comorbid forms of psychopathology: key patterns and future

Comorbid Forms of Psychopathology: Key Patterns and Future ResearchDirections

Magdalena Cerda1,2,3, Aditi Sagdeo2, and Sandro Galea2,3,4

1 Robert Wood Johnson Foundation Health and Society Scholars Program, Department of Epidemiology, University of Michigan,Ann Arbor, MI.2 Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor, MI.3 Center for Urban Epidemiologic Studies, New York Academy of Medicine, New York, NY.4 Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, NY.

Accepted for publication April 29, 2008.

The purpose of this review is to systematically appraise the peer-reviewed literature about clustered forms ofpsychopathology and to present a framework that can be useful for studying comorbid psychiatric disorders. Thereview focuses on four of the most prevalent types of mental health problems: anxiety, depression, conductdisorder, and substance abuse. The authors summarize existing empirical research on the distribution ofconcurrent and sequential comorbidity in children and adolescents and in adults, and they review existingknowledge about exogenous risk factors that influence comorbidity. The authors include articles that useda longitudinal study design and used psychiatric definitions of the disorders. A total of 58 articles met the inclusioncriteria and were assessed. Current evidence demonstrates a reciprocal, sequential relation between mostcomorbid pairs, although the mechanisms that mediate such links remain to be explained. Methodologicalconcerns include the inconsistency of measurement of the disorders across studies, small sample sizes, andrestricted follow-up times. Given the significant mental health burden placed by comorbid disorders, and their highprevalence across populations, research on the key risk factors for clustering of psychopathology is needed.

anxiety; comorbidity; conduct disorder; depression; social conditions; substance-related disorders

INTRODUCTION

Psychiatric comorbidity is the presence, simultaneouslyor in sequence, of two or more disorders in an individualwithin a certain time period (1, 2). Psychiatric comorbidityis a prevalent phenomenon across age groups and types ofpopulations (3–5). The Epidemiologic Catchment Area(ECA) Survey found that 35 percent of respondents withat least one lifetime disorder had one or more additionaldisorders during their lifetime (6); the National ComorbiditySurvey Replication found that 27.7 percent of the respond-ents had two or more disorders during their lifetime (7).

Understanding psychiatric comorbidity has important re-search, clinical, and nosologic implications. From a researchperspective, epidemiologic research conducted on isolateddisorders may underestimate the burden imposed by mentalhealth problems because this group may represent a selectportion of the population that suffers from psychiatric dis-orders (8). Documenting the relation between key psychiat-

ric disorders over the life course can provide a moreaccurate assessment of the psychiatric burden experiencedby different population groups.

With respect to clinical concerns, comorbidity is associ-ated with more severe psychiatric symptoms, more func-tional disability, longer illness duration, less socialcompetence, and higher service utilization (9–11). Giventhe high prevalence of comorbidity and its clinical conse-quences, research efforts that aim to better understand thepatterns of comorbidity over time are clearly essential tohelp guide its prevention.

Investigation into the distribution of comorbidity acrosskey psychiatric disorders can also help us elucidate thestructure of phenotypic psychopathology. The cooccurrenceof disorders has called into question the validity of currentcategorical classifications of mental disorders, with criticssuggesting that a dimensional approach to classifying co-morbid disorders is more appropriate (12). Studying comor-bid patterns can thus help us discern commonalities and

Correspondence to Dr. Magdalena Cerda, Department of Epidemiology, School of Public Health, University of Michigan, 109 South Observatory,

3639 SPH Tower, Ann Arbor, MI 48109-2029 (e-mail: [email protected] or [email protected]).

155 Epidemiol Rev 2008;30:155–177

Epidemiologic Reviews

ª The Author 2008. Published by the Johns Hopkins Bloomberg School of Public Health.

All rights reserved. For permissions, please e-mail: [email protected].

Vol. 30, 2008

DOI: 10.1093/epirev/mxn003

Advance Access publication July 10, 2008

distinctions in the developmental pathways of comorbiddisorders and in the etiologic profiles of each comorbiddisorder—elements that will provide the basis to understandwhether the disorders are different expressions of the sameunderlying disturbance or are distinct disorders in their ownright (13).

This review aims to examine current knowledge on theepidemiology of clustered forms of psychopathology overthe life course, focusing on four of the most common mentaldisorders: anxiety, depression, conduct disorder, and sub-stance abuse. According to the National Comorbidity Sur-vey Replication, 28.8 percent of the US population sufferedfrom an anxiety disorder sometime in their life, whereas20.8 percent suffered from a mood disorder, 14.6 percenthad a substance use disorder, and 9.5 percent presentedsymptoms of conduct disorder (7). The cooccurrence ofthese conditions has been consistently documented in clin-ical and population samples (5).

This paper systematically reviews comorbidity studieswithin a life course framework, focusing exclusively on pro-spective population-based studies. This presents a novel ap-proach to address comorbidity and enables us to studysimilarities and differences in comorbidity patterns at keylife stages as well as to understand how the normative de-velopmental trajectories of disorders may influence themagnitude of comorbidity from childhood to adulthood.Focusing on population-based studies overcomes one ofthe key limitations of clinical studies: selection bias thatmay determine the patterns of comorbidity observed in thesample.

REVIEW OF THE LITERATURE

This review encompasses the peer-reviewed literaturepublished between 1970 and 2007. We limited our reviewto these years to reflect current thinking about psychiatriccomorbidity and to include studies that use methods consid-ered standard today. The literature reviewed was identifiedthrough the Social Science Citation Index and ScienceCitation Index, and it covered studies about concurrent andsequential comorbidity between anxiety, depression, sub-stance use/abuse, and conduct disorder. We restricted ourreview to longitudinal, population-based studies that pro-vided information about the temporal associations betweendisorders. As stated by Kraemer et al. (14), cross-sectionalstudies, particularly those that use lifetime prevalence withmixed-age samples to estimate comorbidity, run the risk ofidentifying instances of ‘‘pseudocomorbidity’’ when no realcomorbidity exists.

The literature on comorbidity is vast and encompassesmany different fields, from medicine to public health topsychology. We restricted our review to psychiatric defini-tions of the disorders because a wider review would havebeen beyond the scope of this project. For example, forconduct disorder, we considered measurement of symptomsor diagnoses of conduct disorder, oppositional defiant dis-order, and antisocial personality disorder but not aggressionor delinquency. The search was limited to English-languagestudies in biomedical research. Keywords and terms used

for the search included primarily the following: 1) for sub-stance abuse: substance abuse, alcohol abuse, drug abuse,cannabis, cocaine, heroin, street drugs, smoking, injectiondrug use; 2) for depression: depression, depression symp-tomatology, depression symptoms, mood disorders; 3) forconduct disorder: conduct disorder, antisocial personalitydisorder, antisocial, opposition defiant disorder, externaliz-ing behavior/disorder; 4) for anxiety disorder: anxiety, anx-iety symptomatology, anxiety symptoms, internalizingdisorder; and 5) for comorbidity: comorbidity, joint trajec-tories, discordant trajectories, concurrent, cooccurrence,risk pathways, chain of risk, launch factors, transitions, riskpathways.

FINDINGS

In this paper, we review concurrent and sequential comor-bidity between the four disorders, and we present availabledata on concurrent comorbidity, sequential comorbidity, andcontributions of each disorder to the deflection of develop-mental trajectories of the other three disorders. This reviewcovers 58 studies, including those using child and adoles-cent samples, and then adult samples, to present availableknowledge on comorbidity at key life stages. The originalsearch provided 103 articles, of which 45 were removedbecause of the absence of a prospective study design.Although studies such as Kessler et al.’s (15) on the NationalComorbidity Survey Replication or the cross-national com-parison of comorbidity conducted by Merikangas et al. (16)offer groundbreaking data on the distribution of comorbiddisorders in population-based samples, they were not in-cluded because they used retrospective measures of disorderonset in cross-sectional study designs.

The studies reviewed here (summarized in tables 1 and 2)approach the issue of causality between comorbid disordersby investigating the relative timing of the onset of eachdisorder and the impact that the prevalence of one disorderhas on the course of the other. The strength of the causal linkdepends on whether the cooccurrence of the disorders per-sists after accounting for common risk factors (3, 17).

Anxiety and substance use

The evidence on the link between anxiety and substanceuse is mixed: most studies have been cross-sectional orretrospective in design and used dimensional rather thandiagnostic measures of anxiety. The few existing prospec-tive, diagnostic studies (18–22) propose several alternativecausal associations between anxiety and substance abuse:1) individuals with an anxiety disorder may be more likelyto develop or sustain a substance use disorder, in part asa way to manage the symptoms of anxiety (21, 23, 24);2) having an anxiety disorder may reduce or delay the onsetof behavioral problems such as substance abuse either be-cause of the high level of anxious social cognitions thatserve as a reminder of the negative consequences of thebehavior or because of behavioral inhibition that resultsfrom anxiety disorders (25); 3) substance use may exertphysiologic effects that increase vulnerabilities to anxiety

156 Cerda et al.

Epidemiol Rev 2008;30:155–177

disorders; or 4) a third variable may promote the develop-ment or maintenance of both (20).

Children and adolescents. Childhood and adole-scence is a critical time to investigate the timing of onsetof the comorbid disorders since anxiety has a relativelyearly age at onset (26). Few studies have looked prospec-tively at the link between anxiety and substance abuseamong the younger age groups. Several studies have foundsupport for the ‘‘self-medication’’ thesis: that persons withanxiety use alcohol and drugs to suppress their anxietysymptoms (18, 22, 27–29). Costello et al. (18) found thatanxiety predicted later substance use disorders among a rep-resentative population sample of adolescent girls, but notamong boys. However, several studies have also found sup-port for the reverse association, in which substance depen-dence promotes the development of anxiety (20, 21).A particularly compelling study that followed children over21 years and obtained prospective reports of anxiety andsubstance dependence found that the association betweenanxiety and substance dependence was largely or entirelynoncausal, while substance dependence predicted anxietydisorder at follow-up. Finally, some evidence also existsfor a negative relation between anxiety and substance use:Pardini et al. (25) followed the Pittsburgh Youth Study sam-ple of boys from early adolescence to young adulthood, andthey found that higher levels of anxiety/withdrawal wereinversely related to symptoms and diagnoses of alcoholuse disorder; a cohort study of a representative populationsample aged 9–13 years at baseline found that anxiety pre-dicted a later onset of smoking (30).

Lack of conclusive evidence is due to the heterogeneity ofsubstance use and anxiety subtypes examined in the fewavailable population-based studies: associations may differbetween subtypes. Kaplow et al. (28) proposed that, in stud-ies that consider anxiety as a general syndrome and fail todifferentiate within its dimensions, the opposing functionsof different types of anxiety may offset each other, resultingin null findings. In fact, Kaplow et al. found opposing effectsfor generalized anxiety disorder and separation anxiety dis-order: whereas generalized anxiety disorder was positivelyassociated with alcohol use initiation, separation anxietydisorder resulted in a lower risk of onset of alcohol use.A study of 2,548 adolescents in Germany also found thatthe association depended on the type of anxiety disorderexamined; although the seven lifetime anxiety disorderstaken as a set predicted subsequent onset of hazardoususe, abuse, and persistence, only panic and social phobiawere independent predictors of hazardous use and abuseand the persistence of an alcohol use disorder (22).

Adults. Studies conducted among adults provide sup-port for a potential reciprocal relation between anxietyand substance abuse. A study of 454 college students fol-lowed over 7 years, for example, found that having an anx-iety disorder was associated with 3.5–5 times higher odds ofdeveloping a new alcohol dependence, while the odds ofdeveloping a new anxiety disorder were about four timeshigher for those diagnosed with alcohol dependence in pre-vious years (24). The direction of the relation between anx-iety and substances may depend on the type of substance.The Kushner et al. study (24) was limited by use of a select

population that may have a more similar mental health sta-tus than the general population. In a study of a randomsample of the adult population, however, Newcomb et al.(31) also found a bidirectional relation between anxiety andsubstance use: polydrug problems in young adulthood in-creased anxiety in adulthood, but anxiety was actually as-sociated with fewer cocaine problems later in life—possiblybecause individuals who experience high levels of arousalwould be less likely to use a drug that exacerbated thatarousal.

Depression and substance abuse

Depressive symptoms are associated with substance usedisorders throughout adolescence and adulthood (25). Aswith anxiety, the directionality and causal nature of the as-sociation remains in question: depression and substance usedisorders may be linked through common confounding fac-tors, or the association may be causal—either that alcoholand illegal substances are used as a form of self-medicationto manage symptoms of depression or that substance usemay increase vulnerability to depression through behavioralchanges or neurophysiologic alterations (32).

Children and adolescents. Several studies have re-ported that depression and substance abuse tend to cooccurin adolescence (18, 19, 21, 33–35). In a study of high schoolstudents in western New York, Windle and Davies (35)found that 24–27 percent of adolescents identified as de-pressed also met the criteria for heavy drinking, whereasRao et al. (34) found that 21.1 percent of adolescent womenwith depression also had a substance use disorder. The na-ture of the association between depression and substanceabuse among adolescents remains in question. Several stud-ies found that depressed adolescents were more vulnerableto substance use; this association has been found for mari-juana (27), smoking (33), alcohol abuse (25), and substanceabuse problems (36). Fergusson et al. (33) and Pardini et al.(25), authors of two of the most promising studies on issuesof depression–substance abuse comorbidity among adoles-cents, highlight the lack of conclusive evidence on this is-sue. Fergusson et al. followed a birth cohort of 1,265children from birth to age 21 years and found that majordepression was associated with increased rates of dailysmoking and nicotine dependence, although the associationcould be due to confounders that affected the developmentof both outcomes. Pardini et al. also followed up 506 boysuntil young adulthood and found that increased depressionin early adolescence was associated with more alcohol usedisorder symptoms and alcohol abuse and dependence di-agnoses by young adulthood, but only for boys with highlevels of conduct problems.

Evidence also exists to support the hypothesis that sub-stance use may increase vulnerability to depression.Hayatbakhsh et al. (21) followed a comparable cohort of3,239 children up to age 21 years and found that symptomsof depression in childhood did not predict use of marijuana,but use of marijuana was associated with depression.A small study of 155 women aged 17–19 years also foundthat substance use disorders predicted worse depression

Comorbid Forms of Psychopathology 157


TABLE 1. Population studies on comorbid patterns between substance abuse, anxiety, depression and conduct disorder among children and adolescents

Comorbidity Citation Sample Follow-up time Measurement of disorders Conclusions

Anxiety andsubstanceuse/abuse

Goodwinet al. (20)

1,265 children (635 males,630 females) fromChristchurch, New Zealand

21 years CIDI* and custom-written surveyitems used; alcohol dependence:DSM-IV* criteria; illicit drugdependence: DSM-IV criteria;nicotine dependence: DSM-IV;anxiety disorders (generalizedanxiety disorders, panic disorder,agoraphobia, social phobia, andspecific phobia): CIDI diagnosticto assess DSM-IV criteria

Young people with anxiety disorders areat increased risk of substancedependence; association is largelynoncausal and is explained by 1)presence of fixed childhood and familyfactors associated with both anxietyand substance dependence; and 2)time-dynamic associations includinglagged effects of substance dependenceand development of comorbiddepression. OR* of substancedependence associated with anxietydisorder 5 1.8 (95% CI*: 1.2, 2.6).


Zimmermanet al. (22)

2,548 adolescents aged14–24 years in Munich,Germany

4 years Diagnostic assessment basedon the computer-assistedversion of the DIA-X/M-CIDI*;DSM-IV anxiety disordersassessed; DSM-IV alcoholuse disorders assessedinclude alcohol abuse withoutdependence and alcoholdependence without alcoholabuse

Anxiety disorders are significantpredictors of the subsequent onsetand persistence of regular andhazardous alcohol use and alcoholuse disorders. Panic disorder andsocial phobia are the only anxietydisorders involved in thesepredictive associations.

Depression andsubstanceuse/abuse

Silberget al. (71)

307 monozygotic and 185dizygotic same-sex maletwin pairs from Virginia;392 monozygotic and187 dizygotic like-sexfemale pairs

4 years Substance use based on ratingsgiven by multiple informantsduring a home interview usingthe DSM-based CAPA,* conductdisorder measured with acomposite rating based onchild’s response to CAPAsymptoms, a diagnosisof conduct disorder andoppositional defiant disorder,and ratings on the Olweusaggression scale; depressionbased on CAPA symptomsassociated with a diagnosisof major depressive episode

Among boys, the cross-laggedcorrelations between conductdisturbance and substance use andbetween substance use and depressionwere approximately equal, indicatingthat a common factor contributed tothese behaviors; among girls, there weredifferences in the pattern of the cross-lagged correlations for conduct disorderand depression, substance use andconduct disorder, and substance useand depression, indicating that conductdisorder preceded substance use andthat depression followed substance use.Among girls, the cross-laggedassociations were also consistent withconduct disorder leading to depression.


Fergussonet al. (33)

Birth cohort of 1,265 children(635 males, 630 females)born in Christchurch,New Zealand in mid-1977

Assessed at birth,4 months, 1 year,annually to age16 years, andat ages 18 and21 years

Smoking based on written survey,nicotine dependence based onDSM-III-R* criteria at age16 years and DSM-IV criteriaat ages 18 and 21 years; majordepression measured withDISC* with DSM-III-R criteriaat age 16 years and CIDI withDSM-IV criteria at ages18 and 21 years

Major depression was associated withincreased rates of daily smoking(IRR* 5 1.19 (95% CI: 1.03, 1.39)).and elevated rates of nicotinedependence (OR 5 1.75 (95%CI: 1.13, 2.70)).

158

Cerdaetal.

EpidemiolR

ev

2008;30:155–177


Raoet al. (34)

155 women aged 17–19years recruited from threehigh schools in LosAngeles, California

Follow-upsconducted every6 months upto 5 years

Axis I disorders assessed with theStructured Clinical Interview forDSM-III-R; self-reporteddepressive symptomsassessed with the BeckDepression Inventory; self-reported problems of substanceuse assessed at the end offollow-up with the Rutgerscollegiate Substance AbuseScreening Test

Depression and substance use disorderwere related concurrently; majordepressive disorder and substanceuse disorder were significantlyassociated in follow-up; substance usedisorder problem score and BeckDepression Inventory (self-reporteddepressive symptoms) during follow-upwere also correlated. Major depressivedisorder before the study did not predictsubstance use disorder during thefollow-up period, whereas priorsubstance use disorder did predictmajor depressive disorder at follow-up.Prior substance use disorder alsopredicted the worst Beck DepressionInventory score during follow-up.


Windle andDavies (35)

975 high school studentsfrom western New YorkState in the first wave,220 added in the secondwave; average age, 15.54years (data collected fromprimary caregivers: 90%mothers)

Time 1: 1 yearsince study; time2: 2 yearssince study

Depression: CES-D* Scalecriterion score of >23,‘‘heavy drinking’’: QualityFrequency Index—adolescentswho consumed >45drinks in the last 30 days

24–27% of adolescents identified asdepressed met the criteria for heavydrinking (33–37% of boys, 16–18.5% ofgirls); 23–27% of adolescent heavydrinkers were depressed (18–20% ofboys, 27–33% of girls). Approximately5% of participants had depression andwere heavy drinkers at either wave.The mixed drinking and depressiongroup showed the lowest stabilitybetween time 1 and time 2: only 20%of those in the mixed group at time1 persisted in this group at time 2.


Aseltineet al. (13)

900 adolescents (9th, 10th,and 11th graders) fromBoston, Massachusetts

3 years Depressive symptoms measuredwith CES-D Scale; measuresof alcohol and drug useadapted from Monitoringthe Future studies

Rate of cooccurring depression andsubstance use problems was 4% (n 5 39);those with prior substance problems weremore than twice as likely to developcooccurring problems (9%) at time 2than were the depressed (4%).

Anxiety, depression,and substanceuse/abuse

Kaplowet al. (28)

1,420 children in the totalstudy from western NorthCarolina; for this analysis,only 936 included whocompleted child orparent report measures ofgeneralized anxiety,separation anxiety, anddepressive symptomatologyas well as reported alcoholuse at all four time points

4 years CAPA used to assesspsychiatric symptoms

Children with early symptoms ofgeneralized anxiety were found to beat increased risk for initiation of alcoholuse (OR 5 1.14 (95% CI: 1.03, 1.25)),whereas children with early symptomsof separation anxiety were at decreasedrisk (OR 5 0.71 (95% CI: 0.51, 0.94));depressive symptomatology wasassociated with increased risk forinitiation of alcohol use in adolescence(OR 5 1.60 (95% CI: 1.29, 1.99)).

Anxiety, depression,and substanceuse/abuse

Hayatbakhshet al. (21)

3,239 Australian young adults Birth to age21 years

Anxiety and depression symptomsin the last 6 months measuredin the 21-year follow-up byusing the Young Adult Self-Report; cannabis useretrospectively assessed,occasional use 5 use ofcannabis once a month or notin the past month, frequentuse 5 use of cannabis everyday or every few days

Frequent use of cannabis, either without(OR 5 2.1 (95% CI: 1.1, 4.0)) or with(OR 5 2.7 (95% CI: 1.8, 4.1)) use ofother illicit drugs, predicts more than atwofold increase in anxiety and depressionin young adults. No association was foundbetween symptoms of anxiety anddepression at age 14 years and eitheroccasional or frequent use of cannabis byyoung adults. Frequent use of cannabisand early onset of use are associated withsymptoms of anxiety and depressionin young adults.

Table continues

Comorbid

Form

sofPsychopathology

159

EpidemiolR

ev

2008;30:155–177

TABLE 1. Continued


Anxiety anddepression

Repettoet al. (51)

579 African-Americanadolescents in Ann Arbor,Michigan, who were atrisk of dropping out of school

4 years Depressive symptoms assessedwith items from the BriefSymptom Inventory; anxietysymptoms assessed with sixitems from the Brief SymptomInventory

Adolescents who presented consistentlyhigh levels of depressive symptomswere more likely to report higher anxietysymptoms compared with adolescentmembers of other trajectories.Adolescents whose depressivesymptoms followed the decreasingtrajectory had higher levels of anxietythan those in the consistently low orincreasing groups.


Steinet al. (2)

2,548 adolescents andyoung adults aged 14–24years from Munich, Germany

One follow-up34–50 monthslater

Munich-CIDI used to diagnoseby using DSM-IV criteria; socialanxiety disorder defined asmeeting DSM-IV criteria perthe Munich-CIDI diagnosticcriteria; depressive disorderdefined as meeting DSM-IVcriteria for one or moreepisodes of major depressionor dysthymia

Persons with a combination of socialanxiety disorder and depression are atthe greatest risk of depression in earlyadulthood. Persons with depressionand social anxiety disorder (current orprevious) were significantly more likely(OR 5 8.7 (95% CI: 4.5, 16.8)) thanpersons with no mental disorder tohave experienced a depressivedisorder during the follow-up period.


Wittchenet al. (26)

3,021 adolescents andyoung adults aged 14–24years at first interview inMunich, Germany

Three wavesover 4–5 years

Diagnostic assessment basedon the computer-assistedversion of the Munich-CIDI;diagnoses made by usingDSM-IV criteria

The proportion of major depressiveepisodes among those with an anxietydisorder was low for those aged 14–17years but increased at ages 18–24years. Almost two thirds of cases withanxiety disorder were comorbid withmajor depression at follow-up(OR 5 3.9 for generalized anxietydisorder, OR 5 3.4 for panic disorder,and OR 5 1.7 for specific phobia).The number of anxiety disorders,baseline impairment associated withanxiety, the frequency of avoidance,and the presence of panic-like attacksall increased the risk of secondarydepression among those with primaryanxiety disorders.


Pineet al. (50)

776 young people (aged9–18 years) in upstateNew York

Assessed in 1983,1985, and 1992:9 years

Used DISC to assess DSM-IIIcriteria from parents and childrenat times 1 and 2; at time 3, DISCused with the children only

Major depression at time 2 predictedgeneralized anxiety disorder at time 3.Overanxious disorder predicted majordepression at time 3. Conduct disorderat time 1 was related to major depressionat time 3 (OR 5 2.54 (95% CI:1.42, 7.68)). Conduct disorder at time2 was related to major depression attime 3 (OR 5 2.38 (95% CI: 1.10, 5.16)).

Conduct disorderand substanceuse/abuse

Cohenet al. (60)

749 adolescents aged13.7 years at the firstinterview from two upstateNew York counties

Interviewed in1983, 1985–1986,and 1991–1994

Substance use disorders andconduct disorder assessed withDISC-I; instrument administeredto both the child and a parent;DSM-III-R diagnoses made

Adolescents with conduct disorder had2.59 times the hazard of initiatingmarijuana use over the study periodin comparison to adolescents withoutconduct disorder. Conduct disorder inearly adolescence was independentlyassociated with significantly elevatedmean symptoms of alcohol abuse/dependency and marijuana usedisorder across the age range.

160

Cerdaetal.

EpidemiolR

ev

2008;30:155–177


Juonet al. (75)

952 African-Americanchildren who began 1stgrade in Woodlawn,Chicago, in 1966–1967and remained in theWoodlawn school duringtheir 1st year

1966–1967;1975–1976;1992–1994

Self-report measures of smokingfrequency and quantity(instrument not specified);substance use disordermeasured by using the modifiedversion of CIDI based onDSM-III-R criteria; majordepressive disorder measuredwith CIDI; antisocial behavioroperationalized as ratings from1st-grade teachers asaggressive or shy and aggressive

Those who had been rated as aggressive(OR 5 2.10) or as both shy andaggressive (OR 5 1.93) in the 1stgrade were more likely to be currentsmokers/early adopters thannonsmokers. Those who hadcomorbidity of major depressivedisorder and drug problems(OR 5 4.70) were more likely to becurrent smokers/early adopters thannonsmokers compared with thosewho had no problems.


Hussonget al. (66)

461 men from the Dunedin,New Zealand, MultidisciplinaryHealth and Development Study

8 years, at ages18, 21, and26 years

Alcohol abuse and marijuanaabuse assessed by symptomsfrom the DIS*; antisocialbehaviors assessed via theself-report offending interview,which ascertains illegalbehaviors and conductproblems; DSM-IV criteria usedfor these measures

Substance abuse exerts both proximal anddistal effects on desistance of antisocialbehavior over young adulthood. Men withgreater substance abuse at the end ofadolescence showed greater antisocialbehavior across young adulthood. Periodsin which men reported greater symptomsof substance abuse corresponded tohigher-than-normal levels ofantisocial behavior.


Marmorsteinand Iacono (67)

289 families of male twinsand 337 families of femaletwins, first assessed whenthe twins were approximately17 years of age in Minnesota

3 years Conduct disorder measured froman adapted version of theStructured Clinical Interview forDSM-III-R Personality Disorders;substance dependencesymptoms assessed by usingadolescents’ reports on amodified version of theSubstance Abuse Module ofCIDI; DSM-III-R criteria used

Youths with late-onset antisocialbehavior had higher rates of substancedependence than controls for all threeclasses of substances. The late-onsetgroup also had more nicotine andcannabis dependence than those withdesisting behavior. Youths withpersisting antisocial behavior hadmore nicotine, alcohol, and cannabisdependence than youths with noantisocial behavior and youths withdesisting antisocial behavior.


Whiteet al. (64)

506 boys in the PittsburghYouth Study

6 years Drug use: assessed with the16-item substance-use scaleadapted from the National YouthSurvey; psychopathologiccharacteristics: assessed withthe Self-Report DelinquencyScale, the Youth Self-Report,and the caretaker-completedDISC-Parent Version; depressedmood 5 number of symptomsfrom the Recent Moods andFeelings Questionnaire

Marijuana use was associated with higheroverall level of conduct disorder but didnot affect the rate of growth of conductdisorder. No association was foundbetween oppositional defiant disorderscores and marijuana use; oppositionaldefiant disorder was a significant predictorof level of alcohol use. Number ofsymptoms of depression was positivelyrelated to level of alcohol use but notmarijuana use.

Conduct disorderand depression

Ingoldsbyet al. (68)

431 school-age children fromDurham, North Carolina;Nashville, Tennessee; Seattle,Washington; centralPennsylvania

5th27th grade Conduct problems: youthcompleted the 32-item ThingsYou Have Done survey at theend of 5th grade, youthcompleted the Self-ReportDelinquency Scale at the end of7th grade; parents rated ChildBehavior Checklist, teacherscompleted the externalizing scaleon the Teacher’s Report Form;depressive symptomatology:youth completed the ReynoldsChild Depression Scale at theend of 5th and 7th grades andparents completed a subscaleof the Child Behavior Checklist

This study did not find support for thehypothesis that conduct problemspredict increases in depression in earlyadolescence. Cases of comorbidconduct problems and depressivesymptoms were significantly persistentover time. A substantial portion of youthin the conduct problems group in 5thgrade moved to the cooccurring groupby 7th grade. Serious depression orconduct problems by 7th gradeoccurred more frequently in comorbidcases than in cases with eitherdisorder alone.

Table continues

Comorbid

Form

sofPsychopathology

161

EpidemiolR

ev

2008;30:155–177

TABLE 1. Continued



Kimet al. (70)

206 boys recruited through4th-grade classes (depressionstudied from ages 14–15to 23–24 years) from Oregon

10 years Depressive symptoms: assessedwith the CES-D Scale; antisocialbehavior in late childhood:antisocial construct from itemsfrom the Teacher Child BehaviorChecklist, telephone interviewscores, peer nominations, homeobservations, and interviewerimpression

Early antisocial behavior wasassociated with depressivesymptoms in adulthood.

Conduct disorder,substance use/abuse, anddepression

Measelleet al. (36)

Community sample of 496female adolescents whovaried in age from 13 to 15years at first assessmentfrom the metropolitan areaof the southwesternUnited States

5 years Depressive symptoms: adaptedversion of the K-SADS*;antisocial behavior symptoms:Externalizing Syndrome of theChild Behavior Checklist;substance abuse symptoms:items adapted from Stice et al.(76); DSM-IV criteria used for all

Initial depression level predictedincreases in substance abuse symptomsand slower decreases in antisocialbehavior; initial levels of antisocialsymptoms predicted increases indepression and substance abuse;initial substance abuse level predictedslower decelerations in antisocialbehavior.

Conduct disorder,anxiety, anddepression

Gregoryet al. (49)

1,037 members of abirth cohort in Dunedin,New Zealand (baselinebeginning at age 11 years)

21 years: follow-ups at ages 11,13, 15, 18, 21,26, 32 years(N 5 972 at thisfollow-up)

At ages 11–15 years, DISC usedto diagnose mental healthaccording to DSM-II; at ages18–32 years, DIS used toassess mental health by usingDSM-IV criteria

Adult anxiety cases were also morelikely to have experienced juveniledepression relative to those withoutadult anxiety. Adults with social phobia,agoraphobia, and posttraumatic stressdisorder were more likely to haveexperienced externalizing spectrumdisorders than those without thesedisorders: adults with posttraumaticstress disorder were likely to havemet diagnostic criteria for conductor oppositional defiant disorder.

Conduct disorder,anxiety, anddepression

Lavigneet al. (61)

510 children aged 2–5years at baseline fromChicago, Illinois

5 years Externalizing and internalizingbehavior problems assessedby the Child Behavior Checklist;DSM-III-R diagnoses conductedwith the Diagnostic Interview forChildren and Adolescents

Single-diagnosis oppositional defiantdisorder at wave 1 was associatedwith later comorbidity of oppositionaldefiant disorder and mood disordersand of comorbidity of oppositionaldefiant disorder and an anxiety disorder.There was moderate-to-high stability foroppositional defiant disorder comorbidwith anxiety and low stability foroppositional defiant disorder comorbidwith mood disorder.

Conduct disorder,anxiety, depression,and substanceuse/abuse

Loeberet al. (74)

Pittsburgh Youth Study ofthree samples of boys ingrades 1 (age 6.9 years),4 (age 10.2 years), and 7(age 13.4 years) (eachsample including 250 ofthe most antisocial boysin each grade plus 250randomly selected others)

Youngest groupassessed atscreening and ateight assessmentwaves spaced at6-month intervals;middle groupassessed atscreening and sixassessment waves;oldest groupassessed atscreening andseven assessmentwaves

Child Behavior Checklist and theDISC-Parent Versionadministered to caretaker;teachers given the TeacherReport Form; boys assessedwith the National Youth Survey40-item Self-ReportedDelinquency Scale and the16-item Substance Use Scalebased on the National YouthSurvey; oppositionaldefiant disorder: DSM-III-Rbased on DISC-Parent Version

Persistent substance use inpreadolescence was predicted bypersistent delinquency and internalizingproblems, while persistent substanceuse in adolescence was predicted bypersistent delinquency only. Comorbidpersistent substance use anddelinquency were predicted byoppositional defiant disorder inmiddle childhood and by persistentinternalizing problems in middle-to-latechildhood.

162

Cerdaetal.

EpidemiolR

ev

2008;30:155–177


Rohdeet al. (77)

1,507 participants selectedfrom high schools inwestern Oregon aged14–18 years

Mean 5 13.8 months Adolescents interviewed with aversion of K-SADS that combinedfeatures of the epidemiologicversion (K-SADS-E) to derivediagnoses based on DSM-II-Rcriteria; alcohol use: foursymptoms for alcohol use andseven symptoms for dependenceassessed; psychopathology:diagnoses clustered amongDSM subdivisions

Of adolescents with alcohol abuse/dependence, 20% had an internalizingdisorder, 35% had an externalizingdisorder, and 45% had both internalizingand externalizing disorders. 87.5% ofthe time, anxiety disorders precededalcohol disorders, and, 80.0% of thetime, disruptive behavior disorderspreceded alcohol disorders. Comorbiditywas associated with an earlier age atonset of alcohol disorder; 16% ofadolescents with a current psychiatricdisorder moved into a more problematicalcohol use group from time 1 to time2 vs. 5.8% of adolescents with nodiagnosis between the assessmentpoints.


Pardiniet al. (25)

506 boys in the PittsburghYouth Study

Average of 12.5years; boysscreened, assessedat ages 13.9, 20.4,and 25.4 years

Depression: assessed with theself-report Recent Moods andFeelings Questionnaire (containsitems with DSM-III criteria);anxiety/withdrawal: assessedwith a scale that combined thechild (youth self-report), teacher(Teacher Report form), andparent-report Child BehaviorChecklist; alcohol use disorders(assessed with the DIS on twooccasions in early adulthood(DSM-III-R diagnoses)

Early conduct disorder symptomspredicted increased alcohol usedisorder symptoms and alcoholdependence diagnosis by youngadulthood. Adolescent boys with highlevels of anxiety/withdrawal had lowerlevels of alcohol use disorder symptomsand were less likely to develop alcoholdependence by young adulthood.Increased depression in earlyadolescence was associated withhigher alcohol use disorder symptomsand alcohol abuse and dependencediagnoses by young adulthood, butonly for boys with high levels ofconduct disorder symptoms.


Costelloet al. (18)

1,420 children fromwestern North Carolina

3–5 years (childrenaged 9–13 yearsassessed annuallyuntil age 16 years)

CAPA used to assesspsychiatric symptoms

Significant comorbidity was foundamong the behavioral disorders andanxiety and depression; depressionwas comorbid with conduct disorderin girls but not boys; depression wascomorbid with substance use disorderin boys but not girls; anxiety predictedlater substance use disorder, butdepression did not; in girls, but notboys, anxiety predicted depressioneven when controlling for currentcomorbidity.


Kruegeret al. (8)

961 members of anunselected birth cohortfrom Dunedin, New Zealand

Aged 18–21 years Mental disorders (in the prior12 months) assessed by usingthe DIS to generate diagnosesaccording to DSM-III-R criteria

A two-factor model provided the bestfit to the data: major depressive disorder,dysthymia, generalized anxiety disorder,agoraphobia, social and simple phobia,and obsessive compulsive disorderwere presumed to reflect internalizingproblems; conduct disorder andmarijuana and alcohol dependencewere presumed to reflect externalizingproblems. Subjects retained theirpositions regarding the latent factorsover the 3-year period.

Table continues

Comorbid

Form

sofPsychopathology

163

EpidemiolR

ev

2008;30:155–177

TABLE 1. Continued



Wittchenet al. (27)

1,395 community subjectsfrom Munich, Germany,aged 14–17 years atbaseline

Three wavesover 10 years

Substance use, abuse, anddependence, as well as mentaldisorders, assessed with theDIA-X/M-CIDI, accordingto DSM-IV

Major depression was associated with ahigher risk of cannabis use and dependence:(OR for cannabis use 5 2.2 (95% CI: 1.3, 3.7);OR for cannabis dependence 5 2.4 (95% CI:1.3, 4.5)). Hypomania and mania wereassociated with incident cannabis use andcannabis dependence (OR for cannabisuse 5 2.4 (95% CI: 1.0, 5.4); OR for cannabisdependence 5 2.7 (95% CI: 1.1, 6.3)). Amonganxiety disorders, only panic-anxiety wasassociated with cannabis use (OR 5 3.1 (95%CI: 1.4, 6.8)). Conduct problems predictedcannabis use (OR 5 2.3 (95% CI: 1.2, 4.2)).


Dierkeret al. (19)

173 high- and low-riskyouths in New Haven,Connecticut (identified onthe basis of parentalhistory of substanceabuse) aged 7–17 years

1988–1998 Child diagnoses based onK-SADS and independentratings from DSM-III-Robtained from a child andadolescent psychiatrist; tobaccoassessed using K-SADS

Psychiatric disorders were significantlyassociated with nicotine dependence:the association between having apsychiatric disorder and dependencevs. lower levels of use were as follows:OR for anxiety 5 4.5 (95% CI: 1.71,11.94), OR for affective disorder 5 10.7(95% CI: 3.79, 30.25), OR for conductdisorder 5 5.0 (95% CI: 1.6, 15.73),OR for oppositional defiantdisorder 5 5.8 (95% CI: 1.80,18.43); in the low-risk population,oppositional defiant disorder was theonly risk factor that predicted transitionto nicotine dependence.


Costelloet al. (30)

1,420 children fromwestern North Carolina

Followed from ages9, 11, and 13 yearsto age 16 years

CAPA used to assesspsychiatric symptoms

Disruptive behavior disorders anddepression were associated with ahigher rate and earlier onset ofsubstance use and abuse in both sexes,but anxiety predicted later onset ofsmoking. Onset of conduct disorderand anxiety was well before the firstuse of any substance. Substance useand abuse of every type were morecommon among girls with behaviordisorders; depressed boys, comparedwith nondepressed boys, hadsignificantly higher rates of every typeof substance use.


Sunget al. (62)

1,420 children aged 9, 11,and 13 years fromNorth Carolina

3–7 years (untilparticipant aged16 years, reviewedannually)

CAPA used in interviews;substance use, abuse, anddependence items used;psychiatric disorders classifiedby using DSM-IV criteria forsymptoms experienced in thepast 3 months

Conduct disorder, particularly at earlierages (14 and 15 years) was thestrongest psychiatric predictor oftransition from substance use to adisorder (OR 5 2.0 (95% CI: 1.2, 3.6)).Boys, but not girls, with a history ofdepression were at increased risk ofsubstance use disorder; anxietyincreased the risk of substance usedisorder in girls at age 16 years.

* CIDI, Composite International Diagnostic Interview; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; OR, odds ratio; CI, confidence interval; DIA-X/M-CIDI;

Munich-Composite International Diagnostic Interview; CAPA, Child and Adolescent Psychiatric Assessment; DSM-III-R, Diagnostic and Statistical Manual of Mental Disorders, Third Edition,

Revised; DISC, Diagnostic Interview Schedule for Children; IRR, incidence rate ratio; CES-D, Center for Epidemiological Studies Depression; DIS, Diagnostic Interview Schedule; K-SADS,

Schedule for Affective Disorders and Schizophrenia and School-Age Children.

164

Cerdaetal.

EpidemiolR

ev

2008;30:155–177

symptoms at follow-up, but depression did not predict sub-stance use disorders at follow-up (34).

Adults. Comorbidity between depression and sub-stance use persists until adulthood: there is a two- to four-fold higher risk of developing one disorder if the other hasoccurred (37, 38). As has been detected in adolescence andyoung adulthood, support also exists for the use of alcoholand illegal substances as a way to manage depressive symp-toms in adulthood (37, 39–41). Several studies also docu-ment a reciprocal association between the two, as was thecase with anxiety (42, 43). Gilman and Abraham (42) fol-lowed up participants from the Epidemiologic CatchmentArea Survey and found that, although the odds of alcoholdependence increased in relation to baseline depressivesymptoms, the odds of major depression also increased inresponse to the number of alcohol symptoms and a diagnosisof alcohol dependence.

Anxiety and depression

High rates of comorbidity have been found between anx-iety and depression. An important fraction of depressioncases also present symptoms of anxiety, to the point thatKrueger et al. (8) proposed that ‘‘pure’’ cases of depressionor anxiety would be unrepresentative of the entire spectrumof these disorders. In a 40-year follow-up of levels of anx-iety and depression, the probability of having anxiety forthose who already suffered from depression ranged from0.54 to 0.98, while, for those who did not suffer from de-pression, the probability ranged from 0.03 to 0.10 (44).Anxiety and depression function in a cohesive manner andhave been classified as internalizing disorders, which arestable over time (8, 45). The comorbidity of these disordershas raised some controversy. While some argue that anxietyand depression are distinct disorders with different etiologicprofiles and high levels of cooccurrence, others interpret theoverlap as indicative of similar psychopathologic processesand shared risks or as different manifestations of the samedisease (26, 46).

Children and adolescents. Strong heterotypic conti-nuity exists between depression and anxiety from childhoodto adulthood (2, 18, 26, 47–51). In the Great Smoky Moun-tains study that followed a cohort of children aged 9–13years for 5 years, anxiety predicted depression and viceversa, even when controlling for current comorbidity (18).A gender-specific analysis indicated, however, that the asso-ciations independent of current comorbidity persistedamong girls only. Another study that followed those aged9–18 years over 9 years found that overanxious disorder inadolescence was associated with adult major depression,while major depression in adolescence predicted general-ized anxiety in adulthood (50). These associations were in-dependent of gender.

The degree of comorbidity between depression and anx-iety may vary as a function of age. Wittchen et al. (26)followed 3,021 respondents aged 14–24 years at baselineover 4 years and found that, although only 9 percent of thosewith an anxiety disorder at ages 14–17 years also had majordepressive episodes, the proportion was considerably higheramong those aged 18–24 years. Moreover, depression in-

creased considerably among those with anxiety disorder atfinal follow-up: almost two thirds of cases were comorbidwith major depression. The variation in comorbidity by agemight be a function of age at onset of the two disorders;whereas anxiety disorders tend to start in childhood andearly adolescence, depressive disorders increase in late ad-olescence and continue to rise in those aged 20 years orolder (2, 26, 52).

The frequency and type of comorbidity differ by disorder.Wittchen et al. (26) found that the odds ratio for presentingmajor depression at follow-up ranged from 1.7 for specificphobia to 3.4 for panic disorder and 3.9 for generalizedanxiety disorder. Pine et al. (50) also found that majordepression predicted generalized anxiety disorder but notsimple or social phobia.

Comorbidity is related to symptom severity, degree ofimpairment, course, and outcome. Wittchen et al. (26)found, in a study of 3,021 respondents aged 14–24 yearsat first interview followed over 4–5 years, that the numberof anxiety disorders present, persistence of anxious avoid-ance behavior, and degree of psychosocial impairment werethe characteristics most strongly associated with develop-ment of secondary depression. Repetto et al. (51) confirmed,in a more restricted study of African-American adolescentsat risk of school dropout, that adolescents who presentedconsistently high levels of depressive symptoms were morelikely to present a higher number of anxiety symptoms.

Adults. Most studies have identified anxiety as a pri-mary disorder that precedes secondary depression, perhapsrelated to the difference in ages at onset as well as to thehigher degree of stability of anxiety. A critical review oflongitudinal studies of representative samples, however, in-dicates that this question remains unsettled and may dependon life stage examined, length of study follow-up, and typeof anxiety disorder studied. A follow-up of a representativepopulation sample over 10 years from early adolescence toyoung adulthood found that social anxiety disorder wastemporally primary relative to depression in most of thecases and that the risk of subsequent depression was twofoldfor individuals with social anxiety disorder in comparison tothose without social anxiety disorder (53). However, a studythat followed up a birth cohort until age 32 years (54) foundthat major depression preceded generalized anxiety disorderalmost as often as generalized anxiety disorder precededmajor depression. Two studies in older adulthood found thatanxiety preceded depression within short time intervals.A study of older Swedish twins found that over two 3-yearintervals, anxiety symptoms led to depressive symptoms,while the opposite did not occur (46). The same phenome-non was reported among the elderly for depression and gen-eralized anxiety disorder: a follow-up of a sample ofcommunity-living elderly over 3 years found that general-ized anxiety disorder either remitted or progressed into de-pression or mixed anxiety/depression, whereas subjects withdepression or depression/generalized anxiety disorder wereunlikely to develop noncomorbid generalized anxiety disor-der at follow-up (12).

Comorbid disorders may be more stable than ‘‘pure’’cases of disease. A study that followed adults aged 19–20years at baseline over 15 years found that substantial



TABLE 2. Population studies on comorbid patterns between substance abuse, anxiety, depression and conduct disorder among adults



de Graafet al. (78)

7,076 adults aged18–64 years fromthe Netherlands

Retrospective; usedage at onset data

Diagnoses based on DSM-III-R*from the CIDI*

The association between major depression and alcohol ordrug dependence was stronger than that for alcohol ordrug abuse. Men were more likely than women toreport major depression temporally secondary to anyother disorder; of those who had ever experienced amood disorder, 46% of males and 57% of femaleshad a history of anxiety disorders, and 43% of malesand 15% of females had a substance use disorder; inmost anxiety-comorbid cases, mood disorder aroseafter anxiety disorder.


Kushneret al. (24)

454 incomingfreshmen at alarge university

At years 1, 4,and 7

Anxiety disorders (generalizedanxiety disorder, social phobia,agoraphobia with panic attacks,panic attacks in the absence ofagoraphobia): assessed with DIS*version III-R to use criteria fromDSM-III; alcohol abuse/dependence:used DIS and DSM-III

Strong evidence was found for reciprocal causalinfluences between anxiety disorders and alcoholdependence; having an anxiety disorder at years1 or 4 quadrupled the risk for new onset of alcoholdependence at year 7.


Newcombet al. (31)

470 adults fromLos Angeles,California

4 years Polydrug problems assessed withitems based on DSM-IV*criteria for drug abuse anddependence; anxiety (includedin the agitation measure)measured by three items ona 5-point Likert rating scale,taken from the HopkinsSymptom Checklist

Early dysphoria directly increased agitation as an adult.Polydrug problems in young adulthood increasedanxiety in adulthood. Social conformity accountedfully for the relation between polydrug problems andagitation. Marijuana problems in early adulthoodincreased later suicidal ideation. Early cocaineproblems increased later hostility. Dysphoria inyoung adulthood increased alcohol problems butdecreased cocaine abuse in later adulthood.


Sartoret al. (79)

1,269 offspring(average age, 20.1years) of maletwins from theVietnam Era TwinRegistry in theUnited States

Retrospective Drinking outcomes measuredwith Semi-StructuredAssessment for the Geneticsof Alcoholism (onset definedas age at which full DSM-IVcriteria were met); lifetimediagnoses for DSM-IV mood,anxiety, and conduct-relateddisorders also obtained

Nicotine dependence (HR* 5 3.91 (95% CI*: 2.48, 6.17))and generalized anxiety disorder (HR 5 3.45 (95% CI:2.08, 5.72)) were robust predictors of progression fromfirst drink to alcohol dependence; conduct disorder(HR 5 1.75 (95% CI: 1.10, 2.77)) and cannabis abuse(HR 5 1.88 (95% CI: 1.22, 2.90)) were also associatedwith rapid transition to alcohol dependence.


Sonntaget al. (29)

2,548 individualsaged 14–15 yearsat baseline fromMunich, Germany

Average, 19.7months

Psychopathologic and diagnosticassessments based on thecomputer-assisted personalinterview version ofthe DIA-X/M-CIDI*

When age/gender/comorbid depressive disorder werecontrolled for, baseline nonusers with at least onesocial fear but no social phobia diagnosis showedelevated odds for an occurrence of nicotinedependence in the follow-up period (OR* 5 3.85(95% CI: 1.34, 11.00)). All baseline, nondependent,regular smokers with social fears but not socialphobia also showed higher odds for nicotinedependence (OR 5 1.5 (95% CI: 1.01, 2.23)).


Bovasso(80)

1,920 individualsin the Baltimore(Maryland) EpidemiologicCatchment Area Survey

15 years DIS assessed symptoms ofDSM-III disorders at baselineand DSM-III-R symptoms atbaseline; outcomes measuredby using the 20-item versionof Goldberg’s General HealthQuestionnaire to assessgeneral distress, DIS forpsychiatric disorders, andDIS for symptoms of substanceabuse/dependence

Anxiety/depression did not predispose individuals tosubstance use, and substance use did not predisposeindividuals to anxiety/depression.

166

Cerdaetal.

EpidemiolR

ev

2008;30:155–177


Crumet al. (40)

1,920 adults fromBaltimore, Maryland(average age, 41.7years at baseline;range, 18–86)

Median 12.6 yearsof follow-up

Depression measured with DISby using DSM-III criteria; alcoholdependence measured as 1)lifetime and 2) new onset;DSM-III criteria used atbaseline and 1 year follow-up,DSM-III-R criteriaused at 12.6-year follow-up

Depressive syndrome was associated with lifetimeprevalence of alcohol dependence: for women,OR 5 1.89 (95% CI: 1.16, 3.06), while, for men,OR 5 1.83 (95% CI: 1.14, 2.93). Depressive syndromewas not associated with onset of alcohol dependence.Having a current or prior anxiety disorder increasedthe odds of alcohol dependence among women:OR 5 2.35 (95% CI: 1.28, 4.32).


Dixit andCrum (41)

1,383 women fromBaltimore, Maryland,ranging in age from18 to �65 years,who did not reportheavy alcohol useat baseline

1 year Center for Epidemiologic StudiesDepression Scale heavy alcoholuse measured as five or moredrinks at least once in the pastmonth; depression measuredwith DIS by using DSM-III criteria

The estimated risk of heavy alcohol use for women withmajor depression, depressive syndrome, and/ordysthymia was 2.22 greater than the risk for womenwithout such a history (95% CI: 1.00, 4.92). Theadjusted relative risk of heavy alcohol useassociated with an incremental number oflifetime-experienced depressive symptomswas estimated to be 1.09 (95% CI: 1.01, 1.18).


Gilman andAbraham (42)

14,480 respondentsnot meeting DSM-IIIcriteria for alcoholdependence orlifetime majordepression, meanage, 43 years (range,18–96), in five UScommunity siteslocated near DukeUniversity; JohnsHopkins University;University of California,Los Angeles;Washington University;and Yale University

1 year Both alcohol dependence andmajor depression measuredwith the DIS by using DSM-IIIcriteria; administeredby nonclinician interviewers

The odds of alcohol dependence associated withnumber of baseline depressive symptoms were,for females, 1–3 symptoms, OR 5 2.75 (95% CI:1.52, 5.00), 4–6 symptoms, OR 5 3.52 (95%CI: 1.86, 6.66), �7 symptoms, OR 5 7.88 (95%CI: 2.86, 21.70); for males, 1–3 symptoms, OR 5 1.50(95% CI: 1.08, 2.08). The odds of depression associatedwith 1–3, 4–6, and �7 alcoholic symptoms,respectively, were, for females, OR 5 1.66 (95% CI:1.05, 2.64), OR 5 3.98 (95% CI: 2.37, 6.69), andOR 5 4.32 (95% CI: 1.92, 9.75); for males, OR 5 1.19(95% CI: 0.76, 1.88) OR 5 2.49 (95% CI: 1.26, 4.92),and OR 5 2.12 (95% CI: 0.90, 5.00). The odds ofmajor depression corresponding to a DSM-III alcoholdependence diagnosis were 3.52, 95% CI: 2.16, 5.72for females and 1.77 (95% CI: 1.08, 2.91) for males.


Hettemaet al. (43)

2,926 male and1,929 female adulttwin subjects aged21–62 years inRichmond, Virginia

Retrospective (basedon age-at-onsetdata)

Lifetime diagnoses for DSM-III-Rmajor depressive disorder,generalized anxiety disorder,panic disorder, phobias,alcohol dependence, andpsychoactive substance usedisorder assessed viatelephone and face-to-facestructured psychiatric interviewsbased on the modified SCID*

Generalized anxiety disorder and major depressivedisorder had the highest same-year onsets (about41%); panic disorder tends to occur after majordepressive disorder in females but before majordepressive disorder in males; the majority ofphobias precede major depressive disorder in bothgenders. Onset of alcohol dependence precedesthat of major depressive disorder in about one thirdof cases; for psychoactive substance use disorder,onset for females was about evenly divided betweenthose occurring before (44%) and after (49%) majordepressive disorder, significantly more males(54%) reported onset before that of major depressivedisorder. Generalized anxiety disorder and panicdisorder cases had the highest risk of developingconcurrent major depressive disorder, with same-year-onset HRs of 38.07 and 12.28, respectively.

Table continues

Comorbid

Form

sofPsychopathology

167

EpidemiolR

ev

2008;30:155–177

TABLE 2. Continued



Kuoet al. (37)

7,477 twins,including 3,302females with amean age whenlast assessed of36.5 years and4,175 males witha mean age of 37.0years in Virginia

Two measurements:waves 1 and 4 forfemale-to-femaletwin pairs and waves1 and 2 for male-to-male and male-to-female twin pairs

DSM-III-R diagnoses ofmajor depression andalcohol dependenceassessed by using adaptedversions of standard structuredinterviews (not specified)

Risk of developing alcohol dependence associated withlifetime or concurrent major depression was higher formales than for females. Risk of alcohol dependence inyear 1 associated with major depression onset in theprior year was higher than that associated withlifetime major depression: twofold for males andfourfold for females. Among females, the risk ofalcohol dependence associated with prior majordepression was higher than that associated withconcurrent major depression (HR 5 5.1 vs. HR 53.1). Lifetime alcohol dependence predictedincreasing HR of major depression for bothgenders, and the effect was stronger for females.Concurrent alcohol dependence had the highesteffects on major depression risk. However,compared with lifetime alcohol dependence,preceding alcohol dependence did not increasemajor depression risk.

Anxiety,depression, andsubstanceuse/abuse

de Graafet al. (81)

4,796 adults fromthe Netherlands

Baseline, 1 year,3 years

CIDI used Of respondents who developed an incident disorder,16.1% developed comorbid disorders; anxiety-comorbid mood disorder was the most frequentincident comorbid disorder among females, andboth anxiety-comorbid and substance use-comorbidmood disorders were the most frequent among males.

Anxiety,depression, andsubstanceuse/abuse

Johnet al. (38)

Random adultpopulation sampleaged 18–64 years(n 5 4,075) fromnorthern Germany(Lubeck andsurroundingcommunities)

Baseline, 30months, and36 months

Nicotine dependence, affectiveand anxiety disordersaccording to DSM-IV assessedwith the computer-assistedDIA-X/M-CIDI

Male former or current daily smokers had higher oddsratios of affective and anxiety disorders: comparedwith male never smokers, male current dailysmokers had 4.5 higher odds of having ananxiety disorder. Having 1–2 nicotine dependencesymptoms was associated with higher odds ofaffective (OR 5 1.7 (95% CI: 1.2, 2.4)) and anxiety(OR 5 1.6 (95% CI: 1.2, 2.2)) disorders; having �3symptoms was associated with 3.6 times higherodds of affective disorders (95% CI: 2.6, 5.0) and3.7 higher odds of anxiety disorders (95% CI:2.8, 5.0).


Beesdoet al. (53)

3,021 individualsfrom Munich,Germany, aged14–24 years atbaseline and 21–34years at follow-up

10 years Individuals assessed withthe DIA-X/M-CIDI

Social anxiety disorder was consistently associated withsubsequent depression, independent of age at onset ofsocial anxiety disorder (relative risk range: 1.49–1.85).


Breslauet al. (48)

1,007 individualsfrom a list of allmembers of a largehealth maintenanceorganization aged21–30 years inMichigan

3.5 years DIS used, following DSM-III-Rcriteria

Higher lifetime prevalence of major depressive disorderin females vs. males was primarily for majordepressive disorder comorbid with anxiety; priorhistory of an anxiety disorder was associated withincreased risk of major depressive disorder in malesand females. Prior anxiety disorder accounted forthe higher prevalence of major depressive disorderin females.

168

Cerdaetal.

EpidemiolR

ev

2008;30:155–177


Breslauet al. (47)

1,007 individualsfrom a list of allmembers of a largehealth maintenanceorganization aged21–30 years inMichigan

3.5 years DIS used, following DSM-III-Rcriteria

Prior anxiety was associated with a higher risk of majordepressive disorder in both sexes. Women were notmore vulnerable than men to depression becauseof an anxiety disorder. Substance use disorder doesnot play the comparable role in men that anxietyplays in women in contributing to depression onset.


de Graafet al. (9)

7,065 adults aged18–65 years inthe Netherlands

Baseline, 1 year,3 years

CIDI used 15.2% of pure mood, 10.5% of anxiety, and 6.8% ofsubstance use disorder cases became comorbid; halfthe transitions from pure substance use disorders wereinto mood-comorbid and half into anxiety-comorbid.


Grantet al. (57)

102 undergraduateswith an average ageof 18.64 years froman undergraduatecollege in theUnited States

1 year Social anxiety measured by usingthe Social Phobia and AnxietyInventory; depressive symptomsmeasured by using the BeckDepression Inventory

22% of the total effect of time 1 social anxiety on time2 depressive symptoms was accounted for byavoidance of expressing emotions. Individuals withsocial anxiety disorder may develop depressivesymptoms to the extent that they withhold strongemotions, which may lead to a feeling of loss of self.


Merikangaset al. (55)

4,547 subjects(2,201 men and2,346 women)aged 19 and 20years representativeof the canton ofZurich, Switzerland

15 years The Structured DiagnosticInterview for Psychopathologicand Somatic Syndromes usedto make diagnoses accordingto DSM-III and DSM-III-Rcriteria for most majorcategories

Concurrent comorbidity between anxiety and depressionwas highly stable across the 15 years of the study(OR range: 3.0–6.0); subjects with anxiety only had amoderate probability of developing depression only(OR range: 1.5–7.8), whereas those with depressiononly did not develop anxiety only (OR range: 0.3–1.8);substantial transitions from pure anxiety or puredepression to comorbid depression/anxiety (OR range:1.5–10.0); women were more likely than men totransition from pure anxiety or depression to comorbidconditions at follow-up.


Moffittet al. (54)

1,037 members of abirth cohort (972 wereassessed by the lastmeasurement) inNew Zealand

21 years (studiedat ages 11, 13,15, 18, 21, 26,and 32 years)

At ages 11, 13, and 15 years,diagnoses with DSM-III criteriamade with the DISC*-ChildVersion and, at older ages,with the DIS (at ages 18 and21 years, DSM-III-R; at ages26 and 32 years, DSM-IV);both generalized anxietydisorder and major depressivedisorder diagnosed

Cumulative prevalence of adult comorbid majordepressive disorder 1 generalized anxiety disorderwas 12% in the cohort; sequential and cumulativecomorbidity was balanced: �37% of cases ofdepression were preceded by anxiety disorders,but 63% were not; the two disorders had strongcumulative lifetime comorbidity from ages 11 to32 years (OR 5 5.3 (95% CI: 3.9, 7.2)); of lifetimemajor depressive disorder cases, 48% had lifetimeanxiety disorder; of lifetime anxiety cases, 72% hadlifetime major depressive disorder.


Wetherellet al. (46)

1,391 Swedish twinswith an average ageof 60.9 years at firstmeasurement

Interviewed in 1987,1990, and 1993

Depressive symptoms assessedwith the Center forEpidemiologic StudiesDepression Scale; anxietyassessed with the StateAnxiety subscale of theState-Trait Personality Inventory

A two-factor model with latent depression and anxietyfactors provided the best fit to the data; the two factorswere highly correlated (0.84). Between 1987–1990 and1990–1993, the parameters for the paths linking anxietyto subsequent depression were significant, whereas thepaths linking depression to subsequent anxiety werenot significant and could be dropped from the model.


Wilhelmet al. (56)

156 young adultsenrolled in apostgraduateteachers’ trainingprogram in theUnited States

15 years DIS and subsequently its newerform, CIDI, used to find casesof depression and anxiety; DISand CIDI used to generateDSM-III and then DSM-III-Rdiagnoses

Subjects with two or more episodes of major depressionhad higher rates of meeting ‘‘major anxiety’’ and ‘‘allanxiety’’ status than those with one episode. Thosewith one episode had higher rates than subjects inthe ‘‘no depression’’ group. Mean age at onset ofsimple phobia was significantly younger among thosewho had multiple episodes of depression. One of thebest predictors to distinguish between ‘‘no depression’’and ‘‘one or more episodes’’ was major anxiety.

Table continues

Comorbid

Form

sofPsychopathology

169

EpidemiolR

ev

2008;30:155–177

TABLE 2. Continued



Schoeverset al. (12)

2,173 subjectsaged 65–84 yearsfrom the AmsterdamStudy of the Elderly

3 years Dutch translation of the Mini-MentalState Examination, the GeriatricMental State Examination, theActivities of Daily Living scale, theInstrumental Activities of DailyLiving scale, and the CambridgeMental Disorders of the ElderlyExamination interview used;diagnoses of depression, organic‘‘caseness,’’ and generalizedanxiety reached by use of theGeriatric Mental State–AGECATsystem

Depression and generalized anxiety disorder together hada significantly worse prognosis (27% remission) than the‘‘pure’’ categories: 4 of 199 subjects with depressiondeveloped generalized anxiety disorder (puregeneralized anxiety disorder), but 27 (13.6%)developed pure depression/generalized anxietydisorder; subjects with pure depression/generalizedanxiety disorder at baseline were more likely tokeep the combined disorder than have it changeinto pure forms.


Nocket al. (82)

3,199 respondentsaged 18–44 yearsin the NationalComorbidity SurveyReplication in theUnited States

Retrospective study(based on age-at-onset data)

Mental disorders assessed byusing CIDI to generatediagnoses accordingto DSM-IV criteria and theICD-10* diagnostic system

Oppositional defiant disorder is associated withsubstantial risk of secondary mood, anxiety, impulse-control, and substance use disorders; 92.4% ofrespondents with lifetime oppositional defiant disordermet criteria for at least one other lifetime disorder;oppositional defiant disorder was significantly comorbidwith agoraphobia without panic disorder (OR 5 2.1)and conduct disorder (OR 5 12.6). The ORs forimpulse control disorders were 4.0–12.6, and theORs for anxiety disorders were 2.1–4.5.


Angstet al. (39)

591 adults (aged 19–20years at baseline)selected from arepresentative sampleof Zurich, Switzerland:two thirds were highscorers in theSymptom Checklist90-R Global SeverityIndex (GSI), and onethird were a randomsample of thosewhose GSI scorewas below the85th percentile

20 years Alcohol abuse and dependenceclassified according to DSM-IVcriteria; major depressiveepisodes defined by DSM-IIIcriteria in the first two interviews(1979 and 1981) and DSM-III-Rcriteria in the last four interviews(1986, 1988, 1993, and 1999)

Major depressive episodes were associated with alcoholdependence but not with alcohol abuse: for men,OR 5 3.8 (95% CI: 2.0, 7.1); for women, OR 5 5.2(95% CI: 1.8, 14.5).


Kratzer andHodgins (63)

6,449 males and6,268 femalesfrom Sweden

30 years Conduct problems in school:teachers rating students on a3-point scale at grades 6 and 9;conduct problems in thecommunity: defined as childrenreferred to the Child WelfareCommittee; mental disorders:Stockholm county registerscreened to documentadmissions to psychiatric wards

Males identified with conduct problems in childhoodwere more likely than the no-conduct-problems groupto develop substance abuse in adulthood; femalesidentified with conduct problems in childhood were atgreater risk than no-conduct-problems females for crimeand for mental disorders in adulthood; a diagnosis ofsubstance abuse accounted for most of the mentaldisorders among conduct-problem-school but notconduct-problem-community females.


Johnsonet al. (69)

658 individualsfrom New York

18–21 years total:assessed in 1983,1985–1986,1991–1993, andfinally in2001–2004

Axis I disorders assessed withDISC-I, the nonpatient versionof the structured clinicalinterview for DSM-IV(SCID-IV-NP) at age 33 years;assessment of personalitydisorders with items from thePersonality DiagnosticQuestionnaire, the StructuredClinical Interview for PersonalityDisorders (SCID-II) and DISC-I

Individuals with a DSM-IV cluster A (paranoid, schizoid,or schizotypal) or cluster C (avoidant, dependent,obsessive-compulsive) panic disorder by mean age 22years were at elevated risk of recurrent or chronicunipolar depression; antisocial, borderline, dependent,depressive, histrionic, and schizotypal panic disordertraits, identified between the ages of 14 and 22 years,were significantly associated with risk of dysthymicdisorder or major depressive disorder.

170

Cerdaetal.

EpidemiolR

ev

2008;30:155–177

transitions occurred from anxiety and depression alone tocomorbidity; an average of 21 percent of those with depres-sion alone and 24 percent of those with anxiety alone de-veloped comorbid anxiety and depression at baseline, and,once comorbidity developed, the recurrence of either disor-der alone was much lower (55).

Specificity also exists in the degree of comorbidity be-tween different types of anxiety disorders and depression.Generalized anxiety disorder has been identified as the typeof anxiety disorder most closely connected with major de-pression (54). Wilhelm et al. (56) followed a cohort of sub-jects involved in teacher training in 1978–1993 and founda stronger association between number of depressive epi-sodes and panic disorder, generalized anxiety disorder, andagoraphobia than between number of depressive episodesand either simple or social phobia. Gregory et al. (49) fol-lowed a representative birth cohort from ages 11 to 32 yearsand found that adult anxiety cases were more likely to haveexperienced juvenile depression than those without adultanxiety, except for specific phobia and panic disorder cases.

The specific nature of the associations between depressionand anxiety may depend on certain interpersonal features ofanxiety disorders. Two studies have found, for example, thatbehavioral inhibition (53) and avoidance of expressing emo-tions (57), rather than lack of assertion or interpersonal de-pendency, explained part of the transition from socialanxiety disorder to depression. Grant et al. (57) hypothesizedthat inhibition of strong emotions to significant others asa strategy to maintain associations may lead to a loss of self,which may increase depressive symptoms.

Conduct disorder and substance use

Conduct disorder has been defined in the Diagnostic andStatistical Manual of Mental Disorders: DSM-IVas a perva-sive pattern of aggressive and deceptive behavior that beginsin adolescence (58). Conduct disorder and substance usehave been classified under the same latent construct of ex-ternalizing disorders (8). Measelle et al. (36) found, for ex-ample, that an externalizing factor accounted for 79 percentand 95 percent of the baseline levels of antisocial behaviorand substance abuse, respectively. Both disorders tend toincrease through late adolescence and peak in young adult-hood, and they are more prevalent among males (59). Sev-eral studies have proposed that a major fraction of theassociation can be attributed to common genetic factors (45).

Children and adolescents. Most studies on conductdisorder and substance use concern adolescence and youngadulthood. Similar to anxiety and depression, the direction-ality of the link between these conditions and conduct seemsto vary across studies (18, 30, 49, 60–62). Cohen et al. (60)followed up 749 adolescents from ages 13 to 24 years andfound that participants with conduct disorder in early ado-lescence experienced significantly elevated rates of alcoholabuse/dependency and marijuana use disorder between earlyadolescence and young adulthood. Kratzer and Hodgins(63) found a similar phenomenon when they followed anunselected birth cohort from pregnancy to 30 years of age:childhood conduct problems were associated with a higherrisk of severe substance abuse. White et al. (64) moved

Conductdisorder

andanxiety

Johnson

etal.(72)

658individuals

from

AlbanyandSaratoga

countie

s,New

York

18–21years

total:

assessedin

1983,

1985–1986,

1991–1993,and

finally

in2001–2004

Axis

Idisorders

assessedwith

DISC-Iatages14–21years,the

nonpatie

ntversionofthe

structuredclinicalinterview

forDSM-IV(SCID-IV-N

P)at

age33years;personality

disorders:Personality

Diagnostic

Questio

nnaire,the

StructuredClinicalInterview

for

PersonalityDisorders

(SCID-II),

andDISC-Iused

Antisocialdisorderbetween14and22years

ofage

wasasso

ciatedwith

elevatedriskofanxiety

disorders

bymeanage33years.Antisocialpersonalitytraits

were

associatedwith

riskofagoraphobia

atage

33years

(OR

51.68(95%

CI:1.14,2.47)).

Conductdisorder

andanxiety

Nock

etal.(82)

3,199adults

aged

�18years

inthe

coterm

inous

Unite

dStates

Retrospective

CIDIto

diagnosebasedon

DSM-IVandICD-10

Conductdisordertypically

precedesmoodand

substanceusedisorders

butoccurs

mostoftenafter

impulseandanxiety

disorders;presenceofconduct

disorderis

associatedwith

anelevatedriskof

substanceusedisorders

(OR

55.9);conductdisorder

predictedthepresenceofcomorbid

anxiety

disorders

(OR

51.5,p5

0.038).

*DSM-III-R

,Diagnostic

andStatisticalM

anualo

fMentalD

isorders,ThirdEdition,Revised;CIDI,Composite

Internatio

nalD

iagnostic

Interview;DIS,Diagnostic

InterviewSchedule;DSM-

IV,Diagnostic

andStatisticalM

anualo

fMentalD

isorders,FourthEdition;HR,hazard

ratio

;CI,confidenceinterval;DIA-X/M

-CIDI;Munich-C

omposite

Internatio

nalD

iagnostic

Interview;OR,

oddsratio

;SCID,StructuredClinicalInterviewforDSM-III-R

;DISC,Diagnostic

InterviewSchedule

forChildren;ICD-10,Internatio

nalS

tatisticalC

lassificatio

nofDiseaseandRelatedHealth

Problems,Tenth

Revision.



beyond examining the influence of conduct disorder on lev-els of substance abuse and investigated its influence on bothmean levels of use and change in use over time. Althoughconduct disorder was associated with higher overall levelsof alcohol and marijuana use throughout the study period, itpredicted change in levels of alcohol use only. Participantswith higher levels of conduct disorder began adolescencewith higher levels of alcohol use, but their level of alcoholuse grew slower than that among participants who had lowerinitial levels of conduct disorder—possibly because of re-gression to the mean.

Some of the most promising studies on comorbidity haveinvestigated the impact that conduct disorder and substanceabuse have on the developmental patterns of the comorbiddisorder. Conduct disorder, for example, has been found toplay an important role in the shift from substance use tosubstance use disorders (30, 65). The Great Smoky Mountainsstudy found that conduct disorder, particularly at earlierages (14 and 15 years), was the strongest psychiatricpredictor of the transition from substance use to a disorder(62). Substance abuse can play a role in crime desistance inyoung adulthood (45, 66). Hussong et al. (66) found thatsubstance abuse may exert both proximal and distal effectson antisocial desistance over young adulthood: 1) substanceabuse early in young adulthood can slow an individual’spattern of criminal desistance relative to the lifetime patternof criminality; and 2) substance abuse can result in ‘‘time-specific’’ elevations in antisocial behavior relative to an in-dividual’s own antisocial trajectory. A community sample offemale adolescents aged 13–15 years found that, althoughinitial levels of antisocial behavior predicted increases insubstance abuse, substance abuse also predicted slower de-celeration in antisocial behavior (36); a study of female andmale twin pairs reported that youths with late-onset andpersisting antisocial behavior had higher rates of substanceuse than youths in groups without a history of antisocialbehavior or a pattern of antisocial desistance (67).

Adults. We found one study that examined the influ-ence of conduct disorder on substance abuse but used retro-spective reports of age at onset for comorbid disorders. TheNational Comorbidity Survey Replication indicated thatlifetime conduct disorder was associated with 5.9 timeshigher odds for substance use disorders (58). Conduct disor-der was more likely to occur before substance use disorders88.5 percent of the time. Although both active and remittedconduct disorder were significant predictors of substance usedisorders, the risk of substance use disorders was signifi-cantly higher for those with active conduct disorder. Thepersistent effect of remitted conduct disorder may indicatethat conduct disorder is a risk marker for unmeasured com-mon causes of substance use disorders or that active conductdisorder has long-lasting consequences that persist even af-ter the disorder has remitted. The higher risk associated withactive conduct disorder may also, however, point to a causalrole that conduct disorder plays in related disorders.

Conduct disorder and anxiety and depressive disorders

The limited number of studies identified that addressedthe connections between conduct disorder and either depres-

sion or anxiety led us to group the review of these disordersinto one section. Factor analyses of internalizing disorders,notably anxiety and depression, and externalizing disorders,of which conduct disorder constitutes an important fraction,have identified distinct developmental patterns for these twogroups of disorders but have also detected a correlationacross time between the two types of disorders (8, 45). Littleis understood about the mechanisms that may connect con-duct disorder with mood disorders. Conduct disorder maydisrupt interpersonal functioning, contributing to greaterconflict with parents and peers and greater social rejectionand academic failure, which may generate a sense of re-peated failure and feelings of anxiety or depression (68).At the same time, depression and anxiety may impede socialdevelopment and generate interpersonal conflict, which maycontribute to the onset of conduct problems. Common etio-logic factors may also underlie both types of disorders (69).

The developmental course of conduct disorder tends to beintertwined with internalizing problems, particularly de-pression (36). For example, a study involving a communitysample of female adolescents followed from early to lateadolescence found that initial depressive and antisocial be-havior symptoms predicted future increases in the other:while initial levels of antisocial symptoms predicted in-creases in depressive symptoms, initial levels of depressionpredicted a slower deceleration in antisocial problems (36).Early antisocial problems have also been found to predictdepressive symptoms in adulthood (69–71). Kim et al. (70)found that, among young men followed from early adoles-cence to young adulthood, early antisocial behavior wasassociated with depressive symptoms in adulthood; Silberget al. (71) detected a correlation between early conductproblems and later depression among adolescent femaletwin pairs. The link may be explained by a failure model,whereby early developmental failures associated withantisocial behavior increase vulnerability to depressivesymptoms (36).

Some evidence also exists of a link between conductdisorder and specific types of anxiety (49, 58, 61, 72).A birth cohort of a representative sample followed fromages 11 to 32 years showed an association between a juvenilehistory of conduct disorder and posttraumatic stress disorderat age 32 years (49). A longitudinal study of a populationsample assessed from ages 14 to 33 years also found that thepresence of antisocial personality disorder traits at ages 14–22years was associated with a higher risk of anxiety disordersby age 33 years, most particularly for agoraphobia (72).The National Comorbidity Survey Replication found mixedevidence about the temporal order between conduct disorderand anxiety: conduct disorder occurred after specific andsocial phobia but prior to all other types of anxiety disorders(58). The findings from the National Comorbidity SurveyReplication require replication within a longitudinal studybecause they are based on retrospective reports of age at onset.

Putting it all together: internalizing and externalizingdisorders as predictors of substance abuse

Most studies on comorbidity assess the interaction be-tween pairs of disorders concurrently and over time, but

172 Cerda et al.


they fail to determine how clusters of disorders may pro-mote onset of psychopathology. One exception has been thearea of youth risk behaviors; selected studies have investi-gated the relative influence of externalizing and internaliz-ing disorders on the prevalence of substance abuse orconduct disorder. Adolescence has long been consideredthe stage at which risk behaviors such as delinquency andsubstance use are initiated and when disorders such as sub-stance abuse become consolidated (36, 73). It is a time whenmultiproblem youth emerge, characterized by persistentsubstance use, persistent delinquency, and/or persistentinternalizing problems (74). This peak in comorbidityhas motivated studies of the intersections between mul-tiple forms of psychopathology in adolescence and youngadulthood.

Loeber et al. (74), for example, used longitudinal datafrom three samples of the Pittsburgh Youth Study to exam-ine the cooccurrence of persistent substance use with de-linquent and persistent internalizing problems in boys aged7–18 years. The joint occurrence of mood dysregulation, inthe form of internalizing problems, and behavioral dysregu-lation, in the form of delinquency, emerged in preadoles-cence and played a significant role in the persistence ofsubstance use. During adolescence, however, only persistentdelinquency predicted persistent substance use. Anotherpublication using the Pittsburgh Youth Study expanded thisinvestigation by examining how the cooccurrence of con-duct disorder and depressive symptoms identified high-riskyouth (25). The authors found that higher levels of depres-sion led to the development of alcohol use disorder symp-toms but among only those subjects who had high levels ofearly conduct disorder symptoms. The interaction betweendepression and conduct disorder may indicate a need tomanage negative affective states without a high regard forsocial norms, or it may indicate a subgroup of individuals atparticularly high risk of developing a severe form of sub-stance use.

Summary

Prospective population studies that investigate the se-quential links between comorbid disorders remain scarce.The bulk of prospective studies assess the influence of theprevalence of one disorder on the future prevalence of thecomorbid disorder. Few studies actually move beyond thistopic to investigate the influence that comorbidity may haveon changes in the state of psychopathology, that is, in shiftsfrom persistence to desistance, or from substance use toabuse, or from disorder-free status to onset. Such informa-tion would provide initial evidence of the specific mecha-nisms by which disorders influence each other. Moreover,differences in the duration of study follow-up, choice of agegroups, and selections of measurement instruments and di-agnostic criteria make comparability across studies moredifficult.

Despite these limitations, several points emerge from ourreview of population-based studies of comorbid disorders.The concurrent and sequential links among conduct disor-der, substance use/abuse, anxiety, and depression are neitherrandom nor a result of bias from help-seeking clinical sam-

ples. The nature of comorbidity is specific to the type ofdisorder under study: evidence exists for a reciprocal rela-tion between substance use and depression, anxiety, andconduct disorder, while the bulk of studies on depressionand anxiety indicate that anxiety may precede onset of de-pression. The studies that address comorbidity of conductdisorder with anxiety and depression indicate that conductdisorder may precede these mood disorders, but the paucityof studies that test a reciprocal association impede us frommaking conclusive statements.

The mechanisms that connect disorders remain in ques-tion. Three hypotheses link comorbid disorders: 1) the as-sociation may be indirectly causal, so that a primarydisorder may exert neurophysiologic, individual, or socialchanges that increase vulnerability to the secondary disor-der; 2) disorders may be directly linked, so a primary dis-order may, for example, exert neurophysiologic changes thatcontribute to onset of the secondary disorder; or 3) a set ofcommon risk factors, such as common genes or the experi-ence of childhood trauma, may explain the connection be-tween two disorders. However, these hypotheses are difficultto distinguish from one another. For example, a social factorsuch as affiliation with deviant peers could predispose ado-lescents to both substance use disorder and conduct disorder(hypothesis 3), but conduct disorder could also lead to as-sociation with deviant peers, which would contribute to sub-stance use disorder (hypothesis 1). A review of the riskfactors associated with the initiation and maintenance ofcomorbid conditions can help illuminate the mechanismsthat underlie comorbidity in psychopathology.

CONCLUSIONS

Understanding patterns of comorbidity is essential in or-der to evaluate psychiatric nosology and fully understandthe developmental trajectories of key forms of psychopa-thology. To date, a number of studies have shown that anx-iety, depression, substance abuse, and conduct disordercluster in individuals across the life course. This criticalreview is one of the first focused on population-based, lon-gitudinal studies of comorbidity between these four keydisorders.

The four psychiatric disorders we reviewed have beenclassified under two latent constructs: externalizing and in-ternalizing disorders. Although factor analyses indicatea higher correlation between disorders within each con-struct, clustering also exists between externalizing and in-ternalizing disorders. Approaching first the issue ofinternalizing disorders, anxiety and depression functionjointly from childhood to adulthood. Comorbidity seemsto be a function of age; it increases from late adolescenceonward, possibly because of the later age at onset of de-pression relative to anxiety. Comorbidity also depends onthe type of anxiety disorder concerned: several studiesfound a higher clustering between major depressive disorderand generalized anxiety disorder, for example, than withother anxiety disorders. In terms of externalizing disorders,conduct disorder and substance abuse show a strong recip-rocal link, particularly in adolescence and young adulthood.



Conduct disorder and substance abuse not only influence theonset and prevalence of the other but also influence the shiftbetween developmental stages of the other disorder, thatis, in shifts from substance use to a substance use disorder,or in the likelihood of moving from conduct disorder todesistance.

Studies have documented a link between disorders acrossthe internalizing/externalizing divide. Studies indicate thatthe relation between substance abuse and anxiety or depres-sion is reciprocal: substance abuse alone or depression/anxiety alone predicts onset of the other disorder. The im-pact of anxiety on substance abuse remains in question:anxiety disorders have been associated with both a higherand a lower likelihood of substance use disorders. The di-rection of the association may depend on the type of anxietydisorder and substance in question. Research has also ad-dressed the question of clustering between conduct disorderand anxiety/depression, although research is less developedin this area. Depression and conduct disorder play a role inshaping the developmental trajectory of the other. Initiallevels of depression have been associated with a slower rateof desistance from conduct disorder, while conduct disorderpredicted increases in depressive symptoms from adoles-cence to adulthood. The links between anxiety and conductdisorder are less understood. The few existing studies focuson adolescence and young adulthood and have indicated thatinitial levels of conduct disorder were associated withhigher levels of anxiety.

Existing research on comorbidity has a number ofstrengths. Prospective studies linking the presence of onedisorder to onset of the other, as well as examining the in-fluence that a disorder may have on developmental shifts incomorbids, are growing, and they offer promise in enablingus to understand the specific ways in which comorbid dis-orders are linked over the life course. Moreover, research isshifting from a focus on a pair of disorders to a more holisticapproach that analyzes how clusters of disorders interactwith each other. We think this line of investigation will bekey in understanding the complexity of comorbid relationsand in estimating the health burden that results from multi-ple comorbidities.

At the same time, the epidemiologic literature falls shortin its consideration of key issues within comorbidity re-search. First, although early research in this area is promis-ing, a need exists for study designs that make it possible totest the timing and specific nature of comorbid relations—that is, shifting from merely investigating whether disordersare ‘‘associated’’ with each other to understanding the waysthat disorders deflect a comorbid pair from specific stages inits normative trajectory. This process involves studying theinfluence of disorders in promoting qualitative shifts in thecomorbid disorder, from alcohol use initiation to abuse todependence, for example. It also implies accounting for thespectrum of relevant comorbid disorders within the samestudy to control for spurious associations and understandhow the presence of a third disorder may influence thecausal link between comorbid pairs.

Second, our understanding is limited in terms of the spe-cific comorbid patterns between subtypes of disorders.Although we recognize that specific types of anxiety disorders,

for example, may show different comorbidity patterns withdepression, or that alcohol abuse may have different levelsof comorbidity with anxiety than illicit drug abuse does, thescarcity of available, population-based, longitudinal studieson each specific subtype combination prevented us frommaking critical distinctions between comorbid patterns.Studies that compare comorbid patterns between disordersubtypes across key developmental stages, from childhoodto adulthood, offer a promising way to address this gap.

Third, one of the key questions that remains relates to thecauses of comorbidity: the direction and mechanisms un-derlying causal links, as well as the potential spurious natureof such links. Investigating factors at multiple levels of in-fluence as potential confounders, mediators, triggers, ormodifiers of comorbid associations would enable us to iden-tify groups at highest risk of the presentation of multipledisorders and distinguish features of the individual and hisor her environment that are most malleable to preventiveintervention. Moreover, it would enable us to establishwhether associations between comorbid disorders are trulycausal or are spurious phenomena resulting from commonrisk factors. Little effort has been invested in systematicallydocumenting the associations between these disordersacross different age groups or in reviewing the factors asso-ciated with the cooccurrence of such comorbid forms ofpsychopathology. Multilevel multivariate models that con-sider the joint influence of factors at the individual, family,and community levels on a set of comorbid disorders pro-vide a promising avenue to attack the question of commonversus distinct etiologies of comorbid disorders.

ACKNOWLEDGMENTS

This project was funded by the Robert Wood JohnsonFoundation Health and Society Scholars Program (M. C.)and the National Institutes of Health (grants DA 017642,DA 022720, MH 08259, and MH 078152 (S. G.)).

Conflict of interest: none declared.

REFERENCES

1. de Graaf R, Bijl RV, Smith F, et al. Risk factors for 12-monthcomorbidity of mood, anxiety, and substance use disorders:findings from the Netherlands Mental Health Survey andIncidence Study. Am J Psychiatry 2002;159:620–9.

2. Stein MB, Fuetsch M, Muller N, et al. Social anxiety disorderand the risk of depression—a prospective community study ofadolescents and young adults. Arch Gen Psychiatry 2001;58:251–6.

3. Angold A, Costello EJ, Erkanli A. Comorbidity. J ChildPsychol Psychiatry 1999;40:57–87.

4. Angst J. Comorbidity of mood disorders: a longitudinal pro-spective study. Br J Psychiatry 1996;168:31–7.

5. Kendler KS, Prescott CA, Myers J, et al. The structure ofgenetic and environmental risk factors for common psychiatricand substance use disorders in men and women. Arch GenPsychiatry 2003;60:929–37.

174 Cerda et al.


6. Bourdon KH, Rae DS, Locke BZ, et al. Estimating theprevalence of mental disorders in U.S. adults from theEpidemiologic Catchment Area Survey. Public Health Rep1992;107:663–8.

7. Kessler RC, Berglund P, Demler O, et al. Lifetime prevalenceand age-of-onset distributions of DSM-IV disorders in theNational Comorbidity Survey Replication. Arch Gen Psychi-atry 2005;62:593–602.

8. Krueger RF, Caspi A, Moffitt TE, et al. The structure andstability of common mental disorders (DSM-III-R): a longitu-dinal-epidemiological study. J Abnorm Psychol 1998;107:216–27.

9. de Graaf R, Bijl RV, ten Have M, et al. Pathways tocomorbidity: the transition of pure mood, anxiety and sub-stance use disorders into comorbid conditions in a longitudinalpopulation-based study. J Affect Disord 2004;82:461–7.

10. Bakken K, Landheim AS, Vaglum P. Axis I and II disorders aslong-term predictors of mental distress: a six-year prospectivefollow-up of substance-dependent patients. BMC Psychiatry2007;7:29–41.

11. Renouf AG, Kovacs M, Mukerji P. Relationship of depressive,conduct, and comorbid disorders and social functioning inchildhood. J Am Acad Child Adolesc Psychiatry 1997;36:998–1004.

12. Schoevers RA, Deeg DJH, van Tilburg W, et al. Depressionand generalized anxiety disorder—co-occurrence and longi-tudinal patterns in elderly patients. Am J Geriatr Psychiatry2005;13:31–9.

13. Aseltine RH, Gore S, Colten ME. The co-occurrence ofdepression and substance abuse in late adolescence. DevPsychopathol 1998;10:549–70.

14. Kraemer HC, Wilson KA, Hayward C. Lifetime prevalenceand pseudocomorbidity in psychiatric research. Arch GenPsychiatry 2006;63:604–8.

15. Kessler RC, Chiu WT, Demler O, et al. Prevalence, severity,and comorbidity of 12-month DSM-IV disorders in theNational Comorbidity Survey Replication. Arch Gen Psychi-atry 2005;62:617–27.

16. Merikangas KR, Angst J, Eaton W, et al. Comorbidity andboundaries of affective disorders with anxiety disorders andsubstance misuse: results of an international task force. Br JPsychiatry Suppl 1996;Jun:58–67.

17. Kushner MG, Abrams K, Borchardt C. The relationshipbetween anxiety disorders and alcohol use disorders: a reviewof major perspectives and findings. Clin Psychol Rev 2000;20:149–71.

18. Costello EJ, Mustillo S, Erkanli A, et al. Prevalence anddevelopment of psychiatric disorders in childhood and ado-lescence. Arch Gen Psychiatry 2003;60:837–44.

19. Dierker LC, Avenevoli S, Merikangas KR, et al. Associationbetween psychiatric disorders and the progression of tobaccouse behaviors. J Am Acad Child Adolesc Psychiatry 2001;40:1159–67.

20. Goodwin RD, Fergusson DM, Horwood LJ. Associationbetween anxiety disorders and substance use disorders amongyoung persons: results of a 21-year longitudinal study.J Psychiatr Res 2004;38:295–304.

21. Hayatbakhsh MR, Najman JM, Jamrozik K, et al. Cannabisand anxiety and depression in young adults: a large pro-spective study. J Am Acad Child Adolesc Psychiatry 2007;46:408–17.

22. Zimmermann P, Wittchen HU, Hofler M, et al. Primary anxietydisorders and the development of subsequent alcohol usedisorders: a 4-year community study of adolescents and youngadults. Psychol Med 2003;33:1211–22.

23. Costello EJ, Egger HL, Angold A. The developmentalepidemiology of anxiety disorders: phenomenology, preva-lence, and comorbidity. Child Adolesc Psychiatr N Am 2005;14:631–48.

24. Kushner MG, Sher KJ, Erickson DJ. Prospective analysis ofthe relation between DSM-III anxiety disorders and alcoholuse disorders. Am J Psychiatry 1999;156:723–32.

25. Pardini D, White HR, Stouthamer-Loeber M. Early adolescentpsychopathology as a predictor of alcohol use disorders byyoung adulthood. Drug Alcohol Depend 2007;88(suppl):S38–49.

26. Wittchen HU, Kessler RC, Pfister H, et al. Why do peoplewith anxiety disorders become depressed? A prospective-longitudinal community study. Acta Psychiatr Scand 2000;102:14–23.

27. Wittchen HU, Frohlich C, Behrendt S, et al. Cannabis use andcannabis use disorders and their relationship to mentaldisorders: a 10-year prospective-longitudinal communitystudy in adolescents. Drug Alcohol Depend 2007;88(suppl):S60–70.

28. Kaplow JB, Curran PJ, Angold A, et al. The prospectiverelation between dimensions of anxiety and the initiationof adolescent alcohol use. J Clin Child Psychol 2001;30:316–26.

29. Sonntag H, Wittchen HU, Hofler M, et al. Are social fears andDSM-IV social anxiety disorder associated with smoking andnicotine dependence in adolescents and young adults? EurPsychiatry 2000;15:67–74.

30. Costello EJ, Erkanli A, Federman E, et al. Development ofpsychiatric comorbidity with substance abuse in adolescents:effects of timing and sex. J Clin Child Psychol 1999;28:298–311.

31. Newcomb MD, Vargas-Carmona J, Galaif ER. Drug problemsand psychological distress among a community sample ofadults: predictors, consequences, or confound? J CommunityPsychol 1999;27:405–29.

32. Chinet L, Plancherel B, Bolognini M, et al. Substance use anddepression. Comparative course in adolescents. Eur ChildAdolesc Psychiatry 2006;15:149–55.

33. Fergusson DM, Goodwin RD, Horwood LJ. Major depressionand cigarette smoking: results of a 21-year longitudinal study.Psychol Med 2003;33:1357–67.

34. Rao U, Daley SE, Hammen C. Relationship between de-pression and substance use disorders in adolescent womenduring the transition to adulthood. J Am Acad Child AdolescPsychiatry 2000;39:215–31.

35. Windle M, Davies PT. Depression and heavy alcohol useamong adolescents: concurrent and prospective relations. DevPsychopathol 1999;11:823–44.

36. Measelle JR, Slice E, Hogansen JM. Developmental trajecto-ries of co-occurring depressive, eating, antisocial and sub-stance abuse problems in female adolescents. J AbnormPsychol 2006;115:524–38.

37. Kuo PH, Gardner CO, Kendler KS, et al. The temporalrelationship of the onsets of alcohol dependence and majordepression: using a genetically informative study design.Psychol Med 2006;36:1153–62.

38. John U, Meyer C, Rumpf HJ, et al. Smoking, nicotinedependence and psychiatric comorbidity—a population-basedstudy including smoking cessation after three years. DrugAlcohol Depend 2004;76:287–95.

39. Angst J, Gamma A, Endrass J, et al. Is the association ofalcohol use disorders with major depressive disorder a conse-quence of undiagnosed bipolar-II disorder? Eur Arch Psychi-atry Clin Neurosci 2006;256:452–7.



40. Crum RM, Storr CL, Chan YF. Depression syndromes withrisk of alcohol dependence in adulthood: a latent classanalysis. Drug Alcohol Depend 2005;79:71–81.

41. Dixit AR, Crum RM. Prospective study of depression and therisk of heavy alcohol use in women. Am J Psychiatry 2000;157:751–8.

42. Gilman SE, Abraham HD. A longitudinal study of the order ofonset of alcohol dependence and major depression. DrugAlcohol Depend 2001;63:277–86.

43. Hettema JM, Prescott CA, Kendler KS. The effects ofanxiety, substance use and conduct disorders on riskof major depressive disorder. Psychol Med 2003;33:1423–32.

44. Murphy JM, Horton NJ, Laird NM, et al. Anxiety anddepression: a 40-year perspective on relationships regardingprevalence, distribution, and comorbidity. Acta PsychiatrScand 2004;109:355–75.

45. Krueger RF, Hicks BM, Patrick CJ, et al. Etiologic connec-tions among substance dependence, antisocial behavior, andpersonality: modeling the externalizing spectrum. J AbnormPsychol 2002;111:411–24.

46. Wetherell JL, Gatz M, Pedersen NL. A longitudinal analysis ofanxiety and depressive symptoms. Psychol Aging 2001;16:187–95.

47. Breslau N, Chilcoat H, Schultz LR. Anxiety disorders and theemergence of sex differences in major depression. J GendSpecif Med 1998;1:33–9.

48. Breslau N, Schultz L, Peterson E. Sex differences in de-pression: a role for preexisting anxiety. Psychiatry Res 1995;58:1–12.

49. Gregory AM, Caspi A, Moffitt TE, et al. Juvenile mentalhealth histories of adults with anxiety disorders. Am JPsychiatry 2007;164:301–8.

50. Pine DS, Cohen P, Gurley D, et al. The risk for early-adulthoodanxiety and depressive disorders in adolescents with anxietyand depressive disorders. Arch Gen Psychiatry 1998;55:56–64.

51. Repetto PB, Caldwell CH, Zimmerman MA. Trajectories ofdepressive symptoms among high risk African-Americanadolescents. J Adolesc Health 2004;35:468–77.

52. Wittchen HU, Nelson CB, Lachner G. Prevalence of mentaldisorders and psychosocial impairments in adolescents andyoung adults. Psychol Med 1998;28:109–26.

53. Beesdo K, Bittner A, Pine DS, et al. Incidence of socialanxiety disorder and the consistent risk for secondary de-pression in the first three decades of life. Arch Gen Psychiatry2007;64:903–12.

54. Moffitt TE, Harrington H, Caspi A, et al. Depression andgeneralized anxiety disorder—cumulative and sequentialcomorbidity in a birth cohort followed prospectively to age 32years. Arch Gen Psychiatry 2007;64:651–60.

55. Merikangas KR, Zhang HP, Avenevoli S, et al. Longitudinaltrajectories of depression and anxiety in a prospectivecommunity study—the Zurich cohort study. Arch Gen Psy-chiatry 2003;60:993–1000.

56. Wilhelm K, Parker G, Dewhurst-Savellis J, et al. Psycholog-ical predictors of single and recurrent major depressiveepisodes. J Affect Disord 1999;54:139–47.

57. Grant DM, Beck JG, Farrow SM, et al. Do interpersonalfeatures of social anxiety influence the development ofdepressive symptoms? Cogn Emot 2007;21:646–63.

58. Nock MK, Kazdin AE, Hiripi E, et al. Prevalence, subtypes,and correlates of DSM-IV conduct disorder in the NationalComorbidity Survey Replication. Psychol Med 2006;36:699–710.

59. Hicks BM, Blonigen DM, Kramer MD, et al. Gender differ-ences and developmental change in externalizing disordersfrom late adolescence to early adulthood: a longitudinal twinstudy. J Abnorm Psychol 2007;116:433–47.

60. Cohen P, Chen H, Crawford TN, et al. Personality disorders inearly adolescence and the development of later substance usedisorders in the general population. Drug Alcohol Depend2007;88(suppl):S71–84.

61. Lavigne JV, Cicchetti C, Gibbons RD, et al. Oppositionaldefiant disorder with onset in preschool years: longitudinalstability and pathways to other disorders. J Am Acad ChildAdolesc Psychiatry 2001;40:1393–400.

62. Sung MJ, Erkanli A, Angold A, et al. Effects of age at firstsubstance use and psychiatric comorbidity on the developmentof substance use disorders. Drug Alcohol Depend 2004;75:287–99.

63. Kratzer L, Hodgins S. Adult outcomes of child conductproblems: a cohort study. J Abnorm Child Psychol 1997;25:65–81.

64. White HR, Xie MG, Thompson W, et al. Psychopathology asa predictor of adolescent drug use trajectories. Psychol AddictBehav 2001;15:210–18.

65. Robins LN, McEvoy L. Conduct problems as predictors ofsubstance abuse. In: Robins LN, Rutter M, eds. Straight anddevious pathways from childhood to adulthood. Cambridge,United Kingdom: Cambridge University Press, 1990:182–204.

66. Hussong AM, Curran PJ, Moffitt TE, et al. Substance abusehinders desistance in young adults’ antisocial behavior. DevPsychopathol 2004;16:1029–46.

67. Marmorstein NR, Iacono WG. Longitudinal follow-up ofadolescents with late-onset antisocial behavior: a pathologicalyet overlooked group. J Am Acad Child Adolesc Psychiatry2005;44:1284–91.

68. Ingoldsby EM, Kohl GO, McMahon RJ, et al. Conductproblems, depressive symptomatology and their co-occurringpresentation in childhood as predictors of adjustment in earlyadolescence. J Abnorm Child Psychol 2006;34:603–21.

69. Johnson JG, Cohen P, Kasen S, et al. Personality disorder traitsassociated with risk for unipolar depression during middleadulthood. Psychiatry Res 2005;136:113–21.

70. Kim HK, Capaldi DM, Stoolmiller M. Depressive symptomsacross adolescence and young adulthood in men: predictionsfrom parental and contextual risk factors. Dev Psychopathol2003;15:469–95.

71. Silberg J, Rutter M, D’Onofrio B, et al. Genetic andenvironmental risk factors in adolescent substance use. J ChildPsychol Psychiatry 2003;44:664–76.

72. Johnson JG, Cohen P, Kasen S, et al. Personality disordersevident by early adulthood and risk for anxiety disordersduring middle adulthood. J Anxiety Disord 2006;20:408–26.

73. Wiesner M, Windle M. Young adult substance use anddepression as a consequence of delinquency trajectories duringmiddle adolescence. J Res Adolesc 2006;16:239–63.

74. Loeber R, Stouthamer-Loeber M, White HR. Developmentalaspects of delinquency and internalizing problems and theirassociation with persistent juvenile substance use betweenages 7 and 18. J Clin Child Psychol 1999;28:322–32.

75. Juon HS, Ensminger ME, Sydnor KD. A longitudinal study ofdevelopmental trajectories to young adult cigarette smoking.Drug Alcohol Depend 2002;66:303–14.

76. Stice E, Barrera M Jr, Chassin L. Prospective differentialprediction of adolescent alcohol use and problem use:examining mechanisms of effect. J Abnorm Psychol 1998;107:616–28.

176 Cerda et al.


77. Rohde P, Lewinsohn PM, Seeley JR. Psychiatric comorbiditywith problematic alcohol use in high school students. J AmAcad Child Adolesc Psychiatry 1996;35:101–9.

78. de Graaf R, Bijl RV, Spijker J, et al. Temporal sequencing oflifetime mood disorders in relation to comorbid anxiety andsubstance use disorders—findings from the NetherlandsMental Health Survey and Incidence Study. Soc PsychiatryPsychiatr Epidemiol 2003;38:1–11.

79. Sartor CE, Lynskey MT, Heath AC, et al. The role of childhoodrisk factors in initiation of alcohol use and progression toalcohol dependence. Addiction 2007;102:216–25.

80. Bovasso G. The long-term treatment outcomes of depressionand anxiety comorbid with substance abuse. J Behav HealthServ Res 2001;28:42–57.

81. de Graaf R, Bijl RV, ten Have M, et al. Rapid onset ofcomorbidity of common mental disorders: findings from theNetherlands Mental Health Survey and Incidence Study(NEMESIS). Acta Psychiatr Scand 2004;109:55–63.

82. Nock MK, Kazdin AE, Hiripi E, et al. Lifetime prevalence,correlates, and persistence of oppositional defiant disorder:results from the National Comorbidity Survey Replication.J Child Psychol Psychiatry 2007;48:703–13.



comorbid forms of psychopathology: key patterns and future

Documents