asmundson, 2010; Lohnberg, 2007). Even though exposure is a promising treatment for this group of patients, there are two complicating issues. The first is that in some studies the effect sizes are quite modest (e.g., George, et al., 2008; Leeuw, et al., 2008). Secondly, it has been noted that there are large individual differences in how well the patients respond to the treatment (Linton, et al., 2008). Consequently, exposure in vivo is an effective treatment for some, but not all, patients with high levels of fear.

This raises the question of possible moderators of the treatment effect. A moderator is traditionally defined as “a third variable that affects the direc-tion and/or strength of the relation between an independent or predictor variable and a dependent or criterion variable” (Baron & Kenny, 1986). In other words, a third variable might suppress the treatment effect and since high levels of catastrophizing have been related to poor outcomes in other CBT treatments, they might also be of importance here. Taken together, CBT is a rather effective treatment for problems with chronic pain but high levels of catastrophizing have been linked to a reduction in the effect of the treatment. Whether this is also true of exposure in vivo, which is specifi-cally developed for patients with high levels of fear, is still to be explored.


Summary Psychological factors are today viewed as integrated features of the pain experience. Pain catastrophizing has been identified as one of the most important psychological determinants for pain and pain-related disability. Links between catastrophizing, pain, and negative outcomes have been found in people with acute pain as well as in patients suffering from chronic conditions. Moreover, catastrophizing has been related to poor treatment outcome. This means that pain patients who catastrophize ap-parently do worse in treatment. To help pain patients who catastrophize, refined methods that directly target catastrophizing might be needed. To develop such methods, it is crucial to have a clear theoretical framework. Several frameworks have emerged to explain catastrophizing from a theo-retical point of view, but there is still sparse empirical support for some of the proposed links.

One question that emerges is whether the link between catastrophizing, pain, and negative outcomes is true regardless of the circumstances. Would a situation where pain clearly is not harmful, and in fact brings conse-quences that are generally viewed as positive, also provoke catastrophiz-ing? Although childbirth contains these features, it also involves extremely intense pain, and the role of catastrophizing in this context remains un-clear. A second question that arises is how high catastrophizing influences the effect of exposure in vivo, a treatment that was specifically developed for patients with pain-related fear, a condition which often goes hand in hand with catastrophizing. One possibility is that patients who catastro-phize are indeed helped by exposure, since the treatment was developed for fearful patients who often have salient catastrophic thoughts. Another possibility is that the same tendency as in other CBT treatments is applica-ble to exposure; that patients with too high levels of catastrophizing actu-ally do worse. A third question concerns the theoretical understanding of catastrophizing. A recurrent feature in existing theoretical frameworks is the link to beliefs that patients hold about their pain. More specifically, catastrophizing has been linked to strong beliefs that back pain is purely caused by medical factors. The misdirected problem solving model is one contemporary model which emphasizes this link, where catastrophizing is framed within a problem solving perspective. As this model is fairly new, empirical studies supporting the proposed links are sparse. In particular, the links between problem framing, catastrophizing, and overt problem solving behavior have not been scrutinized. To find out whether catastro-phizing might be framed within a problem solving perspective is relevant not only in terms of theoretical understanding, but also to provide appro-


priate help for pain patients stuck in a vicious circle of catastrophizing and misdirected attempts to handle the situation.

Aim of the dissertation This dissertation aims to expand our current knowledge about pain catas-trophizing and to advance the theoretical framework around the concept. More specifically, the role of catastrophizing was explored in three areas. Firstly, it was investigated how catastrophizing was related to pain in childbirth, a pain situation which might be considered as unique because of its positive associations. Secondly, catastrophizing was scrutinized in the context of psychological treatment for problems with back pain. More specifically, the question posed was how high levels of catastrophizing were related to outcome in exposure in vivo. Thirdly, the role catastrophiz-ing might play in a problem solving context was explored by linking the concept to problem framing and overt problem solving behavior.

Specific aims and research questions

I. Study I is a prospective study where the objective was to study the role of catastrophizing in pain in childbirth. The central question was whether the reported pain in childbirth and func-tioning postnatally differed between women who catastrophized about labor pain and non-catastrophizers.

II. Study II is a secondary analysis of an RCT on the effectiveness of exposure in vivo for back pain patients with pain-related fear. The main aim was to study possible moderators of out-come in exposure; catastrophizing, anxiety and depressed mood were proposed as possible moderators.

III. Study III is a prospective study where pain catastrophizing was explored from a problem solving perspective. The aim was to study the links between catastrophizing, problem framing and overt problem solving behavior through two possible models of mediation as inferred by two contemporary and complementary theoretical models, the misdirected problem solving model (Ec-cleston & Crombez, 2007) and the fear-anxiety-avoidance model (Asmundson, et al., 2004).


Short description of the studies

STUDY I PAIN IN CHILDBIRTH AND POSTPARTUM RECOVERY -THE ROLE OF CATASTROPHIZING

Introduction Pain in childbirth is one of the most intense pains that women experience in their lives. Consequently, it might well be associated with psychological reactions such as pain catastrophizing. However, childbirth differs from other pain provoking situations because of its positive associations, namely the birth of a child. Also, labor pain is normally expected to be restricted in time, which is different from many situations involving pain. Furthermore, labor pain is an important signal to the mother that the labor is proceeding and hence the pain might have a lower threat value than in other situa-tions. Thus, there is a reason to study the relation between catastrophizing, pain and lower levels of functioning in the context of childbirth.

Aim The aim was to investigate whether the reported pain during childbirth and functioning postnatally differed between women who catastrophize about labor pain (hereafter called “catastrophizers”) and non-catastrophizers. We hypothesized that catastrophizers would anticipate and experience more pain during childbirth than non-catastrophizers. We also hypothesized that recovery, in terms of level of functioning in activities of daily living (ADL), would take longer for catastrophizers than for non-catastrophizers.

Overview of the design In this prospective study, the data collection was made on two occasions: after 34–41 weeks of pregnancy and 2–4 weeks following the birth. Par-ticipants were classified as catastrophizers (n = 38) and non-catastrophizers (n = 44) based on their scores on the Pain Catastrophizing Scale (PCS). Catastrophizers and non-catastrophizers were compared on ratings of an-ticipated and experienced labor pain and on ratings of functioning in ADL at three and seven days after the delivery.

Participants Pregnant women (N = 82) who were giving birth for the first time were recruited through maternal health services. To participate, the women had


to be at least 34 weeks of gestation and planning to give birth vaginally. Only women answering both questionnaires (93%) were included in the analyses. The mean age was 30; range 20-42.

Measurements

Pain catastrophizing Catastrophizing about labor pain was assessed through the Pain Catastro-phizing Scale (PCS) (Sullivan, et al., 1995). In this study, the instructions of the PCS were changed slightly, so that the women were asked to focus specifically on the thoughts they had about labor pain. The PCS has been used in similar ways to assess catastrophizing about genital pain (Pukall, Binik, Khalifé, Amsel, & Abbott, 2002).

Pain Anticipated and experienced labor pain was assessed by the Present Pain Intensity scale, a part of the McGill Pain Questionnaire (Melzack, 1975). Pain ratings were made on two occasions: before (anticipated) and after (experienced) childbirth.

Physical recovery in Activities of Daily Living (ADL) To assess recovery in ADL, the women rated to what degree they were able to do the following, at three and seven days after the delivery, as compared to before the pregnancy: Household chores (e.g. cleaning and cooking), personal care (e.g. hygiene and getting dressed), and physical activities (e.g. taking walks, climbing the stairs, and light exercising). To obtain one score for physical recovery, the average score for the three questions was calcu-lated.

Statistical analysis To explore differences between catastrophizers and non-catastrophizers, independent t-tests were employed for anticipated labor pain, experienced labor pain, and physical recovery at three and seven days after the delivery. Since four t-tests were employed, a Bonferoni correction for multiple tests was made, requiring p < .0125 for statistical significance.

Results The median-split at 20 on the PCS resulted in two distinct groups: women who catastrophized (M = 27.9; SD = 5.6)) and non-catastrophizers (M = 12.4; SD = 5.2).


Pain Catastrophizing women rated both anticipated and experienced labor pain as significantly higher than non-catastrophizers, as shown in table 1.

Table 1 Comparison of pain intensity (scale 0-5) in catastrophizing and non-catastrophizing women.

PC (n = 38)

M(SD)

NPC (n = 44)

M(SD)

t (df)

Anticipated pain 3.79 (0.59) 3.06 (0.65) -5.32* (80)

Experienced pain 3.68 (0.73) 3.09 (0.88) -3.33* (80)

M = mean; SD = standard deviation; df = degrees of freedom; PC = Pain Catastrophizers; NPC = Non Pain Catastrophizers. *p < .0125 Note. The groups have different n because the participants that rated on the median at 20 were treated as non-catastrophizers.

Physical recovery in ADL Catastrophizing women rated their level of functioning in ADL as signifi-cantly lower than non-catastrophizers at three days and at seven days after giving birth, as illustrated in figure 4.

FIGURE 4. Recovery in level of functioning in activities of daily living (ADL; scale 1-6) in catastrophizing and non-catastrophizing women.

Conclusions The results from this study demonstrate that catastrophizing does indeed play a role in pain in childbirth. Both proposed hypotheses were supported:

1

2

3

4

5

6

3 days 7 days

Leve

l of f

unct

ioni

ng in

ac

tiviti

es o

f dai

ly li

ving

(AD

L)

Non PainCatastrophizers

PainCatastrophizers


firstly, women who catastrophized about labor pain anticipated and ex-perienced more intense pain in childbirth than non-catastrophizers. Sec-ondly, women who catastrophized reported lower levels of functioning than non-catastrophizers up to a week after the delivery.

These findings replicate and expand findings from earlier studies, mostly carried out in populations with chronic and pathological pain, where catas-trophizing has been associated with heightened pain and reduced involve-ment in daily activities (for a review, see Sullivan, et al., 2001). In this study, we found the same associations for pain associated with childbirth. Thus, despite the manifest positive facets of childbirth, the relation be-tween catastrophizing and pain outcomes was comparable with other pain situations.

The results from this study show that catastrophizing is indeed associ-ated with negative outcomes despite circumstances where the pain might be framed as “natural” or “harmless”. However, since the framing of the pain was not assessed, it might well be that women who catastrophized framed the pain as harmful or were afraid that something had gone medically wrong. To find out how catastrophizing is related to medical framing of pain, further research on this topic is needed; this will be the focus of study III.

These findings imply possibilities for identifying pregnant women with high levels of catastrophizing during the third trimester, for instance by using screening questionnaires. The next challenge is to develop adequate interventions. In other populations, cognitive-behavioral therapy (CBT) is a successful method which encourages better adjustment to pain. One poten-tial path is to investigate if CBT-techniques (e.g. imaginary exposure) could be adapted as early interventions for women who catastrophize about la-bor pain. However, there is evidence that high levels of catastrophing might hinder the treatment effect of CBT (McCracken & Turk, 2002; Turner, et al., 2007) and there is a need to determine whether this also applies to different types of exposure before these treatments are adapted to suit pregnant women who catastrophize. Exploration of factors that might hinder the treatment effect, which is the research question in study II, will provide valuable information to the development of appropriate interventions for women who catastrophize and are at risk of experiencing childbirth as a negative event.


STUDY II

CATASTROPHIZING MODERATES THE EFFECT OF EXPOSURE IN VIVO FOR BACK PAIN PATIENTS WITH PAIN-RELATED FEAR

Introduction Exposure in vivo is a novel treatment approach in the pain field which has been shown to be effective for back pain patients with high levels of pain-related fear. However, the effects differ notably between studies and one issue raised is that some patients seem to respond better to exposure than others (Linton, et al., 2008). Consequently, the effect of exposure might be suppressed by moderating factors. In other CBT treatments for pain, high levels of catastrophizing, anxiety, and depressed mood have been linked to poor outcome (e.g., McCracken & Turk, 2002; Turner, et al., 2007). How these factors are related to outcome in exposure has not yet been studied.

Aim The main aim was to study possible moderators of outcome in exposure in vivo. Catastrophizing, anxiety, and depression were explored as possible moderators of the treatment effect. The secondary aim was to study changes in psychological variables (depression, anxiety, catastrophizing, and fear-avoidance beliefs) during treatment for patients who improved in terms of function as compared to those who did not improve, to get an indication of what helped or hindered the treatment effect.

Overview of the design This study is based on data from a randomized-controlled trial for patients with non-specific spinal pain (Linton, et al., 2008). Figure 5 presents an overview of the design. After randomization, all participants filled out pretest 1. Thereafter, the exposure group (n = 13) received exposure treat-ment and the waiting-list group (n = 21) continued with the usual treat-ment. After a median of 14 weeks, the waiting-list participants completed pretest 2 and subsequently received exposure. All participants completed the posttest immediately after treatment and the follow-up three months later.


Assessment 1 Assessment 2

Exposure group

Pretest 1

Exposure

Posttest

Waiting- list group

Pretest 1

Pretest 2

Exposure

Posttest

FIGURE 5. Overview of the design of the study.

Participants Participants were recruited via primary care facilities, advertisements in local newspapers, and National Insurance Authority offices. The partici-pants had disabling spinal pain, ongoing or recent sick leave, and substan-tial levels of pain-related fear (TSK > 35). The mean age was 46 in the exposure group and 49 in the waiting-list group.

Measurements

Background and pain Background questions (age, gender, and sick leave) and pain characteristics (intensity, location, and duration) were taken from the Örebro Muscu-loskeletal Pain Screening Questionnaire (ÖMSPQ) (Linton & Halldén, 1998).

Disability Level of functioning in daily activities (e.g. walking or getting dressed) was measured with the Quebec Back Pain Disability Scale (QBPDS) (Kopec, et al., 1995).

Anxiety and depression Anxiety and depression were assessed through the Hospital Anxiety and Depression Scale (HADS) (Zigmond & Snaith, 1983).

Pain catastrophizing Catastrophizing was measured with the Pain Catastrophizing Scale (PCS) (Sullivan, et al., 1995).

Assessment

Three month follow-up Three month follow-up


Fear-avoidance beliefs Fear-avoidance beliefs were assessed through the Tampa Scale of Kinesio-phobia (TSK) (Kori, Miller, & Todd, 1990).

Statistical analysis In order to address the main aim, data from both the exposure group (n = 13) and the waiting-list group (n = 21) were used in the analyses. Three multiple regression analyses were employed for catastrophizing, anxiety, and depression respectively, according to recommendations for the investi-gation of moderational effects (Baron & Kenny, 1986; Holmbeck, 2002). To test whether pretreatment levels of the psychological variables moder-ated the effect of treatment on disability, the interaction between the group and the psychological variable was added to the main variables. If the in-teraction turned out to be significant while the main variables were in-cluded in the model, this would indicate moderation. Subsequently, the nature of the significant interactions was explored with simple regressions, for participants with high and low levels of the psychological variable separately.

In order to tackle the secondary aim, only data from participants who completed the exposure was used (n = 26). The participants were divided into two groups based on a median-split on the percentage change in dis-ability from pretest to posttest (median = -18%). To compare these groups, descriptive statistics were calculated for pretest, posttest, follow-up, and percentage change from pretest to posttest. Because of the small number of participants, it was not tested whether the differences were statistically significant.

Results As can be seen in table 2, of the interaction terms that were entered, only the interaction between catastrophizing and group turned out to be signifi-cant over and above the main effects. This implies that the effect of expo-sure was significantly different depending on the patients’ levels of catas-trophizing, whereas in the waiting-list group catastrophizing was not a predictor of outcome. In other words, catastrophizing was a moderator of treatment outcome in exposure.

In the analysis that investigated anxiety as a possible moderator, only the main effect was significant, meaning that anxiety was a general predic-tor of outcome in both the exposure group and the waiting list group. De-pression was not significantly related to outcome.


Table 2 Multiple regression analyses explaining the treatment effect in reduction on disability depending on pretreatment levels of catastrophizing, anxiety, and depression.

Variables Beta

(standar-

dized)

p value R2 R2

adjusted

Block 1 Group

Catastrophizing

.270

-.083

Ns

Ns

Block 2 Group

Catastrophizing

Group*catastrophizing

.187

.604

-.809

Ns

.048*

.009**

.27

.20

Block 1 Group

Anxiety

.430

.498

.014*

.005**

.28

.23

Block 2 Group

Anxiety

Group*anxiety

.435

.643

-.204

.014*

.008**

Ns

Block 1 Group

Depression

.307

.163

Ns

Ns

Block 2 Group

Depression

Group*depression

.307

.246

-.127

Ns

Ns

Ns

Note. *p < .05, **p < .01

To explore the nature of the significant interaction, the participants were divided into high and low catastrophizers based on a median-split (at score 26) on the PCS. Simple regression analyses for these two groups separately showed that in low catastrophizers, there was a significant effect of group


on treatment outcome, whereas in high catastrophizers, there was no such significant effect. This pattern is shown in figure 6.

High catastrophizers only

Low catastrophzers only

FIGURE 6. Ratings of disability in the exposure group and the waiting-list group for low catastrophizers and high catastrophizers respectively. a Note. These graphs are based on ANOVA’s. b Note. The large difference between the high catastrophizing groups at assessment 1 was only found when dividing the participants into high and low catastrophizers. There was no such difference between the whole exposure and waiting-list group.

In relation to the secondary aim, the descriptive data displayed in table 3 showed that both low change participants (LCP) and high change partici-pants (HCP) scored somewhat lower on anxiety, catastrophizing, and fear-avoidance beliefs after treatment, although these improvements were markedly higher in the HCP group. In depression, the scorings were slightly worse after treatment in the LCP group, whereas HCP also showed improvement on this variable.

30354045505560


Exposure groupWaiting list group

30354045505560



Table 3 Ratings of disability and psychological variables in low change participants and high change participants at pretest, posttest and follow-up, and per-centage change from pretest to posttest.

Variable

Measure

Time point Low change participants (n = 12) M(SD)

High change participants (n = 13) M(SD)

Disabilitya

QBPDS Pretest

Posttest Follow-up Change from pretest to post-test

49.8 (14.8) 48.8 (13.0) 48.1 (16.1) -2%

46.6 (11.6) 22.1 (11.7) 22.9 (16.9) -53%

Depression

HADS

Pretest Posttest Follow-up Change from pretest to post-test

6.8 (4.4) 7.2 (4.2) 8.1 (4.8) +6%

5.5 (4.2) 3.4 (3.4) 3.8 (3.1) -38%

Anxiety

HADS


7.8 (3.1) 6.7 (3.0) 7.4 (3.7) -14%

4.5 (2.8) 3.4 (2.2) 2.5 (1.7) -32%

Catastro-phizing

PCS


25.7 (6.4) 21.2 (8.9) 21.5 (10.0) -18%

18.7 (10.1) 9.5 (6.5) 11.2 (8.6) -49%

Fear-avoidance beliefs

TSK


40.8 (8.0) 35.0 (5.0) 35.5 (7.3) -14%

38.6 (8.1) 26.9 (5.7) 28.9 (6.9) -30%

a Note. The groups (LCP/HCP) were based on the posttest ratings on this measure.


Conclusions In conclusion, catastrophizing moderated the effect of exposure in vivo in this sample. More specifically, high catastrophizing patients were not helped by the treatment whereas low catastrophizers showed marked im-provements. In addition, patients who were high on anxiety tended to have a poor outcome in general, both in the treatment group and in the waiting-list group. Depression was not significantly related to outcome. However, the descriptive data indicated that depressed mood might be related to outcome nevertheless as the high change group did indeed seem to improve on ratings of depressed mood whereas the low change group did not change at all.

This study is the first one to identify catastrophizing as a moderator of treatment outcome in exposure. This is in line with findings from earlier research about moderating factors in other CBT programs for pain (McCracken & Turk, 2002; Turner, et al., 2007). However, it contradicts the hypothesis that exposure would work best for patients with high levels of fear (Leeuw, et al., 2008), which in general is closely linked to catastro-phizing. Instead, our findings point in the opposite direction, as the treat-ment was only effective for patients with low or moderate levels of catas-trophizing (< 26). One possibility is that too high levels of catastrophizing might work as a safety behavior, i.e. a subtle strategy that patients use to “protect” themselves during exposure, which hinders the patient’s ability to fully benefit from exposure (Vlaeyen, de Jong, Leeuw, & Crombez, 2004).

These findings may have clinical implications. First and foremost, as-sessment of patients with spinal pain and pain-related fear should involve catastrophizing. Depressed mood and anxiety should preferably be exam-ined as well, since these variables also seem to play a role for the treatment process and outcome. Presumably treatment success is helped or hindered by other reflections of negative emotionality rather than pure levels of fear-avoidance beliefs, which implies a need to further develop exposure in vivo to benefit all patients with pain-related fear.

Even though the results from this study are informative for improving exposure, the contribution to the conceptual understanding of catastro-phizing is limited. While the results denote that catastrophizing is closely linked to overt behavioral avoidance, they tell us little about how catastro-phizing is related to other factors such as the framing of the pain. It might be that patients who catastrophize frame their pain as being strictly due to medical factors and search for medical explanations and cures for it. This would make them more resistant to psychological treatments in general, and to exposure in particular, as this treatment requires that the patient


engages in several fear-provoking activities that may be interpreted as harmful. If catastrophizing is related to problem framing, it might be im-portant that the clinician identifies and addresses catastrophizing during the exposure. More research about possible links between problem fram-ing, catastrophizing, and overt behavior, which is the topic in study III, will therefore be informative for the further development of treatments for patients with pain-related fear.


STUDY III

UNDERSTANDING CATASTROPHIZING FROM A MISDIRECTED PROBLEM SOLVING PERSPECTIVE

Introduction The misdirected problem solving model has been presented as a framework to explain how worry and catastrophizing might lead to rigid, medically oriented problem solving behaviors in patients with chronic pain (Eccleston & Crombez, 2007). This model underscores that the beliefs patients have about their pain, the problem framing in other words, is closely linked to worry. When the worry progressively intensifies and the outcome is per-ceived as getting worse and worse, it has been defined as catastrophic worry (Davey & Levy, 1998), as captured by the concept catastrophizing. The misdirected problem solving model implies that, in cases where no medical solution exists, patients who catastrophize and have a strict bio-medical problem framing might end up in repeated “misdirected” efforts at finding a solution, or cure, for pain.

The expanded fear-anxiety-avoidance model also underscores the poten-tial maladaptiveness in framing persistent pain in strict biomedical terms, here captured by pain beliefs such as “pain equals damage or serious in-jury”(Asmundson, et al., 2004). In this model, biomedical problem framing is proposed to trigger catastrophizing, which in turn enhances the percep-tion of threat and the motivation to find a solution, or cure, for the pain.

In sum, both of the aforementioned models imply that back pain pa-tients who catastrophize are likely to view the problem as primarily bio-medical and, as a result, attempt to find medical solutions which often prove unsuccessful. The models might be seen as complementary rather than competing. However, the models do imply somewhat different path-ways. In the formation of the misdirected problem solving model, a bio-medical problem frame mediates the relation between catastrophic worry and problem solving behavior, while the fear-anxiety-avoidance model implies that it is the other way around, and that catastrophizing is the me-diator. These links have not been examined empirically.

Aim The aim was to explore if and how catastrophizing was linked to biomedi-cal problem framing and medically oriented problem solving. We examined two possible models of mediation as inferred by the aforementioned theo-retical models. The goal was to increase our understanding of catastrophiz-ing from a problem solving perspective.


Overview of the design This study is based on data from a prospective study about psychological processes in the development of chronic pain problems (Boersma & Lin-ton, 2005). Participants answered questionnaires on two occasions. On the first occasion (timepoint 1, t1), catastrophizing and biomedical problem framing were assessed. On the second occasion (timepoint 2, t2), seven months later, medically oriented problem solving was assessed.

Participants Participants (N = 173) were recruited via advertisements in local newspa-pers. The participants had spinal pain and at least one day of sick-leave the previous year because of the pain problem. The mean age was 57.

Measurements

Background, pain, and sick leave Background variables (age, gender, and nationality), pain characteristics (intensity, location, and duration), and sick leave were assessed with ques-tions taken from the Örebro Musculoskeletal Pain Screening Questionnaire (ÖMSPQ) (Linton & Halldén, 1998).

Catastrophizing Catastrophizing was measured with the Catastrophizing subscale from the Coping Strategies Questionnaire (CSQ) (Rosenstiel & Keefe, 1983).

Biomedical problem framing The Somatic Focus subscale from the Tampa Scale for Kinesiophobia (TSK) (Kori, et al., 1990) was used as a proxy for biomedical problem framing. Even though the TSK was originally developed to assess fear of movement and (re) injury, the Somatic Focus subscale focuses specifically on beliefs about pain signaling underlying serious medical problems. Therefore, this subscale fits well with the concept of biomedical problem framing.

Medically oriented problem solving behavior Health care utilization was used as a proxy for medically oriented problem solving behavior, and was assessed with questions from the Outcome Evaluation Questionnaire (Keefe, Linton, & Lefebvre, 1992).


Function Function was assessed with five questions from the Örebro Musculoskele-tal Pain Screening Questionnaire (ÖMSPQ) (Linton & Halldén, 1998).

Statistical analysis Two possible models of mediation were explored. In the first model (Alter-native 1), which is in line with the misdirected problem solving model, biomedical problem framing was proposed as a mediator of the relation between catastrophizing and medically oriented problem solving behavior (see figure 7). In the second model (Alternative 2), which is in line with the fear-anxiety-avoidance model, catastrophizing was proposed as a mediator of the relation between biomedical problem framing and medically ori-ented problem solving behavior (see figure 8).

FIGURE 7. Alternative 1. Possible model of mediation with biomedical problem framing as a proposed mediator

FIGURE 8. Alternative 2. Possible model of mediation with catastrophizing as a proposed mediator

We used two different approaches to explore whether the data supported any of these possible models of mediation. First, multiple regressions were used to investigate whether the prerequisites of mediation effects were met, using the traditional recommendations by Baron & Kenny (1986). Second, the significance of the indirect effects were assessed with a bootstrapping

Catastrophizing (t1)

Medically oriented problem solving behavior (t2)

Biomedical problem framing (t1)

Catastrophizing (t1)


Medically oriented problem solving behavior (t2)


method (n = 1000 bootstrap resamples) (see, Preacher & Hayes, 2008). Bootstrapping is a nonparametric resampling procedure that generates an approximation of the sampling distribution from the available data set. The bootstrapping distribution is generated by taking a sample (with re-placement) of size n from the full data set. The indirect effects are then calculated in the resamples which results in point estimates and 95% con-fidence intervals are estimated for the indirect effects. We considered point estimates of indirect effects to be significant if zero was not contained in the confidence interval.

Results In the multiple regression analyses, the data best matched Alternative 2 (the fear-anxiety-avoidance model), as the effect of somatic focus (proxy for biomedical problem framing) on health care utilization was reduced to a non-significant level (and not the other way around) when both catastro-phizing and somatic focus were used as predictors in the regression. These results are displayed in table 4.


Table 4 Multiple regression analyses exploring two possible models of mediation.

B SE B β

Alternative 1: Misdirected problem solving model

Step 1

Constant .90 .19

Catastrophizing .05 .01 .31**

Step 2

Constant .57 .32

Catastrophizing

Somatic focus

.04

.04

.01

.03

.26**

.11Ns

Alternative 2: Fear-anxiety-avoidance model

Step 1

Constant .62 .33

Somatic focus .08 .03 .24**

Step 2

Constant .57 .32

Somatic focus

Catastrophizing

.04

.04

.03

.01

.11Ns

.26**

a Note. Alternative 1 R2 = .10 for Step 1; Δ R2 = .01 for Step 2 (Ns). b Note. Alternative 2 R2 = .06 for Step 1: Δ R2 = .05 for Step 2 (p < .01). c Note. Health care utilization was used as the criterion variable in the analyses **p < .01

Subsequently, a bootstrapping method was used to explore the significance of the indirect effects in both possible models of mediation. The analyses showed that in Alternative 1, the indirect effect of catastrophizing on health care utilization through somatic focus as a mediator was not signifi-cant (point estimate = .009; CI = - . 005 - .024; p = .20) In Alternative 2, however, the indirect effect of somatic focus on health care utilization


through catastrophizing as a mediator was significant (point estimate = .042; CI = .015 - .078); p < .01).

Taken together, in both approaches of mediation analyses, the data sup-ported Alternative 2 (the fear-anxiety-avoidance model) where catastro-phizing was proposed as a mediator of the relation between somatic focus and health care utilization.

Conclusions This study provided support for framing pain catastrophizing within a problem solving perspective. The results confirmed the concepts involved in the misdirected problem solving model. However, the direction of the relations was somewhat different than the model proposes.

The mediation analyses supported catastrophizing as a mediator be-tween biomedical problem framing and medically oriented problem solv-ing. In other words, if a patient holds strong beliefs that the pain is due solely to medical factors, this might trigger catastrophizing and the patient is repeatedly seeking medical care in the hope of finding a “cure” or pain relief. This sequential order is not in line with the misdirected problem solving model, but is rather congruent with the expanded fear-anxiety-avoidance model. The latter model implies that pain beliefs, such as “pain means harm or serious injury”, or in other words biomedical problem framing, precedes catastrophizing and medically oriented problem solving.

This study provides support for one theoretical framework around catastrophizing, which posits that it is a “misdirected” attempt to solve a problem. The findings have implications theoretically as well as clinically. Theoretically, the concepts in the misdirected problem solving model were confirmed, but the directions were somewhat different than the model implies. This has important implications for further development of the theoretical models around catastrophizing. Clinically, this brings up the importance of examining how patients with spinal pain view their prob-lem. If the pain is regarded as a definite signal of something medically wrong it is vital for the clinician to assess and possibly also address catas-trophic worry.


General discussion Catastrophizing plays a key role in how pain is experienced and dealt with, not only in persistent pain but also in labor pain, during exposure treat-ment, and in a problem solving context. This dissertation underscores the vital role of catastrophizing in these new areas, giving rise to clinical as well as conceptual issues.

Catastrophizing plays a noteworthy role during childbirth – this was the main finding from study I. Childbirth is a unique situation which generally involves extremely intense acute pain where the role of catastrophizing has practically been unexplored. Research in this area has mainly focused on the broader construct of fear of childbirth, which embraces a woman’s fear for her own or the baby’s health, fear of losing control as well as fear of pain, where the latter is one of the most commonly reported reasons (Eriksson, Westman, & Hamberg, 2006; Geissbuehler & Eberhard, 2002; Sjögren, 1997). Fear of childbirth has been related to a lower pain thres-hold (Saisto, Kaaja, Ylikorkala, & Halesmäki, 2001). The results from study I are in line with these findings but expand them by specifically pointing to pain catastrophizing as a critical component which presumably is related to these other, broader fear constructs.

Catastrophizing is also critical for the outcome of exposure treatment – this was the main finding from study II. Exposure in vivo is a psychological treatment for back pain patients with pain-related fear where the exact impact of catastrophizing has not been known. Earlier studies have hy-pothesized that exposure would be most effective for patients with high levels of fear (George, et al., 2008; Leeuw, et al., 2008). However, the em-pirical data has not supported this hypothesis. Our data from study II rather points in the opposite direction; exposure was not effective for pa-tients with high levels of catastrophizing which is often related to high fear. Our results are in line with findings from studies of other CBT treatments for pain problems, where catastrophizing has indeed been linked to poor outcome (McCracken & Turk, 2002; Turner, et al., 2007). We found this was also the case in exposure. This is informative for the application and further development of exposure as it points to a need for the refinement of treatment methods. It is also informative for the conceptual understanding of catastrophizing as it implies that it is closely linked to overt behavioral avoidance.

Catastrophizing might be framed in a problem solving perspective – this was the main finding from study III. The misdirected problem solving model is a recent attempt to integrate worry, and thereby even catastro-phizing, within a problem solving framework. However, empirical data


about the proposed links has been sparse. Earlier research has specifically stressed the importance of examining in what way catastrophizing is linked to overt problem solving behavior (Crombez, et al., 2008). The findings from study III add to this literature by providing preliminary support for placing catastrophizing within a problem solving framework. However, our data suggested a somewhat different pathway than the model pro-poses, which was more in line with the fear-anxiety-avoidance model. This implies that the theoretical models of catastrophizing are still far from complete. To summarize, on the one hand these findings denote that catas-trophizing might be framed in a problem solving perspective. On the other hand, they imply that there is still room for improvement in existing theo-retical frameworks.

There is a need for further development in the theoretical framework around catastrophizing because of two main reasons. First, the findings from study III highlight a need to refine the links proposed by the misdi-rected problem solving model. Second, existing theoretical models concern-ing catastrophizing do not highlight the function of catastrophizing. Ex-ploring the function would give valuable information about why people catastrophize – in other words, the role of catastrophizing within the indi-vidual.

How are these findings related to theoretical frameworks around catastrophizing? The findings from study I might be explained by the fear-avoidance model (Vlaeyen & Linton, 2000). This model was originally developed to describe the interaction between cognitive and behavioral factors in the develop-ment of problems with persistent back pain. Recent research has shown that the components in this model are also relevant and might signal en-hanced risk in acute pain conditions (for a review, see Leeuw, et al., 2007) as well as in pain-free individuals (Buer & Linton, 2002). Our results repli-cate and extend these findings to pain in childbirth, as one interpretation is that women with fear-avoidant beliefs interpret labor pain as threatening, which in turn evokes catastrophizing. Consequently, they apparently be-come tense and more vigilant to pain signals, which might explain their higher ratings of pain. After childbirth, catastrophizing women had lower ratings of physical recovery suggesting avoidance of physical activities. This expands on findings from an earlier study, where catastrophizing was related to tendencies to avoid pain during the delivery (Van den Bussche, Crombez, Eccleston, & Sullivan, 2007). Our findings denote that this ten-dency continues after childbirth.


The fear-avoidance model might also be used to explain the findings from study II. Apparently, patients with too high levels of catastrophizing were not helped by exposure in vivo. This treatment is partly based on the Pavlovian extinction of fear paradigm (Bouton, 1988) and partly on the notion that many patients with non-specific back pain harbor irrational beliefs about pain and movements as harmful (Vlaeyen, de Jong, Geilen, Heuts, & van Breukelen, 2001). The basic idea is that patients will become less fearful by gradual confrontation to fear-provoking movements. The fear will decrease partly through habituation, and partly through discon-formation of beliefs. Consequently, they will recover in terms of function-ing. This did not happen in high catastrophizing patients. One explanation is that too high levels of catastrophizing might work as a safety behavior, which refers to subtle avoidant strategies that patients use to “protect” themselves from fear-provoking stimulus (Vlaeyen, et al., 2004). Catastro-phizing would then be a form of subtle covert avoidance.

This fits well with the process perspective, where catastrophizing is in-deed viewed as a form of avoidance (Stroebe, et al., 2007). According to this perspective, catastrophic thinking is proposed to serve the same func-tion as other forms of avoidance, namely to down-regulate negative affect; the patient engages in passive, abstract catastrophizing as a way of reduc-ing the intensity of the fear response. Applied to study II, this would mean that patients with high levels of catastrophizing were stuck in these ab-stract, passive patterns of thinking instead of engaging in the exposure treatment, which might have hindered them from getting in contact with the present moment contingencies. Consequently, the habituation did not occur and the treatment effect was never achieved. Hence, the findings from study II might be explained by a process perspective on catastrophiz-ing.

The descriptive data of changes in psychological variables during treat-ment in study II might also be viewed in the light of a process perspective. Low change participants did not improve at all on ratings of depressed mood. Instead, their ratings on the posttest were above the cut-off for “possible cases of depression” (HADS subscale depression > 7) (Herrmann, 1997). Consequently, patients who did not improve by the exposure had substantial levels of both depressed mood and catastrophizing, which sug-gests commonalities between the two. According to the process perspective, catastrophizing is a form of passive repetitive thinking with negative con-tent. Similar patterns of recurrent negative thinking about self and others are well-known and salient features of depressed mood (Lyubomirsky & Nolen-Hoeksema, 1995). Applied to the results of study II, this means that patients who were dwelling on their pain complaints were possibly dwell-


ing on other negative aspects of life as well, and this repetitive negative thinking hindered them from fully participating in and benefitting from the exposure treatment.

The process perspective matches the misdirected problem solving model in that catastrophizing is linked to unconstructive problem solving. From this perspective, inflexible repetitive negative thinking is closely related to inflexibility in overt behavior. In study III, catastrophizing was indeed linked to repeated efforts at finding a medical solution for the pain prob-lem which, in cases where no medical solution actually exists, might be considered as inflexible “misdirected” problem solving behavior. Hence, these findings might be interpreted as additional support for the process perspective.

The results from study III confirmed the variables included in the misdi-rected problem solving model. However, the direction of the results was somewhat different than the model proposes, although this finding needs to be interpreted with caution because of methodological weaknesses (i.e. problem framing and catastrophizing were examined at the same time point). The analyses supported catastrophizing as a mediator between problem framing and medically oriented problem solving behavior. This pattern is in line with the fear-anxiety-avoidance model, where catastro-phizing is proposed to mediate the relation between pain beliefs and defen-sive motivation, which leads to escape or defensive behavior. Pain beliefs are in essence similar to biomedical problem framing; both terms refer to patients’ assumptions that pain is equivalent with harm or injury. In other words, patients who strongly believe that the pain is strictly due to medical factors such as serious injury are more likely to catastrophize than others. This finding also fits well with the appraisal model which views catastro-phizing as a result of underlying beliefs about pain, and perceived (in)capability to cope with the situation. Hence, study III suggest that bio-medical problem framing sets the context for catastrophizing instead of the other way around, which is informative for further development of the theoretical framework around catastrophizing. This finding is elaborated in the model below.

Taken together, the findings from the three studies are to a certain ex-tent explained by existing theoretical frameworks around catastrophizing. As discussed above, the results from study II and study III might further be viewed in light of a process perspective. To date, this perspective applied specifically to pain catastrophizing has only recently begun to be developed (Boersma, Flink, & Linton, 2011). The model below is elaborated in col-laboration with this work and is to be viewed as an initial attempt to inte-grate pain catastrophizing in a process perspective.


FIGURE 9. Model of catastrophizing from a process perspective

This model is proposed as an addition and complement to already existing models, in particular to the fear-anxiety-avoidance model and the misdi-rected problem solving model. Indeed, some parts of this model overlap with the misdirected problem solving model. However, what is different and central to this new model is the framing of catastrophizing as repetitive negative thinking, comparable to rumination or worry. This is inspired by the growing research about repetitive negative thinking as a transdiagnostic

Internal and external context: Predisposing factors (biological, social etc.), history, situational

and environmental demands

PAIN

Negative affect

CATASTROPHIC WORRY: Inflexible, abstract repetitive negative thinking

MOTIVATIONAL WORRY: Flexible, concrete problem-focused thinking

Overt behavior: Inflexible, repetitive

(misdirected) problem solving

Overt behavior: Flexible, concrete

(directed) problem solving


construct, identified across disorders (for a review, see Watkins, 2008). In the proposed model, catastrophizing is captured by the concept of catas-trophic worry, emphasizing that catastrophizing and worry are indistin-guishable from a process perspective. From this perspective, focus is on the function of catastrophizing rather than on the content.

According to this model, internal and external contexts set the stage for these processes. External context refers to the current situation around the individual, for example circumstances at work or at home, whereas key elements of the internal context are the individual’s learning history, state of mood, and more stable dispositional traits (for a review, see Watkins, 2008). One example of an internal context is biomedical problem framing, which refers to a strong belief that pain is due strictly to medical causes. This problem framing might have developed as a result of earlier experi-ences, for example through implicit or explicit messages from the health care system. This affects how the person interprets pain; pain signals might be interpreted as threatening and evoke thoughts about something being seriously wrong in the body (i.e. catastrophic worry), which in turn moti-vate the person to seek medical help to deal with the problem. This is in line with the findings from study III, and is one example of how internal and external contexts are important determinants for how the pain is sub-sequently dealt with.

The model displays how pain might trigger negative affect. When ex-periencing pain, a natural reaction, in fact, is to become distressed (“nega-tive affect”) and to become concerned about what causes and consequences the pain might have (“motivational worry”). A certain amount of worry is vital to prepare and motivate the person to deal with threats (Mathews, 1990). In the model, this is called motivational worry and is displayed in the left pathway. Several problem definitions are considered and solutions are generated and tested out in overt behavior. Finding a medical solution to get rid of pain might be one option, but when there is little chance of success, as is often the case in persistent back pain, it might be more adap-tive to change the problem formulation and consider other options where pain relief is not the main goal. This is in line with the motivational per-spective on coping (Van Damme, Crombez, & Eccleston, 2008). This per-spective has put forward that the most adaptive coping style when suffer-ing from persisting disabling pain is in fact to be flexible and shift between different goals. Furthermore, the considered options ought to be concrete. It has been suggested that, in general, repetitive thinking with a concrete content is constructive whereas repetitive thinking with an abstract content is unconstructive (Watkins, 2008). A concrete content implies that the person formulates several alternatives for dealing with the pain problem,


and is willing to shift between these even if it involves confrontation with fears or a reformulation of the problem and goals.

However, depending on the internal and external context, the negative affect might also end up in a more intense catastrophic worry and a “mis-directed” problem solving pattern, as illustrated in the right pathway in the model. Here, the person gets stuck in repetitive negative thinking with an abstract content. This involves an inflexible goal-formulation such as “get-ting rid of pain”. Indeed, catastrophizing has been related to directed ef-forts at curing pain in cases where the pain problem was insoluble (de Vlieger, et al., 2006). The proposed function of this repetitive negative thinking is to down-regulate, or avoid, negative affect. This is in line with the proposed function of maladaptive worry, namely to avoid the somatic element of the fear response (Borkovec, Alcaine, & Behar, 2004). This fits well with the aforementioned explanation of our results from study II; that patients who catastrophized did not make full contact with the fear re-sponse, which hampered the exposure.

According to a process perspective, it makes little sense to separate dif-ferent forms of repetitive negative thinking such as catastrophizing, (mal-adaptive) worry or rumination because they may all serve the same func-tion, namely to avoid or down-regulate negative affect. The term catastro-phic worry is therefore used as an overall label for this inflexible abstract repetitive negative thinking.

The catastrophic worry goes hand in hand with overt behavior which is also inflexible and repetitive. This might involve repeated unsuccessful visits to health care providers, as in study III, or persistence in avoiding movements, as in study II. These behaviors are naturally adaptive in some circumstances, for example when experiencing acute pain. However, if the person inflexibly persists in the same problem solving behavior without any success, it might be considered as a “misdirected” problem solving behavior.

This model is developed with the results from study III in mind. The data implied that catastrophizing mediated the relation between problem framing and medically oriented problem solving behavior. As can be seen in this new model, internal context such as problem framing is proposed to set the stage for catastrophic worry and inflexible problem solving behav-ior, which is in line with the findings from study III.


Clinical implications and directions for future research The three studies have several clinical implications and set directions for future research. The findings from study I are important information for maternal health services and imply a need to identify catastrophizing women during their pregnancy, possibly by employing screening question-naires. These findings may be viewed in relation to the large number of women suffering from fear of childbirth, about 6-11 % in the industrial-ized countries (for a review, see Saisto & Halmesmäki, 2003). Women with fear of childbirth often report fear of pain (Eriksson, et al., 2006; Geissbuehler & Eberhard, 2002; Sjögren, 1997). Since pain-related fear and catastrophizing are closely related this implies a need to assess whether women who seek help because of fear of childbirth also catastrophize. The next step is to develop appropriate interventions for these groups. Treat-ments for women with fear of childbirth traditionally consist of individual support given by an obstetrician or a midwife (Saisto, 2001). However, the findings from study I indicate that psychological processes are highly involved, which might well require psychological methods to be addressed. To develop psychological interventions that directly target fear of pain and catastrophizing in pregnant women are missions for future research.

The findings from study II imply a need to identify patients with high levels of catastrophizing before initiating exposure in vivo and to monitor their treatment process carefully. Possibly, these patients may need addi-tional interventions that directly target catastrophizing. There have been some attempts to develop treatments specifically targeting catastrophizing (Rodero, García, Casanueva, & Sobradiel, 2008; Thorn, Boothby, & Sulli-van, 2002). However, these interventions have mainly been cognitively oriented, focusing on examining and altering the content of the catastro-phic thinking.

The proposed model of catastrophizing from a process perspective en-tails new and different ways of addressing catastrophizing. If catastrophiz-ing is conceptualized as a form of covert avoidance, a functional treatment approach might be considered (for a description of a functional approach, see e.g., Luoma, Hayes, & Walser, 2007). Instead of altering the content, the intervention would then focus on contextual and functional aspects; to examine when and where catastrophizing occurs, and what it hinders the patient from doing. The explicit aim is not to reduce the repetitive thinking itself, but to reduce its impact on overt behavior. The ultimate aim is that the patient engages in overt behavior that is in line with long-term goals. It is, in theory, well possible to intertwine this approach within the exposure treatment, but the effects need to be empirically examined.


A complementary intervention is to work on altering the abstract con-tent of the catastrophic worry into a more concrete content (i.e. motiva-tional worry). The clinician might assist the patient in formulating and facing unwanted thoughts and feelings. This might be described as a form of interoceptive exposure (McNally, 2007). Interoceptive exposure has shown preliminary but promising results, including on measures of catas-trophizing (Flink, Nicholas, Boersma, & Linton, 2009), though the expo-sure in that study mainly focused on the pain sensations. An alternative would be to encourage the patient to formulate the catastrophizing in con-crete terms and to focus on these, in order to break the internal avoidance.

Another intervention which also has the potential to increase flexibility in thinking and overt behavior is problem solving skills training. This is a traditional approach within CBT, which has successfully been used with a range of psychological and health-related problems (for a review, see Ma-louff, Thorsteinsson, & Schutte, 2007). For example, it has shown good results for people with depression (Unützer, et al., 2002), a condition where repetitive negative thinking is a salient feature. In problem solving skills training, the patient works with increasing flexibility in several ways: in the problem definition, in generating solutions, and in overt behavior through testing out solutions in real life. Consequently, problem solving skills training has the potential to help patients who become stuck in the inflexible “misdirected” problem solving path, although the effects on re-petitive negative thinking need to be empirically tested.

Although the results from study III primarily entail theoretical contribu-tions, there are some clinical implications as well. In particular, the link between biomedical problem framing and catastrophizing is informative for clinicians who meet patients with back pain problems. Asking patients what they believe is the reason for their pain might be the first step to iden-tify catastrophizing. For back pain patients who strongly believe that the pain is strictly due to medical causes, despite indications or direct evidence to the contrary, it is important that the clinician addresses catastrophizing. Most clinicians do this by reassuring the patients that nothing is medically wrong with their back. However, there is surprisingly little evidence that reassurance is effective for calming patients (for a review, see Linton, McCracken, & Vlaeyen, 2008). Furthermore, a recent study found that patients who catastrophized were more often dissatisfied with their medical examination than others (Overmeer & Boersma, 2011). Medical examina-tions and general information are possibly needed, although they might not be enough for patients with high levels of catastrophizing. Another impor-tant aspect in this context is validation - to express acceptance of the pa-tient’s behavior, feelings, and situation as true (Fruzzetti, 2006). Presuma-


bly, a combination of reassurance in the form of concrete, specific informa-tion and a validating response from the clinician is needed, but further research needs to find out the best way to do this.

The last issue for future research is to empirically scrutinize the pro-posed model of catastrophizing. Indeed, study III preliminary supported some of the proposed links but these findings need to be replicated to strengthen the conclusions. Another area to examine empirically is the difference between motivational worry and catastrophic worry. In the model, one proposed difference between these concepts is the level of flexi-bility and concreteness, an assumption which is based on earlier research about repetitive negative thinking. However, whether this is the case spe-cifically in pain catastrophizing has not been explored empirically. More-over, the assumption that catastrophizing is a form of covert avoidance which serves the function of down-regulating negative affect also needs to be empirically tested, preferably in experimental studies. The last question which needs to be addressed empirically is whether pain catastrophizing really is comparable to other forms of repetitive negative thinking such as rumination and (maladaptive) worry. Taken together, the proposed model of catastrophizing from a process perspective is to a certain extent based on earlier research about repetitive negative thinking as a transdiagnostic concept, but the model needs to be empirically tested.

Methodological limitations There are some methodological shortcomings which might limit the con-clusions that may be drawn from these studies.

First, all findings rely on subjective self-report inventories. This might be a threat to the internal validity because of possible response biases (Kazdin, 2003). However, all three studies rely on data from two or more assess-ment points that are separated in time, thereby strengthening the conclu-sions. Furthermore, the majority of the measures utilized are well-known and accepted inventories which earlier research has found to be psycho-metrically sound and reliable tools for assessing the variables of interest. Moreover, most of the concepts of interest in these studies are by definition subjective. For example, catastrophizing and pain are generally assessed through self-reports because they are highly subjective and personal phe-nomenon.

A second limitation concerns the small sample in study II which re-stricted the statistical power and possibly also the reliability of the results. Even though more studies are needed to replicate the findings, the sample size ought to be viewed in light of the fact that exposure in vivo is quite a new intervention in the pain field and only a few randomized controlled


trials have been published. Yet, despite low power we did find significant effects indicating that these are indeed salient. Moreover, our findings fit well with both theory and research in other areas where catastrophizing has been linked to poor treatment outcome.

A third limitation concerns the recruitment of participants. In all three studies, the participants were volunteers and there might have been some self-selection. The main question here is whether it is possible to generalize the findings to clinical settings. However, there are no obvious indications that our participants differ either from pain patients at a clinic or from patients in other studies in levels of pain, disability, and background fea-tures (see e.g., Leeuw, et al., 2008; Woods & Asmundson, 2008). Conse-quently, there are reasons to assume that our findings might well be appli-cable to patients in a clinical setting.

What is to be underscored as one of the main strengths is that all three studies were prospective with at least two points of assessment. Prospective data strengthens the conclusions, especially when investigating possible moderators, as in study II, and mediators, as in study III. Furthermore, psychometrically sound measures have been used and the procedures have followed general recommendations for this type of research (Kazdin, 2003).

Summary and concluding remarks This dissertation has demonstrated that catastrophizing plays a critical role in pain in childbirth, during exposure in vivo, and in a problem solving context. Firstly, catastrophizing mattered for the experience of pain in childbirth, which implies a need to identify catastrophizing women during their pregnancy and provide support, preferably at maternal health ser-vices. Secondly, catastrophizing played the role of moderator of the treat-ment effect during exposure in vivo for back pain patients with pain-related fear. This is important information for the application and further development of this treatment. Moreover, it denotes that catastrophizing is closely linked to overt behavioral avoidance. Thirdly, catastrophizing played a role in a problem solving context. Although this is in line with contemporary models such as the misdirected problem solving model, our findings suggested a somewhat different pathway than that proposed by this model. This implies a need for the further development of the theoreti-cal framework around catastrophizing.

Based partly on the findings from the three studies and partly on existing theoretical frameworks, a somewhat different conceptualization of catas-trophizing has been proposed – a model of catastrophizing from a process perspective. This model is suggested as an addition and complement to


already existing theoretical models. What is new is the focus on the func-tion of catastrophizing rather than on the content. In this model, catastro-phizing is conceptualized as repetitive negative thinking. This repetitive negative thinking is viewed as a form of covert avoidance, with the pro-posed function to down-regulate negative affect.

Conclusions • Catastrophizing plays a vital role in pain perception during child-

birth, in the outcome of exposure treatment, and in a problem solv-ing context

• It is critical to assess and directly address catastrophizing in clinical contexts

• In childbirth: - Maternal health services might benefit from assessing whether women who seek help due to fear of childbirth also catastrophize - Preventative interventions which target catastrophizing in preg-nant women need to be developed and tested

• In exposure treatment: - Catastrophizing should be specifically addressed during exposure in vivo - Back pain patients who catastrophize possibly need additional in-terventions to benefit from exposure

• In a problem solving context: - Clinicians should explore what patients themselves think are the causes of their back pain - If the patient strongly believes that back pain is always due to in-jury or harm, they might have a tendency to catastrophize

• Clear instructions for how clinicians should address catastrophizing need to be developed

• Catastrophizing might be conceptualized as repetitive negative thinking which serves the function to down-regulate negative affect

• The process perspective opens new avenues for developing interven-tions which directly target catastrophizing: - by examining when and where the catastrophizing arises, and what it hinders the patient from doing - by reducing its impact on overt behavior through encouraging the patient to act in line with long-term goals - by problem solving skills training


- by interoceptive exposure • Successful interventions for people who catastrophize would lead to

several gains – for the individual in less suffering and an increased ability to handle pain problems, and for the society in reduced costs for health care for these individuals.


References Asmundson, G. J. G., Norton, P. J., & Vlaeyen, J. W. S. (2004). Fear-

avoidance models of chronic pain: An overview. In G.J.G. Asmundson, J.W.S. Vlaeyen, & G. Crombez (Eds.), Understanding and treating fear of pain (pp. 3–24). Oxford: Oxford University Press.

Atkinson, M., Zibin, S., & Chuang, H. (1997). Characterizing quality of life among patients with chronic mental illness: a critical examination of the self-report methodology. American Journal of Psychiatry, 154(1), 99-105.

Bailey, K. M., Carleton, R. N., Vlaeyen, J., & Asmundson, G. (2010). Treatments addressing pain-related fear and anxiety in patients with chronic musculoskeletal pain: a preliminary review. Cognitive Behav-iour Therapy, 39(1), 46-63.

Barlow, D. H. (2004). Anxiety and its disorders: The nature and treatment of anxiety and panic. London: The Guilford Press.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173-1182.

Beck, A. T. (1976). Cognitive therapy and the emotional disorders. Boston: International Universities Press.

Beck, A. T. (1979). Cognitive therapy of depression. London: The Guilford Press.

Bedard, G., Reid, G., McGrath, P., & Chambers, C. (1997). Coping and self-medication in a community sample of junior high school students. Pain Research and Management, 2, 151-156.

Boersma, K. (2005). Fear and avoidance in the development of a persistent musculoskeletal pain problem. Implications for secondary prevention. Doctoral dissertation, Örebro University - Sweden.

Boersma, K., Flink, I.K., & Linton, S.J. (2011). Pain catastrophizing as negative repetitive thinking. Manuscript in preparation.

Boersma, K., Linton, S., Overmeer, T., Jansson, M., Vlaeyen, J., & de Jong, J. (2004). Lowering fear-avoidance and enhancing function through exposure in vivo: A multiple baseline study across six patients with back pain. Pain, 108(1-2), 8-16.


Boersma, K., & Linton, S. J. (2005). How does persistent pain develop? An analysis of the relationship between psychological variables, pain and function across stages of chronicity. Behaviour Research and Therapy, 43(11), 1495-1507.

Borkovec, T. D., Alcaine, O., & Behar, E. (2004). Avoidance theory of worry and generalized anxiety disorder. In R.G. Heimberg, C.L. Turk, & D.S. Mennin (Eds.), Generalized anxiety disorder: Advances in re-search and practice (pp. 77–108). New York: The Guilford Press.

Bouton, M. E. (1988). Context and ambiguity in the extinction of emo-tional learning: Implications for exposure therapy. Behaviour Research and Therapy, 26(2), 137-149.

Buer, N., & Linton, S. J. (2002). Fear-avoidance beliefs and catastrophiz-ing: occurrence and risk factor in back pain and ADL in the general population. Pain, 99(3), 485-491.

Butler, A. C., Chapman, J. E., Forman, E. M., & Beck, A. T. (2006). The empirical status of cognitive-behavioral therapy: a review of meta-analyses. Clinical Psychology Review, 26(1), 17-31.

Campbell, C. M., Kronfli, T., Buenaver, L. F., Smith, M. T., Berna, C., Haythornthwaite, J. A., et al. (2010). Situational versus dispositional measurement of catastrophizing: associations with pain responses in multiple samples. The Journal of Pain, 11(5), 443-453.

Chaves, J. F., & Brown, J. M. (1987). Spontaneous cognitive strategies for the control of clinical pain and stress. Journal of Behavioral Medicine, 10(3), 263-276.

Cohen, C. E., & Ebbesen, E. B. (1979). Observational goals and schema activation: A theoretical framework for behavior perception. Journal of Experimental Social Psychology, 15(4), 305-329.

Cook, A. J., Brawer, P. A., & Vowles, K. E. (2006). The fear-avoidance model of chronic pain: validation and age analysis using structural equation modeling. Pain, 121(3), 195-206.

Crombez, G., Eccleston, C., Van Hamme, G., & De Vlieger, P. (2008). Attempting to solve the problem of pain: a questionnaire study in acute and chronic pain patients. Pain, 137(3), 556-563.

Davey, G. C. L., & Levy, S. (1998). Catastrophic Worrying: Personal In-adequacy and a Perseverative Iterative Style as Features of the Catastro-phizing Process* 1. Journal of Abnormal Psychology, 107(4), 576-586.


de Jong, J. R., Vlaeyen, J. W. S., Onghena, P., Goossens, M. E. J. B., Geilen, M., & Mulder, H. (2005). Fear of movement/(re) injury in chronic low back pain: education or exposure in vivo as mediator to fear reduction? The Clinical Journal of Pain, 21(1), 9-17.

de Vlieger, P., Bussche, E. V., Eccleston, C., & Crombez, G. (2006). Find-ing a solution to the problem of pain: Conceptual formulation and the development of the Pain Solutions Questionnaire (PaSol). Pain, 123(3), 285-293.

Descartes, R. (1989). The passions of soul. Indianapolis: Hacket.

Dixon, K. E., Thorn, B. E., & Ward, L. C. (2004). An evaluation of sex differences in psychological and physiological responses to experimen-tally-induced pain: a path analytic description. Pain, 112(1-2), 188-196.

Drossman, D. A., Leserman, J., Li, Z., Keefe, F., Hu, Y. J. B., & Toomey, T. C. (2000). Effects of coping on health outcome among women with gastrointestinal disorders. Psychosomatic Medicine, 62(3), 309-317.

Eccleston, C., & Crombez, G. (2007). Worry and chronic pain: A misdi-rected problem solving model. Pain, 132(3), 233-236.

Eccleston, C., Williams, A., & Morley, S. (2009). Psychological therapies for the management of chronic pain (excluding headache) in adults. Cochrane database of systematic reviews (Online)(2), CD007407.

Edwards, R. R., Campbell, C. M., & Fillingim, R. B. (2005). Catastrophiz-ing and experimental pain sensitivity: only in vivo reports of catastro-phic cognitions correlate with pain responses. The Journal of Pain, 6(5), 338-339.

Edwards, R. R., Smith, M. T., Stonerock, G., & Haythornthwaite, J. A. (2006). Pain-related catastrophizing in healthy women is associated with greater temporal summation of and reduced habituation to ther-mal pain. The Clinical Journal of Pain, 22(8), 730-737.

Ellis, A. (1962). Reason and emotion in psychotherapy. Oxford: Stuart.

Eriksson, C., Westman, G., & Hamberg, K. (2006). Content of childbirth related fear in Swedish women and men—Analysis of an open ended question. The Journal of Midwifery & Women’s Health, 51(2), 112-118.

Flink, I. K., Nicholas, M. K., Boersma, K., & Linton, S. (2009). Reducing the threat value of chronic pain: A preliminary replicated single-case


study of interoceptive exposure versus distraction in six individuals with chronic back pain. Behaviour Research and Therapy, 47, 721-728.

Fruzzetti, A. E. (2006). The high conflict couple. Oakland, CA: New Har-binger Publications.

Geissbuehler, V., & Eberhard, J. (2002). Fear of childbirth during preg-nancy: A study of more than 8000 pregnant women. Journal of Psycho-somatic Obstetrics & Gynecology, 23, 229-235.

George, S. Z., Zeppieri, G., Cere, A. L., Cere, M. R., Borut, M. S., Hodges, M. J., et al. (2008). A randomized controlled trial of behavioral physi-cal therapy interventions for acute and sub-acute low back pain. Pain, 140, 145-157.

Giardino, N. D., Jensen, M. P., Turner, J. A., Ehde, D. M., & Cardenas, D. D. (2003). Social environment moderates the association between catas-trophizing and pain among persons with a spinal cord injury. Pain, 106(1-2), 19-25.

Gil, K. M., Thompson, R. J., Keith, B. R., Tota-Faucette, M., Noll, S., & Kinney, T. R. (1993). Sickle cell disease pain in children and adoles-cents: change in pain frequency and coping strategies over time. Journal of Pediatric Psychology, 18(5), 621-637.

Herrmann, C. (1997). Internal experiences with the Hospital Anxiety and Depression Scale - a review of validation data and clinical results. Jour-nal of Psychosomatic Research, 42, 17-41.

Hirsh, A. T., George, S. Z., Riley III, J. L., & Robinson, M. E. (2007). An evaluation of the measurement of pain catastrophizing by the coping strategies questionnaire. European Journal of Pain, 11(1), 75-81.

Holmbeck, G. N. (2002). Post-hoc probing of significant moderational and mediational effects in studies of pediatric populations. Journal of Pedi-atric Psychology, 27(1), 87-96.

IASP (1994). Part III: Pain Terms, A current list with definitions and notes on usage. In H. Merskey & N. Bogduk (Eds.), Classification of Chronic Pain, Second Edition. (pp. 209-214). Seattle: IASP Press.

Kazdin, A. E. (2003). Research design in clinical psychology. Boston: Allyn & Bacon.

Keefe, F. J., Brown, G. K., Wallston, K. A., & Caldwell, D. S. (1989). Cop-ing with rheumatoid arthritis pain: catastrophizing as a maladaptive strategy. Pain, 37(1), 51-56.


Keefe, F. J., Linton, S. J., & Lefebvre, J. C. (1992). The Outcome Evalua-tion Questionnaire: description and initial findings. Cognitive Behav-iour Therapy, 21(1), 19-33.

Keefe, F. J., Lipkus, I., Lefebvre, J. C., Hurwitz, H., Clipp, E., Smith, J., et al. (2003). The social context of gastrointestinal cancer pain: a prelimi-nary study examining the relation of patient pain catastrophizing to pa-tient perceptions of social support and caregiver stress and negative re-sponses. Pain, 103(1-2), 151-156.

Keefe, F. J., Rumble, M. E., Scipio, C. D., Giordano, L. A., & Perri, L. M. (2004). Psychological aspects of persistent pain: Current state of the sci-ence. The Journal of Pain, 5(4), 195-211.

Keogh, E., & Asmundson, G. J. G. (2004). Negative affectivity, catastro-phizing and anxiety sensitivity. In G.J.G. Asmundson, J.W.S. Vlaeyen, & G. Crombez (Eds.), Understanding and treating fear of pain (pp. 91–115). Oxford: Oxford University Press.

Kopec, J. A., Esdaile, J. M., Abrahamowicz, M., Abenhaim, L., Wood-Dauphinee, W., Lamping, D. L., et al. (1995). The Quebec Back Pain Disability Scale. Spine, 20, 341-352.

Kori, S. H., Miller, R. P., & Todd, D. D. (1990). Kinesiophobia: a new view of chronic pain behavior. Pain Management, 3, 35-43.

Krismer, M., & Van Tulder, M. (2007). Low back pain (non-specific). Best Practice & Research Clinical Rheumatology, 21(1), 77-91.

Lackner, J. M., & Gurtman, M. B. (2004). Pain catastrophizing and inter-personal problems: a circumplex analysis of the communal coping model. Pain, 110(3), 597-604.

Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer Publishing Company.

Leeuw, M., Goossens, M., Linton, S. J., Crombez, G., Boersma, K., & Vlaeyen , J. W. S. (2007). The fear-avoidance model of musculoskeletal pain: current state of scientific evidence. Journal of Behavioral Medi-cine, 30(1), 77-94.

Leeuw, M., Goossens, M. E. J. B., van Breukelen, G. J. P., de Jong, J. R., Heuts, P. H. T. G., Smeets, R. J. E. M., et al. (2008). Exposure in vivo versus operant graded activity in chronic low back pain patients: Re-sults of a randomized controlled trial. Pain, 138, 192-207.


Linton, S. J., Boersma, K., Jansson, M., Overmeer, T., Lindblom, K., & Vlaeyen, J. W. S. (2008). A randomized controlled trial of exposure in vivo for patients with spinal pain reporting fear of work-related activi-ties. European Journal of Pain, 12(6), 722-730.

Linton, S. J., & Halldén, K. (1998). Can we screen for problematic back pain? A screening questionnaire for predicting outcome in acute and subacute back pain. The Clinical Journal of Pain, 14(3), 209-215.

Linton, S. J., McCracken, L. M., & Vlaeyen, J. W. S. (2008). Reassurance: help or hinder in the treatment of pain. Pain, 134(1-2), 5-8.

Linton, S. J., Nicholas, M. K., MacDonald, S., Boersma, K., Bergbom, S., Maher, C., et al. (2011). The role of depression and catastrophizing in musculoskeletal pain. European Journal of Pain, 15(4), 416-422.

Lohnberg, J. A. (2007). A review of outcome studies on cognitive-behavioral therapy for reducing fear-avoidance beliefs among individu-als with chronic pain. Journal of Clinical Psychology in Medical Set-tings, 14(2), 113-122.

Lu, Q., Tsao, J. C. I., Myers, C. D., Kim, S. C., & Zeltzer, L. K. (2007). Coping predictors of children's laboratory-induced pain tolerance, in-tensity, and unpleasantness. The Journal of Pain, 8(9), 708-717.

Luoma, J. B., Hayes, S. C., & Walser, R. D. (2007). Learning ACT: An Acceptance & Commitment Therapy skills-training manual for thera-pists. Oakland, CA: New Harbinger Publications.

Lyubomirsky, S., & Nolen-Hoeksema, S. (1995). Effects of self-focused rumination on negative thinking and interpersonal problem solving. Journal of Personality and Social Psychology, 69(1), 176-190.

Main, C. J., Sullivan, M. J. L., & Watson, P. J. (2008). Pain management: Practical applications of the biopsychosocial perspective in clinical and occupational settings. Amsterdam: Elsevier Health Sciences.

Malouff, J. M., Thorsteinsson, E. B., & Schutte, N. S. (2007). The efficacy of problem solving therapy in reducing mental and physical health problems: a meta-analysis. Clinical Psychology Review, 27(1), 46-57.

Mathews, A. (1990). Why worry? The cognitive function of anxiety. Be-haviour Research and Therapy, 28(6), 455-468.

McCracken, L. M., & Turk, D. C. (2002). Behavioral and cognitive-behavioral treatment for chronic pain: outcome, predictors of outcome, and treatment process. Spine, 27(22), 2564-2573.


McNally, R. J. (2007). Mechanisms of exposure therapy: how neuroscience can improve psychological treatments for anxiety disorders. Clinical Psychology Review, 27, 750-759.

Melzack, R. (1975). The McGill Pain Questionnaire: major properties and scoring methods. Pain, 1(3), 277-299.

Melzack, R. (1993). Labour pain as a model of acute pain. Pain, 53(2), 117-120.

Meyer, K., Sprott, H., & Mannion, A. F. (2008). Cross-cultural adapta-tion, reliability, and validity of the German version of the Pain Catas-trophizing Scale. Journal of Psychosomatic Research, 64(5), 469-478.

Miró, J., Nieto, R., & Huguet, A. (2008). The Catalan version of the Pain Catastrophizing Scale: a useful instrument to assess catastrophic think-ing in whiplash patients. The Journal of Pain, 9(5), 397-406.

Nicassio, P. M., Schoenfeld-Smith, K., Radojevic, V., & Schuman, C. (1995). Pain coping mechanisms in fibromyalgia: relationship to pain and functional outcomes. Journal of Rheumatology, 22(8), 1552-1558.

Nicholas, M. K. (2008). Pain management in musculoskeletal conditions. Best Practice & Research Clinical Rheumatology, 22(3), 451-470.

Nicholas, M. K. (2009). First things first: Reductions in catastrophising before fear of movement. Pain, 145(1), 6-7.

Nolem-Hoeksema, S. (2005). Lost in thought: Rumination and depression. Paper presented at the Annual Conference of the American Psychologi-cal Association, Washington DC.

Ohman, A. (2000). Fear and anxiety: Evolutionary, cognitive, and clinical perspectives. Handbook of emotions, 2, 573-593.

Osman, A., Barrios, F. X., Gutierrez, P. M., Kopper, B. A., Merrifield, T., & Grittmann, L. (2000). The Pain Catastrophizing Scale: further psy-chometric evaluation with adult samples. Journal of Behavioral Medi-cine, 23(4), 351-365.

Osman, A., Barrios, F. X., Kopper, B. A., Hauptmann, W., Jones, J., & O'Neill, E. (1997). Factor structure, reliability, and validity of the Pain Catastrophizing Scale. Journal of Behavioral Medicine, 20(6), 589-605.

Overmeer, T., & Boersma, K. (2011). The relationship between psycho-logical risk factors and treatment experience. Manuscript in prepara-tion.


Peters, M. L., Vlaeyen, J. W. S., & Weber, W. E. J. (2005). The joint con-tribution of physical pathology, pain-related fear and catastrophizing to chronic back pain disability. Pain, 113(1-2), 45-50.

Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strate-gies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879-891.

Pukall, C. F., Binik, Y. M., Khalifé, S., Amsel, R., & Abbott, F. V. (2002). Vestibular tactile and pain thresholds in women with vulvar vestibulitis syndrome* 1. Pain, 96(1-2), 163-175.

Quartana, P. J., Campbell, C. M., & Edwards, R. R. (2009). Pain catas-trophizing: a critical review. Expert Review of Neurotherapeutics, 9(5), 745-758.

Rodero, B., García, J., Casanueva, B., & Sobradiel, N. (2008). Imagined exposure as treatment of catastrophizing in fibromyalgia: a pilot study. Actas españolas de psiquiatría, 36(4), 223-226.

Rosenstiel, A. K., & Keefe, F. J. (1983). The use of coping strategies in chronic low back pain patients: relationship to patient characteristics and current adjustment. Pain, 17(1), 33-44.

Saisto, T. (2001). Obstetric, psychosocial, and pain-related background, and treatment of fear of childbirth. Doctoral dissertation, University of Helsinki - Finland.

Saisto, T., & Halmesmäki, E. (2003). Fear of childbirth: a neglected di-lemma. Acta Obstetricia et Gynecologica Scandinavica, 82(3), 201-208.

Saisto, T., Kaaja, R., Ylikorkala, O., & Halesmäki, E. (2001). Reduced pain tolerance during and after pregnancy in women suffering from fear of childbirth. Pain, 93, 123-127.

Severeijns, R., Vlaeyen, J. W. S., & van den Hout, M. A. (2006). Do we need a communal coping model of pain catastrophizing? An alternative explanation. Pain, 11(1), 226-229.

Severeijns, R., Vlaeyen, J. W. S., van den Hout, M. A., & Weber, W. E. J. (2001). Pain catastrophizing predicts pain intensity, disability, and psy-chological distress independent of the level of physical impairment. The Clinical Journal of Pain, 17(2), 165-172.

Sjögren, B. (1997). Reasons for anxiety about childbirth in 100 pregnant women. Journal of Psychosomatic Obstetrics & Gynecology, 18, 266-272.


Spanos, N. P., Radtke-Bodorik, H. L., Ferguson, J. D., & Jones, B. (1979). The effects of hypnotic susceptibility, suggestions for analgesia, and the utilization of cognitive strategies on the reduction of pain. Journal of Abnormal Psychology, 88(3), 282-292.

Stroebe, M., Boelen, P. A., van den Hout, M., Stroebe, W., Salemink, E., & van den Bout, J. (2007). Ruminative coping as avoidance. European Archives of Psychiatry and Clinical Neuroscience, 257(8), 462-472.

Sullivan, M. J. L., Adams, H., & Sullivan, M. E. (2004). Communicative dimensions of pain catastrophizing: social cueing effects on pain behav-iour and coping. Pain, 107(3), 220-226.

Sullivan, M. J. L., Bishop, S. R., & Pivik, J. (1995). The Pain Catastrophiz-ing Scale: development and validation. Psychological Assessment, 7, 524-532.

Sullivan, M. J. L., Rodgers, W. M., Wilson, P. M., Bell, G. J., Murray, T. C., & Fraser, S. N. (2002). An experimental investigation of the rela-tion between catastrophizing and activity intolerance. Pain, 100, 47-53.

Sullivan, M. J. L., Thorn, B., Haythornthwaite, J. A., Keefe, F. J., Martin, M., Bradley, L. A., et al. (2001). Theoretical perspectives on the relation between catastrophizing and pain. The Clinical Journal of Pain, 17, 52-64.

Swinkels-Meewisse, I. E. J., Roelofs, J., Oostendorp, R. A. B., Verbeek, A. L. M., & Vlaeyen, J. W. S. (2006). Acute low back pain: pain-related fear and pain catastrophizing influence physical performance and per-ceived disability. Pain, 120(1-2), 36-43.

Thorn, B., Ward, L., Sullivan, M., & Boothby, J. (2003). Communal cop-ing model of catastrophizing: conceptual model building. Pain, 106(1), 1-2.

Thorn, B. E., Boothby, J. L., & Sullivan, M. J. L. (2002). Targeted treat-ment of catastrophizing for the management of chronic pain. Cognitive and Behavioral Practice, 9(2), 127-138.

Turk, D. C., & Okifuji, A. (2001). Pain terms and taxonomies of pain. In J. J. Bonica, J. D. Loeser, C. R. Chapman, D. C. Turk & S. H. Butler (Eds.), Bonica's management of pain. Hagerstwon, MD: Lippincott Williams & Wilkins.

Turner, J. A., & Aaron, L. A. (2001). Pain-related catastrophizing: What is it? The Clinical Journal of Pain, 17(1), 65-71.


Turner, J. A., Holtzman, S., & Mancl, L. (2007). Mediators, moderators, and predictors of therapeutic change in cognitive-behavioral therapy for chronic pain. Pain, 127, 276-286.

Turner, J. A., Jensen, M. P., Warms, C. A., & Cardenas, D. D. (2002). Catastrophizing is associated with pain intensity, psychological distress, and pain-related disability among individuals with chronic pain after spinal cord injury. Pain, 98(1-2), 127-134.

Unützer, J., Katon, W., Callahan, C. M., Williams, J. W., Hunkeler, E., Harpole, L., et al. (2002). Collaborative care management of late-life depression in the primary care setting. Journal of the American Medical Association, 288(22), 2836-2845.

Waddell, G. (2004). The back pain revolution. Amsterdam: Elsevier Health Sciences.

Van Damme, S., Crombez, G., & Eccleston, C. (2008). Coping with pain: a motivational perspective. Pain, 139(1), 1-4.

Van den Bussche, E., Crombez, G., Eccleston, C., & Sullivan, M. J. L. (2007). Why women prefer epidural analgesia during childbirth: the role of beliefs about epidural analgesia and pain catastrophizing. Euro-pean Journal of Pain, 11(3), 275-282.

van Tulder, M. W., Ostelo, R., Vlaeyen, J. W. S., Linton, S. J., Morley, S. J., & Assendelft, W. J. J. (2001). Behavioral treatment for chronic low back pain: a systematic review within the framework of the Cochrane Back Review Group. Spine, 26(3), 270-281.

Watkins, E. R. (2008). Constructive and unconstructive repetitive thought. Psychological Bulletin, 134(2), 163-206.

Wideman, T. H., Adams, H., & Sullivan, M. J. L. (2009). A prospective sequential analysis of the fear-avoidance model of pain. Pain, 145(1-2), 45-51.

Vlaeyen, J. W. S., Crombez, G., & Linton, S. J. (2009). The fear-avoidance model of pain: We are not there yet. Pain, 146(1), 222-223.

Vlaeyen, J. W. S., de Jong, J., Geilen, M., Heuts, P. H. T. G., & van Breu-kelen, G. (2001). Graded exposure in vivo in the treatment of pain-related fear: a replicated single-case experimental design in four patients with chronic low back pain. Behaviour Research and Therapy, 39(2), 151-166.


Vlaeyen, J. W. S., de Jong, J., Leeuw, M., & Crombez, G. (2004). Fear reduction in chronic pain: graded exposure in vivo with behavioral ex-periments. In G.J.G. Asmundson, J.W.S. Vlaeyen, & G. Crombez (Eds.), Understanding and treating fear of pain (pp. 313-346). Oxford: Oxford University Press.

Vlaeyen, J. W. S., de Jong, J. R., Geilen, M., Heuts, P., & van Breukelen, G. (2001). Graded exposure in vivo in the treatment of pain-related fear: a replicated single-case experimental design in four patients with low back pain. Behavioral Research and Therapy, 39, 151-166.

Vlaeyen, J. W. S., & Linton, S. J. (2000). Fear-avoidance and its conse-quences in chronic musculoskeletal pain: A state of the art. Pain, 85, 317-332.

Vlaeyen, J. W. S., & Morley, S. (2005). Cognitive-behavioral treatments for chronic pain: what works for whom? The Clinical Journal of Pain, 21(1), 1-8.

Woods, M. P., & Asmundson, G. J. G. (2008). Evaluating the efficacy of graded in vivo exposure for the treatment of fear in patients with chronic back pain: a randomized controlled clinical trial. Pain, 136(3), 271-280.

Yap, J. C., Lau, J., Chen, P. P., Gin, T., Wong, T., Chan, I., et al. (2008). Validation of the Chinese Pain Catastrophizing Scale (HK PCS) in Pa-tients with Chronic Pain. Pain Medicine, 9(2), 186-195.

Zigmond, A. S., & Snaith, R. P. (1983). The Hospital Anxiey and Depres-sion Scale. Acta Psychiatrica Scandinavica, 67, 361-370.

Pain in childbirth and postpartum recovery – The role of catastrophizing

Ida K. Flink a,*, Magdalena Z. Mroczek b, Michael J.L. Sullivan c, Steven J. Linton a

aCenter for Health and Medical Psychology, Department of Behavioural, Social and Legal Sciences – Psychology, Örebro University, SwedenbOccupational and Environmental Health, Stockholm Centre for Public Health, Stockholm County Council, SwedencDepartments of Psychology, Medicine and Neuroscience, McGill University, Montreal, Canada

a r t i c l e i n f o

Article history:Received 29 November 2007Received in revised form 26 March 2008Accepted 17 April 2008Available online 2 June 2008

Keywords:CatastrophizingPainPCSChildbirthPhysical recovery

a b s t r a c t

This prospective study investigated how pain catastrophizing was related to labor pain intensity andphysical recovery after childbirth. Eighty-eight women giving birth for the first time completed the firstquestionnaire before delivery. Eighty-two of those returned the second questionnaire after delivery. Par-ticipants were classified as catastrophizers (n = 38) or non-catastrophizers (n = 44) based on their scoreson the Pain Catastrophizing Scale. Comparison of the groups showed that catastrophizers anticipated andexperienced more intense pain (p < .0125) and had poorer physical recovery (p < .0125), measured as thelevel of self-reported functioning in activities of daily living, than non-catastrophizers. These resultsextend the association between catastrophizing and pain, to pain and recovery in childbirth and providesupport for the fear-avoidance model. It is concluded that pain catastrophizing plays a role in the expe-rience of pain in childbirth and postpartum recovery. Further research is needed to identify appropriateinterventions for catastrophizing women during the latter part of pregnancy.� 2008 European Federation of Chapters of the International Association for the Study of Pain. Published

by Elsevier Ltd. All rights reserved.

1. Introduction

Pain during childbirth is one of the most excruciating painexperiences that women encounter in their lives (Melzack, 1993).Although there is great variability, about 60% of the women whogive birth for the first time describe their pain as severe or extre-mely severe (Melzack, 1993). Consequently, anticipation of theremarkably intense pain of childbirth may be associated with psy-chological reactions, such as pain catastrophizing.

Catastrophizing emerges in the literature as one of the most po-tent predictors of pain in general (Vlaeyen and Linton, 2000; Keefeet al., 2004; Sullivan et al., 2002; Severeijns et al., 2001). It has beenassociated with heightened pain across several pain populations,and accounts for 7–31% of the variance in pain ratings (Sullivanet al., 1995, 2001). Catastrophizing has also been associated witha number of pain-related outcomes, including reduced involve-ment in daily activities (Keefe et al., 1989) and activity intolerance(Sullivan et al., 2002). Moreover, in the fear-avoidance model,catastrophizing is described as a cognitive precursor to pain-re-lated fear and tendencies to avoid pain, which may enhance painintensity and hinder resumption of physical activity (Vlaeyen andLinton, 2000). The role of catastrophizing has mainly been studiedin chronic and pathological pain conditions and its influences onchildbirth remains unclear.

Melzack (1993) claimed that childbirth provides an excellentmodel of acute pain with a clear beginning, progress and ending.However, labor pain is different from pathological pain, both inthe physiological (Catheline et al., 2006) and the psychological pro-cess (Olofsson, 2003). Usually pain is a signal of a potential threat,which should be diminished or eliminated. In contrast, labor paindoes not usually mean that something is wrong but is an importantsignal to the mother that labor is proceeding. Furthermore, child-birth differs from other types of pain with respect to its positiveoutgrowths; a child is born. Most other pain conditions (e.g. injury,illness) that have been studied do not have these positive associa-tions. Thus, although catastrophizing repeatedly has been associ-ated with heightened pain due to injuries, there is good reasonto query if childbirth might represent a unique case where catas-trophizing does not predict adverse pain outcomes.

Up to now, only one study has investigated catastrophizing spe-cifically in relation to labor pain (Van den Bussche et al., 2007).Catastrophizing was positively associated with fear of being over-whelmed by pain and tendencies to avoid pain during delivery.However, the relation between catastrophizing and labor painintensity was not studied, nor whether the tendencies to avoidpain persisted after childbirth.

This study was designed to assess whether the reported pain ofchildbirth and functioning postnatally differs between women whocatastrophize about labor pain, catastrophizers, and non-catastro-phizers. We hypothesized that catastrophizers will anticipate andexperience more pain during childbirth than non-catastrophizers.

1090-3801/$34.00 � 2008 European Federation of Chapters of the International Association for the Study of Pain. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.ejpain.2008.04.010

* Corresponding author. Tel.: +46 19 303 740.E-mail address: [email protected] (I.K. Flink).

European Journal of Pain 13 (2009) 312–316

Contents lists available at ScienceDirect

European Journal of Pain

journal homepage: www.EuropeanJournalPain.com

STUDY I

Pain in childbirth and postpartum recovery – The role of catastrophizing

Ida K. Flink a,*, Magdalena Z. Mroczek b, Michael J.L. Sullivan c, Steven J. Linton a

aCenter for Health and Medical Psychology, Department of Behavioural, Social and Legal Sciences – Psychology, Örebro University, SwedenbOccupational and Environmental Health, Stockholm Centre for Public Health, Stockholm County Council, SwedencDepartments of Psychology, Medicine and Neuroscience, McGill University, Montreal, Canada


Article history:Received 29 November 2007Received in revised form 26 March 2008Accepted 17 April 2008Available online 2 June 2008

Keywords:CatastrophizingPainPCSChildbirthPhysical recovery

a b s t r a c t

This prospective study investigated how pain catastrophizing was related to labor pain intensity andphysical recovery after childbirth. Eighty-eight women giving birth for the first time completed the firstquestionnaire before delivery. Eighty-two of those returned the second questionnaire after delivery. Par-ticipants were classified as catastrophizers (n = 38) or non-catastrophizers (n = 44) based on their scoreson the Pain Catastrophizing Scale. Comparison of the groups showed that catastrophizers anticipated andexperienced more intense pain (p < .0125) and had poorer physical recovery (p < .0125), measured as thelevel of self-reported functioning in activities of daily living, than non-catastrophizers. These resultsextend the association between catastrophizing and pain, to pain and recovery in childbirth and providesupport for the fear-avoidance model. It is concluded that pain catastrophizing plays a role in the expe-rience of pain in childbirth and postpartum recovery. Further research is needed to identify appropriateinterventions for catastrophizing women during the latter part of pregnancy.� 2008 European Federation of Chapters of the International Association for the Study of Pain. Published

by Elsevier Ltd. All rights reserved.

1. Introduction

Pain during childbirth is one of the most excruciating painexperiences that women encounter in their lives (Melzack, 1993).Although there is great variability, about 60% of the women whogive birth for the first time describe their pain as severe or extre-mely severe (Melzack, 1993). Consequently, anticipation of theremarkably intense pain of childbirth may be associated with psy-chological reactions, such as pain catastrophizing.

Catastrophizing emerges in the literature as one of the most po-tent predictors of pain in general (Vlaeyen and Linton, 2000; Keefeet al., 2004; Sullivan et al., 2002; Severeijns et al., 2001). It has beenassociated with heightened pain across several pain populations,and accounts for 7–31% of the variance in pain ratings (Sullivanet al., 1995, 2001). Catastrophizing has also been associated witha number of pain-related outcomes, including reduced involve-ment in daily activities (Keefe et al., 1989) and activity intolerance(Sullivan et al., 2002). Moreover, in the fear-avoidance model,catastrophizing is described as a cognitive precursor to pain-re-lated fear and tendencies to avoid pain, which may enhance painintensity and hinder resumption of physical activity (Vlaeyen andLinton, 2000). The role of catastrophizing has mainly been studiedin chronic and pathological pain conditions and its influences onchildbirth remains unclear.

Melzack (1993) claimed that childbirth provides an excellentmodel of acute pain with a clear beginning, progress and ending.However, labor pain is different from pathological pain, both inthe physiological (Catheline et al., 2006) and the psychological pro-cess (Olofsson, 2003). Usually pain is a signal of a potential threat,which should be diminished or eliminated. In contrast, labor paindoes not usually mean that something is wrong but is an importantsignal to the mother that labor is proceeding. Furthermore, child-birth differs from other types of pain with respect to its positiveoutgrowths; a child is born. Most other pain conditions (e.g. injury,illness) that have been studied do not have these positive associa-tions. Thus, although catastrophizing repeatedly has been associ-ated with heightened pain due to injuries, there is good reasonto query if childbirth might represent a unique case where catas-trophizing does not predict adverse pain outcomes.

Up to now, only one study has investigated catastrophizing spe-cifically in relation to labor pain (Van den Bussche et al., 2007).Catastrophizing was positively associated with fear of being over-whelmed by pain and tendencies to avoid pain during delivery.However, the relation between catastrophizing and labor painintensity was not studied, nor whether the tendencies to avoidpain persisted after childbirth.

This study was designed to assess whether the reported pain ofchildbirth and functioning postnatally differs between women whocatastrophize about labor pain, catastrophizers, and non-catastro-phizers. We hypothesized that catastrophizers will anticipate andexperience more pain during childbirth than non-catastrophizers.

1090-3801/$34.00 � 2008 European Federation of Chapters of the International Association for the Study of Pain. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.ejpain.2008.04.010

* Corresponding author. Tel.: +46 19 303 740.E-mail address: [email protected] (I.K. Flink).





We also hypothesized that recovery, in terms of level of function-ing in activities of daily living, would take longer for catastrophiz-ers than for non-catastrophizers.

2. Method

2.1. Design

In this prospective study, the data collection was made on twooccasions; 34–41 weeks of pregnancy and 2–4 weeks followingbirth, as shown in Fig. 1.

2.2. Participants

Pregnant women were recruited through maternal health ser-vices in the county of Örebro in Sweden. To participate, the womenhad to be carrying their first child and be at least at 34 weeks ofgestation. Women who planned to give birth by caesarean sectionwere excluded from the study but those who had an acute caesar-ean section in advanced labor were included. Participants wereconsecutively recruited by their midwife and informed consentswere obtained. Eighty-eight women answered the first question-naire and eighty-two of those returned the second questionnaire(response rate 93%). Only participants answering both question-naires (n = 82) were included in the analyses. The mean age ofthe sample was 29, 6; range 20–42.

The study was conducted according to current ethical principlesfor clinical research stated in the Declaration of Helsinki (WorldMedication Association, 1997) and was approved by the maternalhealth care board in the county of Örebro.

2.3. Measures

Participants answered two questionnaires; one before and oneafter delivery. Items from several standardized self-report invento-ries were included in the questionnaires. Background variables in-cluded age, estimated and actual date of delivery, planned andactual use of analgesics. The reliability and internal consistencyof the scales and items was assessed by calculating Cronbach’s al-pha coefficient. Questionnaires were endorsed by midwives inmaternity health care and tested in a pilot study (N = 2). The wo-men in the pilot study provided comments on how to improvethe questionnaires and were not included in the data analysis.

Pain catastrophizing: Pain catastrophizing was measured withthe Pain Catastrophizing Scale (PCS) (Sullivan et al., 1995), de-signed to assess various dimensions of catastrophizing about pain.The PCS consists of 13 statements describing various thoughts andfeelings that people may experience when they are in pain (e.g. ‘‘Ikeep thinking how badly I want the pain to stop”, ‘‘There is nothingI can do to stop the intensity of the pain”). Respondents were askedto rank each statement on a five-point scale, with respect to the de-gree to which they have these thoughts and feelings when they are

in pain (0 = not at all; 4 = all the time). In our study, the womenwere asked to focus specifically on the thoughts they had about la-bor pain. The PCS has been used in similar ways to measure catas-trophizing about genital pain (Pukall et al., 2002). In the presentstudy the PCS had good internal consistency (a = 0.925). Catastro-phizing about labor pain was measured before delivery.

Pain: The Present Pain Intensity scale, a part of McGill PainQuestionnaire (MPQ) (Melzack, 1975), was used to assess antici-pated and experienced labor pain. It is a six-point numerical scaleused for rating pain intensity (0 = none, 1 = mild, 2 = discomforting,3 = distressing, 4 = horrible, 5 = excruciating). The MPQ is the mostcommonly used measure of labor pain (Stockman and Altmaier,2001).

Pain ratings were measured on two occasions; before (antici-pated pain) and after delivery (experienced pain). On each occa-sion, the women were asked to rate both the average and themaximum pain during labor. Labor pain has been measured in sim-ilar ways in earlier research (Lang et al., 2005). The average score ofthe two items (average/maximum) was calculated for both occa-sions (anticipated/experienced), respectively. The internal consis-tency of the two items was good (a = 0.791 for anticipated painand a = 0.825 for experienced pain).

Physical recovery in ADL: To measure physical recovery afterdelivery we assessed the level of functioning in Activities of DailyLiving (ADL). Most standardized instruments for measuring ADLhave been developed for chronic conditions, such as stroke orchronic backpain (Buer and Linton, 2002). In our study the levelof functioning in ADL was defined as the degree to which the wo-man estimated being able to do activities within the following,three and seven days after delivery, compared to their level beforethe pregnancy: Household chores (e.g. cleaning, cooking), personalcare (e.g. hygiene, getting dressed) and physical activities (e.g. tak-ing walks, climbing the stairs, light exercising). Responses were gi-ven on an index between 1 and 6 (1 = a lot worse than beforepregnancy, 6 = as good as, or better than, before pregnancy).

To obtain one score for physical recovery the average score forthe three questions was calculated for the two occasions respec-tively. Cronbach’s alpha coefficients indicated good internal consis-tency and high reliability for the items three days (a = 0.881) andseven days (a = 0.894) after delivery.

Use of analgesics: To explore the planned and actual use of anal-gesics during labor, two items were employed: ‘‘Do you plan to useanalgesics?” (before delivery) and ‘‘Did you use analgesics duringdelivery?” (after delivery). For both items the answers ‘‘yes” or‘‘no” were given. Participants were also asked to write down thetype of analgesics planned and actually used.

2.4. Procedure

Midwifes in maternal health services provided the participantswith information about the study, a consent form, the first ques-tionnaire and a pre-paid envelope to return the questionnaire.The follow up questionnaire was mailed one week after the date

Fig. 1. Overview of the design of the study. M1 = first measurement, 1–8 weeks before expected delivery; M2 = second measurement, 2–4 weeks after expected delivery;R = reminder, if needed.

I.K. Flink et al. / European Journal of Pain 13 (2009) 312–316 313

STUDY I

We also hypothesized that recovery, in terms of level of function-ing in activities of daily living, would take longer for catastrophiz-ers than for non-catastrophizers.

2. Method

2.1. Design

In this prospective study, the data collection was made on twooccasions; 34–41 weeks of pregnancy and 2–4 weeks followingbirth, as shown in Fig. 1.

2.2. Participants

Pregnant women were recruited through maternal health ser-vices in the county of Örebro in Sweden. To participate, the womenhad to be carrying their first child and be at least at 34 weeks ofgestation. Women who planned to give birth by caesarean sectionwere excluded from the study but those who had an acute caesar-ean section in advanced labor were included. Participants wereconsecutively recruited by their midwife and informed consentswere obtained. Eighty-eight women answered the first question-naire and eighty-two of those returned the second questionnaire(response rate 93%). Only participants answering both question-naires (n = 82) were included in the analyses. The mean age ofthe sample was 29, 6; range 20–42.

The study was conducted according to current ethical principlesfor clinical research stated in the Declaration of Helsinki (WorldMedication Association, 1997) and was approved by the maternalhealth care board in the county of Örebro.

2.3. Measures

Participants answered two questionnaires; one before and oneafter delivery. Items from several standardized self-report invento-ries were included in the questionnaires. Background variables in-cluded age, estimated and actual date of delivery, planned andactual use of analgesics. The reliability and internal consistencyof the scales and items was assessed by calculating Cronbach’s al-pha coefficient. Questionnaires were endorsed by midwives inmaternity health care and tested in a pilot study (N = 2). The wo-men in the pilot study provided comments on how to improvethe questionnaires and were not included in the data analysis.

Pain catastrophizing: Pain catastrophizing was measured withthe Pain Catastrophizing Scale (PCS) (Sullivan et al., 1995), de-signed to assess various dimensions of catastrophizing about pain.The PCS consists of 13 statements describing various thoughts andfeelings that people may experience when they are in pain (e.g. ‘‘Ikeep thinking how badly I want the pain to stop”, ‘‘There is nothingI can do to stop the intensity of the pain”). Respondents were askedto rank each statement on a five-point scale, with respect to the de-gree to which they have these thoughts and feelings when they are

in pain (0 = not at all; 4 = all the time). In our study, the womenwere asked to focus specifically on the thoughts they had about la-bor pain. The PCS has been used in similar ways to measure catas-trophizing about genital pain (Pukall et al., 2002). In the presentstudy the PCS had good internal consistency (a = 0.925). Catastro-phizing about labor pain was measured before delivery.

Pain: The Present Pain Intensity scale, a part of McGill PainQuestionnaire (MPQ) (Melzack, 1975), was used to assess antici-pated and experienced labor pain. It is a six-point numerical scaleused for rating pain intensity (0 = none, 1 = mild, 2 = discomforting,3 = distressing, 4 = horrible, 5 = excruciating). The MPQ is the mostcommonly used measure of labor pain (Stockman and Altmaier,2001).

Pain ratings were measured on two occasions; before (antici-pated pain) and after delivery (experienced pain). On each occa-sion, the women were asked to rate both the average and themaximum pain during labor. Labor pain has been measured in sim-ilar ways in earlier research (Lang et al., 2005). The average score ofthe two items (average/maximum) was calculated for both occa-sions (anticipated/experienced), respectively. The internal consis-tency of the two items was good (a = 0.791 for anticipated painand a = 0.825 for experienced pain).

Physical recovery in ADL: To measure physical recovery afterdelivery we assessed the level of functioning in Activities of DailyLiving (ADL). Most standardized instruments for measuring ADLhave been developed for chronic conditions, such as stroke orchronic backpain (Buer and Linton, 2002). In our study the levelof functioning in ADL was defined as the degree to which the wo-man estimated being able to do activities within the following,three and seven days after delivery, compared to their level beforethe pregnancy: Household chores (e.g. cleaning, cooking), personalcare (e.g. hygiene, getting dressed) and physical activities (e.g. tak-ing walks, climbing the stairs, light exercising). Responses were gi-ven on an index between 1 and 6 (1 = a lot worse than beforepregnancy, 6 = as good as, or better than, before pregnancy).

To obtain one score for physical recovery the average score forthe three questions was calculated for the two occasions respec-tively. Cronbach’s alpha coefficients indicated good internal consis-tency and high reliability for the items three days (a = 0.881) andseven days (a = 0.894) after delivery.

Use of analgesics: To explore the planned and actual use of anal-gesics during labor, two items were employed: ‘‘Do you plan to useanalgesics?” (before delivery) and ‘‘Did you use analgesics duringdelivery?” (after delivery). For both items the answers ‘‘yes” or‘‘no” were given. Participants were also asked to write down thetype of analgesics planned and actually used.

2.4. Procedure

Midwifes in maternal health services provided the participantswith information about the study, a consent form, the first ques-tionnaire and a pre-paid envelope to return the questionnaire.The follow up questionnaire was mailed one week after the date

Fig. 1. Overview of the design of the study. M1 = first measurement, 1–8 weeks before expected delivery; M2 = second measurement, 2–4 weeks after expected delivery;R = reminder, if needed.


of estimated delivery (see Fig. 1). Women completed the first ques-tionnaire some time between the 34th and 41st week of their preg-nancy, and the second 2–4 weeks after delivery. If no response wasobtained within 2 weeks, a reminder phone-call was made.

Based on their PCS scores, a median-split divided the partici-pants into two groups; catastrophizers and non-catastrophizers.PCS has been used similarly in previous studies (Sullivan et al.,1997; Sullivan and Neish, 1999).

2.5. Data analysis

To evaluate the data, descriptive statistics were calculated anddistributions checked. To explore the differences between groups,comparisons of the means were made with independent t-testsfor the variables anticipated labor pain, experienced labor painand physical recovery at three and seven days after delivery. Sincefour t-tests were employed, a Bonferoni correction for multipletests was made, requiring p < .0125 for statistical significance.

Statistical analyses were performed with SPSS 14.0. Missing val-ues were estimated by calculating the mean score for the scaleused.

3. Results

3.1. Pain catastrophizing

Table 1 displays mean and median scores for the PCS. The med-ian for the whole group (20) was used to classify the participants ascatastrophizers or non-catastrophizers.

Of the drop-outs (i.e. the six women who did not return the sec-ond questionnaire), five scored above the median on the PCS(catastrophizers), and one scored below the median (non-catastrophizer).

3.2. Group characteristics

As shown in Table 2, there were no marked differences betweencatastrophizing and non-catastrophizing women regarding age,number of women giving birth by acute caesarean sections oruse of analgesics during the delivery.

3.3. Pain

Table 3 displays anticipated and experienced pain intensity forpain catastrophizers and non-catastrophizers (0 = none, 5 = excru-ciating). As can be seen, catastrophizing women rated their pain assignificantly higher than non-catastrophizers on both occasions.The magnitude of the differences is large in anticipated pain (Co-hens d = 1.17) and moderate in experienced pain (Cohens d = .74).

3.4. Physical recovery after delivery

Fig. 2 displays recovery in terms of level of functioning in activ-ities of daily living (ADL) after childbirth for pain catastrophizers

and non-catastrophizers (1 = a lot worse than before the preg-nancy, 6 = as good as, or better than, before the pregnancy).

As illustrated in Fig. 2, non-catastrophizers rated their ADLfunction significantly higher than catastrophizers at three days[M = 2.53, SD = 1.13; M = 3.59, SD = 1.38; t(80) = 3.76, p = .00] andat seven days [M = 3.58, SD = 1.21; M = 4.71, SD = 1.07;t(80) = 4.49, p = .00] after delivery. The magnitude of the differ-ences in recovery is large after three days (Cohens d = .83) as wellas after seven days (Cohens d = .99).

4. Discussion

This study was designed to investigate how pain catastrophiz-ing is associated with labor pain and physical recovery after child-birth. We found that women who catastrophized about labor painanticipated and experienced more pain than non-catastrophizers,supporting our first hypothesis. The second hypothesis was also

Table 1Scores on the Pain Catastrophizing Scale (PCS)

PCS – total PC NPC

N 82 38 44Median 20 25.50 13M (SD) 19.57 (9.46) 27.92 (5.57) 12.36 (5.22)Min 2 21 2Max 40 40 20

Note. Maximum score on the PCS is 52. PC = pain catastrophizers; NPC = non-paincatastrophizers.M = mean; SD = standard deviation; Min = minimum score; Max = maximum score.

Table 2Background variables and use of analgesics

PC(n = 38)

NPC(n = 44)

Age, M (SD) 29.6 29.7Acute caesarean sections, number 3 5Duration of labor 11.9 14.6Planned to use analgesics, yes/no/no answer 32/1/5 40/2/2Actual use of analgesics, yes/no 36/2 43/1Use of epidural analgesia, number 23 26Use of natural/light analgesia (e.g. laughing gas,

acupuncture), number11 14

Use of opiate analgesic drugs (e.g. morphine), number 2 3

M = mean; SD = standard deviation; PC = pain catastrophizers; NPC = non-paincatastrophizers.

Table 3Comparison of pain intensity in catastrophizing and non-catastrophizing women.N = 82 (scale 0–5)

PC (n = 38) M (SD) NPC (n = 44) M (SD) T (df)

Anticipated pain 3.79 (0.59) 3.06 (0.65) �5.32* (80)Experienced pain 3.68 (0.73) 3.09 (0.88) �3.33* (80)

M = mean; SD = standard deviation; df = degrees of freedom; PC = pain catastro-phizers; NPC = non-pain catastrophizers.Note: The same basic results were obtained using the non-parametric MannWhitney U-test.* p < .0125.

1

2

3

4

5

6

3 days 7 days

Lev

el o

f fu

ncti

onin

g in

ac

tiviti

es o

fda

ily li

ving

(A

DL

)

Non PainCatastrophizers

PainCatastrophizers

Fig. 2. Recovery in level of functioning in activities of daily living (ADL) incatastrophizing and non-catastrophizing women. N = 82 (scale 1–6). Note. Thisrelation remained the same even after controlling for pain experienced during theweek after delivery in a hierarchical multiple regression analysis.

314 I.K. Flink et al. / European Journal of Pain 13 (2009) 312–316

supported; non-catastrophizers rated their recovery after child-birth better than catastrophizers.

These findings replicate and expand findings from earlier stud-ies, mostly carried out in populations with chronic and pathologi-cal pain, where catastrophizing has been associated withheightened pain (Sullivan et al., 2001, 1995) and reduced involve-ment in daily activities (Keefe et al., 1989). In this study, we foundthe same associations for pain associated with childbirth. Thus, de-spite the positive facets of childbirth (e.g. a new life), the relationbetween catastrophizing and pain outcomes was comparable withother pain situations.

Furthermore, our study provided an opportunity to observecatastrophizing women over time, as catastrophizing was mea-sured when the women were pregnant and outcomes were mea-sured a number of weeks later, after the child was born. In thisway, the present study contributes to the growing amount of pro-spective research that has emerged recently within the area (Pica-vet et al., 2002; Severeijns et al., 2005).

The results may be explained by the fear-avoidance model ofpain. According to this model, pain experience is influenced by arange of emotional, cognitive, biological and behavioural factors(Vlaeyen and Linton, 2000). The central idea is that some individu-als tend to interpret pain stimuli as threatening, which generatescatastrophic thoughts, tension, vigilance, fear of pain and attemptsto avoid the pain. The model was elaborated to describe the devel-opment of chronic pain but there is also some evidence that somepain-free individuals may have a tendency to harbour fear-avoid-ance beliefs, which become activated when they approach pain(Buer and Linton, 2002). Applied to the present study, women withthis tendency may interpret labor pain as menacing, which evokescatastrophizing thoughts. They may then focus their attention onthe pain, leading to an overestimation of its intensity (i.e. theyanticipate more pain) and an underestimation of their own copingabilities. In the model, this vigilance is associated with increasedmuscle tension which may also increase pain intensity, and induceattempts to avoid the pain. Catastrophizing has indeed been posi-tively associated with tendencies to avoid pain during delivery(Van den Bussche et al., 2007). In our study, the women alsoseemed to show a tendency to avoid certain movements afterchildbirth, resulting in a slower resumption of daily activities; i.e.walking, cooking and personal hygiene. Thus, our results indicatethat the fear-avoidance model may also be relevant to the painof childbirth.

Our findings may be viewed in relation to the increasing num-ber of women requiring caesarean sections because of fear of child-birth (Wax et al., 2004). About 6–11% of the pregnant women inthe industrialized countries suffer from this fear (Saisto and Hal-mesmäki, 2003; Waldenström et al., 2006). Fear of pain duringdelivery is one of the most common reasons for fear of childbirth(Eriksson et al., 2006; Sjögren, 1997; Geissbuehler and Eberhard,2002). Also, fear of childbirth has been found to be related to a low-er pain threshold (Saisto et al., 2001). Our findings provide a possi-ble explanation for these relationships. Although fear was notmeasured in the present study, it is conceivable that catastrophiz-ing women also have fear of pain. However, while catastrophizingand fear of pain are overlapping constructs, catastrophizing hasbeen found to provide a unique predictor for pain, beyond fearper se (Sullivan et al., 2004). Therefore, catastrophizing may actas a mediator, explaining fear of childbirth and a lower painthreshold among these women.

Both groups of women in this study managed to predict the le-vel of pain they later experienced during delivery. This contradictsearlier findings, where women who were giving birth for the firsttime underestimated the actual pain experienced during childbirth(Fridh and Gaston-Johansson, 1990; Paech, 1991; Waldenström,1999). On the other hand, catastrophizers have been found to exag-

gerate the menace of pain, while understating their own capacityto handle the situation (Sullivan et al., 2001). However, in the pres-ent study both catastrophizers and non-catastrophizers had a fairlyrealistic picture of how intense their pain would actually be.

Catastrophizing may be an important indicator for identifyingmothers who may have difficulties with labor and recovery afterdelivery. This coincides with earlier studies, where catastrophizinghas been found to predict heightened pain intensity and activityintolerance in pain-free individuals (Sullivan et al., 2002). How-ever, even though catastrophizing preceded childbirth in our study,this does not necessarily infer a causal relationship between catas-trophizing and outcomes.

The limitations of this study need to be considered. First,although the response rate was high in comparison with otherstudies about related topics (Van den Bussche et al., 2007), a sub-stantial proportion of women giving birth in the county during thisperiod were unable to participate because the questionnaires wereonly accessible in Swedish. This leads to a possible selection biaswhich might have influenced our results, as immigrant womenmake up about 10% of the women living in the county. Second,all measures used were subjective self-report inventories. A moreobjective measurement for rating physical recovery would havebeen appropriate, since the identified differences may be due toreporting differences rather than physical, i.e. the catastrophizersonly rate their level of functioning as lower than the non-catastro-phizers. One factor that may have influenced the rating is changesin cognition and mood that many women experience during latepregnancy and shortly after childbirth (Russell et al., 2001). Third,because there are large individual differences in labor pain, otherfactors may be influencing our results. One major determinant isprior births (Melzack, 1993). To minimize the impact of earlierexperiences, possibly affecting both catastrophizing and pain, onlywomen giving birth for the first time participated in the study. An-other influencing factor is use of methods to relieve the pain. How-ever, there were no marked differences between catastrophizingand non-catastrophizing women in use of analgesics. This repli-cates earlier findings where no difference was found betweencatastrophizers and non-catastrophizers in use of epidural analge-sia (Van den Bussche et al., 2007), one of the most widespread andeffective methods for relief of labor pain (Capogna et al., 1996).

Other medical factors (e.g. use of oxytocin/syntocinon) areimportant in relation to childbirth and postpartum recovery andmay have influenced our results. However, since the main focusof this study was women’s experiences, we focused on their self-reports on standardized questionnaires. Fourth, among the womenwho did not answer the second questionnaire, five out of six werecatastrophizers, and we do not know how this might have influ-enced the results. Fifth, the distinction between statistical and clin-ical significance should be addressed. The differences identified inthis study have moderate to large effect sizes. Further studies arerequired to understand the impact of the findings in a clinicalsetting.

In this first study, pain catastrophizing was clearly associatedwith a heightened labor pain experience and a lower level ofresumption of activities after childbirth. These findings suggestpossibilities for identifying pregnant women with high levels ofcatastrophizing during the third trimester, for instance by usingscreening questionnaires. If further research confirms the role ofcatastrophizing in this group the next challenge will be to developadequate interventions. In other populations, cognitive-behav-ioural therapy (CBT) is a successful method of treating catastro-phizing and encourages better adjustment to pain (Keefe et al.,1990; Sullivan et al., 2001). One potential path is to investigate ifCBT-techniques (e.g. imaginary exposure, cognitive restructuring)could be adapted as an early intervention for women who catastro-phize about their labor pain. Hopefully, appropriate interventions


STUDY I

supported; non-catastrophizers rated their recovery after child-birth better than catastrophizers.

These findings replicate and expand findings from earlier stud-ies, mostly carried out in populations with chronic and pathologi-cal pain, where catastrophizing has been associated withheightened pain (Sullivan et al., 2001, 1995) and reduced involve-ment in daily activities (Keefe et al., 1989). In this study, we foundthe same associations for pain associated with childbirth. Thus, de-spite the positive facets of childbirth (e.g. a new life), the relationbetween catastrophizing and pain outcomes was comparable withother pain situations.

Furthermore, our study provided an opportunity to observecatastrophizing women over time, as catastrophizing was mea-sured when the women were pregnant and outcomes were mea-sured a number of weeks later, after the child was born. In thisway, the present study contributes to the growing amount of pro-spective research that has emerged recently within the area (Pica-vet et al., 2002; Severeijns et al., 2005).

The results may be explained by the fear-avoidance model ofpain. According to this model, pain experience is influenced by arange of emotional, cognitive, biological and behavioural factors(Vlaeyen and Linton, 2000). The central idea is that some individu-als tend to interpret pain stimuli as threatening, which generatescatastrophic thoughts, tension, vigilance, fear of pain and attemptsto avoid the pain. The model was elaborated to describe the devel-opment of chronic pain but there is also some evidence that somepain-free individuals may have a tendency to harbour fear-avoid-ance beliefs, which become activated when they approach pain(Buer and Linton, 2002). Applied to the present study, women withthis tendency may interpret labor pain as menacing, which evokescatastrophizing thoughts. They may then focus their attention onthe pain, leading to an overestimation of its intensity (i.e. theyanticipate more pain) and an underestimation of their own copingabilities. In the model, this vigilance is associated with increasedmuscle tension which may also increase pain intensity, and induceattempts to avoid the pain. Catastrophizing has indeed been posi-tively associated with tendencies to avoid pain during delivery(Van den Bussche et al., 2007). In our study, the women alsoseemed to show a tendency to avoid certain movements afterchildbirth, resulting in a slower resumption of daily activities; i.e.walking, cooking and personal hygiene. Thus, our results indicatethat the fear-avoidance model may also be relevant to the painof childbirth.

Our findings may be viewed in relation to the increasing num-ber of women requiring caesarean sections because of fear of child-birth (Wax et al., 2004). About 6–11% of the pregnant women inthe industrialized countries suffer from this fear (Saisto and Hal-mesmäki, 2003; Waldenström et al., 2006). Fear of pain duringdelivery is one of the most common reasons for fear of childbirth(Eriksson et al., 2006; Sjögren, 1997; Geissbuehler and Eberhard,2002). Also, fear of childbirth has been found to be related to a low-er pain threshold (Saisto et al., 2001). Our findings provide a possi-ble explanation for these relationships. Although fear was notmeasured in the present study, it is conceivable that catastrophiz-ing women also have fear of pain. However, while catastrophizingand fear of pain are overlapping constructs, catastrophizing hasbeen found to provide a unique predictor for pain, beyond fearper se (Sullivan et al., 2004). Therefore, catastrophizing may actas a mediator, explaining fear of childbirth and a lower painthreshold among these women.

Both groups of women in this study managed to predict the le-vel of pain they later experienced during delivery. This contradictsearlier findings, where women who were giving birth for the firsttime underestimated the actual pain experienced during childbirth(Fridh and Gaston-Johansson, 1990; Paech, 1991; Waldenström,1999). On the other hand, catastrophizers have been found to exag-

gerate the menace of pain, while understating their own capacityto handle the situation (Sullivan et al., 2001). However, in the pres-ent study both catastrophizers and non-catastrophizers had a fairlyrealistic picture of how intense their pain would actually be.

Catastrophizing may be an important indicator for identifyingmothers who may have difficulties with labor and recovery afterdelivery. This coincides with earlier studies, where catastrophizinghas been found to predict heightened pain intensity and activityintolerance in pain-free individuals (Sullivan et al., 2002). How-ever, even though catastrophizing preceded childbirth in our study,this does not necessarily infer a causal relationship between catas-trophizing and outcomes.

The limitations of this study need to be considered. First,although the response rate was high in comparison with otherstudies about related topics (Van den Bussche et al., 2007), a sub-stantial proportion of women giving birth in the county during thisperiod were unable to participate because the questionnaires wereonly accessible in Swedish. This leads to a possible selection biaswhich might have influenced our results, as immigrant womenmake up about 10% of the women living in the county. Second,all measures used were subjective self-report inventories. A moreobjective measurement for rating physical recovery would havebeen appropriate, since the identified differences may be due toreporting differences rather than physical, i.e. the catastrophizersonly rate their level of functioning as lower than the non-catastro-phizers. One factor that may have influenced the rating is changesin cognition and mood that many women experience during latepregnancy and shortly after childbirth (Russell et al., 2001). Third,because there are large individual differences in labor pain, otherfactors may be influencing our results. One major determinant isprior births (Melzack, 1993). To minimize the impact of earlierexperiences, possibly affecting both catastrophizing and pain, onlywomen giving birth for the first time participated in the study. An-other influencing factor is use of methods to relieve the pain. How-ever, there were no marked differences between catastrophizingand non-catastrophizing women in use of analgesics. This repli-cates earlier findings where no difference was found betweencatastrophizers and non-catastrophizers in use of epidural analge-sia (Van den Bussche et al., 2007), one of the most widespread andeffective methods for relief of labor pain (Capogna et al., 1996).

Other medical factors (e.g. use of oxytocin/syntocinon) areimportant in relation to childbirth and postpartum recovery andmay have influenced our results. However, since the main focusof this study was women’s experiences, we focused on their self-reports on standardized questionnaires. Fourth, among the womenwho did not answer the second questionnaire, five out of six werecatastrophizers, and we do not know how this might have influ-enced the results. Fifth, the distinction between statistical and clin-ical significance should be addressed. The differences identified inthis study have moderate to large effect sizes. Further studies arerequired to understand the impact of the findings in a clinicalsetting.

In this first study, pain catastrophizing was clearly associatedwith a heightened labor pain experience and a lower level ofresumption of activities after childbirth. These findings suggestpossibilities for identifying pregnant women with high levels ofcatastrophizing during the third trimester, for instance by usingscreening questionnaires. If further research confirms the role ofcatastrophizing in this group the next challenge will be to developadequate interventions. In other populations, cognitive-behav-ioural therapy (CBT) is a successful method of treating catastro-phizing and encourages better adjustment to pain (Keefe et al.,1990; Sullivan et al., 2001). One potential path is to investigate ifCBT-techniques (e.g. imaginary exposure, cognitive restructuring)could be adapted as an early intervention for women who catastro-phize about their labor pain. Hopefully, appropriate interventions


developed in collaboration with midwives may reduce women’spain perception during labor and hasten their recoverypostpartum.

Acknowledgement

The authors want to thank CarinaWestergren and the midwivesin Örebro läns landsting for their help with the recruitment of theparticipants.

References

Buer N, Linton SJ. Fear-avoidance beliefs and catastrophizing: occurrence and riskfactor in back pain and ADL in the general population. Pain 2002;99:485–91.

Capogna G, Alahuhta S, Celleno D, De Vlieger H, Moreira J, Morgan B, et al. Maternalexpectations and experiences of labour pain and analgesia: a multicentre studyof nulliparous women. Int J Obstet Anasth 1996;5:229–35.

Catheline G, Touquet B, Besson J-M, Lombard M-C. Parturation in the rat: aphysiological pain model. Anesthesiology 2006;104:1257–65.

Eriksson C, Westman G, Hamberg K. Content of childbirth-related fear in Swedishwomen and men – analysis of an open-ended question. J Midwifery Wom Heal2006;51:112–8.

Fridh G, Gaston-Johansson F. Do primiparas and multiparas have realisticexpectations of labor. Acta Obstet Gyn Scan 1990;69:103–9.

Geissbuehler V, Eberhard J. Fear of childbirth during pregnancy: a study of morethan 8000 pregnant women. J Psychosom Obst Gyn 2002;23:229–35.

Keefe FJ, Brown GK, Wallston KA, Caldwell DS. Coping with rheumatoid arthritis:catastrophizing as a maladaptive strategy. Pain 1989;37:51–6.

Keefe FJ, Caldwell DS, Williams DA, Gil KM, Mitchell D, Robertson C, et al. Paincoping skills training in the management of osteoarthritic knee pain: acomparative study. Behav Ther 1990;21:49–62.

Keefe FJ, Rumble ME, Scipio CD, Giordano LA, Perri LM. Psychological aspects ofpersistent pain: current state of the science. J Pain 2004;5:195–211.

Lang AJ, Sorrell JT, Rodgers CS, Lebeck MM. Anxiety sensitivity as a predictor of laborpain. Eur J Pain 2005;10:263–70.

Melzack R. Labour pain as a model of acute pain. Pain 1993;53:117–20.Melzack R. The McGill pain questionnaire: major properties and scoring methods.

Pain 1975;1:277–99.Olofsson N. Förlossningssmärta – och dess behandling [Labor pain – and the

treatment]. Lund: Studentlitteratur; 2003.Paech MJ. The King Edward Memorial Hospital 1000 mother survey of methods of

pain relief in labour. Anaesth Intensitve Care 1991;19:393–9.Picavet HS, Vlaeyen JWS, Schouten JS. Pain catastrophizing and kinesiophobia:

predictors of chronic low back pain. Am J Epidemiol 2002;156:1028–34.

Pukall CF, Binik YM, Khalifé S, Amsel R, Abbott FV. Vestibular tactile and painthresholds in women with vulvar vestibulitis syndrome. Pain 2002;96:163–75.

Russell JA, Douglas AJ, Ingram CD. Brain preparations for maternity – adaptivechanges in behavioral and neuroendocrine systems during pregnancy andlactation. An overview. Prog Brain Res 2001;133:1–38.

Saisto T, Halmesmäki E. Fear of childbirth: a neglected dilemma. Acta Obst Gyn Scan2003;82:201–8.

Saisto T, Kaaja R, Ylikorkala O, Halesmäki E. Reduced pain tolerance during and afterpregnancy in women suffering from fear of labor. Pain 2001;93:123–7.

Severeijns R, Vlaeyen JWS, van den Hout MA, Picavet SJ. Pain catastrophizing andconsequences of musculoskeletal pain: a prospective study in the Dutchcommunity. J Pain 2005;6:125–32.

Severeijns R, Vlaeyen JWS, van den Hout MA, Weber WEJ. Pain catastrophizingpredicts pain intensity, disability, and psychological distress independent of thelevel of physical impairment. Clin J Pain 2001;17:165–72.

Sjögren B. Reasons for anxiety about childbirth in 100 pregnant women. JPsychosom Obst Gyn 1997;18:266–72.

Stockman AF, Altmaier EM. Relation of self-efficacy to reported pain and painmedication usage during labor. J Clin Psychol Med S 2001;8:161–6.

Sullivan MJL, Bishop SR, Pivik J. The Pain Catastrophizing Scale: development andvalidation. Psychol Assessment 1995;7:524–32.

Sullivan MJL, Neish N. The effects of disclosure on pain during dental hygienetreatment: the moderating role of catastrophizing. Pain 1999;79:155–63.

Sullivan MJL, Rodgers WM, Wilson PM, Bell GJ, Murray TC, Fraser SN. Anexperimental investigation of the relation between catastrophizing andactivity intolerance. Pain 2002;100:47–53.

Sullivan MJL, Rouse D, Bishop S, Johnston S. Thought suppression, catastrophizingand pain. Cognitive Ther Res 1997;21:555–68.

Sullivan MJL, Thorn B, Haythornthwaite JA, Keefe FJ, Martin M, Bradley LA, et al.Theoretical perspectives on the relation between catastrophizing and pain. ClinJ Pain 2001;17:52–64.

Sullivan MJL, Thorn B, Rodgers WM, Ward C. Path model of psychologicalantecendents to pain experience. Clin J Pain 2004;20:164–73.

Van den Bussche E, Crombez G, Eccleston C, Sullivan MJL. Why women preferepidural analgesia during childbirth: the role of beliefs about epidural analgesiaand pain catastrophizing. Eur J Pain 2007;11:275–82.

Vlaeyen JWS, Linton SJ. Fear-avoidance and its consequences in chronicmusculoskeletal pain: a state of the art. Pain 2000;85:317–32.

Waldenström U. Experience of labor and birth in 1111 women. J Psychosom Res1999;47:471–82.

Waldenström U, Hildingsson I, Ryding EL. Antenatal fear of childbirth and itsassociation with subsequent caesarean section and experience of childbirth.BJOG – Int J Obstet Gy 2006;113:638–46.

Wax JR, Cartin A, Pinette MG, Blackstone J. Patient choice cesarian: an evidence-based review. Obstet Gynecol Surv 2004;59:601–16.

World Medication Association. Declaration of Helsinki. JAMA – J Am Med Assoc1997;277:925–6.


STUDY II

STUDY IICatastrophizing moderates the effect of exposure in vivo for back pain patients

with pain-related fear

Ida K. Flink *, Katja Boersma, Steven J. LintonCenter for Health and Medical Psychology (CHAMP), School of Law, Psychology and Social Work, 701 82 Örebro, Sweden


Article history:Received 3 July 2009Received in revised form 20 January 2010Accepted 2 February 2010Available online 9 March 2010

Keywords:Exposure in vivoBack painFear-avoidanceRandomized-controlled trialCatastrophizing

a b s t r a c t

This investigation was an initial attempt to explore psychological factors that might help or hinder theeffect of exposure in vivo for patients with musculoskeletal pain and pain-related fear. The study wasbased on data from a randomized-controlled trial for patients with non-specific spinal pain (Lintonet al., 2008).First, catastrophizing, anxiety, and depression were studied as possible treatment moderators. We

found evidence that catastrophizing was a moderator of treatment outcome in exposure. When furtherexploring the nature of the relationship between catastrophizing and outcome, the results showed thatthe exposure was effective only for patients with low or moderate levels of catastrophizing. High catastr-ophizers did not improve from the treatment. On the other hand, anxiety was a general predictor of pooroutcome, and not a specific moderator of outcome in exposure. In contrast, depression was not signifi-cantly related to outcome.Next, patients were divided into high change participants and low change participants based on their

improvement in disability after treatment in order to investigate the change in psychological variablesduring treatment. Descriptive data indicated that high change participants had large improvementsacross treatment on depression, anxiety, catastrophizing, and fear-avoidance beliefs whereas low changeparticipants virtually did not change at all on these variables across treatment.These findings denote that catastrophizing is a moderator of treatment outcome in exposure whereas

several psychological variables might be important for the treatment process.� 2010 European Federation of International Association for the Study of Pain Chapters. Published by

Elsevier Ltd. All rights reserved.

1. Introduction

Exposure in vivo is a novel treatment approach for patients withmusculoskeletal pain (MSP) and high levels of pain-related fear.Fear and subsequent avoidance of activities are key factors thatmay perpetuate pain and dysfunction in a subgroup of patients,as proposed in the fear-avoidance model (Vlaeyen and Linton,2000). Several studies have indeed shown that exposure is effectivefor reducing fear, catastrophic thoughts, disability and pain inthese patients (Vlaeyen et al., 2001; Boersma et al., 2004; de Jonget al., 2005; Leeuw et al., 2008; Linton et al., 2008; Woods andAsmundson, 2008). A recent review concluded that exposure isthe treatment of choice for patients with pain-related fear (Lohn-berg, 2007). This conclusion was based on the first studies of expo-sure which showed promising effects in replicated single-caseexperimental studies (Vlaeyen et al., 2001, 2002a, 2002b; Lintonet al., 2002; Boersma et al., 2004; de Jong et al., 2005). However,one of these studies noted substantial individual differences in

how well the patients responded to the treatment (Boersmaet al., 2004).

Recently, four RCTs further examined the effect of exposurein vivo for patients with MSP, and the effects were not as impres-sive as those reported in earlier studies (George et al., 2008; Leeuwet al., 2008; Linton et al., 2008; Woods and Asmundson, 2008).Exposure resulted in moderate to large effect sizes for reducingpain-related fear, catastrophizing, disability, and pain. But whencompared to other commonly used treatments, the effects weremore modest (George et al., 2008; Leeuw et al., 2008; Woods andAsmundson, 2008). While the unclear results may partly due to alack of power due to small samples, a more compelling explana-tion, which indeed has been noticed (Linton et al., 2008), is thatsome patients respond better to exposure than others. Conse-quently, the effect of exposure might be suppressed by moderatingfactors (Baron and Kenny, 1986; Vlaeyen and Morley, 2005).

One hypothesis has been that patients with high levels of pain-related fear would benefit most from exposure, but data havefailed to support this (George et al., 2008; Leeuw et al., 2008). Incognitive behavioral treatment packages for MSP, where exposureis not included, high catastrophizing, anxiety, and depression have

1090-3801/$36.00 � 2010 European Federation of International Association for the Study of Pain Chapters. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.ejpain.2010.02.003

* Corresponding author. Tel.: +46 19 303 740; fax: +46 19 303 484.E-mail address: [email protected] (I.K. Flink).





instead been linked to poor outcome (McCracken and Turk, 2002;Turner et al., 2007). However, how these factors relate to outcomein exposure has not yet been studied.

This secondary analysis of an RCT on the effectiveness of expo-sure treatment has two aims. The main aim is to study possiblemoderators of outcome in exposure in vivo. Catastrophizing, anxi-ety, and depression will be explored as possible moderators of thetreatment effect. Pain-related fear will not be included, as onlyfearful patients participated. The second aim, which has a moreexploratory character, is to study changes in psychological vari-ables during treatment for patients who improve in disability aftertreatment as compared to those who do not improve, to get anindication of what helps or hinders treatment success.

2. Method

2.1. Overview of the design

This study is based on data from a randomized-controlled trialfor patients with non-specific spinal pain (Linton et al., 2008).The purpose was to study the effects of exposure in vivo for painpatients with pain-related fear and disability as an addition to theirusual medical treatment. Fig. 1 presents an overview of the designof the study. After randomization, all participants filled out pretest1. The exposure group then received exposure treatment and thewaiting-list continued usual treatment. After a median of14 weeks, the waiting-list participants completed pretest 2 andsubsequently received exposure. All participants completed theposttest immediately after treatment and the follow-up threemonths later. The study was approved by the Örebro UniversityHospital’s Board of Research Ethics.

2.2. Participants

Participants were recruited through local primary care facilitiesand via advertisements in local newspapers and National InsuranceAuthority offices. The inclusion criteria were: (1) 18–60 years old,(2) disabling spinal pain, (3) current or previous (within the last3 months) sick leave, (4) substantial levels of pain-related fear(TSK > 35), and (5) no red flags, e.g. fractures or infections(Nachemson and Jonsson, 2000). Two hundred and twenty threepeople fulfilled the initial criteria and were mailed a baseline ques-tionnaire. Forty six returned the questionnaire within the limit of3 weeks, fulfilled the inclusion criteria, and were randomized toexposure (N = 21) or to the waiting list (N = 25). Four of the wait-ing-list participants did not complete the second pretest (pretest2). In addition, 16 participants did not complete the posttest (expo-sure group, N = 8; waiting list, N = 8), and three did not answer thefollow-up. Analyses showed that non-completers and completersdid not significantly differ on background variables (age, gender,

nationality, and work status), pain characteristics (intensity, dura-tion, and location), disability, and psychological variables (pain-re-lated fear and catastrophizing). For a description of reasons fornon-completion, see the original study (Linton et al., 2008).

2.3. Treatment

The treatment was based on the exposure protocol developedby Vlaeyen and colleagues (Vlaeyen et al., 2004). Four clinical psy-chologists, with support from a physiotherapist, administered thetreatment which consisted of 13–15 individual sessions includingassessment, psychoeducation, and graded exposure trainingin vivo. Goals were developed, consisting of activities that usuallyare carried out in daily life, e.g. playing with the children or carry-ing bags from the store. During the exposure, the patients gradu-ally confronted feared movements, e.g. lifting, bending, andtwisting, in order to decrease fear and to reach the patients’ goals.Between the sessions, the patients completed homework thatincorporated the movements into activities at home or at work.For a detailed description of the treatment, see the original study(Linton et al., 2008).

2.4. Measures

Swedish versions of all questionnaires were used.

2.4.1. Background variables and pain characteristicsQuestions from the Örebro Musculoskeletal Pain Screening

Questionnaire (ÖMPSQ) (Linton and Boersma, 2003) were used toassess background variables (age, gender, and sick leave) and paincharacteristics (intensity, location, and duration).

2.4.2. DisabilityThe Quebec Back Pain Disability Scale (QBPDS) (Kopec et al.,

1995) was used to assess level of functioning in daily activities(e.g. walking, get dressed). In the QBPDS, the responders are askedhow difficult it is to perform 20 different activities with their cur-rent pain (e.g. climb stairs, walk 300–400 m, carry two bags offood). The answers are given on a six-point scale (0 = not difficultat all; 5 = impossible to do). The QBPDS have shown sufficientvalidity and reliability (Kopec et al., 1995).

2.4.3. Anxiety and depressionThe Hospital Anxiety and Depression Scale (HADS) (Zigmond

and Snaith, 1983) was used to assess anxiety and depression. TheHADS consists of 14 statements reflecting common symptoms ofanxiety (e.g. ‘‘I feel tense or ‘wound up’”) and depression (e.g. ‘‘Ihave lost interest in my appearance”). The responders rate to whatdegree they agree with the statements on a four-point scale(0 = not at all; 3 = very much indeed). The HADS has sound

Exposuregroup

Pretest 1 Exposure Posttest Follow-up

Waiting-listgroup

Pretest 1 Pretest 2 Exposure Posttest Follow-up

Assessment 1 Assessment 2 Assessment 3 Follow-upassessment

Fig. 1. Overview of the design of the study.


psychometric properties, including the Swedish version (Herr-mann, 1997; Lisspers et al., 1997; Bjelland et al., 2002).

2.4.4. Pain catastrophizingThe pain catastrophizing scale (PCS) (Sullivan et al., 1995) was

used to assess pain-related catastrophic thoughts. The PCS reflects13 thoughts and feelings that may arise when people have pain(e.g. ‘‘I keep thinking about how much it hurts”, ‘‘There is nothingI can do to reduce the intensity of the pain”). The answers are givenon a five-point scale, where the respondents rate to what extentthey have these thoughts and feelings when experiencing pain(0 = not at all; 4 = all the time). The PCS has been shown to havegood psychometric properties (Sullivan et al., 1995).

2.4.5. Fear-avoidance beliefsThe Tampa Scale of Kinesiophobia (TSK) (Kori et al., 1990) was

used to assess fear-avoidance beliefs. The TSK consists of 17 state-ments (e.g. ‘‘Simply being careful that I do not make anythingworse is the safest thing I can do to prevent my pain from worsen-ing”). The respondents are asked to what extent they agree withthe statement on a four-point scale (1 = strongly disagree;4 = strongly agree). The questionnaire has shown good validityand reliability, including the Swedish version (Vlaeyen et al.,1995; Goubert et al., 2004; Lundberg et al., 2004; Roelofs et al.,2004).

2.5. Statistical analyses

In order to address the question of possible treatment modera-tors, data from both the exposure group and the waiting-list groupwere used in the analyses (exposure group, N = 13, pretest 1 andposttest; waiting-list group, N = 21, pretests 1 and 2). Three multi-ple regression analyses were conducted for catastrophizing, anxi-ety, and depression, respectively, according to recommendationsfor investigation of moderational effects (Baron and Kenny, 1986;Holmbeck, 2002). To test whether pretreatment levels of thepsychological variables moderated the effect of treatment ondisability, an interaction term was created for each psychologicalvariable (group � psychological variable). The interaction termwas added to the main variables, and if it turned out to besignificant while the main variables were included in the model,this would indicate moderation. Subsequently, the nature of thesignificant interactions was explored with simple regressions, forparticipants with high and low levels of the psychological variableseparately.

To reduce multicollinearity between the independent variableand the interaction variable, the variables were centered prior tothe analyses.

In order to tackle the second aim, i.e. to explore changes duringtreatment for patients who improved on disability as compared tothose who did not, only data from the participants that completedthe exposure treatment were used (exposure group, N = 13, pretest1 and posttest; waiting-list group, N = 13, pretest 2 and posttest).In these analyses, all participants were treated as one group andno distinction was made between the exposure group and thewaiting list. Data was missing for one participant on the posttestratings of disability. Consequently, this participant was not in-cluded in the analyses. The participants were divided into twogroups (high change participants, HCP and low change participants,LCP), based on a median-split on the percentage disability changefrom pretest to posttest. The median change was an improvementof 18%. Those who scored above the median (disability decreasedmore than 18%) were classified as HCP, and those who scored be-low the median (disability decreased less than 18%) were classifiedas LCP. To study changes in psychological variables during treat-ment for HCP and LCP, descriptive statistics were calculated for

posttest and follow-up on disability, depression, anxiety, catastro-phizing, and fear-avoidance beliefs, and percentage change in thesevariables from pretest to posttest. Because of the small number ofparticipants, it was not tested whether the differences were statis-tically significant.

Statistical analyses were performed with SPSS 16.0. Missing val-ues were estimated by calculating the mean score for the scale.

3. Results


As shown in Table 1, there were no marked differences betweenparticipants in the exposure group and the waiting-list group onbackground variables. The groups were similar regarding age,nationality and gender. They were also similar on pain characteris-tics. There was a significant difference between the groups on sicklisting, where the exposure group had more days of sick listingthan the waiting-list group (v2 = 4.2; df = 1; p = .04).

3.2. Treatment moderators

Multiple regression analyses were performed to investigatewhether improvement on disability was moderated by pretreat-ment levels of catastrophizing, anxiety, and depression respec-tively. As can be seen in Table 2, of the interaction terms thatwere entered, the interaction between catastrophizing and groupwas the only one that turned out to be significant over and abovethe main effects. This implies that the effect of exposure was signif-icantly different depending on the patients’ levels of catastrophiz-ing, whereas in the waiting-list group catastrophizing was not apredictor of outcome. In other words, catastrophizing was a mod-erator of treatment outcome in the exposure group. In the analysisthat investigated anxiety as a possible moderator, only the main ef-fects were significant, meaning that in this sample anxiety was ageneral predictor of poor outcome for both the exposure groupand the waiting-list group. In the analysis that investigated depres-sion as a possible moderator, neither main effects nor the interac-tion effect was significant, suggesting that depression was notsignificantly related to outcome.

To further explore the nature of the significant interaction, theparticipants were divided into high and low catastrophizers based

Table 1Background variables for participants in the exposure group and the waiting-listgroup.

Exposure group(n = 13)

Waiting-list group(n = 21)

Age, M (SD) 46 (9.9) 49 (7.3)Pain intensity, M (SD) 6.5 (1.1) 6.3 (1.5)Gender, women/men (%) 62/38 48/52

Pain location (more than one location possible)Neck (%) 77 91Shoulder (%) 62 76Upper back (%) 54 62Lower back (%) 77 76

Pain duration<3 months (Acute) (%) 0 03–12 months (Subacute)

(%)17 24

>12 months (Chronic) (%) 83 76

Sick listing last 12 monthsa

1–180 days (%) 31 67180–365 days (%) 69 33

a Note: There was a significant difference between the exposure group and thewaiting-list group on sick listing.


STUDY II

instead been linked to poor outcome (McCracken and Turk, 2002;Turner et al., 2007). However, how these factors relate to outcomein exposure has not yet been studied.

This secondary analysis of an RCT on the effectiveness of expo-sure treatment has two aims. The main aim is to study possiblemoderators of outcome in exposure in vivo. Catastrophizing, anxi-ety, and depression will be explored as possible moderators of thetreatment effect. Pain-related fear will not be included, as onlyfearful patients participated. The second aim, which has a moreexploratory character, is to study changes in psychological vari-ables during treatment for patients who improve in disability aftertreatment as compared to those who do not improve, to get anindication of what helps or hinders treatment success.

2. Method

2.1. Overview of the design

This study is based on data from a randomized-controlled trialfor patients with non-specific spinal pain (Linton et al., 2008).The purpose was to study the effects of exposure in vivo for painpatients with pain-related fear and disability as an addition to theirusual medical treatment. Fig. 1 presents an overview of the designof the study. After randomization, all participants filled out pretest1. The exposure group then received exposure treatment and thewaiting-list continued usual treatment. After a median of14 weeks, the waiting-list participants completed pretest 2 andsubsequently received exposure. All participants completed theposttest immediately after treatment and the follow-up threemonths later. The study was approved by the Örebro UniversityHospital’s Board of Research Ethics.

2.2. Participants

Participants were recruited through local primary care facilitiesand via advertisements in local newspapers and National InsuranceAuthority offices. The inclusion criteria were: (1) 18–60 years old,(2) disabling spinal pain, (3) current or previous (within the last3 months) sick leave, (4) substantial levels of pain-related fear(TSK > 35), and (5) no red flags, e.g. fractures or infections(Nachemson and Jonsson, 2000). Two hundred and twenty threepeople fulfilled the initial criteria and were mailed a baseline ques-tionnaire. Forty six returned the questionnaire within the limit of3 weeks, fulfilled the inclusion criteria, and were randomized toexposure (N = 21) or to the waiting list (N = 25). Four of the wait-ing-list participants did not complete the second pretest (pretest2). In addition, 16 participants did not complete the posttest (expo-sure group, N = 8; waiting list, N = 8), and three did not answer thefollow-up. Analyses showed that non-completers and completersdid not significantly differ on background variables (age, gender,

nationality, and work status), pain characteristics (intensity, dura-tion, and location), disability, and psychological variables (pain-re-lated fear and catastrophizing). For a description of reasons fornon-completion, see the original study (Linton et al., 2008).

2.3. Treatment

The treatment was based on the exposure protocol developedby Vlaeyen and colleagues (Vlaeyen et al., 2004). Four clinical psy-chologists, with support from a physiotherapist, administered thetreatment which consisted of 13–15 individual sessions includingassessment, psychoeducation, and graded exposure trainingin vivo. Goals were developed, consisting of activities that usuallyare carried out in daily life, e.g. playing with the children or carry-ing bags from the store. During the exposure, the patients gradu-ally confronted feared movements, e.g. lifting, bending, andtwisting, in order to decrease fear and to reach the patients’ goals.Between the sessions, the patients completed homework thatincorporated the movements into activities at home or at work.For a detailed description of the treatment, see the original study(Linton et al., 2008).

2.4. Measures

Swedish versions of all questionnaires were used.

2.4.1. Background variables and pain characteristicsQuestions from the Örebro Musculoskeletal Pain Screening

Questionnaire (ÖMPSQ) (Linton and Boersma, 2003) were used toassess background variables (age, gender, and sick leave) and paincharacteristics (intensity, location, and duration).

2.4.2. DisabilityThe Quebec Back Pain Disability Scale (QBPDS) (Kopec et al.,

1995) was used to assess level of functioning in daily activities(e.g. walking, get dressed). In the QBPDS, the responders are askedhow difficult it is to perform 20 different activities with their cur-rent pain (e.g. climb stairs, walk 300–400 m, carry two bags offood). The answers are given on a six-point scale (0 = not difficultat all; 5 = impossible to do). The QBPDS have shown sufficientvalidity and reliability (Kopec et al., 1995).

2.4.3. Anxiety and depressionThe Hospital Anxiety and Depression Scale (HADS) (Zigmond

and Snaith, 1983) was used to assess anxiety and depression. TheHADS consists of 14 statements reflecting common symptoms ofanxiety (e.g. ‘‘I feel tense or ‘wound up’”) and depression (e.g. ‘‘Ihave lost interest in my appearance”). The responders rate to whatdegree they agree with the statements on a four-point scale(0 = not at all; 3 = very much indeed). The HADS has sound

Exposuregroup

Pretest 1 Exposure Posttest Follow-up

Waiting-listgroup

Pretest 1 Pretest 2 Exposure Posttest Follow-up

Assessment 1 Assessment 2 Assessment 3 Follow-upassessment

Fig. 1. Overview of the design of the study.


psychometric properties, including the Swedish version (Herr-mann, 1997; Lisspers et al., 1997; Bjelland et al., 2002).

2.4.4. Pain catastrophizingThe pain catastrophizing scale (PCS) (Sullivan et al., 1995) was

used to assess pain-related catastrophic thoughts. The PCS reflects13 thoughts and feelings that may arise when people have pain(e.g. ‘‘I keep thinking about how much it hurts”, ‘‘There is nothingI can do to reduce the intensity of the pain”). The answers are givenon a five-point scale, where the respondents rate to what extentthey have these thoughts and feelings when experiencing pain(0 = not at all; 4 = all the time). The PCS has been shown to havegood psychometric properties (Sullivan et al., 1995).

2.4.5. Fear-avoidance beliefsThe Tampa Scale of Kinesiophobia (TSK) (Kori et al., 1990) was

used to assess fear-avoidance beliefs. The TSK consists of 17 state-ments (e.g. ‘‘Simply being careful that I do not make anythingworse is the safest thing I can do to prevent my pain from worsen-ing”). The respondents are asked to what extent they agree withthe statement on a four-point scale (1 = strongly disagree;4 = strongly agree). The questionnaire has shown good validityand reliability, including the Swedish version (Vlaeyen et al.,1995; Goubert et al., 2004; Lundberg et al., 2004; Roelofs et al.,2004).

2.5. Statistical analyses

In order to address the question of possible treatment modera-tors, data from both the exposure group and the waiting-list groupwere used in the analyses (exposure group, N = 13, pretest 1 andposttest; waiting-list group, N = 21, pretests 1 and 2). Three multi-ple regression analyses were conducted for catastrophizing, anxi-ety, and depression, respectively, according to recommendationsfor investigation of moderational effects (Baron and Kenny, 1986;Holmbeck, 2002). To test whether pretreatment levels of thepsychological variables moderated the effect of treatment ondisability, an interaction term was created for each psychologicalvariable (group � psychological variable). The interaction termwas added to the main variables, and if it turned out to besignificant while the main variables were included in the model,this would indicate moderation. Subsequently, the nature of thesignificant interactions was explored with simple regressions, forparticipants with high and low levels of the psychological variableseparately.

To reduce multicollinearity between the independent variableand the interaction variable, the variables were centered prior tothe analyses.

In order to tackle the second aim, i.e. to explore changes duringtreatment for patients who improved on disability as compared tothose who did not, only data from the participants that completedthe exposure treatment were used (exposure group, N = 13, pretest1 and posttest; waiting-list group, N = 13, pretest 2 and posttest).In these analyses, all participants were treated as one group andno distinction was made between the exposure group and thewaiting list. Data was missing for one participant on the posttestratings of disability. Consequently, this participant was not in-cluded in the analyses. The participants were divided into twogroups (high change participants, HCP and low change participants,LCP), based on a median-split on the percentage disability changefrom pretest to posttest. The median change was an improvementof 18%. Those who scored above the median (disability decreasedmore than 18%) were classified as HCP, and those who scored be-low the median (disability decreased less than 18%) were classifiedas LCP. To study changes in psychological variables during treat-ment for HCP and LCP, descriptive statistics were calculated for

posttest and follow-up on disability, depression, anxiety, catastro-phizing, and fear-avoidance beliefs, and percentage change in thesevariables from pretest to posttest. Because of the small number ofparticipants, it was not tested whether the differences were statis-tically significant.

Statistical analyses were performed with SPSS 16.0. Missing val-ues were estimated by calculating the mean score for the scale.

3. Results


As shown in Table 1, there were no marked differences betweenparticipants in the exposure group and the waiting-list group onbackground variables. The groups were similar regarding age,nationality and gender. They were also similar on pain characteris-tics. There was a significant difference between the groups on sicklisting, where the exposure group had more days of sick listingthan the waiting-list group (v2 = 4.2; df = 1; p = .04).

3.2. Treatment moderators

Multiple regression analyses were performed to investigatewhether improvement on disability was moderated by pretreat-ment levels of catastrophizing, anxiety, and depression respec-tively. As can be seen in Table 2, of the interaction terms thatwere entered, the interaction between catastrophizing and groupwas the only one that turned out to be significant over and abovethe main effects. This implies that the effect of exposure was signif-icantly different depending on the patients’ levels of catastrophiz-ing, whereas in the waiting-list group catastrophizing was not apredictor of outcome. In other words, catastrophizing was a mod-erator of treatment outcome in the exposure group. In the analysisthat investigated anxiety as a possible moderator, only the main ef-fects were significant, meaning that in this sample anxiety was ageneral predictor of poor outcome for both the exposure groupand the waiting-list group. In the analysis that investigated depres-sion as a possible moderator, neither main effects nor the interac-tion effect was significant, suggesting that depression was notsignificantly related to outcome.

To further explore the nature of the significant interaction, theparticipants were divided into high and low catastrophizers based

Table 1Background variables for participants in the exposure group and the waiting-listgroup.

Exposure group(n = 13)

Waiting-list group(n = 21)

Age, M (SD) 46 (9.9) 49 (7.3)Pain intensity, M (SD) 6.5 (1.1) 6.3 (1.5)Gender, women/men (%) 62/38 48/52

Pain location (more than one location possible)Neck (%) 77 91Shoulder (%) 62 76Upper back (%) 54 62Lower back (%) 77 76

Pain duration<3 months (Acute) (%) 0 03–12 months (Subacute)

(%)17 24

>12 months (Chronic) (%) 83 76

Sick listing last 12 monthsa

1–180 days (%) 31 67180–365 days (%) 69 33

a Note: There was a significant difference between the exposure group and thewaiting-list group on sick listing.


on a median-split (at score 26) on the Pain Catastrophizing Scale.Simple regression analyses for these two groups separately showedthat for low catastrophizers (n = 17) was there a significant effect

of group on treatment outcome (ß = .488; p = .047). For highcatastrophizers (n = 17) there was no significant effect of groupon treatment outcome (ß = �.136; p = .616). These results implythat the exposure was more effective than waiting in reducing dis-ability for patients who were low on catastrophizing. For patientswho were high on catastrophizing there was no difference in out-come between those who received exposure and those who werein the waiting-list group. Fig. 2 displays the results visually.

3.3. Description of changes in disability and psychological variables

To describe changes during treatment for patients who im-proved and those who did not, all patients that received exposurewere included in the analyses and no distinction was made be-tween when they received the exposure (i.e. exposure group orafter the waiting period for the waiting-list group). The patientswere divided into low change participants (LCP) and high changeparticipants (HCP) based on a median-split on reduction in disabil-ity. Table 3 displays scorings on disability and psychological vari-ables at pretest, posttest, and follow-up for LCP and HCP, as wellas percentage change during treatment. On the disability measure,which was used to create the groups (LCP/HCP), LCP scored almostthe same at pretest and posttest (�2%), whereas HCP decreasedtheir scorings markedly (53%). From posttest to follow-up, the dis-ability scorings remained stable for both groups. As can be seen inTable 3, LCP scored somewhat higher on depression at posttestthan at pretest (+6%), whereas HCP decreased their scorings(�38%). LCP decreased their scorings of anxiety from pretest toposttest (�14%), but at follow-up their ratings had increased andwere almost back at the pretest level. Also HCP scored lower onanxiety at posttest than at pretest (�32%), and their ratings wereeven lower on the follow-up. LCP decreased their ratings of

Table 2Multiple regression analyses explaining the treatment effect in reduction on disabilitydepending on pretreatment levels of catastrophizing, anxiety, and depression.

Variables Beta (standardized) p Value R2

Block 1 Group .270 NsCatastrophizing �.083 Ns

Block 2 Group .187 NsCatastrophizing .604 .048a .27Group � catastrophizing �.809 .009b

Block 1 Group .430 .014a

Anxiety .498 .005b .28Block 2 Group .435 .014a

Anxiety .643 .008b

Group � anxiety �.204 Ns

Block 1 Group .307 NsDepression .163 Ns

Block 2 Group .307 NsDepression .246 NsGroup � depression �.127 Ns

* p < .05.** p < .01.

Low catastrophizers only

30

35

40

45

50

55

60


Exposure group

Waiting list group


30

35

40

45

50

55

60


Exposure group

Waiting list group

Fig. 2. Ratings of disability in the exposure group and the waiting-list group for lowcatastrophizers and high catastrophizers respectively. Note. These graphs are basedon ANOVA’s.

Table 3Ratings of disability and psychological variables in low change participants and highchange participants at pretest, posttest and follow-up, and percentage change frompretest to posttest.

Variable Measure Timepoint Low changeparticipants(n = 12) M (SD)

High changeparticipants(n = 13) M (SD)

Disabilitya QBPDS Pretest 49.8 (14.8) 46.6 (11.6)Posttest 48.8 (13.0) 22.1 (11.7)Follow-up 48.1 (16.1) 22.9 (16.9)Changefrompretest toposttest

�2% �53%

Depression HADS Pretest 6.8 (4.4) 5.5 (4.2)Posttest 7.2 (4.2) 3.4 (3.4)Follow-up 8.1 (4.8) 3.8 (3.1)Changefrompretest toposttest

+6% �38%

Anxiety HADS Pretest 7.8 (3.1) 4.5 (2.8)Posttest 6.7 (3.0) 3.4 (2.2)Follow-up 7.4 (3.7) 2.5 (1.7)Changefrompretest toposttest

�14% �32%

Catastrophizing PCS Pretest 25.7 (6.4) 18.7 (10.1)Posttest 21.2 (8.9) 9.5 (6.5)Follow-up 21.5 (10.0) 11.2 (8.6)Changefrompretest toposttest

�18% �49%

Fear-avoidancebeliefs

TSK Pretest 40.8 (8.0) 38.6 (8.1)

Posttest 35.0 (5.0) 26.9 (5.7)Follow-up 35.5 (7.3) 28.9 (6.9)Changefrompretest toposttest

�14% �30%

a Note: The groups (LCP/HCP) were based on the posttest ratings on this measure.


catastrophizing from pretest to posttest (�18%), but the decreasewas more than twice as large for HCP (�49%). The same patternis seen on fear-avoidance beliefs, where LCP decreased their scor-ings from pretest to posttest (�14%), and HCP decreased twice asmuch (�30%). In sum, these data indicate that LCP and HCP scoredlower on anxiety, catastrophizing, and fear-avoidance beliefs aftertreatment, but the improvements were markedly higher in the HCPgroup. In depression, the scorings in the LCP group were slightlyworse after treatment, whereas HCP improved also on thisvariable.

4. Discussion

This investigation is an initial attempt to explore psychologicalfactors that might help or hinder the effect of exposure in vivo forpatients with MSP and pain-related fear.

We found evidence that catastrophizing was a moderator oftreatment outcome in exposure. Indeed, the effect of exposurewas dependent on the patients’ pretreatment levels of catastro-phizing, whereas in the waiting-list group catastrophizing wasnot a predictor of outcome. When further exploring the nature ofthe relation between catastrophizing and outcome, the resultsshowed that the exposure was effective only for patients withlow and moderate levels of catastrophizing. For high catastrophiz-ers, there was no difference in outcome for patients who receivedexposure and those who did not.

On the other hand, anxiety was a predictor of poor outcomeboth in the exposure group and in the waiting-list group suggest-ing that it is a general predictor of poor outcome, and not a specificmoderator of outcome in exposure. However, depression was notsignificantly related to outcome.

Patients were divided into high change participants and lowchange participants based on their improvement in disability aftertreatment in order to investigate the change in psychologicalvariables during treatment. Descriptive data were employed. Theresults indicated that high change participants had large improve-ments across treatment on depression, anxiety, catastrophizing,and fear-avoidance beliefs whereas low change participants virtu-ally did not change at all on these variables across treatment. Thisdenotes that several psychological variables might be importantfor the treatment process in exposure in vivo for pain-related fear.

The finding that catastrophizing moderated the treatment effectis in line with findings from earlier research. In other cognitivebehavioral therapies (CBT) for patients with MSP, high catastro-phizing has been a predictor of poor outcome (McCracken andTurk, 2002; Turner et al., 2007). Furthermore, reductions in catas-trophizing during CBT have been related to treatment success (Jen-sen et al., 2001; Smeets et al., 2006; Vowles et al., 2007). One studyof exposure in vivo found that reductions in catastrophizing med-iated treatment outcome (Leeuw et al., 2008), but this has not beenreplicated (George et al., 2008). The current study is the first one toidentify catastrophizing as a moderator of treatment outcome inexposure.

One hypothesis has been that patients with high levels of pain-related fear would benefit most from exposure, but the data havenot supported this hypothesis (George et al., 2008; Leeuw et al.,2008). Our findings rather point in the opposite direction, as thetreatment in the current study was helpful only for patients withlow or moderate levels of catastrophizing (<26). Even thoughcatastrophizing and pain-related fear have been considered to betheoretically different constructs (Sullivan et al., 2001), there ishigh overlap between them (Sullivan et al., 1995) and it is thuscontradictory to the earlier hypothesis that high catastrophizersdid not improve at all from the exposure. As only patients with adocumented high level of pain-related fear (TSK > 35) were in-

cluded, fear was not included in the analyses as a possible moder-ator. However, it is noteworthy that there were no salient pretestdifferences on the TSK between participants who improved andthose who did not improve.

Catastrophizing might not only be a moderator of the treatmenteffect in exposure but is possibly also a mediator, as data from anearlier study have suggested (Leeuw et al., 2008). Our data goes inline with this, as the descriptive data indicate that catastrophizingdecreased markedly in high change participants (�49% from pre-to posttest), whereas it did not decrease as much in the low changegroup (�18%). Clear reductions of catastrophizing during treat-ment might thus be crucial for improvement on disability. Onepossibility is that too high levels of catastrophizing trigger safetybehaviors, i.e. subtle strategies that patients use to ‘‘protect” them-selves, which hinders the patient to fully benefit from the exposure(Vlaeyen et al., 2004).

Anxiety was found to be a general predictor of poor outcome.Participants who scored high on anxiety tended to have a poor out-come, irregardless of receiving treatment or not. The descriptivedata showed that before the treatment started, low change partic-ipants scored on average above the cut-off (>7) for ‘‘possible casesof anxiety” (Herrmann, 1997). The initial average ratings ondepression were somewhat lower, but after the exposure it had in-creased to the level of ‘‘possible cases of depression” (HADS sub-scale depression >7) for the low change participants, and atfollow-up it had increased even more. Despite that initial scoringson depression were unrelated to outcome, this implies that depres-sion might be related to outcome anyway. However, the results donot illuminate whether low mood hinders treatment success, or ifit is a result of treatment failure. In light of the fear-avoidancemodel (Vlaeyen and Linton, 2000), which provides a theoretical ba-sis for exposure, one would predict that depression would increasewith failure since the model highlights the negative spiral effect.

One limitation in this study is the small sample which restrictsits statistical power. As all other RCTs on the effects of exposure formusculoskeletal pain (George et al., 2008; Leeuw et al., 2008; Lin-ton et al., 2008; Woods and Asmundson, 2008), this investigationhas a limited N. Consequently, the results should be interpretedwith caution. One result of the small N might be the surprisinglylarge difference between high catastrophizers in the exposuregroup and the waiting-list group which can be noted in Fig. 2. De-spite the potential problem with power, the moderation analysesrevealed significant effects. This indicates that the identified rela-tions indeed are important. The limited power might howeverwithhold effects that are not that obvious but nevertheless impor-tant or restrict generalization of the findings. More research is thusneeded to replicate the results.

Only descriptive data is provided to answer the second aim be-cause of the limited N, and the findings should therefore bedeemed preliminary. Nevertheless, the results provide importantindications to investigate in future research. Another potential lim-itation concerning the second aim is the division of participantsinto groups of high change and low change based on a median-spliton disability change. A division of groups based on a clinical rele-vant change on disability would presumably have resulted in morereliable groups. But as no normative data are available for the Que-bec scale, we found the current procedure defensible. This wassupported by the data as the procedure resulted in two cleargroups of participants; those who improved on the main outcomevariable ‘‘disability” during treatment and those who did not.

The results do not tell anything about the process of change, norabout the interaction between variables. The data indicated thatthere were several pretreatment differences between high changeparticipants and low change participants, and these differenceswere even larger after treatment. But this does not clarify whenduring treatment the differences increased, i.e. when did the high


STUDY II

on a median-split (at score 26) on the Pain Catastrophizing Scale.Simple regression analyses for these two groups separately showedthat for low catastrophizers (n = 17) was there a significant effect

of group on treatment outcome (ß = .488; p = .047). For highcatastrophizers (n = 17) there was no significant effect of groupon treatment outcome (ß = �.136; p = .616). These results implythat the exposure was more effective than waiting in reducing dis-ability for patients who were low on catastrophizing. For patientswho were high on catastrophizing there was no difference in out-come between those who received exposure and those who werein the waiting-list group. Fig. 2 displays the results visually.

3.3. Description of changes in disability and psychological variables

To describe changes during treatment for patients who im-proved and those who did not, all patients that received exposurewere included in the analyses and no distinction was made be-tween when they received the exposure (i.e. exposure group orafter the waiting period for the waiting-list group). The patientswere divided into low change participants (LCP) and high changeparticipants (HCP) based on a median-split on reduction in disabil-ity. Table 3 displays scorings on disability and psychological vari-ables at pretest, posttest, and follow-up for LCP and HCP, as wellas percentage change during treatment. On the disability measure,which was used to create the groups (LCP/HCP), LCP scored almostthe same at pretest and posttest (�2%), whereas HCP decreasedtheir scorings markedly (53%). From posttest to follow-up, the dis-ability scorings remained stable for both groups. As can be seen inTable 3, LCP scored somewhat higher on depression at posttestthan at pretest (+6%), whereas HCP decreased their scorings(�38%). LCP decreased their scorings of anxiety from pretest toposttest (�14%), but at follow-up their ratings had increased andwere almost back at the pretest level. Also HCP scored lower onanxiety at posttest than at pretest (�32%), and their ratings wereeven lower on the follow-up. LCP decreased their ratings of

Table 2Multiple regression analyses explaining the treatment effect in reduction on disabilitydepending on pretreatment levels of catastrophizing, anxiety, and depression.

Variables Beta (standardized) p Value R2

Block 1 Group .270 NsCatastrophizing �.083 Ns

Block 2 Group .187 NsCatastrophizing .604 .048a .27Group � catastrophizing �.809 .009b

Block 1 Group .430 .014a

Anxiety .498 .005b .28Block 2 Group .435 .014a

Anxiety .643 .008b

Group � anxiety �.204 Ns

Block 1 Group .307 NsDepression .163 Ns

Block 2 Group .307 NsDepression .246 NsGroup � depression �.127 Ns

* p < .05.** p < .01.

Low catastrophizers only

30

35

40

45

50

55

60


Exposure group

Waiting list group


30

35

40

45

50

55

60


Exposure group

Waiting list group

Fig. 2. Ratings of disability in the exposure group and the waiting-list group for lowcatastrophizers and high catastrophizers respectively. Note. These graphs are basedon ANOVA’s.

Table 3Ratings of disability and psychological variables in low change participants and highchange participants at pretest, posttest and follow-up, and percentage change frompretest to posttest.

Variable Measure Timepoint Low changeparticipants(n = 12) M (SD)

High changeparticipants(n = 13) M (SD)

Disabilitya QBPDS Pretest 49.8 (14.8) 46.6 (11.6)Posttest 48.8 (13.0) 22.1 (11.7)Follow-up 48.1 (16.1) 22.9 (16.9)Changefrompretest toposttest

�2% �53%

Depression HADS Pretest 6.8 (4.4) 5.5 (4.2)Posttest 7.2 (4.2) 3.4 (3.4)Follow-up 8.1 (4.8) 3.8 (3.1)Changefrompretest toposttest

+6% �38%

Anxiety HADS Pretest 7.8 (3.1) 4.5 (2.8)Posttest 6.7 (3.0) 3.4 (2.2)Follow-up 7.4 (3.7) 2.5 (1.7)Changefrompretest toposttest

�14% �32%

Catastrophizing PCS Pretest 25.7 (6.4) 18.7 (10.1)Posttest 21.2 (8.9) 9.5 (6.5)Follow-up 21.5 (10.0) 11.2 (8.6)Changefrompretest toposttest

�18% �49%

Fear-avoidancebeliefs

TSK Pretest 40.8 (8.0) 38.6 (8.1)

Posttest 35.0 (5.0) 26.9 (5.7)Follow-up 35.5 (7.3) 28.9 (6.9)Changefrompretest toposttest

�14% �30%

a Note: The groups (LCP/HCP) were based on the posttest ratings on this measure.


catastrophizing from pretest to posttest (�18%), but the decreasewas more than twice as large for HCP (�49%). The same patternis seen on fear-avoidance beliefs, where LCP decreased their scor-ings from pretest to posttest (�14%), and HCP decreased twice asmuch (�30%). In sum, these data indicate that LCP and HCP scoredlower on anxiety, catastrophizing, and fear-avoidance beliefs aftertreatment, but the improvements were markedly higher in the HCPgroup. In depression, the scorings in the LCP group were slightlyworse after treatment, whereas HCP improved also on thisvariable.

4. Discussion

This investigation is an initial attempt to explore psychologicalfactors that might help or hinder the effect of exposure in vivo forpatients with MSP and pain-related fear.

We found evidence that catastrophizing was a moderator oftreatment outcome in exposure. Indeed, the effect of exposurewas dependent on the patients’ pretreatment levels of catastro-phizing, whereas in the waiting-list group catastrophizing wasnot a predictor of outcome. When further exploring the nature ofthe relation between catastrophizing and outcome, the resultsshowed that the exposure was effective only for patients withlow and moderate levels of catastrophizing. For high catastrophiz-ers, there was no difference in outcome for patients who receivedexposure and those who did not.

On the other hand, anxiety was a predictor of poor outcomeboth in the exposure group and in the waiting-list group suggest-ing that it is a general predictor of poor outcome, and not a specificmoderator of outcome in exposure. However, depression was notsignificantly related to outcome.

Patients were divided into high change participants and lowchange participants based on their improvement in disability aftertreatment in order to investigate the change in psychologicalvariables during treatment. Descriptive data were employed. Theresults indicated that high change participants had large improve-ments across treatment on depression, anxiety, catastrophizing,and fear-avoidance beliefs whereas low change participants virtu-ally did not change at all on these variables across treatment. Thisdenotes that several psychological variables might be importantfor the treatment process in exposure in vivo for pain-related fear.

The finding that catastrophizing moderated the treatment effectis in line with findings from earlier research. In other cognitivebehavioral therapies (CBT) for patients with MSP, high catastro-phizing has been a predictor of poor outcome (McCracken andTurk, 2002; Turner et al., 2007). Furthermore, reductions in catas-trophizing during CBT have been related to treatment success (Jen-sen et al., 2001; Smeets et al., 2006; Vowles et al., 2007). One studyof exposure in vivo found that reductions in catastrophizing med-iated treatment outcome (Leeuw et al., 2008), but this has not beenreplicated (George et al., 2008). The current study is the first one toidentify catastrophizing as a moderator of treatment outcome inexposure.

One hypothesis has been that patients with high levels of pain-related fear would benefit most from exposure, but the data havenot supported this hypothesis (George et al., 2008; Leeuw et al.,2008). Our findings rather point in the opposite direction, as thetreatment in the current study was helpful only for patients withlow or moderate levels of catastrophizing (<26). Even thoughcatastrophizing and pain-related fear have been considered to betheoretically different constructs (Sullivan et al., 2001), there ishigh overlap between them (Sullivan et al., 1995) and it is thuscontradictory to the earlier hypothesis that high catastrophizersdid not improve at all from the exposure. As only patients with adocumented high level of pain-related fear (TSK > 35) were in-

cluded, fear was not included in the analyses as a possible moder-ator. However, it is noteworthy that there were no salient pretestdifferences on the TSK between participants who improved andthose who did not improve.

Catastrophizing might not only be a moderator of the treatmenteffect in exposure but is possibly also a mediator, as data from anearlier study have suggested (Leeuw et al., 2008). Our data goes inline with this, as the descriptive data indicate that catastrophizingdecreased markedly in high change participants (�49% from pre-to posttest), whereas it did not decrease as much in the low changegroup (�18%). Clear reductions of catastrophizing during treat-ment might thus be crucial for improvement on disability. Onepossibility is that too high levels of catastrophizing trigger safetybehaviors, i.e. subtle strategies that patients use to ‘‘protect” them-selves, which hinders the patient to fully benefit from the exposure(Vlaeyen et al., 2004).

Anxiety was found to be a general predictor of poor outcome.Participants who scored high on anxiety tended to have a poor out-come, irregardless of receiving treatment or not. The descriptivedata showed that before the treatment started, low change partic-ipants scored on average above the cut-off (>7) for ‘‘possible casesof anxiety” (Herrmann, 1997). The initial average ratings ondepression were somewhat lower, but after the exposure it had in-creased to the level of ‘‘possible cases of depression” (HADS sub-scale depression >7) for the low change participants, and atfollow-up it had increased even more. Despite that initial scoringson depression were unrelated to outcome, this implies that depres-sion might be related to outcome anyway. However, the results donot illuminate whether low mood hinders treatment success, or ifit is a result of treatment failure. In light of the fear-avoidancemodel (Vlaeyen and Linton, 2000), which provides a theoretical ba-sis for exposure, one would predict that depression would increasewith failure since the model highlights the negative spiral effect.

One limitation in this study is the small sample which restrictsits statistical power. As all other RCTs on the effects of exposure formusculoskeletal pain (George et al., 2008; Leeuw et al., 2008; Lin-ton et al., 2008; Woods and Asmundson, 2008), this investigationhas a limited N. Consequently, the results should be interpretedwith caution. One result of the small N might be the surprisinglylarge difference between high catastrophizers in the exposuregroup and the waiting-list group which can be noted in Fig. 2. De-spite the potential problem with power, the moderation analysesrevealed significant effects. This indicates that the identified rela-tions indeed are important. The limited power might howeverwithhold effects that are not that obvious but nevertheless impor-tant or restrict generalization of the findings. More research is thusneeded to replicate the results.

Only descriptive data is provided to answer the second aim be-cause of the limited N, and the findings should therefore bedeemed preliminary. Nevertheless, the results provide importantindications to investigate in future research. Another potential lim-itation concerning the second aim is the division of participantsinto groups of high change and low change based on a median-spliton disability change. A division of groups based on a clinical rele-vant change on disability would presumably have resulted in morereliable groups. But as no normative data are available for the Que-bec scale, we found the current procedure defensible. This wassupported by the data as the procedure resulted in two cleargroups of participants; those who improved on the main outcomevariable ‘‘disability” during treatment and those who did not.

The results do not tell anything about the process of change, norabout the interaction between variables. The data indicated thatthere were several pretreatment differences between high changeparticipants and low change participants, and these differenceswere even larger after treatment. But this does not clarify whenduring treatment the differences increased, i.e. when did the high


change group get better? Possibly one element changed first whichbrought a change in the others, but the changes may also have oc-curred at the same time. To reveal the process of change and theinteraction between variables, future studies would preferably in-clude weekly measures during treatment. This would assist both inextending the theoretical framework for exposure for pain-relatedfear, and in developing more effective treatments.

In conclusion, catastrophizing moderated the effect of exposurein vivo in this sample. High catastrophizing patients were nothelped by the treatment. In addition, patients who were high onanxiety tended to have a poor outcome in general, both in thetreatment group and in the waiting-list group. The descriptive dataindicated that depression also played an important role for out-come, even though it was not significantly related to outcome inthe moderation analyses. These findings may have clinical implica-tions. Assessment of patients with MSP should include these vari-ables. Presumably treatment success is helped or hindered bycatastrophizing and other reflections of emotional distress ratherthan levels of mere fear-avoidance beliefs, which implies a needto further develop exposure in vivo to benefit all MSP patients withpain-related fear.

References

Baron RM, Kenny DA. The moderator–mediator variable distinction in socialpsychological research: conceptual, strategic, and statistical considerations. JPers Soc Psych 1986;51:1173–82.

Bjelland I, Dahl AA, Haug TT, Neckelmann D. The validity of the hospital anxiety anddepression scale: an updated literature review. J Psychosom Res2002;52:69–77.

Boersma K, Linton SJ, Overmeer T, Jansson M, Vlaeyen JWS, de Jong JR. Loweringfear-avoidance and enhancing function through exposure in vivo: a multiplebaseline study across six patients with back pain. Pain 2004;108:8–16.

de Jong JR, Vlaeyen JWS, Onghena P, Cuypers C, den Hollander M, Ruijgrok J.Reduction of pain-related fear in complex regional pain syndrome type 1: theapplication of graded exposure in vivo. Pain 2005;116:264–75.

George SZ, Zeppieri G, Cere AL, Cere MR, Borut MS, Hodges MJ, et al. A randomizedcontrolled trial of behavioral physical therapy interventions for acute and sub-acute low back pain. Pain 2008;140:145–57.

Goubert L, Crombez G, Van Damme S, Vlaeyen JW, Bijttebier P, Roelofs J.Confirmatory factor analysis of the Tampa Scale for Kinesiophobia: invarianttwo factor model across low back pain patients and fibromyalgia patients. Clin JPain 2004;20.

Herrmann C. Internal experiences with the hospital anxiety and depression scale – areview of validation data and clinical results. J Psychosom Res 1997;42:17–41.

Holmbeck GN. Post-hoc probing of significant moderational and mediational effectsin studies of pediatric populations. J Ped Psych 2002;27:87–96.

Jensen MP, Turner JA, Romano JM. Changes in beliefs, catastrophizing, and copingare associated with improvement in multidisciplinary pain treatment. J ConsClin Psych 2001;69:655–62.

Kopec JA, Esdaile JM, Abrahamowicz M, Abenhaim L, Wood-Dauphinee W, LampingDL, et al. The Quebec back pain disability scale. Spine 1995;20:341–52.

Kori SH, Miller RP, Todd DD. Kinesiophobia: a new view of chronic pain behavior.Pain Manage 1990;3:35–43.

Leeuw M, Goossens MEJB, van Breukelen GJP, de Jong JR, Heuts PHTG, Smeets RJEM,et al. Exposure in vivo versus operant graded activity in chronic low back painpatients: results of a randomized controlled trial. Pain 2008;138:192–207.

Linton SJ, Boersma K. Early identification of patients at risk of developing apersistent back problem: the predictive validity of the Örebro musculoskeletalpain screening questionnaire. Clin J Pain 2003;19:80–6.

Linton SJ, Boersma K, Jansson M, Overmeer T, Lindblom K, Vlaeyen JWS. Arandomized controlled trial of exposure in vivo for patients with spinal painreporting fear of work-related activities. Eur J Pain 2008;12:722–30.

Linton SJ, Overmeer T, Jansson M, Vlaeyen JWS, de Jong JR. Graded in vivo exposuretreatment for fear-avoidant pain patients with functional disability: a casestudy. Cogn Behav Therapy 2002;31:49–58.

Lisspers J, Nygren A, Söderman E. Hospital anxiety and depression scale (HAD):some psychometric data for a Swedish sample. Acta Psych Scand1997;96:281–6.

Lohnberg J. A review of outcome studies on cognitive-behavioral therapy forreducing fear-avoidance beliefs among individuals with chronic pain. J ClinPsych Med Set 2007;14:113–22.

Lundberg MKE, Stymf J, Carlsson SV. A psychometric evaluation of the Tampa Scalefor Kinesiophobia: a physiotherapeutic perspective. Physiother Theory Pract2004;20:121–33.

McCracken LM, Turk D. Behavioral and cognitive-behavioral treatment for chronicpain: outcome, predictors of outcome, and treatment process. Spine2002;27:2564–73.

Nachemson A, Jonsson E. Neck and back pain: the scientific evidence of causes,diagnosis, and treatment. Philadelphia: Lippincott Williams & Wilkins; 2000.

Roelofs J, Goubert L, Peters ML, Vlaeyen JW, Crombez G. The Tampa Scale forKinesiophobia: further examination of psychometric properties in patients withchronic low back pain and fibromyalgia. Eur J Pain 2004;8:495–502.

Smeets RJEM, Vlaeyen JWS, Kester ADM, Knotterus JA. Reduction of paincatastrophizing mediates the outcome of both physical and cognitive-behavioral treatment in chronic low back pain. J Pain 2006;7:261–71.

Sullivan MJL, Bishop SR, Pivik J. The pain catastrophizing scale: development andvalidation. Psychol Assess 1995;7:524–32.

Sullivan MJL, Thorn B, Haythornthwaite JA, Keefe FJ, Martin M, Bradley LA, et al.Theoretical perspectives on the relation between catastrophizing and pain. ClinJ Pain 2001;17:52–64.

Turner JA, Holtzman S, Mancl L. Mediators, moderators, and predictors oftherapeutic change in cognitive-behavioral therapy for chronic pain. Pain2007;127:276–86.

Vlaeyen JW, de Jong J, Geilen M, Heuts PHTG, van Breukelen GJ. The treatment offear of movement/(re)injury in chronic low back pain: further evidence on theeffectiveness of exposure in vivo. Clin J Pain 2002a;18:251–61.

Vlaeyen JW, de Jong JR, Onghena P, Kerckhoffs-Hanssen M, Kole-Snijders AMJ. Canpain-related fear be reduced? The application of cognitive-behavioral exposurein vivo. Pain Res Manage 2002b;7:144–53.

Vlaeyen JW, Morley S. Cognitive-behavioral treatments for chronic pain. Whatworks for whom? Clin J Pain 2005;21:1–8.

Vlaeyen JWS, de Jong J, LeeuwM, Crombez G. Fear reduction in chronic pain: gradedexposure in vivo with behavioral experiments. In: Asmundson GJ, Vlaeyen JWS,Crombez G, editors. Understanding and treating fear of pain. Oxford: OxfordUniversity Press; 2004.

Vlaeyen JWS, de Jong JR, Geilen M, Heuts P, van Breukelen G. Graded exposurein vivo in the treatment of pain-related fear: a replicated single-caseexperimental design in four patients with low back pain. Behav Res Therapy2001;39:151–66.

Vlaeyen JWS, Kole-Snijders AMJ, Boeren RGB, Rotteveel RGB, Ruesink R, HeutsPHTG. The role of fear of movement/(re)injury in pain disability. J Occup Rehab1995;5:235–52.

Vlaeyen JWS, Linton SJ. Fear-avoidance and its consequences in chronicmusculoskeletal pain: a state of the art. Pain 2000;85:317–32.

Woods MP, Asmundson GJG. Evaluating the efficacy of graded in vivo exposure forthe treatment of fear in patients with chronic back pain: a randomizedcontrolled trial. Pain 2008;136:271–80.

Vowles KE, McCracken LM, Eccleston C. Processes of change in treatment for chronicpain: the contributions of pain, acceptance, and catastrophizing. Eur J Pain2007;11:779–87.

Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psych Scand1983;67:361–70.


STUDY III

Catastrophizing moderates the effect of exposure in vivo for back pain patientswith pain-related fear

Ida K. Flink *, Katja Boersma, Steven J. LintonCenter for Health and Medical Psychology (CHAMP), School of Law, Psychology and Social Work, 701 82 Örebro, Sweden


Article history:Received 3 July 2009Received in revised form 20 January 2010Accepted 2 February 2010Available online 9 March 2010

Keywords:Exposure in vivoBack painFear-avoidanceRandomized-controlled trialCatastrophizing

a b s t r a c t

This investigation was an initial attempt to explore psychological factors that might help or hinder theeffect of exposure in vivo for patients with musculoskeletal pain and pain-related fear. The study wasbased on data from a randomized-controlled trial for patients with non-specific spinal pain (Lintonet al., 2008).First, catastrophizing, anxiety, and depression were studied as possible treatment moderators. We

found evidence that catastrophizing was a moderator of treatment outcome in exposure. When furtherexploring the nature of the relationship between catastrophizing and outcome, the results showed thatthe exposure was effective only for patients with low or moderate levels of catastrophizing. High catastr-ophizers did not improve from the treatment. On the other hand, anxiety was a general predictor of pooroutcome, and not a specific moderator of outcome in exposure. In contrast, depression was not signifi-cantly related to outcome.Next, patients were divided into high change participants and low change participants based on their

improvement in disability after treatment in order to investigate the change in psychological variablesduring treatment. Descriptive data indicated that high change participants had large improvementsacross treatment on depression, anxiety, catastrophizing, and fear-avoidance beliefs whereas low changeparticipants virtually did not change at all on these variables across treatment.These findings denote that catastrophizing is a moderator of treatment outcome in exposure whereas

several psychological variables might be important for the treatment process.� 2010 European Federation of International Association for the Study of Pain Chapters. Published by

Elsevier Ltd. All rights reserved.

1. Introduction

Exposure in vivo is a novel treatment approach for patients withmusculoskeletal pain (MSP) and high levels of pain-related fear.Fear and subsequent avoidance of activities are key factors thatmay perpetuate pain and dysfunction in a subgroup of patients,as proposed in the fear-avoidance model (Vlaeyen and Linton,2000). Several studies have indeed shown that exposure is effectivefor reducing fear, catastrophic thoughts, disability and pain inthese patients (Vlaeyen et al., 2001; Boersma et al., 2004; de Jonget al., 2005; Leeuw et al., 2008; Linton et al., 2008; Woods andAsmundson, 2008). A recent review concluded that exposure isthe treatment of choice for patients with pain-related fear (Lohn-berg, 2007). This conclusion was based on the first studies of expo-sure which showed promising effects in replicated single-caseexperimental studies (Vlaeyen et al., 2001, 2002a, 2002b; Lintonet al., 2002; Boersma et al., 2004; de Jong et al., 2005). However,one of these studies noted substantial individual differences in

how well the patients responded to the treatment (Boersmaet al., 2004).

Recently, four RCTs further examined the effect of exposurein vivo for patients with MSP, and the effects were not as impres-sive as those reported in earlier studies (George et al., 2008; Leeuwet al., 2008; Linton et al., 2008; Woods and Asmundson, 2008).Exposure resulted in moderate to large effect sizes for reducingpain-related fear, catastrophizing, disability, and pain. But whencompared to other commonly used treatments, the effects weremore modest (George et al., 2008; Leeuw et al., 2008; Woods andAsmundson, 2008). While the unclear results may partly due to alack of power due to small samples, a more compelling explana-tion, which indeed has been noticed (Linton et al., 2008), is thatsome patients respond better to exposure than others. Conse-quently, the effect of exposure might be suppressed by moderatingfactors (Baron and Kenny, 1986; Vlaeyen and Morley, 2005).

One hypothesis has been that patients with high levels of pain-related fear would benefit most from exposure, but data havefailed to support this (George et al., 2008; Leeuw et al., 2008). Incognitive behavioral treatment packages for MSP, where exposureis not included, high catastrophizing, anxiety, and depression have

1090-3801/$36.00 � 2010 European Federation of International Association for the Study of Pain Chapters. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.ejpain.2010.02.003

* Corresponding author. Tel.: +46 19 303 740; fax: +46 19 303 484.E-mail address: [email protected] (I.K. Flink).





STUDY III

1

British Journal of Health Psychology (2011)C� 2011 The British Psychological Society

TheBritishPsychologicalSociety

www.wileyonlinelibrary.com

Understanding catastrophizing from amisdirected problem-solving perspective

Ida K. Flink∗, Katja Boersma, Shane MacDonaldand Steven J. LintonCenter for Health and Medical Psychology (CHAMP), School of Law, Psychologyand Social Work, Orebro University, Sweden

Objectives. The aim is to explore pain catastrophizing from a problem-solvingperspective. The links between catastrophizing, problem framing, and problem-solvingbehaviour are examined through two possible models of mediation as inferred by twocontemporary and complementary theoretical models, the misdirected problem solvingmodel (Eccleston & Crombez, 2007) and the fear-anxiety-avoidance model (Asmundson,Norton, & Vlaeyen, 2004).

Design. In this prospective study, a general population sample (n = 173) withperceived problems with spinal pain filled out questionnaires twice; catastrophizingand problem framing were assessed on the first occasion and health care seeking (as aproxy for medically oriented problem solving) was assessed 7 months later.

Methods. Two different approaches were used to explore whether the datasupported any of the proposed models of mediation. First, multiple regressions wereused according to traditional recommendations for mediation analyses. Second, abootstrapping method (n = 1000 bootstrap resamples) was used to explore thesignificance of the indirect effects in both possible models of mediation.

Results. The results verified the concepts included in the misdirected problem solvingmodel. However, the direction of the relations was more in line with the fear-anxiety-avoidance model. More specifically, the mediation analyses provided support for viewingcatastrophizing as a mediator of the relation between biomedical problem framing andmedically oriented problem-solving behaviour.

Conclusion. These findings provide support for viewing catastrophizing from aproblem-solving perspective and imply a need to examine and address problem framingand catastrophizing in back pain patients.

Disabling spinal pain is a common problem in Western societies. It affects about 85% ofworkers at some point in their working years, and for about 10% it develops into a long-term condition (Nachemson, Waddell, & Norlund, 2000; Waddell, Aylward, & Sawney,

∗Correspondence should be addressed to Ida K. Flink, Center for Health and Medical Psychology (CHAMP), School of Law,Psychology and Social work, 701 82 Orebro, Sweden (e-mail: [email protected]).

DOI:10.1111/j.2044-8287.2011.02044.x

2 Ida K. Flink et al.

2002). When suffering from disabling pain, a natural reaction is to worry about the causesand how to solve the problem. A certain amount of worry is adaptive in preparing us forfuture threats, but only if it does not get too intense (Mathews, 1990). When the worryprogressively intensifies and the outcome is perceived as getting worse and worse, it hasbeen defined as catastrophic worry (Davey & Levy, 1998), as captured by the conceptcatastrophizing. Catastrophizing is widely studied in the area of musculoskeletal pain,and has been identified as one of the most important psychological variables predictinglong-term disability (Sullivan et al., 2001). Thus, while worry might fill an importantmotivational function initially, in the long run it may aggravate problems, especiallywhen it takes the form of catastrophizing.

The misdirected problem solving model [see Figure 1 (Eccleston & Crombez, 2007)]was recently presented as an attempt to provide a framework to explain how worryand catastrophizing might lead to rigid, medically oriented problem-solving behaviours.According to this model, a patient who catastrophizes is likely to frame the pain as abiomedical problem which, in the case that no medical solution actually exists, resultsin the patient repeatedly seeking medical care without getting better. In the case ofspinal pain, many times there is no medical solution of the problem. Thus, to keep onsearching for a cure will unlikely solve the problem and in this sense an individuals’efforts become ‘misdirected’. As the problem remains unsolved, the catastrophic worryincreases, which in turn results in further biomedical problem framing. Consequently, themisdirected problem solving model poses that catastrophizing is part of an unsuccessfulproblem-solving behaviour, involving repeated fruitless efforts that are directed towardscuring or getting rid of pain.

This model was developed in relation and addition to the expanded fear-anxiety-avoidance model [see Figure 2 (Asmundson, Norton, & Vlaeyen, 2004)]. This modelalso underscores the potential maladaptiveness in framing a long-term pain problemin biomedical terms, here captured by pain beliefs such as ‘pain equals damage orserious injury’. In this model, a biomedical problem framing is proposed to set the stage

Catastrophicworry

Pain

Hypervigilance

Biomedicalproblem frame

Problem solvingbehavior

Problem unsolved

Perseverance loop

Catastrophicworry

Pain

Hypervigilance


Problem- solvingbehaviour

Problem unsolved

Perseverance loop

Figure 1. A stylized version of the misdirected problem solving model, focusing on the perseveranceloop (Eccleston & Crombez, 2007).Note. This figure has been reproduced with permission of the International Association for the Studyof Pain (IASP). The figure may not be reproduced for any other purpose without permission.

Misdirected problem-solving perspective 3

Injury ororganic

pathologyPain

perception

Paincatastrophizing

Nocatastrophizing No fear No anxiety Recovery

Disuse/deconditioning

Fearof pain

DEFEN

SIVE

MO

TIVATIONAU

TON

OM

IC

ARO

USAL

AUTO

NO

MIC

ARO

USAL

THREATPERCEPTION

Pain-relatedanxiety

PREVENTATIVE

MO

TIVATION

HYPERVIGILANCE

Escape/defensivebehaviour

Avoidance/preventativebehaviour

No escape/defensivebehaviour

No avoidance/preventativebehaviour

Predisposingrisk factors

Painbeliefs

Figure 2. The fear-anxiety-avoidance model (Asmundson et al., 2004).Note. C© Oxford University Press (www.oup.com) This figure has been reproduced from Fear-avoidancemodels of chronic pain: An overview by permission of Oxford University Press.

for catastrophizing which in turn enhances the perception of threat and thereby themotivation to find a medical solution, or cure, for the pain.

In summary, both of the aforementioned models suggest that pain patients whocatastrophize are likely to view the problem as primarily biomedical and try to findmedical solutions for it, which in the case of spinal pain often prove unsuccessful. Themodels might be seen as complementary rather than competing, because they bothfocus on the emotions, cognitions, and overt behaviour involved in solving the problemof long-term pain. However, the models do imply somewhat different pathways. Inthe formation of the misdirected problem solving model, a biomedical problem framemediates the relation between catastrophic worry and problem-solving behaviour whilethe fear-anxiety-avoidance model implies that it is the other way around. In this model,catastrophizing mediates the relation between a biomedical problem frame (i.e., painbeliefs) and problem-solving behaviour (i.e., defence motivation/defensive behaviour).Neither of the paths has, however, been empirically tested, and earlier research hasstressed the importance of examining in what way catastrophizing is linked to overtproblem-solving behaviour (Crombez, Eccleston, Van Hamme, & De Vlieger, 2008).Even though the sequential order of the fear-avoidance model has been scrutinized(e.g., Leeuw et al., 2007; Wideman, Adams & Sullivan, 2009), the specific link betweenpain beliefs and catastrophizing has not been investigated empirically, and nor has thesequential order of the misdirected problem solving model. Thus, these complementarymodels implicitly suggest somewhat different pathways to explain the links betweenproblem framing, catastrophizing, and problem-solving behaviour but these have notbeen examined empirically.

The current study aims to explore if and how catastrophizing is linked to biomedicalproblem framing and medically oriented problem solving. We will do this using mediation

STUDY III


2002). When suffering from disabling pain, a natural reaction is to worry about the causesand how to solve the problem. A certain amount of worry is adaptive in preparing us forfuture threats, but only if it does not get too intense (Mathews, 1990). When the worryprogressively intensifies and the outcome is perceived as getting worse and worse, it hasbeen defined as catastrophic worry (Davey & Levy, 1998), as captured by the conceptcatastrophizing. Catastrophizing is widely studied in the area of musculoskeletal pain,and has been identified as one of the most important psychological variables predictinglong-term disability (Sullivan et al., 2001). Thus, while worry might fill an importantmotivational function initially, in the long run it may aggravate problems, especiallywhen it takes the form of catastrophizing.

The misdirected problem solving model [see Figure 1 (Eccleston & Crombez, 2007)]was recently presented as an attempt to provide a framework to explain how worryand catastrophizing might lead to rigid, medically oriented problem-solving behaviours.According to this model, a patient who catastrophizes is likely to frame the pain as abiomedical problem which, in the case that no medical solution actually exists, resultsin the patient repeatedly seeking medical care without getting better. In the case ofspinal pain, many times there is no medical solution of the problem. Thus, to keep onsearching for a cure will unlikely solve the problem and in this sense an individuals’efforts become ‘misdirected’. As the problem remains unsolved, the catastrophic worryincreases, which in turn results in further biomedical problem framing. Consequently, themisdirected problem solving model poses that catastrophizing is part of an unsuccessfulproblem-solving behaviour, involving repeated fruitless efforts that are directed towardscuring or getting rid of pain.

This model was developed in relation and addition to the expanded fear-anxiety-avoidance model [see Figure 2 (Asmundson, Norton, & Vlaeyen, 2004)]. This modelalso underscores the potential maladaptiveness in framing a long-term pain problemin biomedical terms, here captured by pain beliefs such as ‘pain equals damage orserious injury’. In this model, a biomedical problem framing is proposed to set the stage

Catastrophicworry

Pain

Hypervigilance


Problem solvingbehavior

Problem unsolved

Perseverance loop

Catastrophicworry

Pain

Hypervigilance


Problem- solvingbehaviour

Problem unsolved

Perseverance loop

Figure 1. A stylized version of the misdirected problem solving model, focusing on the perseveranceloop (Eccleston & Crombez, 2007).Note. This figure has been reproduced with permission of the International Association for the Studyof Pain (IASP). The figure may not be reproduced for any other purpose without permission.


Injury ororganic

pathologyPain

perception

Paincatastrophizing

Nocatastrophizing No fear No anxiety Recovery

Disuse/deconditioning

Fearof pain

DEFEN

SIVE

MO

TIVATIONAU

TON

OM

IC

ARO

USAL

AUTO

NO

MIC

ARO

USAL

THREATPERCEPTION

Pain-relatedanxiety

PREVENTATIVE

MO

TIVATION

HYPERVIGILANCE

Escape/defensivebehaviour

Avoidance/preventativebehaviour

No escape/defensivebehaviour

No avoidance/preventativebehaviour

Predisposingrisk factors

Painbeliefs

Figure 2. The fear-anxiety-avoidance model (Asmundson et al., 2004).Note. C© Oxford University Press (www.oup.com) This figure has been reproduced from Fear-avoidancemodels of chronic pain: An overview by permission of Oxford University Press.

for catastrophizing which in turn enhances the perception of threat and thereby themotivation to find a medical solution, or cure, for the pain.

In summary, both of the aforementioned models suggest that pain patients whocatastrophize are likely to view the problem as primarily biomedical and try to findmedical solutions for it, which in the case of spinal pain often prove unsuccessful. Themodels might be seen as complementary rather than competing, because they bothfocus on the emotions, cognitions, and overt behaviour involved in solving the problemof long-term pain. However, the models do imply somewhat different pathways. Inthe formation of the misdirected problem solving model, a biomedical problem framemediates the relation between catastrophic worry and problem-solving behaviour whilethe fear-anxiety-avoidance model implies that it is the other way around. In this model,catastrophizing mediates the relation between a biomedical problem frame (i.e., painbeliefs) and problem-solving behaviour (i.e., defence motivation/defensive behaviour).Neither of the paths has, however, been empirically tested, and earlier research hasstressed the importance of examining in what way catastrophizing is linked to overtproblem-solving behaviour (Crombez, Eccleston, Van Hamme, & De Vlieger, 2008).Even though the sequential order of the fear-avoidance model has been scrutinized(e.g., Leeuw et al., 2007; Wideman, Adams & Sullivan, 2009), the specific link betweenpain beliefs and catastrophizing has not been investigated empirically, and nor has thesequential order of the misdirected problem solving model. Thus, these complementarymodels implicitly suggest somewhat different pathways to explain the links betweenproblem framing, catastrophizing, and problem-solving behaviour but these have notbeen examined empirically.

The current study aims to explore if and how catastrophizing is linked to biomedicalproblem framing and medically oriented problem solving. We will do this using mediation


analyses where catastrophizing and a proxy for biomedical problem framing will beassessed at the first occasion, and a proxy for medically oriented problem solving will beassessed 7 months later. In this way, we will examine two possible models of mediation asimplied by aforementioned theoretical models. The goal is to increase our understandingof catastrophizing from a problem-solving perspective.

MethodsDesignThis study is based on data from a prospective study about psychological processesin the development of chronic pain problems (Boersma & Linton, 2005). Participantsanswered questionnaires on two occasions. On the first occasion the questions addressedbackground variables, pain characteristics, sick leave, and psychological processes. Onthe second occasion, 7 months later, the questions focused on function and health careutilization which was used as a proxy for medically oriented problem solving. Onlyparticipants who answered both questionnaires were included in the study (n = 173).The study was approved by the Orebro University Hospital’s Board on Research Ethics.

ParticipantsParticipants were recruited via advertisements in local newspapers in three counties inmiddle Sweden. The inclusion criteria were (1) 20–60 years old, (2) perceived problemswith spinal pain, and (3) at least 1 day of sick leave the previous year because of thepain problem. The questionnaires were sent to the participants via mail. For detailsconcerning recruitment and dropouts, see the original study (Boersma & Linton, 2005).Table 1 displays background variables (age, gender, and nationality), pain characteristics(intensity, location, and duration), and sick leave for the participants.

MeasuresSwedish versions of all questionnaires were used.

Table 1. Background variables, pain characteristics, and sick leave for the participants

Age, M (SD) 57 (9.1)Gender, women/men 69% / 31%Nationality, born in Sweden 90%Pain intensity, M (SD) 5.6 (2.3)

Pain location (more than one location possible)Neck 73%Shoulder 68%Upper back 52%Lower back 74%

Pain duration�3 months (acute) 7%3–12 months (subacute) 17%�12 months (chronic) 75%

Sick leave last 12 months1–14 days 35%14–180 days 28%�180 days 34%


Background variables, sick leave, and pain characteristicsQuestions from the Orebro Musculoskeletal Pain Screening Questionnaire (OMPSQ;Linton & Hallden, 1998) were used to assess background variables (age, gender, andnationality), sick leave (‘How many days of work have you missed because of painduring the past 12 months?’), and pain characteristics (intensity, location, and duration).The ratings of pain intensity (‘How would you rate the pain that you have had duringthe past week?’) were made on an 11-point scale (0 = no pain; 10 = pain as bad as itcould be). The information about pain location (‘Where do you have pain?’) was givenby checking the appropriate sites (neck, shoulder, upper back, and/or lower back).The information about pain duration was given by checking the appropriate site by10 possible options, ranging from 0 to 1 week to >52 weeks. The OMPSQ has shownto be a reliable and valid instrument for assessing psychosocial factors in people withmusculoskeletal pain and disability (Linton & Bersma, 2003; Westman, Linton, Ohrvik,Wahlen, & Leppert, 2008).

CatastrophizingThe catastrophizing subscale from the Coping Strategies Questionnaire (CSQ; Rosentiel,1983) was used to assess pain-related catastrophic thoughts. This subscale consists ofsix items reflecting thoughts and feelings that might arise when people experience pain(e.g., ‘I worry all the time whether it will end’, ‘It’s awful and I feel that it overwhelmsme’). The answers are given on a seven-point scale, where the respondents rate to whatextent they have these thoughts and feelings when they experience pain (0 = never;6 = always). The catastrophizing scale has shown appropriate construct validity(Stewart, Harvey, & Evans, 2001), and the Swedish version has shown good internalconsistency (alpha range .7–.8) but somewhat less satisfactory test–retest reliability(correlations range .4–.9) (Jensen & Linton, 1993).

Biomedical problem framingThe somatic focus subscale from the Tampa Scale for Kinesiophobia (TSK) (Kori, Miller, &Todd, 1990) was used as a proxy for biomedical problem framing. The TSK was originallydeveloped to assess fear of movement and (re)injury. However, several investigations ofthe factor structure of the TSK have confirmed a two-factor solution with two distinctsubscales: somatic focus and activity avoidance (Goubert et al., 2004; Roelofs et al.,2007; Roelofs, Goubert, Peters, Vlaeyen, & Crombez, 2004). The construct validity ofthe somatic focus subscale has been supported by showing that it differs from the activityavoidance subscale in correlations with self-report measures of pain-related fear, paincatastrophizing, and disability (Roelofs et al., 2004).While the activity avoidance subscalefocuses on beliefs that pain-provoking activities should be avoided, the somatic focussubscale focuses on beliefs about pain signalling underlying serious medical problems.Therefore, this subscale fits well with the concept of biomedical problem framing.Specifically, the somatic focus subscale consists of five items: ‘My body is telling me I havesomething dangerously wrong’, ‘I wouldn’t have this much pain if there weren’t some-thing potentially dangerous going on in my body’, ‘My accident has put my body at riskfor the rest of my life’, ‘Pain always mean I have injured my body’, and ‘People aren’ttaking my medical condition seriously enough’. The answers are given on a four-pointscale (0 = strongly disagree; 4 = strongly agree). The questionnaire has shown goodpsychometric properties (Goubert et al., 2004; Roelofs et al., 2004; Vlaeyen et al., 1995).

STUDY III


analyses where catastrophizing and a proxy for biomedical problem framing will beassessed at the first occasion, and a proxy for medically oriented problem solving will beassessed 7 months later. In this way, we will examine two possible models of mediation asimplied by aforementioned theoretical models. The goal is to increase our understandingof catastrophizing from a problem-solving perspective.

MethodsDesignThis study is based on data from a prospective study about psychological processesin the development of chronic pain problems (Boersma & Linton, 2005). Participantsanswered questionnaires on two occasions. On the first occasion the questions addressedbackground variables, pain characteristics, sick leave, and psychological processes. Onthe second occasion, 7 months later, the questions focused on function and health careutilization which was used as a proxy for medically oriented problem solving. Onlyparticipants who answered both questionnaires were included in the study (n = 173).The study was approved by the Orebro University Hospital’s Board on Research Ethics.

ParticipantsParticipants were recruited via advertisements in local newspapers in three counties inmiddle Sweden. The inclusion criteria were (1) 20–60 years old, (2) perceived problemswith spinal pain, and (3) at least 1 day of sick leave the previous year because of thepain problem. The questionnaires were sent to the participants via mail. For detailsconcerning recruitment and dropouts, see the original study (Boersma & Linton, 2005).Table 1 displays background variables (age, gender, and nationality), pain characteristics(intensity, location, and duration), and sick leave for the participants.

MeasuresSwedish versions of all questionnaires were used.

Table 1. Background variables, pain characteristics, and sick leave for the participants

Age, M (SD) 57 (9.1)Gender, women/men 69% / 31%Nationality, born in Sweden 90%Pain intensity, M (SD) 5.6 (2.3)

Pain location (more than one location possible)Neck 73%Shoulder 68%Upper back 52%Lower back 74%

Pain duration�3 months (acute) 7%3–12 months (subacute) 17%�12 months (chronic) 75%

Sick leave last 12 months1–14 days 35%14–180 days 28%�180 days 34%


Background variables, sick leave, and pain characteristicsQuestions from the Orebro Musculoskeletal Pain Screening Questionnaire (OMPSQ;Linton & Hallden, 1998) were used to assess background variables (age, gender, andnationality), sick leave (‘How many days of work have you missed because of painduring the past 12 months?’), and pain characteristics (intensity, location, and duration).The ratings of pain intensity (‘How would you rate the pain that you have had duringthe past week?’) were made on an 11-point scale (0 = no pain; 10 = pain as bad as itcould be). The information about pain location (‘Where do you have pain?’) was givenby checking the appropriate sites (neck, shoulder, upper back, and/or lower back).The information about pain duration was given by checking the appropriate site by10 possible options, ranging from 0 to 1 week to >52 weeks. The OMPSQ has shownto be a reliable and valid instrument for assessing psychosocial factors in people withmusculoskeletal pain and disability (Linton & Bersma, 2003; Westman, Linton, Ohrvik,Wahlen, & Leppert, 2008).

CatastrophizingThe catastrophizing subscale from the Coping Strategies Questionnaire (CSQ; Rosentiel,1983) was used to assess pain-related catastrophic thoughts. This subscale consists ofsix items reflecting thoughts and feelings that might arise when people experience pain(e.g., ‘I worry all the time whether it will end’, ‘It’s awful and I feel that it overwhelmsme’). The answers are given on a seven-point scale, where the respondents rate to whatextent they have these thoughts and feelings when they experience pain (0 = never;6 = always). The catastrophizing scale has shown appropriate construct validity(Stewart, Harvey, & Evans, 2001), and the Swedish version has shown good internalconsistency (alpha range .7–.8) but somewhat less satisfactory test–retest reliability(correlations range .4–.9) (Jensen & Linton, 1993).

Biomedical problem framingThe somatic focus subscale from the Tampa Scale for Kinesiophobia (TSK) (Kori, Miller, &Todd, 1990) was used as a proxy for biomedical problem framing. The TSK was originallydeveloped to assess fear of movement and (re)injury. However, several investigations ofthe factor structure of the TSK have confirmed a two-factor solution with two distinctsubscales: somatic focus and activity avoidance (Goubert et al., 2004; Roelofs et al.,2007; Roelofs, Goubert, Peters, Vlaeyen, & Crombez, 2004). The construct validity ofthe somatic focus subscale has been supported by showing that it differs from the activityavoidance subscale in correlations with self-report measures of pain-related fear, paincatastrophizing, and disability (Roelofs et al., 2004).While the activity avoidance subscalefocuses on beliefs that pain-provoking activities should be avoided, the somatic focussubscale focuses on beliefs about pain signalling underlying serious medical problems.Therefore, this subscale fits well with the concept of biomedical problem framing.Specifically, the somatic focus subscale consists of five items: ‘My body is telling me I havesomething dangerously wrong’, ‘I wouldn’t have this much pain if there weren’t some-thing potentially dangerous going on in my body’, ‘My accident has put my body at riskfor the rest of my life’, ‘Pain always mean I have injured my body’, and ‘People aren’ttaking my medical condition seriously enough’. The answers are given on a four-pointscale (0 = strongly disagree; 4 = strongly agree). The questionnaire has shown goodpsychometric properties (Goubert et al., 2004; Roelofs et al., 2004; Vlaeyen et al., 1995).


Catastrophizing(t1)

Medically oriented problem solving behaviour (t2)


Figure 3. Possible model of mediation with biomedical problem framing as a proposed mediator.

The Swedish version has shown good reliability including stability over time and internalconsistency (alpha .81) (Lundberg, Stymf, & Carlsson, 2004).

Medically oriented problem-solving behaviourHealth care utilization was used as a proxy for medically oriented problem-solvingbehaviour, and was assessed with questions from the Outcome Evaluation Questionnaire(Keefe, Linton, & Lefebvre, 1992). The participants were asked which of the followinghealth care providers they had visited because of their pain problem during the last 7months (i.e., since the first questionnaire): medical doctor, physiotherapist, specialistcare, or other (e.g., chiropractic or acupuncture). The total number of health careproviders was used an index of health care utilization.

FunctionFunction was assessed with five questions from the OMPSQ (Linton & Hallden, 1998).The respondents are asked to what degree they are able to participate in certain dailyactivities (e.g., ‘I can do ordinary household chores’, ‘I can do the weekly shopping’).Ratings are made on an 11-point scale (0 = cannot do it because of pain; 10 = can do itwithout pain being a problem). The questions have shown high reliability and validity(Boersma & Linton, 2005; Linton, 1990).

Statistical analysisFirst, the data were summarized and inspected through descriptive and correlationalstatistics.

Next, we explored two possible models of mediation. In the first model, which isin line with the misdirected problem solving model, biomedical problem framing wasproposed as a mediator of the relation between catastrophizing and medically orientedproblem-solving behaviour (see Figure 3). In the second model, which is in line withthe fear-anxiety-avoidance model, catastrophizing was proposed as a mediator of therelation between biomedical problem framing and medically oriented problem-solvingbehaviour (see Figure 4).

Catastrophizing(t1)



Figure 4. Possible model of mediation with catastrophizing as a proposed mediator.


We used two different approaches to explore whether the data supported anyof these possible models of mediation. First, multiple regressions were used toinvestigate whether the prerequisites of mediation effects were met, using the tradi-tional recommendations by Baron & Kenny (1986). Second, the significance of theindirect effects were assessed with a bootstrapping method (n = 1000 bootstrapresamples) (see Preacher & Hayes, 2008). Bootstrapping is a non-parametric resamplingprocedure that generates an approximation of the sampling distribution from theavailable data set. More specifically, the bootstrapping distribution is generated bytaking a sample (with replacement) of size n from the full data set. The indirecteffects are then calculated in the resamples that results in point estimates and 95%confidence intervals are estimated for the indirect effects. We considered point estimatesof indirect effects to be significant if zero was not contained in the confidenceinterval.

Statistical analyses were performed with SPSS 18.0. Missing values were estimated bycalculating the mean score for the scale.

ResultsDescriptive and correlational statisticsTable 2 displays means, standard deviations, and Pearson’s product moment correlationsbetween the primary variables in the study. Somatic focus was used as a proxy forbiomedical problem framing and health care utilization was used as a proxy for medicallyoriented problem-solving behaviour. The participants had moderate levels of bothcatastrophizing, with a mean rating of 14.0 (SD = 7.5), and somatic focus, with amean rating of 12.6 (SD = 3.7). The participants were also moderately disabled, with amean rating of 28.8 (SD = 14.0) on function. The mean ratings of health care utilizationwas 1.6 (SD = 1.2), which gives the mean number of health care providers that theparticipants had visited between timepoint 1 and 2.

As can be seen in Table 2, all primary variables in the study were correlatedwith coefficients varying between .24 and .50. The most pronounced correlationswere found between catastrophizing and somatic focus, and between somatic focusand function. The latter correlation coefficient was negative, which means that ahigher level of somatic focus at timepoint 1 indicated lower levels of function attimepoint 2.

Table 2. Means (M), standard deviations (SD), and correlations of the primary variables in the study(n = 173)

Range M (SD) 2 3 4

1. Catastrophizing (CSQ) 0–36 14.0 (7.5) .50∗∗ −.39∗∗ .31∗∗

2. Somatic focus (TSK) 0–20 12.6 (3.7) - −.46∗∗ .24∗∗

3. Function (OMSPQ)a 0–50 28.8 (14.0) - −.36∗∗

4. Health care utilization (OEQ)a 0–4 1.6 (1.2) -

Notes. CSQ, coping strategies questionnaire; TSK, Tampa Scale of Kinesiophobia; OEQ, outcomeevaluation questionnaire; OMPSQ, Orebro musculoskeletal pain screening questionnaire.aThese scorings are based on ratings at timepoint 2.∗∗p � .01

STUDY III


Catastrophizing(t1)



Figure 3. Possible model of mediation with biomedical problem framing as a proposed mediator.

The Swedish version has shown good reliability including stability over time and internalconsistency (alpha .81) (Lundberg, Stymf, & Carlsson, 2004).

Medically oriented problem-solving behaviourHealth care utilization was used as a proxy for medically oriented problem-solvingbehaviour, and was assessed with questions from the Outcome Evaluation Questionnaire(Keefe, Linton, & Lefebvre, 1992). The participants were asked which of the followinghealth care providers they had visited because of their pain problem during the last 7months (i.e., since the first questionnaire): medical doctor, physiotherapist, specialistcare, or other (e.g., chiropractic or acupuncture). The total number of health careproviders was used an index of health care utilization.

FunctionFunction was assessed with five questions from the OMPSQ (Linton & Hallden, 1998).The respondents are asked to what degree they are able to participate in certain dailyactivities (e.g., ‘I can do ordinary household chores’, ‘I can do the weekly shopping’).Ratings are made on an 11-point scale (0 = cannot do it because of pain; 10 = can do itwithout pain being a problem). The questions have shown high reliability and validity(Boersma & Linton, 2005; Linton, 1990).

Statistical analysisFirst, the data were summarized and inspected through descriptive and correlationalstatistics.

Next, we explored two possible models of mediation. In the first model, which isin line with the misdirected problem solving model, biomedical problem framing wasproposed as a mediator of the relation between catastrophizing and medically orientedproblem-solving behaviour (see Figure 3). In the second model, which is in line withthe fear-anxiety-avoidance model, catastrophizing was proposed as a mediator of therelation between biomedical problem framing and medically oriented problem-solvingbehaviour (see Figure 4).

Catastrophizing(t1)



Figure 4. Possible model of mediation with catastrophizing as a proposed mediator.


We used two different approaches to explore whether the data supported anyof these possible models of mediation. First, multiple regressions were used toinvestigate whether the prerequisites of mediation effects were met, using the tradi-tional recommendations by Baron & Kenny (1986). Second, the significance of theindirect effects were assessed with a bootstrapping method (n = 1000 bootstrapresamples) (see Preacher & Hayes, 2008). Bootstrapping is a non-parametric resamplingprocedure that generates an approximation of the sampling distribution from theavailable data set. More specifically, the bootstrapping distribution is generated bytaking a sample (with replacement) of size n from the full data set. The indirecteffects are then calculated in the resamples that results in point estimates and 95%confidence intervals are estimated for the indirect effects. We considered point estimatesof indirect effects to be significant if zero was not contained in the confidenceinterval.

Statistical analyses were performed with SPSS 18.0. Missing values were estimated bycalculating the mean score for the scale.

ResultsDescriptive and correlational statisticsTable 2 displays means, standard deviations, and Pearson’s product moment correlationsbetween the primary variables in the study. Somatic focus was used as a proxy forbiomedical problem framing and health care utilization was used as a proxy for medicallyoriented problem-solving behaviour. The participants had moderate levels of bothcatastrophizing, with a mean rating of 14.0 (SD = 7.5), and somatic focus, with amean rating of 12.6 (SD = 3.7). The participants were also moderately disabled, with amean rating of 28.8 (SD = 14.0) on function. The mean ratings of health care utilizationwas 1.6 (SD = 1.2), which gives the mean number of health care providers that theparticipants had visited between timepoint 1 and 2.

As can be seen in Table 2, all primary variables in the study were correlatedwith coefficients varying between .24 and .50. The most pronounced correlationswere found between catastrophizing and somatic focus, and between somatic focusand function. The latter correlation coefficient was negative, which means that ahigher level of somatic focus at timepoint 1 indicated lower levels of function attimepoint 2.

Table 2. Means (M), standard deviations (SD), and correlations of the primary variables in the study(n = 173)

Range M (SD) 2 3 4

1. Catastrophizing (CSQ) 0–36 14.0 (7.5) .50∗∗ −.39∗∗ .31∗∗

2. Somatic focus (TSK) 0–20 12.6 (3.7) - −.46∗∗ .24∗∗

3. Function (OMSPQ)a 0–50 28.8 (14.0) - −.36∗∗

4. Health care utilization (OEQ)a 0–4 1.6 (1.2) -

Notes. CSQ, coping strategies questionnaire; TSK, Tampa Scale of Kinesiophobia; OEQ, outcomeevaluation questionnaire; OMPSQ, Orebro musculoskeletal pain screening questionnaire.aThese scorings are based on ratings at timepoint 2.∗∗p � .01


Table 3. Multiple regression analyses exploring two possible models of mediation

B SE B �

Alternative 1: Misdirected problem solving modelStep 1Constant .90 .19Catastrophizing .05 .01 .31∗∗

Step 2Constant .57 .32Catastrophizing .04 .01 .26∗∗

Somatic focus .04 .03 .11Ns

Alternative 2: Fear-anxiety-avoidance modelStep 1Constant .62 .33Somatic focus .08 .03 .24∗∗

Step 2Constant .57 .32Somatic focus .04 .03 .11Ns

Catastrophizing .04 .01 .26∗∗

Notes.aAlternative 1 R2 = .10 for Step 1; � R2 = .01 for Step 2 (Ns).bAlternative 2 R2 = .06 for Step 1: � R2 = .05 for Step 2 (p � .01).c Health care utilization was used as the criterion variable in the analyses.∗∗p � .01.

Mediation analysesMultiple regression analyses were performed to investigate whether the traditionalprerequisites of mediation effects were met in the two possible models of mediationdescribed in the statistical analysis section. In both models, health care utilizationwas used as the dependent variable as a proxy for medically oriented problem-solvingbehaviour. In the first model (Alternative 1: Misdirected problem solving model), somaticfocus was a proposed mediator between catastrophizing and health care utilization, andin the second model (Alternative 2: Fear-anxiety-avoidance model) catastrophizing wasa proposed mediator between somatic focus and health care utilization. As can be seenin Table 3, the data best matched Alternative 2 as the effect of somatic focus on healthcare utilization was reduced to a non-significant level (� = .11 Ns) (and not the otherway around) when both catastrophizing and somatic focus were used as predictors inthe regression.

Subsequently, a bootstrapping method (n = 1000 bootstrap resamples) was used toexplore the significance of the indirect effects in both possible models of mediation. Theanalyses showed that in Alternative 1, the indirect effect of catastrophizing on healthcare utilization through somatic focus as a mediator was not significant (point estimate= .009; CI [–. 005, – .024], p = .20). In Alternative 2, however, the indirect effectof somatic focus on health care utilization through catastrophizing as a mediator wassignificant (point estimate = .042; CI [.015, – .078], p < .01).

In sum, in both approaches of mediation analyses, the data supported Alternative 2where catastrophizing was proposed as a mediator of the relation between somatic focusand health care utilization.


DiscussionThe current study is an attempt to empirically investigate the links between problemframing, catastrophizing, and problem-solving behaviour. On basis of two complemen-tary theoretical models which differ somewhat in their implicit suggestions of whatdrives what, we examined possible mediators using proxies for biomedical problemframing and medically oriented problem-solving behaviour. The goal was to explorewhether catastrophizing might be understood from a problem-solving perspective.

We found evidence that biomedical problem framing, catastrophizing, and medicallyoriented problem-solving behaviour were related. This partially supports the misdirectedproblem solving model (Eccleston & Crombez, 2007), which proposes that theseconcepts are related and interact to worsen the pain problem. The model implies thatif an individual who experiences pain worries extensively and frames the problem inbiomedical terms, this often results in medically oriented problem-solving behaviours,that is the individual relies on medical solutions and seeks health care to cure the pain.Our data indeed support the link between these concepts but the direction of therelations was somewhat different than the model proposes.

Our results indicate that the way people view their pain problem set the stagefor catastrophic worry. The mediation analyses supported catastrophizing as a mediatorbetween biomedical problem framing and medically oriented problem solving. This is notin line with the sequential order of the misdirected problem solving model, which impliesthat it is the other way around; that biomedical problem framing mediates the relationbetween catastrophizing and problem-solving behaviour. The current findings are rathercongruent with the expanded fear-anxiety-avoidance model (Asmundson et al., 2004),which implies that pain beliefs, such as ‘pain means harm or serious injury’, or in otherwords biomedical problem framing, might set the stage for catastrophizing and medicallyoriented problem-solving behaviours. In other words, the current findings indicate thatif patients view their pain as primarily a medically problem, then catastrophic thoughtsabout what is wrong might arise. In turn, the patients may seek medical health servicesas an attempt to solve the problem.

It is important to note that efforts at finding medical solutions for the pain are notadaptive or maladaptive per se. To try to find a solution for pain is generally relatedto better functioning (Keefe, Rumble, Scipio, Giordano, & Perri, 2004) but when it isdifficult to find a medical solution, such as in the case of spinal pain, a more adaptivereaction is to engage in other goals (Eccleston & Crombez, 2007). Based on our data,we cannot draw any conclusions about whether the repeated efforts for finding medicalsolutions for the pain problem were adaptive or not. There is, however, a relationbetween health care visits and poor function at the second assessment. On the onehand, this could be a sign that the efforts for finding medical solutions have not beensuccessful. On the other hand, it might also be that the patients seek more health carebecause of their poor function. A third possibility is that the contact with the health careprovider in fact has increased the patients’ feelings of helplessness, which in turn leadto further health care visits.

A related issue when trying to understand catastrophizing from a problem-solvingperspective is that we cannot tell whether catastrophizing is a precursor or also aconsequence of medically oriented problem-solving behaviour. The question of causalityis a well-known issue when examining bidirectional models; since they are circular itis difficult to tease out what drives what. To clarify that, experimental studies thatmanipulate these concepts are needed. The current design indeed has limitations when

STUDY III


Table 3. Multiple regression analyses exploring two possible models of mediation

B SE B �

Alternative 1: Misdirected problem solving modelStep 1Constant .90 .19Catastrophizing .05 .01 .31∗∗

Step 2Constant .57 .32Catastrophizing .04 .01 .26∗∗

Somatic focus .04 .03 .11Ns

Alternative 2: Fear-anxiety-avoidance modelStep 1Constant .62 .33Somatic focus .08 .03 .24∗∗

Step 2Constant .57 .32Somatic focus .04 .03 .11Ns

Catastrophizing .04 .01 .26∗∗

Notes.aAlternative 1 R2 = .10 for Step 1; � R2 = .01 for Step 2 (Ns).bAlternative 2 R2 = .06 for Step 1: � R2 = .05 for Step 2 (p � .01).c Health care utilization was used as the criterion variable in the analyses.∗∗p � .01.

Mediation analysesMultiple regression analyses were performed to investigate whether the traditionalprerequisites of mediation effects were met in the two possible models of mediationdescribed in the statistical analysis section. In both models, health care utilizationwas used as the dependent variable as a proxy for medically oriented problem-solvingbehaviour. In the first model (Alternative 1: Misdirected problem solving model), somaticfocus was a proposed mediator between catastrophizing and health care utilization, andin the second model (Alternative 2: Fear-anxiety-avoidance model) catastrophizing wasa proposed mediator between somatic focus and health care utilization. As can be seenin Table 3, the data best matched Alternative 2 as the effect of somatic focus on healthcare utilization was reduced to a non-significant level (� = .11 Ns) (and not the otherway around) when both catastrophizing and somatic focus were used as predictors inthe regression.

Subsequently, a bootstrapping method (n = 1000 bootstrap resamples) was used toexplore the significance of the indirect effects in both possible models of mediation. Theanalyses showed that in Alternative 1, the indirect effect of catastrophizing on healthcare utilization through somatic focus as a mediator was not significant (point estimate= .009; CI [–. 005, – .024], p = .20). In Alternative 2, however, the indirect effectof somatic focus on health care utilization through catastrophizing as a mediator wassignificant (point estimate = .042; CI [.015, – .078], p < .01).

In sum, in both approaches of mediation analyses, the data supported Alternative 2where catastrophizing was proposed as a mediator of the relation between somatic focusand health care utilization.


DiscussionThe current study is an attempt to empirically investigate the links between problemframing, catastrophizing, and problem-solving behaviour. On basis of two complemen-tary theoretical models which differ somewhat in their implicit suggestions of whatdrives what, we examined possible mediators using proxies for biomedical problemframing and medically oriented problem-solving behaviour. The goal was to explorewhether catastrophizing might be understood from a problem-solving perspective.

We found evidence that biomedical problem framing, catastrophizing, and medicallyoriented problem-solving behaviour were related. This partially supports the misdirectedproblem solving model (Eccleston & Crombez, 2007), which proposes that theseconcepts are related and interact to worsen the pain problem. The model implies thatif an individual who experiences pain worries extensively and frames the problem inbiomedical terms, this often results in medically oriented problem-solving behaviours,that is the individual relies on medical solutions and seeks health care to cure the pain.Our data indeed support the link between these concepts but the direction of therelations was somewhat different than the model proposes.

Our results indicate that the way people view their pain problem set the stagefor catastrophic worry. The mediation analyses supported catastrophizing as a mediatorbetween biomedical problem framing and medically oriented problem solving. This is notin line with the sequential order of the misdirected problem solving model, which impliesthat it is the other way around; that biomedical problem framing mediates the relationbetween catastrophizing and problem-solving behaviour. The current findings are rathercongruent with the expanded fear-anxiety-avoidance model (Asmundson et al., 2004),which implies that pain beliefs, such as ‘pain means harm or serious injury’, or in otherwords biomedical problem framing, might set the stage for catastrophizing and medicallyoriented problem-solving behaviours. In other words, the current findings indicate thatif patients view their pain as primarily a medically problem, then catastrophic thoughtsabout what is wrong might arise. In turn, the patients may seek medical health servicesas an attempt to solve the problem.

It is important to note that efforts at finding medical solutions for the pain are notadaptive or maladaptive per se. To try to find a solution for pain is generally relatedto better functioning (Keefe, Rumble, Scipio, Giordano, & Perri, 2004) but when it isdifficult to find a medical solution, such as in the case of spinal pain, a more adaptivereaction is to engage in other goals (Eccleston & Crombez, 2007). Based on our data,we cannot draw any conclusions about whether the repeated efforts for finding medicalsolutions for the pain problem were adaptive or not. There is, however, a relationbetween health care visits and poor function at the second assessment. On the onehand, this could be a sign that the efforts for finding medical solutions have not beensuccessful. On the other hand, it might also be that the patients seek more health carebecause of their poor function. A third possibility is that the contact with the health careprovider in fact has increased the patients’ feelings of helplessness, which in turn leadto further health care visits.

A related issue when trying to understand catastrophizing from a problem-solvingperspective is that we cannot tell whether catastrophizing is a precursor or also aconsequence of medically oriented problem-solving behaviour. The question of causalityis a well-known issue when examining bidirectional models; since they are circular itis difficult to tease out what drives what. To clarify that, experimental studies thatmanipulate these concepts are needed. The current design indeed has limitations when


examining mediating factors since both the independent factor and the mediator weremeasured at the same timepoint. Further, it would have been of large interest to followthese people from their very first pain episode to get more information about how theproposed feedback loops develop. Another option which would have strengthened ourconclusions would be to control for health care visits at baseline. However, as there isstill sparse empirical support for these models, the current findings are informative as afirst step and serve to set a direction for future research.

Another methodological limitation is that our findings rely solely on self-report data.An addition of an objective measure of health care visits could have verified the reliabilityof these self-reports, and therewith strengthened the results. To check the stability ofour data, however, we repeated the same analyses with the total number of health carevisits (as opposed to the number of different providers visited) as the outcome variable,and the results remained the same. A reservation should also be made concerningthe use of TSK to assess biomedical problem framing. There have been conflictingfindings concerning its factor structure (see e.g., Lundberg et al., 2004; Vlaeyen et al.,1995) and the measure might therefore not be considered as stable. However, to ourknowledge there are no established measures for assessing problem framing and whenscrutinizing the somatic focus subscale in the TSK, the questions have high face validityand were, therefore, considered as the best available proxy for biomedical problemframing.

The prospective design of this study is indeed one of the strengths. The time spanof 7 months allowed analyses of how these components interacted over time. Anotherstrength is that, apart from a relatively high mean age of 57, the sample appeared tobe representative of patients suffering from spinal pain problems, with moderate levelsof catastrophizing and disability. The level of somatic focus was also comparable withother samples suffering from pain problems (Roelofs et al., 2004). This denotes that thefindings might be generalizable to other patients with spinal pain problems.

The findings from the present study have implications theoretically as well asclinically. Theoretically, the link between the concepts included in the misdirectedproblem solving model was confirmed, but the data suggests that the directions aresomewhat different than the model implies. This might be important for refinement ofthis model. The findings also provide a framework for catastrophizing, and suggest onepossible path that links catastrophizing to poor outcome in patients, namely as a mediatorbetween problem framing and problem-solving behaviour. Clinically, this brings up theimportance of examining how patients with spinal pain view their problem. If pain isregarded as a definite signal of something medically wrong, it is vital for the clinicianto address catastrophic worry. Most clinicians do this by reassuring the patients thatnothing is medically wrong in their back. However, there is surprisingly little evidencethat reassurance is effective for calming patients (for a review, see Linton, McCracken, &Vlaeyen, 2008). General information is possibly needed, but it might not be enough forpatients with high levels of catastrophizing. Another important aspect in this context isvalidation; to express acceptance of the patient’s behaviour, feelings, and situation astrue (Fruzzetti, 2006). Presumably, a combination of reassurance in the form of concrete,specific information and a validating response from the clinician is needed, but furtherresearch needs to find out the best way to do this.

This study has contributed to the understanding of catastrophizing from a problem-solving perspective. In conclusion, catastrophizing was linked to how the pain problemwas framed and dealt with. More specifically, the data indicated that catastrophic worrymediates the relation between biomedical problem framing and medically oriented


problem-solving behaviour. Future studies would preferably examine the long-termconsequences of this problem-solving behaviour.

AcknowledgementsThe authors sincerely want to thank Maarten Van Zalk for helpful comments about thestatistical analyses.

ReferencesAsmundson, G. J., Norton, P. J., & Vlaeyen, J. W .S. (2004). Fear-avoidance models of chronic pain:

An overview. In G. J. Asmundson, J.W. S. Vlaeyen, & G. Crombez (Eds.), Understanding andtreating fear of pain (pp. 3–24). Oxford: Oxford University Press.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psycho-logical research: Conceptual, strategic, and statistical considerations. Journal of Personalityand Social Psychology, 51(6), 1173–1182. doi:10.1037/0022-3514.51.6.1173

Boersma, K., & Linton, S. J. (2005). How does persistent pain develop? An analysis of therelationship between psychological variables, pain and function across stages of chronicity.Behaviour Research and Therapy, 43, 1495–1507. doi:10.1016/j.brat.2004.11.006

Crombez, G., Eccleston, C., Van Hamme, G., & De Vlieger, P. (2008). Attempting to solve theproblem of pain: A questionnaire study in acute and chronic pain patients. Pain, 137, 556–563. doi:10.1016/j.pain.2007.10.020

Davey, G. C. L., & Levy, S. (1998). Catastrophic worrying: Personal inadequacy and a perseverativeiterative style as features of the catastrophizing process. Journal of Abnormal Psychology,107(4), 576–586. doi:10.1037/0021-843X.107.4.576

Eccleston, C., & Crombez, G. (2007). Worry and chronic pain: A misdirected problem solvingmodel. Pain, 132, 233–236. doi:10.1016/j.pain.2007.09.014

Fruzzetti, A. E. (2006). The high conflict couple. Oakland, CA: New Harbinger Publications.Goubert, L., Crombex, G., Van Damme, S., Vlaeyen, J. W. S., Bijttebier, P., & Roelofs, J. (2004).

Confirmatory factor analysis of the Tampa Scale for Kinesiophobia: Invariant two factormodel across low back pain patients and fibromyalgia patients. Clinical Journal of Pain, 20,103–110. doi:10.1097/00002508-200403000-00007

Jensen, I., & Linton, S. J. (1993). Coping Strategies Questionnaire (CSQ): Reliability of the Swedishversion of the CSQ. Cognitive Behaviour Therapy, 22(3), 139–145.

Keefe, F., Linton, S. J., & Lefebvre, J. (1992). The outcome evaluation questionnaire: Descriptionand initial findings. Cognitive Behaviour Therapy, 21(1), 19–33.

Keefe, F., Rumble, M., Scipio, C., Giordano, L., & Perri, L. (2004). Psychological aspects of persistentpain: Current state of the science. The Journal of Pain, 5(4), 195–211. doi:10.1016/j.jpain.2004.02.576

Kori, S. H., Miller, R. P., & Todd, D. D. (1990). Kinesiophobia: A new view of chronic pain behavior.Pain Management, 3 35–43.

Leeuw, M., Houben, R., Severeijns, R., Picavet, H. S. J., Schouten, E. G. W., & Vlaeyen, J. W. S.(2007). Pain-related fear in low back pain: A prospective study in the general population.European Journal of Pain, 11(3), 256–266. doi:10.1016/j.ejpain.2006.02.009

Linton, S. J. (1990). Activities of daily living scale for patients with chronic pain. Perceptual andMotor Skills, 71, 722. doi:10.2466/PMS.71.7.722-722

Linton, S. J., & Boersma, K. (2003). Early identification of patients at risk of developing a persistentback problem: The predictive validity of the Orebro musculoskeletal pain questionnaire.Clinical Journal of Pain, 19, 80–86. doi:10.1097/00002508-200303000-00002

Linton, S. J., & Hallden, K. (1998). Can we screen for problematic back pain? A screeningquestionnaire for predicting outcome in acute and subacute back pain. The Clinical Journalof Pain, 14(3), 209–215. doi:10.1097/00002508-199809000-00007

STUDY III


examining mediating factors since both the independent factor and the mediator weremeasured at the same timepoint. Further, it would have been of large interest to followthese people from their very first pain episode to get more information about how theproposed feedback loops develop. Another option which would have strengthened ourconclusions would be to control for health care visits at baseline. However, as there isstill sparse empirical support for these models, the current findings are informative as afirst step and serve to set a direction for future research.

Another methodological limitation is that our findings rely solely on self-report data.An addition of an objective measure of health care visits could have verified the reliabilityof these self-reports, and therewith strengthened the results. To check the stability ofour data, however, we repeated the same analyses with the total number of health carevisits (as opposed to the number of different providers visited) as the outcome variable,and the results remained the same. A reservation should also be made concerningthe use of TSK to assess biomedical problem framing. There have been conflictingfindings concerning its factor structure (see e.g., Lundberg et al., 2004; Vlaeyen et al.,1995) and the measure might therefore not be considered as stable. However, to ourknowledge there are no established measures for assessing problem framing and whenscrutinizing the somatic focus subscale in the TSK, the questions have high face validityand were, therefore, considered as the best available proxy for biomedical problemframing.

The prospective design of this study is indeed one of the strengths. The time spanof 7 months allowed analyses of how these components interacted over time. Anotherstrength is that, apart from a relatively high mean age of 57, the sample appeared tobe representative of patients suffering from spinal pain problems, with moderate levelsof catastrophizing and disability. The level of somatic focus was also comparable withother samples suffering from pain problems (Roelofs et al., 2004). This denotes that thefindings might be generalizable to other patients with spinal pain problems.

The findings from the present study have implications theoretically as well asclinically. Theoretically, the link between the concepts included in the misdirectedproblem solving model was confirmed, but the data suggests that the directions aresomewhat different than the model implies. This might be important for refinement ofthis model. The findings also provide a framework for catastrophizing, and suggest onepossible path that links catastrophizing to poor outcome in patients, namely as a mediatorbetween problem framing and problem-solving behaviour. Clinically, this brings up theimportance of examining how patients with spinal pain view their problem. If pain isregarded as a definite signal of something medically wrong, it is vital for the clinicianto address catastrophic worry. Most clinicians do this by reassuring the patients thatnothing is medically wrong in their back. However, there is surprisingly little evidencethat reassurance is effective for calming patients (for a review, see Linton, McCracken, &Vlaeyen, 2008). General information is possibly needed, but it might not be enough forpatients with high levels of catastrophizing. Another important aspect in this context isvalidation; to express acceptance of the patient’s behaviour, feelings, and situation astrue (Fruzzetti, 2006). Presumably, a combination of reassurance in the form of concrete,specific information and a validating response from the clinician is needed, but furtherresearch needs to find out the best way to do this.

This study has contributed to the understanding of catastrophizing from a problem-solving perspective. In conclusion, catastrophizing was linked to how the pain problemwas framed and dealt with. More specifically, the data indicated that catastrophic worrymediates the relation between biomedical problem framing and medically oriented


problem-solving behaviour. Future studies would preferably examine the long-termconsequences of this problem-solving behaviour.

AcknowledgementsThe authors sincerely want to thank Maarten Van Zalk for helpful comments about thestatistical analyses.

ReferencesAsmundson, G. J., Norton, P. J., & Vlaeyen, J. W .S. (2004). Fear-avoidance models of chronic pain:

An overview. In G. J. Asmundson, J.W. S. Vlaeyen, & G. Crombez (Eds.), Understanding andtreating fear of pain (pp. 3–24). Oxford: Oxford University Press.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psycho-logical research: Conceptual, strategic, and statistical considerations. Journal of Personalityand Social Psychology, 51(6), 1173–1182. doi:10.1037/0022-3514.51.6.1173

Boersma, K., & Linton, S. J. (2005). How does persistent pain develop? An analysis of therelationship between psychological variables, pain and function across stages of chronicity.Behaviour Research and Therapy, 43, 1495–1507. doi:10.1016/j.brat.2004.11.006

Crombez, G., Eccleston, C., Van Hamme, G., & De Vlieger, P. (2008). Attempting to solve theproblem of pain: A questionnaire study in acute and chronic pain patients. Pain, 137, 556–563. doi:10.1016/j.pain.2007.10.020

Davey, G. C. L., & Levy, S. (1998). Catastrophic worrying: Personal inadequacy and a perseverativeiterative style as features of the catastrophizing process. Journal of Abnormal Psychology,107(4), 576–586. doi:10.1037/0021-843X.107.4.576

Eccleston, C., & Crombez, G. (2007). Worry and chronic pain: A misdirected problem solvingmodel. Pain, 132, 233–236. doi:10.1016/j.pain.2007.09.014

Fruzzetti, A. E. (2006). The high conflict couple. Oakland, CA: New Harbinger Publications.Goubert, L., Crombex, G., Van Damme, S., Vlaeyen, J. W. S., Bijttebier, P., & Roelofs, J. (2004).

Confirmatory factor analysis of the Tampa Scale for Kinesiophobia: Invariant two factormodel across low back pain patients and fibromyalgia patients. Clinical Journal of Pain, 20,103–110. doi:10.1097/00002508-200403000-00007

Jensen, I., & Linton, S. J. (1993). Coping Strategies Questionnaire (CSQ): Reliability of the Swedishversion of the CSQ. Cognitive Behaviour Therapy, 22(3), 139–145.

Keefe, F., Linton, S. J., & Lefebvre, J. (1992). The outcome evaluation questionnaire: Descriptionand initial findings. Cognitive Behaviour Therapy, 21(1), 19–33.

Keefe, F., Rumble, M., Scipio, C., Giordano, L., & Perri, L. (2004). Psychological aspects of persistentpain: Current state of the science. The Journal of Pain, 5(4), 195–211. doi:10.1016/j.jpain.2004.02.576

Kori, S. H., Miller, R. P., & Todd, D. D. (1990). Kinesiophobia: A new view of chronic pain behavior.Pain Management, 3 35–43.

Leeuw, M., Houben, R., Severeijns, R., Picavet, H. S. J., Schouten, E. G. W., & Vlaeyen, J. W. S.(2007). Pain-related fear in low back pain: A prospective study in the general population.European Journal of Pain, 11(3), 256–266. doi:10.1016/j.ejpain.2006.02.009

Linton, S. J. (1990). Activities of daily living scale for patients with chronic pain. Perceptual andMotor Skills, 71, 722. doi:10.2466/PMS.71.7.722-722

Linton, S. J., & Boersma, K. (2003). Early identification of patients at risk of developing a persistentback problem: The predictive validity of the Orebro musculoskeletal pain questionnaire.Clinical Journal of Pain, 19, 80–86. doi:10.1097/00002508-200303000-00002

Linton, S. J., & Hallden, K. (1998). Can we screen for problematic back pain? A screeningquestionnaire for predicting outcome in acute and subacute back pain. The Clinical Journalof Pain, 14(3), 209–215. doi:10.1097/00002508-199809000-00007


Linton, S. J., McCracken, L. M., & Vlaeyen, J. W. S. (2008). Reassurance: Help or hinder in thetreatment of pain. Pain, 134, 5–8. doi:10.1016/j.pain.2007.10.002

Lundberg, M. K. E., Stymf, J., & Carlsson, S.V. (2004). A psychometric evaluation of the Tampa Scalefor Kinesiophobia: A physiotherapeutic perspective. Physiotherapy Theory and Practice, 20,121–133. doi:10.1080/09593980490453002

Mathews, A. (1990). Why worry? The cognitive function of anxiety. Behaviour Research andTherapy, 28(6), 455–468. doi:10.1016/005-7967(90)90132-3

Nachemson, A., Waddell, G., & Norlund, A. I. (2000). Epidemiology of neck and low back pain.In A. Nachemson & E. Jonsson (Eds.), Neck and back pain: The scientific evidence of causes,diagnosis, and treatment. Philadelphia: Lippincott Williams & Wilkins.

Preacher, K., & Hayes, A. (2008). Asymptotic and resampling strategies for assessing and comparingindirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879–891.doi:10.3758/BRM.40.3.879

Roelofs, J., Goubert, L., Peters, M. L., Vlaeyen, J. W. S., & Crombez, G. (2004). The Tampa Scalefor Kinesiophobia: Further examination of psychometric properties in patients with chroniclow back pain and fibromyalgia. European Journal of Pain, 8, 495–502. doi:10.1016/j.ejpain.2003.11.016

Roelofs, J., Sluiter, J. K., Frings-Dresen, M. H. W., Goossens, M., Thibault, P., Boersma, K., &Vlaeyen, J. W. S. (2007). Fear of movement and (re) injury in chronic musculoskeletal pain:Evidence for an invariant two-factor model of the Tampa Scale for Kinesiophobia across paindiagnoses and Dutch, Swedish, and Canadian samples. Pain, 131(1–2), 181–190. doi:10.1016/j.pain.2007.01.008

Rosentiel, K. (1983). Coping strategy questionnaire. Pain, 17, 33–43.Stewart, M., Harvey, S., & Evans, I. (2001). Coping and catastrophizing in chronic pain: A

psychometric analysis and comparison of two measures. Journal of Clinical Psychology, 57(1),131–138. doi:10.1002/jclp.10227

Sullivan, M. J. L., Thorn, B., Haythornthwaite, J. A., Keefe, F. J., Martin, M., Bradley, L. A., &Lefebvre, J. C. (2001). Theoretical perspectives on the relation between catastrophizing andpain. The Clinical Journal of Pain, 17, 52–64. doi:10.1097/00002508-200103000-00008

Vlaeyen, J. W. S., Kole-Snijders, A. M. J., Boeren, R. G. B., Rotteveel, R. G. B., Ruesink,R., & Heuts, P .H. T. G. (1995). The role of fear of movement/(re)injury in pain disability.Journal of Occupational Rehabilitation, 5, 235–252. doi:10.1007/BF02109988

Waddell, G., Aylward, M., & Sawney, P. (2002). Back pain, incapacity for work and socialsecurity benefits: An international literature review and analysis. London: Royal Society ofMedicine Press.

Westman, A., Linton, S. J., Ohrvik, J., Wahlen, P., & Leppert, J. (2008). Do psychosocial factorspredict disability and health at a 3-year follow-up for patients with non-acute musculoskeletalpain? European Journal of Pain, 12, 641–649. doi:10.1016/j.ejpain.2007.10.007

Wideman, T. H., Adams, H., & Sullivan, M. J. L. (2009). A prospective sequential analysis of thefear-avoidance model of pain. Pain, 145, 45–51. doi:10.1016/j.pain.2009.04.022

Received 29 March 2011; revised version received 3 June 2011

Publications in the series Örebro Studies in Psychology

0.* Andersson [Andershed], Anna-Karin, The Rhythm of Adolescence – Morningness – Eveningness and Adjustment from a Developmental Perspective. 2001.

1. Andershed, Henrik, Antisocial Behavior in Adolescence – The Role of Individual Characteristics. 2002.

2. Trost, Kari, A new look at parenting during adolescence: Reciprocal interactions in everyday life. 2002.

3. Jensen, Eva, (Mis)understanding and Learning of Feedback Relations in a Simple Dynamic System. 2004.

4. Wester Herber, Misse, Talking to me? – Risk Communication to a diverse Public. 2004.

5. Boersma, Katja, Fear and avoidance in the development of a persistent musculoskele tal pain problem. Implications for secondary prevention. 2005.

6. Jansson, Markus, Insomnia: Psychological Mechanisms and Early Intervention. A Cognitive-Behavioral Perspective. 2005.

7. Lindblom, Karin, Utbrändhet i normalbefolkningen. Arbets- och individfaktorers relation i utvecklingen mot eller återhämtning från utbrändhet. 2006. (Vetenskaplig uppsats)

8. Almqvist, Lena, Children’s Health and Developmental Delay: Positive Functioning in Every-day Life. 2006.

9. Johansson, Peter, Understanding psychopathy through the study of long-term violent offenders. 2006.

10. Persson, Andreas, Leisure in Adolescence: Youth’s activity choices and why they are linked to problems for some and not others. 2006.

11. Ojala, Maria, Hope and worry: Exploring young people’s values, emotions, and behavior regarding global environmental problems. 2007.

12. larsson, Mats, Human Iris Characteristics as Biomarkers for Personality. 2007.

* Finns sedan tidigare utgiven i serien "Örebro Studies".

13. pakalniskiene, Vilmante, Harsh or Inept Parenting, Youth Characteristics and Later Adjustment. 2008.

14. skoog, therése, On the developmental significance of female pubertal timing. 2008.

15. brav, Agneta, Industrial Work Groups. The Impact of Job Design, Leader Support and Group Processes on Initiative and Self-organization. 2008.

16. besic, nejra, At First Blush: The Impact of Shyness on Early Adolescents’ Social Worlds. 2009.

17. persson, stefan, Adolescents’ role in democratic “parenting”. 2009.

18. Lillvist, Anne, The applicability of a functional approach to social competence in preschool children in need of special support. 2010.

19. Kakihara, Fumiko, Incorporating Adolescents’ Interpretations and Feelings about Parents into Models of Parental Control. 2010.

20. MacDonald, Shane, Stress, musculoskeletal pain and insomnia. Distinct difficulties or variations of the same problem? 2011.

21. Flink, Ida, Stuck in Mind: The role of Catastrophizing in Pain. 2011

stuck in mind - diva portaloru.diva-portal.org/smash/get/diva2:445438/fulltext01.pdfthank you steven...

Documents