predictors of response to problem-solving treatment of depression in primary care

17
BEHAVIOR THERAPY 33,511-527,2002 Predictors of Response to Problem-Solving Treatment of Depression in Primary Care MARK T. HEGEL JAMES E. BARRETT Dartmouth Medical School JOHN E. CORNELL South Texas Veterans Affairs Health Care System, University of Texas Health Science Center, San Antonio THOMAS E. OXMAN Dartmouth Medical School Problem-solving treatment of depression is a brief intervention specifically designed for primary care (PST-PC). To date, predictors of an optimal response to PST-PC have not been studied. In primary care, knowing such factors is essential for proper triage decisions. Patient, therapist, and process variables were evaluated for their ability to predict remission of minor depression or dysthymia in patients treated with PST-PC. The most salient predictors of remission were the ability to understand the PST-PC rationale and to apply the PST-PC procedure in early treatment sessions, having a cognitive-behavioral therapist, and, for dysthymia, having a lower depres- sion severity level at baseline. These results provide preliminary evidence of some factors associated with an optimal response to PST-PC and also present challenges for the ability to broadly disseminate the intervention. Modifications to the existing PST-PC training program and directions for future research are discussed. A number of studies have demonstrated the central role of the primary care medical setting in detecting and managing mental disorders (Regier et al., Supported in part by grants from the John A. Hartford Foundation of New York and the John D. and Catherine T. MacArthur Foundation. We thank Anjana Sengupta for her helpful comments on earlier versions of the manuscript and Cynthia Hewitt who assisted with manuscript prepara- tion. This material is the result of work supported in part with resources and the use of facilities at South Texas Veterans Health Care System. The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs. Address correspondence to Mark T. Hegel, Department of Psychiatry, Dartmouth-Hitchcock Medical Center, One Medical Center Drive, Lebanon, NH 03756; e-mall: mark.t.hegel@dartmouth. edu. 511 005-7894/02/0511~)52751.00/0 Copyright 2002 by Associationfor Advancement of Behavior Therapy All rights for reproductionin any formreserved.

Upload: mark-t-hegel

Post on 13-Sep-2016

214 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Predictors of response to problem-solving treatment of depression in primary care

BEHAVIOR THERAPY 33,511-527,2002

Predictors of Response to Problem-Solving Treatment of Depression in Primary Care

MARK T. HEGEL

JAMES E . BARRETT

Dartmouth Medical School

JOHN E . CORNELL

South Texas Veterans Affairs Health Care System, University o f Texas Health Science Center, San Antonio

THOMAS E . OXMAN

Dartmouth Medical School

Problem-solving treatment of depression is a brief intervention specifically designed for primary care (PST-PC). To date, predictors of an optimal response to PST-PC have not been studied. In primary care, knowing such factors is essential for proper triage decisions. Patient, therapist, and process variables were evaluated for their ability to predict remission of minor depression or dysthymia in patients treated with PST-PC. The most salient predictors of remission were the ability to understand the PST-PC rationale and to apply the PST-PC procedure in early treatment sessions, having a cognitive-behavioral therapist, and, for dysthymia, having a lower depres- sion severity level at baseline. These results provide preliminary evidence of some factors associated with an optimal response to PST-PC and also present challenges for the ability to broadly disseminate the intervention. Modifications to the existing PST-PC training program and directions for future research are discussed.

A number of studies have demonstrated the central role of the primary care medical setting in detecting and managing mental disorders (Regier et al.,

Supported in part by grants from the John A. Hartford Foundation of New York and the John D. and Catherine T. MacArthur Foundation. We thank Anjana Sengupta for her helpful comments on earlier versions of the manuscript and Cynthia Hewitt who assisted with manuscript prepara- tion. This material is the result of work supported in part with resources and the use of facilities at South Texas Veterans Health Care System. The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs.

Address correspondence to Mark T. Hegel, Department of Psychiatry, Dartmouth-Hitchcock Medical Center, One Medical Center Drive, Lebanon, NH 03756; e-mall: mark.t.hegel@dartmouth. edu.

511 005-7894/02/0511~)52751.00/0 Copyright 2002 by Association for Advancement of Behavior Therapy

All rights for reproduction in any form reserved.

Page 2: Predictors of response to problem-solving treatment of depression in primary care

512 HEGEL ET AL.

1993), particularly depression (Barrett, Barrett, Oxman, & Gerber, 1988; Broadhead, Blazer, George, & Tse, 1990; Katon & Schulberg, 1992; Wil- liams, Kerber, Mulrow, Medina, & Aguilar, 1995). The majority of persons with depression are treated in primary care due to patient preference, the lack of referral options, barriers created by a separate mental health care system, and stigma associated with referral to the mental health sector (Brody, Khaliq, & Thompson, 1997; Fortney, Rost, & Zhang, 1998; Solberg, Korsen, Oxman, Kischer, & Bartels, 1999). Pharmacological intervention is the prin- cipal treatment used for depression in this setting, but at times antidepres- sants are contraindicated or are ineffective (Katon et al., 1994; Kendrick, 1996; Williams et al., 1999). In addition, for minor depression, the most com- mon type of depressive disorder in primary care (Williams et al., 1995), the effectiveness of antidepressants remains uncertain (Katon et al., 1996).

Even when medications are indicated, many primary care patients prefer counseling over medications (Brody et al., 1997; Priest, Vize, Roberts, & Tylee, 1996; Scott & Freeman, 1992). In 1993 the USPHS Agency for Health Care Policy and Research (AHCPR) Depression Guideline Panel (AHCPR, 1993) endorsed the value of psychological treatments in treating depression but emphasized the paucity of data regarding these interventions when applied to depressed primary care patients.

Psychological interventions used in primary care are designed for the spe- cial demands of the primary care setting (Arean, Hegel, & Reynolds, 2001). Because of the high volume of patients and brief appointment times, these interventions are designed to be practical and brief in session length and duration of treatment. They tend to focus on psychoeducation, self-help strat- egies, and coping skills training. Ideally, they should not depend upon mental health professionals for administration as this limits the ability to broadly dis- seminate the intervention.

A problem-solving model of treatment designed for primary care was adapted from traditional cognitive-behavioral interventions for the mental health sector (e.g., Hawton & Kirk, 1989; Nezu, Nezu, & Perri, 1989) by a group of investigators in the United Kingdom (Catalan et al., 1991). To dis- tinguish this version of problem-solving treatment from traditional mental health models, we refer to it as Problem-Solving Treatment for Primary Care (PST-PC).

An overview of PST-PC, including its procedures and training program, has been published elsewhere (Hegel, Barrett, & Oxman, 2000). Its major tenets are briefly reviewed here. PST-PC for depression has three main goals. First, the patient is educated about the relationship between depressive symp- toms and problems in living and the attending rationale for learning adaptive problem-solving skills; second, the patient's problems are delineated through spontaneous report and questioning about key areas of living (e.g., finances, relationships, work problems); and third, an attempt is made to solve the problems in a structured way using the PST-PC intervention. The intervention consists of seven stages: defining and clarifying the problem, establishing a

Page 3: Predictors of response to problem-solving treatment of depression in primary care

PREDICTORS OF RESPONSE TO PST-PC FOR DEPRESSION 513

realistic goal, generating multiple alternative solutions, implementing decision- making guidelines, choosing a solution, implementing the solution, and eval- uating the outcome. Sessions last approximately 1 hour for the first visit and 30 minutes for each subsequent visit, for a total of four to six visits. An effort is made to implement the entire problem-solving intervention for at least one problem each session.

In order to make it suitable for the demands of the primary care setting (e.g., high patient volume and limited staff resources), PST-PC was designed differently from traditional problem-solving therapies. PST-PC sessions fit into 30-minute clinic visits (versus 1- to 2-hour visits) and its total length of treatment is 4 to 6 sessions (versus 10 to 15 sessions). In order to condense the protocol and to make it more feasible for primary care staff such as nurses to learn and administer, PST-PC also deletes the "problem-orientation train- ing" that is emphasized in traditional problem-solving therapy models (e.g., Nezu et al., 1989). In addition, rather than teach the problem-solving stages in a step-by-step fashion over the course of treatment visits, the entire PST- PC model is taught and put into practice during the very first session. The reduced number of treatment visits and the consequently reduced opportunity to practice the problem-solving stages independently is made up for by the increased frequency of practice with the entire model across all sessions. In randomized controlled studies, this manual-based intervention has been shown to be equivalent to antidepressant medication for major depression in primary care as well as being effective for other common emotional syn- dromes seen mostly in primary care settings (Catalan et al., 1991; Mynors- Wallis, Davies, Grey, Barbour, & Gath, 1997; Mynors-Wallis, Gath, Day, & Baker, 2000; Mynors-Wallis, Gath, Lloyd-Thomas, & Tomlinson, 1995).

Although successful in proving its efficacy, the previous studies made no attempt to examine variables that may predict patient response to PST-PC. Triage decisions must rely on the best data available for what types of prob- lems and patients are best suited for a particular course of treatment. Whether to manage the patient in primary care or to refer that particular patient to the specialty sector (mental health or otherwise) is a crucial decision for the pri- mary care provider. Because of the brevity of PST-PC and its potential reli- ance upon non-mental health professionals for its delivery, the intervention may require specific traits on the part of the patient or the therapist. For example, higher education level, activation for homework activities, or a lower level of depression severity may be necessary patient characteristics in order for the intervention to be effective. For the therapist, it is not clear what level of training or experience is necessary and this has direct relevance for the ability to broadly disseminate the intervention. Especially in primary care, where triage decisions must be made regarding referral to the specialty sector, understanding the salient predictors of response to this abbreviated form of treatment would be very helpful.

Traditional cognitive-behavioral therapies (from which PST-PC is derived) are associated with several strong response predictors. These include home-

Page 4: Predictors of response to problem-solving treatment of depression in primary care

514 HEGEL ET AL.

work compliance (Bums & Nolen-Hoeksema, 1991; Burns & Spangler, 2000; Fennell & Teasdale, 1987) and the belief in the rationale and credibility of the treatment (Fennell & Teasdale; Frank & Frank, 1991; Kazdin & Krouse, 1983; Morrison & Shapiro, 1987; Nan, Caputo, & Borkovec, 1974; Rokke, Carter, Rehm, & Veltum, 1990). For interpersonal therapy, the patient's beliefs about the causes of depression (e.g., internal versus external causes, biological versus psychological, etc.) are predictive of outcome (Spanier, Frank, McEachran, Grochnocinski, & Kupfer, 1996).

Although serving as a valuable starting point, these findings from the spe- cialty mental health sector have limitations for generalizing to primary care. The studies have been conducted with patients suffering major depression treated in the psychiatric setting. Depressed patients treated in primary care are categorically and qualitatively different. The majority of primary care depressive disorders are less severe conditions such as minor depression or dysthymia (Barrett et al., 1988; Cole & Bellavance, 1997; Katon & Schulberg, 1992; Sireling, Freeling, Paykel, & Rao, 1985). Also, the research is largely for traditional mental health models of psychotherapy or cognitive-behavioral therapy. These treatments take longer and are dependent on mental health pro- fessionals for delivery, making it less likely to be administered in primary care.

A recent study in the U.S. examined outcomes for minor depression and dysthymia for primary care patients treated with medication or PST-PC (Bar- rett et al., 1999; Barrett et al., 2001; Williams et al., 2000). This report focuses on those patients treated with PST-PC. We hypothesized that some specific patient variables (depression severity and belief in a psychological cause of their depression), therapist variables (cognitive-behavioral versus a non-cognitive-behavioral theoretical orientation), and treatment process vari- ables (homework compliance and the ability to understand the rationale for PST-PC and to apply it independently) would predict response to PST-PC. Because this was the first large-scale study of PST-PC for minor depression and dysthymia in a primary care sample conducted in the U.S., we also exam- ined demographic and diagnostic variables (age, gender, education level, and the diagnoses of minor depression versus dysthymia).

We also hypothesized that certain interactions would mediate the effects of particular variables. Specifically, we hypothesized that having a CBT thera- pist for patients with lower education would confer an advantage for this group over patients with lower education having a non-CBT therapist (due to our assumption of a higher competency level for CBT therapists), whereas a higher baseline depression severity level would confer a disadvantage toward remission. In addition, we hypothesized that the specific interaction set of understanding the rationale and applying the PST-PC process and homework compliance would offset potentially deleterious effects of other variables.

It is important to note that the analyses in this study were designed for prognostic purposes rather than prescriptive purposes. That is, the intent was to identify variables associated with efficacious PST-PC for the primary care population studied (What patients are helped most by this treatment?).

Page 5: Predictors of response to problem-solving treatment of depression in primary care

PREDICTORS OF R E S P O N S E TO PST-PC FOR D E P R E S S I O N 515

Prognostic analyses stand in contrast to predictive analyses, which require a comparison across two or more interventions in order to identify variables assisting with optimal treatment selection or patient-treatment "matching" (What treatment would be most helpful for this patient?) (Hollon & Najavits, 1988). In primary care, the professionals perhaps most likely to administer PST-PC are medical nurses or medical social workers. They may be able to be trained in one very pragmatic model of therapy, such as PST-PC, but it is unlikely that they can be trained to deliver a wide variety of therapeutic alter- natives. If a more sophisticated or tailored form of therapy is required, the patient will more than likely be referred to the specialty mental health sector for this. Thus, to assist in the triage process, the most important question to answer is whether this particular patient is likely to benefit from PST-PC as delivered in the primary care setting (What are the best prognostic indicators for the success of PST-PC?).

Method The analyses described are based on combined data from two parallel multi-

site randomized placebo controlled trials with two age groups (age 60 +; 18 to 59) conducted in primary care. The studies compared PST-PC and an anti- depressant medication for minor depression and dysthymia. Our methods have been described in detail elsewhere (Barrett et al., 1999) and are summa- rized here. The institutional review boards for each participating site approved the study and all participants provided informed consent.

Patients and Setting

Patients aged 18 and older were recruited through referral and screening in four cities at community, Veterans Affairs, and academic-affiliated primary care clinics. The four participating centers were chosen for geographic diver- sity and diversity of clinical populations. To be eligible, patients had to meet DSM-IV criteria for dysthymia or minor depression (American Psychiatric Association, 1987) and score of 10 or greater on the 17-item Hamilton Depres- sion Rating Scale (HAMD; Hamilton, 1960). Diagnoses were made using the PRIME-MD, a diagnostic instrument designed for use in primary care (Spitzer, Williams, & Kroenke, 1994). The HAMD and PRIME-MD were adminis- tered by experienced psychologists or psychiatrists who were trained to com- petency criteria prior to study initiation and who followed the same patients throughout the course of the study.

Patients were excluded for the following psychiatric diagnoses and clinical conditions: major depression, psychosis, schizophrenia or schizo-affective disorder, bipolar affective disorder, alcohol or other substance abuse within the past 6 months, antisocial personality disorder, borderline personality dis- order, serious suicidal risk, moderate or severe cognitive impairment (Fol- stein Mini-Mental State score -<23; Folstein, Folstein, & McHugh, 1975), or medical illness with a prognosis of less than 6 months to live. Patients in cur- rent mental health treatment were also excluded.

Page 6: Predictors of response to problem-solving treatment of depression in primary care

516 H E G E L ET AL.

Procedures

Patients were randomly assigned to clinical management with placebo or antidepressant medication (paroxetine hydrochloride), or to PST-PC. All patients were scheduled for six treatment sessions occurring over 11 weeks. Treatment sessions took place in the primary care setting. For patients assigned to PST-PC, visits occurred at weeks 1,2, 4, 6, 8, and 10. Antidepressant med- ication was prohibited.

PST-PC Therapists There were 11 PST-PC therapists: seven Ph.D. clinical psychologists and 4

master's-prepared therapists. Previous research had demonstrated that PST- PC could be effectively administered by non-mental health personnel such as nurses (Mynors-Wallis et al., 2000; Mynors-Wallis et al., 1997). However, due to the limited availability of the initial PST-PC trainer from the U.K. (Laurence Mynors-Wallis) and the consequent need to shorten the training period, it was decided to use mental health professionals for this first test of PST-PC in the U.S. The first author (MTH), having been trained by Mynors- Wallis, subsequently trained, either partially or in total, the remainder of ther- apists (approximately 50% of total training).

All therapists received PST-PC training that consisted of a short theoretical course, role-playing in clinical scenarios, watching a training videotape, and reading a treatment manual. Subsequently, therapists treated four practice patients for six sessions each. During the study, PST-PC sessions were audio- taped and a small number of randomly chosen treatment sessions were reviewed by MTH to facilitate discussion on monthly group conference calls. A collegial group consultation model was utilized as opposed to a super- vision model. No additional supervision was provided.

Sample Composition In the original study, 218 patients were randomized to PST-PC (Barrett et al.,

2001; Williams et al., 2000). The sample for the current analyses consisted of 179 patients (82% of the original sample) who had received an adequate expo- sure to PST-PC (four or more treatment sessions). Patients were designated as either responders or nonresponders based on remission status (HAMD -< 6 at Week 11; Frank et al., 1991). Of the 179 PST-PC patients who received ade- quate treatment exposure, 94 (52%) achieved remission status at Week 11.

Variables Examined

The predictor variables examined fell into three major categories: treat- ment process measures, therapist therapeutic orientation, and patient measures. We chose to evaluate variables with previous empirical support but that also are likely to be modifiable and/or practical to assess in primary care, such as demographic variables, homework compliance, treatment credibility and beliefs regarding the causes of depression.

Page 7: Predictors of response to problem-solving treatment of depression in primary care

PREDICTORS OF R E S P O N S E TO PST-PC FOR D E P R E S S I O N 517

Treatment Process Measures

Therapists completed ratings of treatment process variables at each treat- ment visit, from which the following two single-item measures were chosen for analysis:

Homework compliance (range = 0 to 10; 0 = did none of the homework, 10 = did all of the homework). Therapists based their ratings on the percent- age of homework tasks completed. For example, if assigned two homework tasks and the patient only completed one of these (50% completed), the rating would be a 5.

Ability to understand the rationale and apply the PST-PC process (Rationale/ Apply PST-PC) (range = 0 to 10; 0 = does not understand or apply theprocess at all, 10 = understands and applies the process completely). Therapists esti- mated the score based on the patient's ability to track the PST-PC sequence in session, verbalize a comprehension of the PST-PC model, competently uti- lize the PST-PC worksheet, and apply the technique independently between sessions. As an aid to this rating, patients were prompted every session to independently recall the seven PST-PC stages and were asked to gradually take on more responsibility for guiding the problem-solving process as treat- ment sessions progressed.

The ratings on these items were averaged (Sessions 2 through 4 for home- work compliance and Sessions 1 through 3 for Rationale/Apply PST-PC) and used as the predictive variable for outcome. By using data from early in treat- ment, we hoped to reduce the likelihood of therapist ratings being biased by the patient's response to treatment and thereby increase the likelihood of detecting a causal rather than associative relationship between the predictor variables and therapeutic outcome (Bums & Spangler, 2000). By averaging ratings across three sessions, our intent was to produce a stable index of com- pliance and degree of understanding on the part of the patient.

Therapist Therapeutic Orientation

Therapeutic orientation for therapists was assessed prior to study initiation. Therapists completed a background questionnaire, which included a question asking them to indicate their primary therapeutic orientation ' (i.e., "What therapeutic model do you primarily use to guide your practice of psychother- apy?") Choices included cognitive-behavioral, psychodynamic, humanistic, family systems, eclectic, and other. There were six therapists with a cogni- tive-behavioral (CBT) background and five with non-CBT backgrounds (two psychodynamic, two humanistic, and one eclectic). A CBT orientation versus a non-CBT orientation (CBT therapist) was included as a dichotomous vari- able. There was no difference in total years of therapy experience between the two groups (mean CBT = 9.83, SD = 12.14, range = 1 to 34; mean non-CBT = 12.80, SD = 10.75, range = 1 to 26; t = .43, df = 9, us).

Therapist adherence and competency ratings were completed for occasional treatment sessions over the course of the study by the first author (45 sessions across 10 therapists from a potential 1,074 sessions across 11 therapists) for the

Page 8: Predictors of response to problem-solving treatment of depression in primary care

518 HEGEL ET AL.

purpose of facilitating a monthly group conference call among therapists. All therapists began the study above the designated minimal adherence/competence levels and, based on the limited data available, all therapists were judged to have improved over the course of the study (Hegel et al., 2000). It would have been desirable to evaluate the relationship between therapist adherence/competence and treatment outcome. However, this was not evaluated for several reasons. First, the number of sessions rated were relatively small compared to the actual number completed and were not chosen in a random fashion (i.e., ther- apists were aware that the third session would be sent for possible review). Thus, we were not confident that these sessions accurately represented a typ- ical performance for the therapists. Second, MTH was a therapist on the study himself and treated approximately 15% of study patients receiving at least four PST-PC sessions. The analyses reported in this study are based on this entire sample of patients. MTH did not rate his own performance (nor should he have) and therefore data were not available for all therapists and patients in the study. Finally, because MTH was a member of the therapist team and was not a formal supervisor, we were concerned that his ratings may have been biased.

Patient Measures

Diagnosis. Minor depression or dysthymia were predictor variables from the original study. Each is included as a dichotomous variable.

Depression severity level. A 20-item, self-report depression scale consist- ing of the 13 items from the Hopkins Symptom Checklist Depression Scale assessed depression severity and seven additional depression-related items were added to increase responsiveness (HSCL-D-20; Lipman, Covi, & Shapiro, 1979; Katon, von Korff, & Lin, 1995). Items were averaged, yielding scores from 0 to 4, with higher scores indicating more severe symptoms. The scale was administered at the baseline assessment. The decision to use the HSCL- D-20 as the covariate for depression severity rather than the scale used to determine inclusion in the study (HAMD --- 10) was based on the fact that a HAMD score --- 10 artificially restricted these scores, thus reducing power for analyses.

Patient Attitudes and Beliefs Scale (PAB). The PAB is a 20-item, self-report measure developed for the NIMH Treatment of Depression Collaborative Research Program (Elkin et al., 1989; Spanier et al., 1996). The PAB was designed to assess patient beliefs about the causes of depression along three factors: biological, cognitive, and external. Patients are asked to indicate to what degree they agree or disagree with a statement (e.g., "An imbalance of certain substances in my brain is a cause of my problems"--biological; "Pes- simistic attitudes about many things are a cause of my problems"--cognitive; "Marital or family problems are a cause of my problems" -- external) using a 7-point Likert format (0 to 6; 0 = very strongly disagree, 6 = very strongly agree).

A two-factor exploratory principal factor solution with a promax rotation

Page 9: Predictors of response to problem-solving treatment of depression in primary care

PREDICTORS OF RESPONSE TO PST-PC FOR DEPRESSION 519

accounted for 91.1% of the common variance among the PAB items in our sample. Psychological beliefs and external causes items loaded on a single psychological causes factor that accounted for 75.6% of the common vari- ance and biological causes loaded on the second factor. We used these two factors (psychological cause and biological cause) in our predictive model.

Age, gender, and education. In the original study the sample was, by design, two-thirds age 60 and over, offering an opportunity to study the impact of older age on response to treatment. Age (60+; 18 to 59) was added as a dichotomous variable. Gender and education (->13 years versus <13 years) were also included as dichotomous variables.

Results Relationships between 11-week remission status and each predictive vari-

able were tested via chi-square tests for dichotomous variables and students' t test for continuous data (Table 1). The following variables showed a signifi- cant relationship (p < .05) to remission status at 11 weeks: higher education; patient belief in a biological cause of depression; better homework compli- ance; better ability to understand the rationale and apply PST-PC; and having a CBT therapist. Variables that showed a trend toward a relationship with remission at Week 11 (p --< .10) were lower baseline depression severity and lesser belief in a psychological cause of depression. For diagnosis, age, and gender there was no relationship to 11-week remission status.

Variables with relationships to I 1-week remission status of p <- .10 were included in logistic regression models as potential predictors of 11-week remission status. Although they did not distinguish differences in the bivari- ate analysis, gender and diagnosis were retained in the model because they reflected recruitment differences across sites and relevant aspects of the

T A B L E 1 RELATIONSHIP OF PREDICTOR VARIABLES TO REMISSION STATUS (STUDENT'S t TEST)

Remitted at Week 11 (HAMD --< 6)

Yes No Predictor Variable (n = 94) (n = 85) p-value

Age (-->60 years) (%) 47.8 Gender (Female) (%) 55.2 Education (-> 13 years) (%) 58.8 Diagnosis (Dysthymia) (%) 53.0 Baseline depression level: M (SD) 1.34 (0.62) Biological cause: M (SD) 2.99 (0.89) Psychological cause: M (SD) 3.48 (0.89) Homework compliance: M (SD) 7.50 (2.16) Rationale/apply PST-PC: M (SD) 7.20 (2.60) CBT therapist (%) 60.3

52.2 44.8 41.2 47.0

1.53 (0.73) 3.41 (1.11) 3.71 (.91) 6.45 (2.67) 5.75 (2.97)

39.7

.9975

.4885

.0517

.8835

.0591

.0067

.0844

.0054

.0014

.0006

Page 10: Predictors of response to problem-solving treatment of depression in primary care

520 HEGEL ET AL.

sampling design. A goodness-of-fit test showed that the model adequately fit the data.

In analyzing our predicted interaction effects, only the interactions involv- ing baseline severity contributed significantly to the model (p = .039). Base- line severity interacted with homework compliance and diagnosis to predict remission, and therefore these interaction terms were introduced into the model.

Upon reintroducing into the model the two interaction terms of baseline severity with diagnostic group and baseline severity with homework compli- ance, only the baseline severity and diagnostic group interaction remained significant (Odds Ratio = 0.274; p < .05). Other significant predictors of remission (p < .05) included rationale/apply PST-PC (Odds Ratio = 1.183; p < .05) and CBT therapist (Odds Ratio = 3.936; p < 0.01). (See Table 2.)

To understand the nature of the interaction between severity of depressive symptoms and diagnosis, we examined the remission rates for three levels of baseline severity by diagnosis. Patients scoring in the lower third of the dis- tribution were assigned to the low severity group, patients scoring in the mid- dle third of the distribution were assigned to the moderate severity group, and patients scoring in the upper third of the distribution were assigned to the high severity group. Table 3 shows the significant interaction effect observed for baseline severity and dysthymia in which dysthymics of low severity were most likely to remit with PST-PC (72.2%). Moderate and severely depressed dysthymics were less likely to enter remission (48.3% and 36.4%, respec- tively); chi-square (2, N = 98) = 9.2675, p = .0097. There was no effect of

TABLE 2 LOGISTIC REGRESSION FOR PREDICTORS OF REMISSION AND INTERACTION TERMS

INVOLVING BASELINE SEVERITY

Standard Odds Predictor Coefficient Error Ratio 95% C.I.

Intercept - 1.564 2.026 Age (-->60 years) 0.086 0.434 1.090 0.465 2.553 Gender (Female) 0.115 0.397 1.122 0.515 2.444 Education (->13 years) 0.059 0.423 1.061 0.463 2.431 Diagnosis (Dysthymia) 1.900 0.933 6.682 1.073 41.608 Baseline severity - 1.393 1.086 0.248 0.030 2.087 Biological cause -0.121 0.226 0.887 0.569 1.382 Psychological cause 0.082 0.252 1.085 0.663 1.777 Homework compliance -0.088 0.261 0.916 0.612 1.372 Rationale/apply PST-PC 0.168"* 0.079 1.183 1.014 1.380 CBT therapist 1.370*** 0.538 3.936 1.370 11.306 Severity × Diagnosis -1.295"* 0.612 0.274 0.082 0.909 Severity × Compliance 0.198 0.136 1.219 0.934 1.592

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

Page 11: Predictors of response to problem-solving treatment of depression in primary care

PREDICTORS OF RESPONSE TO PST-PC FOR DEPRESSION 521

T A B L E 3 REMISSION STATUS STRATIFIED BY DIAGNOSTIC GROUP AND BASELINE LEVEL

OF DEPRESSION SYMPTOM SEVERITY

Remitted at Week 11 (HAMD -< 6) Depression Diagnostic Group Severity Yes No p-value

Dysthymia High b 36.4% 63.6% Moderate b 48.3 % 51.7 % Low a 72.2% 27.8%

Minor depression High s 53.8% 46.2% Moderate a 51.7 % 48.3 % Low a 47.8% 52.2%

.0097

.9136

Note. Differing superscripts denote significant differences (chi-square).

severity on remission status for minor depression (high = 53.8%; moderate = 51.7%; low = 47.8%); chi-square (2, N = 78) = 0.1806,p = .9136.

Discussion Our findings showed that three variables most strongly predicted remission

in response to PST-PC for minor depression or dysthymia treated in primary care. These predictors were the ability of the patient to understand the ratio- nale and apply the PST-PC procedure early in treatment, receiving treatment from a CBT therapist, and lower depression severity levels for dysthymics at the start of treatment.

Those patients who better understood the rationale and were able to apply the problem-solving procedure early in treatment were more likely to achieve remission. These findings are consistent with other treatment studies for depression, for psychotherapy and pharmacotherapy interventions alike, in which approximately 50% of improvement occurs during the first few weeks of treatment, often before the so-called "active" components of treatment (e.g., homework tasks) are applied or have time to take effect (Ilardi & Craig- head, 1994). Based on this finding, we recommend evaluating the patient's ability to understand and apply PST-PC in the first few sessions and, if found lacking, consider the option of pharmacotherapy or a referral for a more tailored psychotherapy in the mental health sector. Whether these latter inter- ventions will be effective for this population cannot be answered from the current study and will require further investigation.

Compliance with specific homework tasks and education level were not as strong independent predictors of remission, although bivariate analyses repli- cated the findings from previous research for their role in predicting response to CBT interventions. A surprising finding from these analyses was that a pri- mary belief in a biological cause of depression did not mitigate the likelihood of responding to PST-PC, and the primary belief in a psychological cause of

Page 12: Predictors of response to problem-solving treatment of depression in primary care

522 H E G E L ET AL.

depression did not confer an advantage for responding to PST-PC. Therefore, selecting patients to receive, or not receive, PST-PC based on their beliefs about the etiology of their depression does not seem crucial.

Having a CBT therapist was associated with a higher likelihood of achiev- ing remission. As a manual-based behavioral skills intervention, it is natural to expect that a CBT therapist would be more effective in delivering the PST- PC intervention than a non-CBT therapist. However, previous studies of PST-PC from the U.K. showed that nurses and general practitioners could effectively deliver PST-PC for major depression (Mynors-Wallis et al., 2000; Mynors-Wallis et al., 1997; Mynors-Wallis et al., 1995). It is therefore of concern that one possible implication from our findings is that highly trained CBT therapists are necessary to deliver PST-PC in primary care.

There are two likely explanations for our findings. First, the non-CBT ther- apists may have needed a more closely supervised training experience and follow-up supervision. Mynors-Wallis and colleagues have reported from their work (e.g., Mynors-Wallis et al., 2000) that midlevel professionals (i.e., nurses), although effective PST-PC therapists, required diligent training in order to achieve high-level outcomes. Therapist skill level and adherence to the treatment model have been shown in previous research to predict outcome for CBT (DeRubeis & Feeley, 1990) and interpersonal therapy (O'Malley, Foley, & Rounsaville, 1988) of depression. In our training program therapists submitted to the trainers for rating the audiotapes of sessions they believed to be satisfactory exemplars of proper PST-PC, and were required to have two such tapes rated as satisfactory by the trainers before starting with study patients. In hindsight, this rather lenient teaching method may not have been adequate for non-CBT therapists. On average, the non-CBT therapists per- formed above the minimum criterion for adherence/competence. However, on average, the CBT therapists performed at a significantly higher level across the study (Hegel et al., 2000). It is possible that one or more non-CBT therapists were responsible for the reduced outcomes for this group, or that the set minimal criteria for competency were too lenient. Without an analysis of the relationship between treatment integrity and outcomes, we cannot definitively answer this question.

Second, in light of the fact that understanding the rationale for treatment was predictive of outcome, it may be that the CBT therapists in our study were more effective not only in the technical provision of PST-PC but in developing the rationale and understanding of the treatment on the part of the patient. That is, familiarity with the underlying theoretical and conceptual processes for PST-PC may have been greater and even "second nature" to the CBT therapists, thereby rendering the treatment procedures, including rationale building, that much easier to employ. As a caveat regarding these possible conclusions, however, we must also point out that all six of the CBT thera- pists were doctoral-level psychologists, whereas only one of the other five therapists was at an equivalent level of training (the others having master's- level training), thus creating a degree of confounding between training level

Page 13: Predictors of response to problem-solving treatment of depression in primary care

PREDICTORS OF R E S P O N S E TO PST-PC FOR D E P R E S S I O N 523

and theoretical orientation. Previous studies have provided equivocal evi- dence of a relationship between level of training and outcomes (Atkins & Christensen, 2001). Based on this body of research and our own experience, we believed that a CBT orientation, in combination with specific intervention training, would be most important for determining the quality of the interven- tion, rather than training level, and ran our analyses accordingly. The possi- bility of a training level effect for our outcomes does exist and will require further research to delineate.

In response to these findings we have concluded that non-CBT therapists will require more intensive training and have thus instituted some major changes in our PST-PC training program for non-CBT therapists. We have enhanced the initial training phase by requiring five training cases of six ses- sions each for which half of the sessions are tape recorded, rated, and super- vised. We have also enhanced the introductory component of the first session in order to better establish the rationale and understanding of the interven- tion. We now dedicate half of the initial training workshop to teaching this crucial introductory stage, which occurs within the first 15 minutes of treat- ment. In addition, although not included in the original PST-PC model, we have added an abbreviated form of "problem orientation training" as the liter- ature indicates this may enhance treatment outcomes (Nezu et al., 1989).

This new training program is currently being evaluated in a multisite depression treatment study in primary care (Unutzer et al., 2001). In this study, the therapist pool consists primarily of nurses (N = 17). A strong cor- relation has been found between the number of therapy training sessions and adherence/competence ratings (r = 0.78). In fact, the therapists achieved a level of performance during their last two training cases that was actually superior to the CBT therapists in the current study (F = 5.91, p < .001; Hegel et al., in press).

That PST-PC was most effective for dysthymics with low baseline severity but not for minor depressed patients with low severity is not a surprising find- ing for several reasons. Primary care patients with minor depression have a 50% likelihood of spontaneous remission within 1 month of the index visit (Williams et al., 1995) as well as a high placebo response rate (49% to 66%; Barrett et al., 2001; Williams et al., 2000). These factors make it very difficult for any intervention (medication or psychosocial) to show a treatment effect in the minor depression population. To date, treatment effects have only been demonstrated in the subsamples with severe functional impairments (Barrett et al.; Williams et al., 2000). By contrast, dysthymia by definition is a chronic condition that does not spontaneously remit and is less vulnerable to placebo response (40% to 44%; Barrett et al.; Williams et al., 2000), thus allowing treatment effects to be more readily detected. Considering the chronic nature of dysthymia, often starting early in life, it is probable that the more severe forms of the disorder have interfered with the learning of adaptive coping skills, including problem-solving abilities. A vicious cycle may therefore occur in which poor coping increases depression and vice-versa. The less

Page 14: Predictors of response to problem-solving treatment of depression in primary care

524 HEGEL ET AL.

severe dysthymic patient likely brings more coping resources to bear on the treatment process, which is an advantage that is perhaps particularly valuable for a brief therapy.

A few additional limitations of this study should be noted. First, we did not measure the actual process of problem-solving skill acquisition. Previous research has shown associative, although not causal, links between enhanced problem-solving skills and response to treatment for a traditional problem- solving therapy model (Arean et al., 1993). Without proof of this specific treatment effect for PST-PC, we do not know whether it is entirely effective due to problem-solving skills acquisition or perhaps partly as a "behavioral activation" treatment for depression as well. Second, we did not analyze these predictor variables for patients enrolled in the other arms of the study (medication and placebo) and therefore do not know whether their predictive capacity was specific to PST-PC or for response to treatment in general (e.g., medication). However, with the exception of the Diagnosis X Severity inter- action, only the more psychotherapy-specific variables (homework compli- ance, understand/apply, CBT therapist) were strong independent predictors of treatment outcome. Corollaries for medication treatment do not readily exist, and when they do (e.g., pill-taking compliance) are different conceptually. Therefore, we tend not to believe that these variables are generalizable as a tendency to respond to any intervention, although this possibility should be tested. Finally, the current findings are relevant only for the sample of patients receiving an "adequate" trial of PST-PC (i.e., four or more treatment sessions). These results cannot be generalized to patients receiving less treat- ment or who drop out of treatment.

To our knowledge, this is the first study to examine response predictors for a primary care psychological intervention for depression. The results provide insights into variables that are important for predicting higher-level outcomes with PST-PC but also raise questions about the ability to fully disseminate the intervention into primary care. Future research should continue to study the necessary background, training, and supervision requirements for PST-PC therapists as well as additional predictors of response, including problem- solving skills acquisition.

References Agency for Health Care Policy and Research Depression Guideline Panel, U.S. Dept of Health

and Human Services. (1993). Clinical practice guideline, depression in primary care: Vol. 2. Treatment of major depression. AHCPR Publication No. 93-0551. Washington, DC: U.S. Government Printing Office.

American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disor- ders. Washington, DC: Author.

Arean, P. A., Hegel, M. T., & Reynolds, C. F. (2001). Treating depression in older medical patients with psychotherapy. Journal of Clinical Geropsychology, 7, 93-104.

Arean, P. A., Perri, M. G., Nezu, A. M., Schein, R. L., Christopher, F., & Joseph, T. X. (1993). Comparative effectiveness of social problem-solving therapy and reminiscence therapy as

Page 15: Predictors of response to problem-solving treatment of depression in primary care

PREDICTORS OF RESPONSE TO PST-PC FOR DEPRESSION 525

treatment for depression in older adults. Journal of Consulting and Clinical Psychology, 61, 1003-1010.

Atkins, D. C., & Christensen, A. (2001). Is professional training worth the bother? A review of the impact of psychotherapy training on client outcome. Australian Psychologist, 36, 122-131.

Barrett, J. E., Barrett, J. A., Oxman, T. E., & Gerber, R D. (1988). The prevalence of psychiatric disorders in a primary care practice. Archives of General Psychiatry, 45, 1100-1106.

Barrett, J. E., Williams, J. W., Oxman, T. E., Frank, E., Katon, W., Sullivan, M., Hegel, M. T., Comell, J. E., & Sengupta, A. S. (2001). Treatment of dysthymia and minor depression in primary care: A randomized trial in patients aged 18 to 59 years. Journal of Family Prac- tice, 50,405-412.

Barrett, J. E., Williams, J. W., Oxman, T. E., Katon, W., Frank, E., Hegel, M. T., Sullivan, M., & Schulberg, H. C. (1999). The Treatment Effectiveness Project: A comparison of paroxetine, problem-solving therapy, and placebo in the treatment of minor depression and dysthymia in primary care patients: Background and research plan. General Hospital Psychiatry, 21, 260-273.

Broadhead, W. E., Blazer, D. G., George, L. K., & Tse, C. K. (1990). Depression, disability days, and days lost from work in a prospective epidemiologic survey. Journal of the Amer- ican Medical Association, 264, 2524-2528.

Brody, D. S., Khaliq, A. A., & Thompson, T. L. (1997). Patients' perspectives on the manage- ment of emotional distress in primary care settings. Journal of General Internal Medicine, 12,403-406.

Bums, D. D., & Nolen-Hoeksema, S. (1991). Coping styles, homework assignments and the effectiveness of cognitive-behavioral therapy. Journal of Consulting and Clinical Psychol- ogy, 59,305-311.

Bums, D. D., & Spangler, D. L. (2000). Does psychotherapy homework lead to improvement in depression in cognitive-behavioral therapy or does improvement lead to increased home- work compliance? Journal of Consulting and Clinical Psychology, 68, 46-56.

Catalan, J., Garb, D. H., Anastasiades, R, Bond, S. A., Day, A., & Hall, L. ( 1991). Evaluation of a brief psychological treatment for emotional disorders in primary care. Psychological Medicine, 21, 1013-1018.

Cole, M. G., & Bellavance, F. (1997). The prognosis of depression in old age. The American Journal of Geriatric Psychiatry, 5, 4-14.

DeRubeis, R., & Feeley, M. (1990). Determinants of change in cognitive therapy of depression. Cognitive Therapy and Research, 14,469-482.

Elkin, I., Shea, M. T., Watkins, J. T., Imber, S. D., Sotsky, S. M., Colins, J. F., Glass, D. R., Pilkonis, P. A., Leber, W. R., Docherty, J. P., Fiester, S. J., & Parloff, M. B. (1989). NIMH treatment of depression collaborative research program: General effectiveness of treat- ments. Archives of General Psychiatry, 46, 971-982.

Fennell, M. J., & Teasdale, J. D. (1987). Cognitive therapy for depression: Individual differ- ences and the process of change. Cognitive Therapy and Research, 11,253-271.

Folstein, M., Folstein, S., & McHugh, P. (1975). "Mini Mental State": A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12,189.

Fortney, J., Rost, K., & Zhang, M. (1998). A joint choice model of the decision to seek depres- sion treatment and choice of provider sector. Medical Care, 36, 307-320.

Frank, E., Prien, R., Jarrett, R., & Keller, M. B. (1991). Conceptualization and rationale for con- sensus definitions of terms in major depressive disorder: Remission, recovery, relapse, and recurrence. Archives of General Psychiatry, 48, 851-855.

Frank, J. D., & Frank, J. B. (1991). Persuasion and healing. Baltimore: Johns Hopkins Univer- sity Press.

Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery & Psy- chiatry, 23, 56-62.

Page 16: Predictors of response to problem-solving treatment of depression in primary care

526 HEGEL ET AL.

Hawton, K., & Kirk, J. (1989). Problem-solving. In K. Hawton, P. Salkovskis, J. Kirk, & D. Clark (Eds.), Cognitive behaviour therapy for psychiatric problems: A practical guide. Oxford: Oxford Medical Press.

Hegel, M. T., Barrett, J. E., & Oxman, T. E. (2000). Training therapists in problem-solving treatment of depressive disorders in primary care: Lessons learned from the "Treatment Effectiveness Project." Families, Systems & Health, 18,423-439.

Hegel, M. T., Imming, J., Cyr-Provost, M., Hitchcock Noel, M., Arean, P. A., & UntRzer, J. (in press). Role of behavioral health professionals in a collaborative stepped care treatment model for depression in primary care: Project IMPACT. Families, Systems & Health.

Hollon, S. D., & Najavits, L. (1988). Review of empirical studies of cognitive therapy. In A. J. Frances & R. E. Hales (Eds.), American Psychiatric Press review of psychiatry (Vol. 7, pp. 643-666). Washington, DC: American Psychiatric Press.

Ilardi, S. S., & Craighead, W. E. (1994). The role of nonspecific factors in cognitive-behavior therapy for depression. Clinical Psychology: Science and Practice, 1,138-156.

Katon, W., Lin, E., von Korff, M., Bush, T., Walker, E., Simon, G., & Robinson, P. (1994). The pre- dictors of persistence of depression in primary care. Journal of Affective Disorders, 31, 81-90.

Katon, W., Robinson, P., von Korff, M., Lin, E., Bush, T., Ludman, E., Simon, G. & Walker, E. (1996). A multifaceted intervention to improve treatment of depression in primary care. Archives of General Psychiatry, 53,924-932.

Katon, W., & Schulberg, H. C. (1992). Epidemiology of depression in primary care. General Hospital Psychiatry, 14,237-247.

Katon, W., von Korff, M., & Lin, E. (1995). Collaborative management to achieve treatment guideline: Impact on depression in primary care. Journal of the American Medical Associ- ation, 273, 1026-1031.

Kazdin, A. E., & Krouse, R. (1983). The impact of variations in treatment rationales on expect- ancies for therapeutic change. Behavior Therapy, 14,657-671.

Kendrick, T. (1996). Prescribing antidepressants in general practice: Watchful waiting for minor depression, full dose treatment for major depression. British Medical Journal, 31,829- 830.

Lipman, R. S., Covi, L., & Shapiro, A. K. (1979). The Hopkins Symptoms Checklist (HSCL): Factors derived from the HSCL-90. Journal of Affective Disorders, 1, 9-24.

Morrison, L. A., & Shapiro, D. A. (1987). Expectancy and outcome in prescriptive vs. explor- atory psychotherapy. British Journal of Clinical Psychology, 26, 59-60.

Mynors-Wallis, L. M., Davies, I., Gray, A., Barbour, F., & Gath, D. (1997). A randomized con- trolled trial and cost analysis of problem-solving treatment for emotional disorders given by community nurses in primary care. British Journal of Psychiatry, 170, 113-119.

Mynors-Wallis, L. M., Gath, D. H., Day, A., & Baker, F. (2000). Randomised controlled trial of problem solving treatment, antidepressant medication, and combined treatment for major depression in primary care. British Medical Journal, 320, 26-30.

Mynors-Wallis, L., Gath, D. H., Lloyd-Thomas, A. R., & Tomlinson, D. (1995). Randomised controlled trial comparing problem-solving treatment with amitriptyline and placebo for major depression in primary care. British Medical Journal, 310,441-445.

Nau, S. D., Caputo, J. A., & Borkovec, T. D. (1974). The relationship between credibility of therapy and simulated therapeutic effects. Journal of Behavior Therapy and Experimental Psychiatry, 5, 129-133.

Nezu,A. M., Nezu, C. M., & Perri, M. G. (1989). Problem-solving therapy for depression: The- ory, research, and clinical guidelines. New York: Wiley.

O'Malley, S., Foley, S., & Rounsaville, B. (1988). Therapist competence and patient outcome in interpersonal psychotherapy of depression. Journal of Consulting and Clinical Psychology, 56,496-501.

Priest, R., Vize, C., Roberts, A., & Tylee, A. (1996). Lay people's attitudes to treatment of depression. British Medical Journal, 313,838-859.

Page 17: Predictors of response to problem-solving treatment of depression in primary care

PREDICTORS OF RESPONSE TO PST-PC FOR DEPRESSION 527

Regier, D. A., Narrow, W. E., Rae, R. S., Manderscheid, R. W., Locke, B. Z., & Goodwin, F. K. (1993). The de facto U.S. mental and addictive disorders services system: Epidemiologic catchment area prospective I-year prevalence rates of disorders and services. Archives of General Psychiatry, 50, 85-94.

Rokke, P. D., Carter, A. S., Rehm, L. P., & Veltum, L. G. (1990). Comparative credibility of cur- rent treatments for depression. Psychotherapy, 27, 235-242.

Scott, A., & Freeman, C. (1992). Edinburgh primary care depression study: Treatment outcome, patient satisfaction, and cost after 16 weeks. British Medical Journal, 304,883-887.

Sireling, L. I., Freeling, P., Paykel, E. S., & Rao, B. M. (1985). Depression in general practice: Clinical features and comparison with outpatients. British Journal of Psychiatry, 147, 119-126.

Solberg, L. I., Korsen, N., Oxman, T. E., Kischer, L. R., & Barrels, S. (1999). The need for a system in the care of depression. The Journal of Family Practice, 48, 973-979.

Spanier, C., Frank, E., McEachran, A. B., Grochocinski, V. J., & Kupfer, D. J. (1996). The pro- phylaxis of depressive episodes in recurrent depression following discontinuation of drug therapy: Integrating psychological and biological factors. Psychological Medicine, 26, 461475.

Spitzer, R., Williams, J. W., Kroenke, K., Linzer, M., deGruy, E V., III, Hahn, S. R., Brody, D., & Johnson, J. G. (1994). Utility of a new procedure for diagnosing mental disorders in pri- mary care: The PRIME-MD 1000 Study. Journal of the American Medical Association, 14, 1749-1756.

Unutzer, J., Katon, W., Williams, J. W., Callahan, C., Harpole, L., Hunkeler, E. M., Hoffing, M., Arean, P., Hegel, M. T., Schoenbaum, M., Oishi, S. M., & Langston, C. A. (2001). Improv- ing primary care for depression in late life: The design of a multi-center randomized trial. Medical Care, 39, 785-799.

Williams, J. W., Barrett, J. E., Oxman, T. E., Frank, E., Katon, W., Sullivan, M., Coruell, J. E., & Sengupta, A. (2000). Treatment of dysthymia and minor depression in primary care: A randomized controlled trial in older adults. Journal of the American Medical Association, 284, 1519-1526.

Williams, J. W., Kerber, C. A., Mulrow, C. D., Medina, A., & Aguilar, C. (1995). Depressive disorders in primary care: Prevalence, functional disability, and identification. Journal of General lnternal Medicine, 10, 7-12.

Williams, J. W., Rost, K., Dietrich, A., Ciotti, M., Zysansky, S., & Cornell, J. (1999). Primary care physicians' approach to depressive disorders: Effects of physician specialty and prac- tice structure. Archives of Family Medicine, 8, 58-67.

RECEIVED: December 3, 2001 ACCEPTED: May 8, 2002