0
We're unable to sign you in at this time. Please try again in a few minutes.
Retry
We were able to sign you in, but your subscription(s) could not be found. Please try again in a few minutes.
Retry
There may be a problem with your account. Please contact the AMA Service Center to resolve this issue.
Contact the AMA Service Center:
Telephone: 1 (800) 262-2350 or 1 (312) 670-7827  *   Email: subscriptions@jamanetwork.com
Error Message ......
Original Article |

Familiality of Symptom Dimensions in Depression FREE

Ania Korszun, PhD, MD, MRCPsych; Valentina Moskvina, PhD; Shyama Brewster, BSc Joint (Hons); Nick Craddock, MB, PhD, MRCPsych; François Ferrero, MD; Michael Gill, MD, MRCPsych, FTCD; Ian Richard Jones, MB, PhD, MRCPsych; Lisa Anne Jones, BSc (Hons), PhD; Wolfgang Maier, MD; Ole Mors, MD, PhD; Michael J. Owen, PhD, FRCPsych; Martin Preisig, MD, MPH; Theodore Reich, MD; Marcella Rietschel, MD; Anne Farmer, MD, FRCPsych; Peter McGuffin, MB, PhD, FRCP, FRCPsych
[+] Author Affiliations

From the Department of Psychological Medicine, University of WalesCollege of Medicine, Cardiff (Dr Korszun, Moskvina, Craddock, and Owen), GlaxoSmithKlineMedical Genetics, Middlesex (Ms Brewster), Department of Psychiatry, Universityof Birmingham, Birmingham (Drs I. R. Jones and L. A. Jones), and Medical Research Council Social Genetic and Developmental PsychiatryCentre, Institute of Psychiatry, King's College, London (Drs Farmer and McGuffin), England; Department of Psychiatry, University of Geneva,Geneva (Dr Ferrero), and Departement Universitaire de Psychiatrie Adult, Prilly-Lausanne(Dr Preisig), Switzerland; Department of Psychiatry, Trinity Centre for HealthSciences, St James' Hospital, Dublin, Ireland (Dr Gill); Department of Psychiatry,University of Bonn, Bonn (Dr Maier), and Division of Genetic Epidemiologyin Psychiatry, Central Institute of Mental Health, Mannheim (Dr Rietschel),Germany; Department of Psychiatric Demography, Psychiatric Hospital in Aarhus,Aarhus, Denmark (Dr Mors); and Department of Psychiatry, Washington UniversitySchool of Medicine, St Louis, Mo (Dr Reich).


Arch Gen Psychiatry. 2004;61(5):468-474. doi:10.1001/archpsyc.61.5.468.
Text Size: A A A
Published online

Background  Depression is a clinically heterogeneous disorder thought to result from multiple genes interacting with environmental and developmental components. A dimensional rather than a categorical approach to depressive phenotype definition may be more useful for identification of susceptibility genes.

Objectives  To perform an exploratory factor analysis on a range of depressive and anxiety symptoms in a large, well-defined sample of depressed siblings, as well as a confirmatory factor analysis in a separate large group of unrelated depressed subjects, and to analyze correlations of identified symptom dimensions between depressed siblings.

Design  Subjects (N = 1034), including 475 sibling pairs, with a history of at least 2 depressive episodes were recruited from the Depression Network Study, a large-scale multicenter collection of families affected by recurrent unipolar depression. Subjects were interviewed using the Schedules for Clinical Assessment in Neuropsychiatry (SCAN) and diagnosed according to the DSM-IV and the International Classification of Diseases, 10th Revision, using a computerized scoring program (CATEGO5). Factor analysis was carried out on 26 depression symptom items, including 4 anxiety screening items. Confirmatory factor analysis was performed on an independent sample of 485 depressed individuals.

Results  Four interpretable factors were identified: (1) mood symptoms and psychomotor retardation; (2) anxiety; (3) psychomotor agitation, guilt, and suicidality; and (4) appetite gain and hypersomnia. For each symptom group, a quantitative scale was constructed, and correlations between siblings were calculated. There was a moderate degree of sibling homotypia for some depressive symptoms, and factors 1, 2, and 3 showed significant positive familial correlation (0.145 [P = .001], 0.335 [P<.001], and 0.362 [P<.001], respectively).

Conclusions  This is the first study of large, well-defined samples of depressed subjects in whom symptom dimensions have been derived and then confirmed using independent material. The significant correlations between siblings for 3 of the dimensions suggest substantial familial, perhaps genetic, etiologies.

Figures in this Article

For the last 30 years, depression has been conceptualized mainly asa single syndrome that represents the final common pathway of a range of etiologicalfactors. Both current classification systems of psychiatric disorders reflectthis view (DSM-IV and InternationalClassification of Diseases, 10th Revision [ICD-10]). However, depression comprises disparate symptoms, with disturbancesof mood, thinking, sleep, appetite, and motor activity that do not all occurin the same distribution in every depressed individual. There is also considerableclinical overlap between symptoms of depression and anxiety. A unified biologicalexplanation of depression has not been established, and this has led to questionsabout the validity of present classification systems and to conjectures aboutbiologically distinct subcategories of depression.1

Depression is a clinically heterogeneous disorder thought to resultfrom the interaction of genetic and environmental factors.24 However,before susceptibility genes can be identified, it is crucial to achieve anoptimal definition of depressive phenotypes. Broadly speaking, 2 approacheshave been used for this: a categorical one in which individuals are fittedinto subcategories that are separate and mutually exclusive, or a dimensionalone in which symptoms are grouped together within different symptom complexesor "symptom dimensions" that can coexist to different degrees in individualpatients. In this study, we explore using a dimensional approach to delineatethe genetic architecture of depression.

Factor analysis is a useful tool for discovering structures in multivariatedata and allows the major groupings, or dimensions, of correlated symptomsto be identified by looking at the highest factor loading for each symptom.Exploratory factor analysis starts from the raw data without any preconceptions,yielding suggested factors that are consistent with the data but must be justifiedby a plausible interpretation. Confirmatory factor analysis, on the otherhand, is used to confirm or reject an assumed factor structure by testingit against independent data.

In the past, the construct of depression has been subjected to variousfactor analytic studies exploring different concepts of subtypes of depression,ranging from a 2-factor model of neurotic vs endogenous depression57 to a 10-factor model,8 with more recent focus on the atypical depressivesubtype.9 This approach has renewed potentialin light of current understanding of a genetic contribution to the etiologyof depression. For instance, a particular symptom dimension might representthe action of a contributory gene or group of genes, and a particular combinationof such genes may result in a characteristic phenotype. Identification ofsymptom dimensions could thus lead to definition of a more appropriate classificationsystem for depression, particularly for genetic studies.

Subsequent demonstration that symptom dimensions identified in thisway are correlated in pairs of affected relatives would provide external validationof the proposed factor structure and suggest usefulness for genetic studies.This type of approach has been used to explore the classification and heritabilityof schizophrenia,10,11 but therehave been fewer studies of the familiality of symptom dimensions in unipolardepression. Although previous work has shown familiality between categoricalsubtypes of major depression,12,13 toour knowledge, there has not been a large systematic study of sibling correlationsof depressive dimensions derived by factor analysis.

The aims of this study were therefore (1) to identify depressive symptomdimensions by performing an exploratory factor analysis on a range of depressiveand anxiety symptoms in a large, well-defined sample of depressed siblingsdiagnosed using accepted standardized criteria; (2) to perform a confirmatoryfactor analysis in a separate large group of unrelated depressed subjectsassessed using the same clinical research methods to validate the resultsof the exploratory factor analysis; and (3) to provide further validationof the identified symptom dimensions by analyzing their correlations betweendepressed siblings.

SUBJECTS

The main part of the analysis was carried out on subjects recruitedfor a large international multicenter genetic study of siblings with depression(depression network study [DeNT]) conducted at the following 8 clinical centers:St Louis, MO, London, England, Cardiff, Wales, Birmingham, England, Dublin,Ireland, Lausanne, Switzerland, Aarhus, Denmark, and Bonn, Germany. Confirmatoryfactor analysis was carried out on subjects recruited from a multicenter case-controldepression study (depression case-control study [DeCC]) conducted in Birmingham,Cardiff, and London. Ethical approval was first obtained from the appropriatelocal ethics committees in each of the countries involved, and every participantgave written informed consent.

Both studies used similar methods for subject ascertainment. Subjectswere identified from psychiatric clinics, hospitals, and general medical practicesand from volunteers responding to media advertisements. White subjects olderthan 18 years were included if they had experienced 2 or more episodes ofunipolar depression of at least moderate severity, separated by at least 2months of remission, as defined by DSM-IV and ICD-10.In the DeNT study, subjects were included if theyhad at least 1 full sibling older than 18 years meeting the same inclusioncriteria. Subjects were excluded if either sibling was adopted or if theywere the monozygotic twin of any other sibling in the study.

Exclusion criteria for the DeCC and DeNT included a history of psychoticsymptoms that were mood incongruent or present when there was no evidenceof a mood disturbance, intravenous drug use with a lifetime diagnosis of dependency,depression occurring solely in relation to alcohol or substance abuse, ordepression secondary to medical illness or medication. Subjects were alsoexcluded from both studies if there was a clear diagnosis of bipolar disorder,schizophrenia, schizoaffective disorder, or transient psychotic disordersin first- or second-degree relatives.

CLINICAL ASSESSMENT

All subjects were interviewed using the Schedules for Clinical Assessmentin Neuropsychiatry (SCAN),14 a set of instrumentsvalidated in assessing, measuring, and classifying the symptoms of major adultpsychiatric disorders. Subjects identified their 2 worst episodes of depression,and SCAN items were rated from the worst and second worst episodes. Most itemsare coded on an ordinal scale indicating the presence and severity of items(general rating of anxiety, general rating of phobias, sleep problem withdepressed mood, and morning depression, 0-1; hypersomnia and appetite gain,0-2; suicide or self-harm, 0-4; and the remainder of items, 0-3). The ratingsfrom the SCAN interviews were entered into a computerized scoring program,CATEGO5, which provides diagnoses according to DSM-IV and ICD-10 operational definitions.

AGREEMENT BETWEEN RATERS ACROSS SITES

All interviewers from each site undertaking the DeNT study attendeda 4-day SCAN training course in the United Kingdom. Additional interraterreliability meetings were held regularly at each site, and annually the interviewersfrom all sites took part in a joint interrater reliability exercise, witha mean κ across centers of 0.77 (range, 0.63-0.89), giving a substantiallevel of interrater agreement.15 Raters fromthe 3 United Kingdom sites undertaking the DeCC study also participated ina 4-day SCAN training program and regular local interrater reliability meetings.In addition, intersite joint audiotape rating sessions were undertaken viatelephone conferencing.

DATA ANALYSIS
Correlation of Symptoms Between Episodes

Spearman rank correlation coefficients were calculated for depressivesymptoms between worst and second worst episodes.

Correlation of Symptoms Between Siblings

To remove the effects of sex and age on the depressive symptoms, eachsymptom was adjusted by age and sex in the DeCC and DeNT samples. The SASPROC MIXED (SAS Institute, Cary, NC) procedure was used to fit a linear modelwith age as a covariate and sex as a fixed effect. The residuals were consideredas continuous variables and used for further analysis. Data were also analyzedwith symptoms adjusted for the age at first onset of depression; however,as no significant effect of age at onset was found, all results are presentedbased on age and sex adjustment only.

Intraclass correlations were calculated for adjusted depressive symptomsbetween affected sibling pairs. Intraclass correlations were derived as (MSb − MSw)/(MSb + [k − 1]MSw), where MSb and MSw are mean squares between and withinsiblings, respectively, obtained using analysis of variance for the random-effectsmodel,11 and k is the number of subjects inthe class (k = 2 for sib pairs). In families with more than 2 members, eachsibling pair contributed 1 independent pair, each trio (proband, sibling 1,and sibling 2) contributed 2 independent pairs (proband-sibling 1 and proband-sibling2), each quartet contributed 3 independent pairs, and so on.

EXPLORATORY FACTOR ANALYSIS

Factor analysis was performed on 26 adjusted depression symptom itemsfrom the SCAN interview questions, representing a broad range of depressivesymptoms, and 4 screening questions for the presence of anxiety disorders,using SAS PROC FACTOR. To assess the effect of latent dependencies betweensib pairs stemming from the same family, we adjusted each symptom for thefamily (random) effect in addition to the sex and age adjustment. The SASPROC MIXED procedure was used to fit a mixed model with age as a covariate,sex as a fixed effect, and family effect as a random effect. The residualswere used for the factor analysis. Initial factors were extracted using theprincipal components method, and rotations were then performed by the PROMAXmethod. To simplify interpretation, different rotations are used: orthogonalrotation is used if the assumption is that factors are uncorrelated, and oblique(including PROMAX) rotation is used if the factors are correlated with eachother, as in the present analysis. The number of meaningful factors was determinedby the scree plot.

CONFIRMATORY FACTOR ANALYSIS

Confirmatory maximum likelihood factor analysis, using structural equationmodeling, was performed to test the factor construction obtained from theDeNT data against the DeCC data (Figure 1). The PROC CALIS in SAS was used to carry out the confirmatoryfactor analysis. Several goodness-of-fit measures were used to assess differentversions of the construction: the goodness-of-fit index (GFI) and the GFIadjusted for degrees of freedom,16 comparativefit index,17 nonnormed fit index,18 androot-mean-square error of approximation.19 Toprove that a model fits the data, the accepted standard requirements for thegoodness-of-fit indices are as follows: GFI, GFI adjusted for degrees of freedom,and comparative fit index greater than 0.95; nonnormed fit index greater than0.9; and root-mean-square error of approximation less than 0.05.2023

Place holder to copy figure label and caption

The linear model used in the confirmatory factor analysis. Four-factormodel with cross-loadings and factor and error covariances. Unbroken linesconnect symptom boxes with factors and errors. Dashed lines indicate cross-loadings;curved lines, factor covariances; and e1 through e26, error covariances (forclarity, the error covariances are not plotted).

Graphic Jump Location
FAMILIALITY OF SYMPTOM DIMENSIONS

A quantitative scale was constructed for each of the symptom dimensionsidentified. All sex- and age-adjusted symptom items were used to constructquantitative scales, and subjects were scored by calculating the weightedmean of items present for each symptom dimension, with corresponding factorloadings as weight. To identify the significance of the familial effect, intraclasscorrelations were calculated.

DEPRESSION NETWORK STUDY

Individuals (N = 1034) were recruited from all 8 sites, and all individualswere included in the factor analyses. In some cases, families had to be excluded(eg, because of noncompliance or insufficient severity of depression in thesibling). Analysis was conducted on 403 families and 156 single subjects,with 403 probands and 475 siblings yielding 486 sib pairs. The sex distributionwas 31% men and 69% women. The overall age range at assessment was 18 to 80years; the mean ± SD age was 45 ± 12 years. The mean ±SD age at onset of the first episode of depression was 24 ± 12 years(range, 3-74 years). The mean ± SD period between assessment and theworst episode of depression was 9.1 ± 9.8 years. Table 1 shows the characteristics of the sample obtained from eachsite.

Table Graphic Jump LocationTable 1. Sex and Age of Subjects Recruited at Each Site
DEPRESSION CASE-CONTROL STUDY

Four hundred eighty-five depressed individuals (31.9% men and 68.1%women) were recruited from the 3 United Kingdom sites. The age range at assessmentwas 18 to 82 years (mean ± SD age, 47 ± 12 years). Both studieswere designed for genetic analysis. Therefore, to simplify later linkage andassociation analyses, subjects were restricted to those with white parentsand grandparents. The frequency and severity of clinically significant depressivesymptoms (as defined by SCAN) in all subjects are shown in Table 2. The severity was calculated as the ratio of the mean scoreto the maximum score for each symptom.

Table Graphic Jump LocationTable 2. Frequency and Severity of Symptoms in Depression Network Study(DeNT) and Depression Case-Control Study (DeCC) Samples
DATA ANALYSIS
Correlations of Symptoms Between Episodes

Spearman rank correlation coefficients for all symptoms between episodesin the DeNT sample were highly significant (P<.001).Therefore, further analyses were carried out on symptoms reported during theworst episodes.

Correlations of Symptoms Between Siblings

For the between-siblings correlation analysis, the data consisted of346 pairs, 46 trios, 8 quartets, 2 quintets, and 1 sextet. The number of independentpairs from sibling pairs, trios, quartets, quintets, and sextets was therefore346, 92, 24, 8, and 5, respectively. Table3 shows the correlation coefficients for depressive symptoms, correctedfor age and sex effect, between siblings. Restlessness (0.307), anxiety symptoms(0.260-0.306), loss of libido (0.295), and irritability (0.258) showed thehighest correlations.

Table Graphic Jump LocationTable 3. Intraclass Correlations (ICC) of Depressive Symptoms Adjustedfor Age and Sex Between Siblings
EXPLORATORY FACTOR ANALYSIS

To ensure that artifactual groupings did not result from the selectionof several SCAN items that address similar symptoms (eg, initial insomniaand middle sleep period insomnia, problems with thinking, and problems withconcentration), only one item was used for each type of symptom. Psychoticsymptoms (delusions of guilt or worthlessness, delusions of catastrophe, hypochondriacaldelusions in the context of depression, and auditory hallucinations with affectivestate) were excluded from the analysis because they occurred in too few subjects.

All 1034 subjects were used for the exploratory factor analysis. Thescree plot indicated 4 substantive factors, which accounted for 39% of thevariance. The 4 symptom dimensions are shown in Table 4 and comprised the following symptom groupings. Factor 1comprises the mood symptoms, including depressed mood, anhedonia, loss ofhope, loss of reactivity, loss of interest, and low self-esteem. It also includespsychomotor retardation symptoms, with inefficient thinking and loss of energyand libido. Factor 2 comprises the anxiety dimension, including general ratingof anxiety, free-floating anxiety, anxious foreboding with autonomic symptoms,and general rating of phobias. Factor 3 comprises psychomotor agitation withrestlessness and irritability, pathological guilt and guilty ideas of reference,suicidality, and morning worsening of depressed mood. Factor 4 includes appetitegain and hypersomnia negatively correlated with appetite loss and early waking.

Table Graphic Jump LocationTable 4. Rotated Factor Pattern (PROMAX Rotation Method)*

There was cross-loading of suicidality between factors 1 and 3 and cross-loadingof hypersomnia between factors 1 and 4.

CONFIRMATORY FACTOR ANALYSIS

Subjects from the DeCC sample (n = 485) were used in the confirmatoryfactor analysis of the hypothesized factor structure derived from the exploratoryfactor analysis of the DeNT data set. The best fit (GFI, 0.99; GFI adjustedfor df, 0.9; comparative fit index, 0.99; nonnormedfit index, 0.87; and root-mean-square error of approximation, 0.036) was obtainedfor the 4-factor model, which included cross-loadings, correlations betweenfactors, and error covariances. The improvement in model fit after includingthe error covariances suggests that some of the errors are correlated andindicates that the data may also contain a nonlinear finer structure. Nevertheless,the linear modeling yields acceptable fit results and provides a good approximation.

FAMILIALITY OF SYMPTOM DIMENSIONS

Intraclass correlations between sibs were calculated for the 4 factors.Factor 1 showed a low but significant correlation of 0.145 (P = .001). Factors 2 and 3 showed highly significant moderate correlationsof 0.335 (P<.001) and 0.362 (P<.001), respectively. The correlation between siblings for factor4 was 0.075 (P = .052).

Both samples were from genetic studies designed to ascertain individualswith moderate to severe recurrent depression. This is reflected by the typeand frequency of symptoms occurring in this study group, with a 62% frequencyof suicidality and a high frequency of depressive cognitions and disturbancesin thinking. Depressed mood, loss of mood reactivity, and anhedonia were almostubiquitous symptoms, which would be expected as these are core requirementsfor a diagnosis of depression. However, there was only a small proportionof subjects with psychotic symptoms, reflecting not only the fact that psychoticdepression is rare24 but also the method ofrecruitment. Subjects were recruited from outpatient rather than inpatientsettings, and many volunteered in response to media advertisements. Consequently,they represent a moderately ill group, rather than the more severe type ofillness more commonly associated with psychotic symptoms.

There was a high correlation of symptoms between the worst and secondworst episodes within individuals; therefore, further analysis was carriedout on symptoms occurring in the worst episodes. Kendler and colleagues25 have shown that, although there are limitations tousing retrospectively acquired data, depression that is sufficiently severeor disabling as to require treatment tends to be more memorable and thereforemore reliably reported. Furthermore, it has been demonstrated that ratingsfrom a past episode of depression are comparable to contemporary accountsderived from case notes.26 On the other hand,memories of symptoms that occurred during a severe episode of depression mayconfound memory of other episodes, resulting in recall of the same featuresfor all episodes.27 Therefore, in the presentstudy, the use of data from only the worst episodes produced a more robustanalysis.

Reliability was further increased by inclusion of only those individualswho had experienced 2 or more depressive episodes. In this sample, there wasa slightly greater preponderance of women (2.2:1) than is usually found incommunity samples, in which a 1.7 times greater depression risk for womenhas been reported.28 This probably also reflectsrecruitment methods and the fact that women are more likely to volunteer toparticipate in research studies. The higher mean age of 45 years in this studyreflects the fact that only subjects with recurrent depression were included.Overall, the samples used in the present study were representative of moderateto severe nonpsychotic depression.

After eliminating the possible effects of age, age at onset of depression,and sex, factor analysis of a range of SCAN items identified 4 interpretablefactors. The first factor (18% of the variance) comprised depressed mood symptomsthat are almost ubiquitous among individuals with depression, as well as symptomsassociated with psychomotor retardation and loss of libido and self-esteem.The second factor, consisting of the 4 anxiety symptoms, formed an independentdimension of depression accounting for 9% of the variance. There is considerableoverlap between depressive and anxiety symptoms, and anxiety as a symptom29 is the norm rather than the exception in major depression;58% of individuals with a lifetime episode of major depression also meet criteriafor an anxiety disorder.30 In this study, thefrequency of anxiety symptoms ranged from 23% (anxious foreboding) to 58%(general rating of anxiety). Interestingly, there was no cross-loading betweenthe anxiety factor and any other dimension.

The third factor identifies a strong grouping of signs of agitation,such as irritability and restlessness, with suicidality and other depressivecognitions, in particular, guilt. Morning worsening of depressed mood is alsopart of this dimension. The association of suicidality and agitation is interesting;although ideally psychomotor agitation needs to be documented by an observer,irritability and subjective restlessness are indicators of its presence. Parker1 has proposed a hierarchical model of depression inwhich there are separate neurobiological processes generating differing clinicalfeatures. He further postulates that psychomotor disturbance is a distinctcomponent associated with "melancholic" and more severe depression. Suicidalityis an indicator of more severe depression, at least in the sense that it increasesthe risk of mortality from depression, and recognition of this suicidality-agitationdimension has important clinical implications if the presence of agitationpredicts suicide.

The fourth factor comprised increased appetite and hypersomnia negativelycorrelated with early awakening and appetite loss. There has been recent interestin the atypical depressive subtype (ie, hypersomnia with increased weightand appetite),31 and a study13 ofsubtypes of depression in twins found appetite and weight to be among themost discriminating symptoms and findings and identified an atypical depressivesubgroup. In our analysis, we focused on appetite changes rather than weightchanges, as the former is a more reliable measure and not as dependent onthe length of a depressive episode. The strong separation of increased appetiteand sleep into a separate dimension adds some support to the existence ofa separate atypical dimension.

Three of the 4 symptom dimensions showed a significant correlation betweensiblings, adding validation of this factor structure. Factor 1 showed a lowlevel of correlation, but as this largely comprises a group of symptoms thatare obligatory for a diagnosis of depression and thus almost ubiquitous inthis sample and in any sample defined under our present classification systems,this dimension may be more indicative of severity rather than representinga particular phenotypic component, and, whereas heritability contributes tothe latter, severity is not familial.

Factors 2 (anxiety) and 3 (psychomotor agitation, guilt, and suicidality)showed a highly significant moderate degree of correlation (0.335 and 0.362,respectively) between siblings. There was also significant correlation betweensiblings of several individual symptoms (Table 2), but with lower correlations overall than for symptom dimensions,suggesting that dimensions could be more robust phenotypic markers. However,although our findings support the existence of an atypical symptom dimension,it does not appear to have a familial etiology. This dimension and the individualsymptoms of hypersomnia and appetite gain show low sibling correlation.

The correlations between siblings for symptom dimensions reported hereinare much higher than those found in similar factor analyses of schizophrenia10,13 and are more likely to reflect geneticor environmental factors shared between the depressed siblings, rather thanmodifying factors. Genetic effects are the most important contributor to familialaggregation,3 and if the factor structure shownin this study represents different components of genetic liability, then suchfactors could be used in genetic studies to identify more homogeneous subsamplesof depression.

Limitations of this study are that subjects were ascertained in differentways at the various sites, relying on advertisement in some centers, and inothers being mainly recruited from psychiatric clinics; therefore, findingscannot necessarily be generalized to other studies. This study was carriedout on white subjects only; therefore, results cannot be extrapolated to otherethnic groups. Furthermore, there are potential sources of bias in that volunteeringmay be more likely in sibling pairs who have more similar types of depressivesymptoms, and that subjects came from families with more than 1 sib with depression.However, the confirmatory analysis demonstrated similar findings in an independentgroup of single depressed subjects (DeCC). A further limitation, as alreadydiscussed, is that we made cross-sectional assessments of psychiatric symptoms;although a longitudinal or repeated assessment would have been preferable,assessments were retrospective and therefore subject to memory bias.

There has been a long-standing and largely unresolved debate as to whetherdepression is best classified as a collection of syndromes or as a singleentity in which cases differ mainly in terms of severity.32 Inthis study of large, well-defined samples of depressed subjects, symptom dimensionshave been derived and then confirmed using independent material. We also foundhighly significant correlations between siblings. Caution is required wheninterpreting correlation between siblings, but recent twin studies3,33 suggest that shared environmentaleffects in depression are small or nonexistent. We can therefore concludethat the dimensions corresponding to factors 1, 2, and 3 have substantialfamilial, perhaps genetic, etiologies. Although this is not the same as finding2 causally distinct syndromes, the identification of depressive symptom dimensionsprovides the potential for a more refined phenotypic definition for moleculargenetic studies of depression using a quantitative trait locus approach. Furthermore,such dimensions may prove useful in psychopharmacological research, in whichit has been pointed out that the development of new drugs to treat depressionwould be facilitated by dissecting the current "monolithic" definition ofthe disorder into component symptom complexes.34

Corresponding author and reprints: Ania Korszun, PhD, MD, MRCPsych,Department of Psychological Medicine, University of Wales College of Medicine,Heath Park, Cardiff CF14 4XN, United Kingdom (e-mail: akorszun@umich.edu).

Submitted for publication June 2, 2003; final revision received January13, 2004; accepted January 20, 2004.

The DeNT study is funded by grants from GlaxoSmithKline Clinical Geneticsand the DeCC study by the Medical Research Council, United Kingdom.

We acknowledge the following individuals: Sian Caesar; Carly Cooper,BSc; Miriam Craddock, BSc; Fiona Cule, MA; Subodh Dave, MRCPsych; Laura Dean,BSc; Caroline Drain, MHS; Emma Dunn, BSc; Magda Gross, BSc; Andrea Hough,BSc; Izabella Jurewicz, MRCPsych; Svetlana Kovalenko, MD; Louisa Lyon, BSc;Liz McGillivray, BSc; Vicky Swainson, MSc; and Debra Woolway.

Parker  G Classifying depression. Am J Psychiatry. 2000;1571195- 1203
PubMed Link to Article
Kendler  KS Major depression and the environment: a psychiatric genetic perspective. Pharmacopsychiatry. 1998;315- 9
PubMed Link to Article
Sullivan  PFNeale  MCKendler  KS Genetic epidemiology of major depression: review and meta-analysis. Am J Psychiatry. 2000;1571552- 1562
PubMed Link to Article
McGuffin  POwen  MJedGottesman  IIed Psychiatric Genetics and Genomics.  Oxford, England Oxford University Press2002;
Kendell  RE The classification of depressive illness. Br J Psychiatry. 1970;117347- 348
PubMed
Rassaby  EPaykel  ES Factor patterns in depression. J Affect Disord. 1979;1187- 194
PubMed Link to Article
MacFadyen  HW The classification of depressive disorders, I: a review of statisticallybased classification studies. J Clin Psychol. 1975;31380- 394
PubMed Link to Article
Gullion  CMRush  AJ Toward a generalizable model of symptoms in major depressive disorder. Biol Psychiatry. 1998;44959- 972
PubMed Link to Article
Sullivan  PFKessler  RCKendler  KS Latent class analysis of lifetime depressive symptoms in the NationalComorbidity Survey. Am J Psychiatry. 1998;1551398- 1406
PubMed
Cardno  AGJones  LAMurphy  KCSanders  RDAsherson  POwen  MJMcGuffin  P Dimensions of psychosis in affected sibling pairs. Schizophr Bull. 1999;25841- 850
Link to Article
Wickham  HWalsh  CAsherson  PTaylor  CSigmundson  TGill  MOwen  MJMcGuffin  PMurray  RSham  P Familiality of symptom dimensions in schizophrenia. Schizophr Res. 2001;47223- 232
PubMed Link to Article
Leckman  JFWeissman  MMPrusoff  BACaruso  KAMerikangas  KRPauls  DLKidd  KK Subtypes of depression. Arch Gen Psychiatry. 1984;41833- 838
PubMed Link to Article
Sullivan  PFPrescott  CAKendler  KS The subtypes of major depression in a twin registry. J Affect Disord. 2002;68273- 284
PubMed Link to Article
Wing  JKBabor  TBrugha  TBurke  JCooper  JEGiel  RJablenski  ARegier  DSartorius  N SCAN: Schedules for Clinical Assessment in Neuropsychiatry. Arch Gen Psychiatry. 1990;47589- 593
PubMed Link to Article
Everitt  BS Making Sense of Statistics in Psychology.  Oxford, England Oxford University Press1996;291
Joreskog  KGSorbom  D LISREL VI: Analysis of Linear Structural Relationshipsby Maximum Likelihood, Instrumental Variables, and Least Squares Methods. 4th Uppsula, Sweden Dept of Statistics, University of Uppsula1986;
Bentler  PM Comparative fit indexes in structural models. Psychol Bull. 1990;107238- 246
PubMed Link to Article
Bentler  PMBonett  DG Significance tests and goodness of fit in the analysis of covariancestructures. Psychol Bull. 1980;88588- 606
Link to Article
Steiger  JH Structural model evaluation and modification: an interval estimationapproach. Multivariate Behav Res. 1990;25173- 180
Link to Article
Marsh  HWBalla  JWHau  K-T An evaluation of incremental fit indices: a clarification of mathematicaland empirical properties. Marcoulides  GASchumacker  REedsAdvancedStructural Equation Modeling Issues and Techniques. Mahwah, NJ LawrenceErlbaum Associates Inc1996;315- 354
Browne  MWCudeck  R Alternative ways of assessing model fit. Bollen  KALong  JSedsTesting StructuralEquation Models. Newbury Park, Calif Sage Publications1993;136- 161
Byrne  BM Structural Equation Modeling With LISREL, PRELIS,and SIMPLIS: Basic Concepts, Applications, and Programming.  Mahwah, NJ Lawrence Erlbaum Associates Inc1998;
Bollen  KALong  JS Testing Structural Equation Models.  Newbury Park, Calif Sage Publications1993;
Johnson  JHorwath  EWeissman  MM The validity of major depression with psychotic features based on acommunity study. Arch Gen Psychiatry. 1991;481075- 1081
PubMed Link to Article
Kendler  KSNeale  MCKessler  RCHeath  ACEaves  LJ The lifetime history of major depression in women. Arch Gen Psychiatry. 1993;50863- 870
PubMed Link to Article
McGuffin  PKatz  RAldrich  J Past and present state examination: the assessment of "lifetime ever"psychopathology. Psychol Med. 1986;16461- 465
PubMed Link to Article
Rice  JPRochberg  NEndicott  JLavori  PWMiller  C Stability of psychiatric diagnoses. Arch Gen Psychiatry. 1992;49824- 830
PubMed Link to Article
Kessler  RCMcGonagle  KASwartz  MBlazer  DGNelson  CB Sex and depression in the National Comorbidity Survey, I: lifetimeprevalence, chronicity and recurrence. J Affect Disord. 1993;2985- 96
PubMed Link to Article
Ninan  PTBerger  J Symptomatic and syndromal anxiety and depression. Depress Anxiety. 2001;1479- 85
PubMed Link to Article
Kessler  RCNelson  CBMcGonagle  KALiu  JSwartz  MBlazer  DG Comorbidity of DSM-III-R major depressivedisorder in the general population: results from the US National ComorbiditySurvey. Br J Psychiatry Suppl. June1996;17- 30
PubMed
Posternak  MAZimmerman  M Partial validation of the atypical features subtype of major depressivedisorder. Arch Gen Psychiatry. 2002;5970- 76
PubMed Link to Article
Farmer  AMcGuffin  P The classification of the depressions: contemporary confusion revisited. Br J Psychiatry. 1989;155437- 443
PubMed Link to Article
McGuffin  PKatz  RWatkins  SRutherford  J A hospital-based twin register of the heritability of DSM-IV unipolar depression. Arch Gen Psychiatry. 1996;53129- 136
Link to Article
Hyman  SEFenton  WS Medicine: what are the right targets for psychopharmacology? Science. 2003;299350- 351
PubMed Link to Article

Figures

Place holder to copy figure label and caption

The linear model used in the confirmatory factor analysis. Four-factormodel with cross-loadings and factor and error covariances. Unbroken linesconnect symptom boxes with factors and errors. Dashed lines indicate cross-loadings;curved lines, factor covariances; and e1 through e26, error covariances (forclarity, the error covariances are not plotted).

Graphic Jump Location

Tables

Table Graphic Jump LocationTable 1. Sex and Age of Subjects Recruited at Each Site
Table Graphic Jump LocationTable 2. Frequency and Severity of Symptoms in Depression Network Study(DeNT) and Depression Case-Control Study (DeCC) Samples
Table Graphic Jump LocationTable 3. Intraclass Correlations (ICC) of Depressive Symptoms Adjustedfor Age and Sex Between Siblings
Table Graphic Jump LocationTable 4. Rotated Factor Pattern (PROMAX Rotation Method)*

References

Parker  G Classifying depression. Am J Psychiatry. 2000;1571195- 1203
PubMed Link to Article
Kendler  KS Major depression and the environment: a psychiatric genetic perspective. Pharmacopsychiatry. 1998;315- 9
PubMed Link to Article
Sullivan  PFNeale  MCKendler  KS Genetic epidemiology of major depression: review and meta-analysis. Am J Psychiatry. 2000;1571552- 1562
PubMed Link to Article
McGuffin  POwen  MJedGottesman  IIed Psychiatric Genetics and Genomics.  Oxford, England Oxford University Press2002;
Kendell  RE The classification of depressive illness. Br J Psychiatry. 1970;117347- 348
PubMed
Rassaby  EPaykel  ES Factor patterns in depression. J Affect Disord. 1979;1187- 194
PubMed Link to Article
MacFadyen  HW The classification of depressive disorders, I: a review of statisticallybased classification studies. J Clin Psychol. 1975;31380- 394
PubMed Link to Article
Gullion  CMRush  AJ Toward a generalizable model of symptoms in major depressive disorder. Biol Psychiatry. 1998;44959- 972
PubMed Link to Article
Sullivan  PFKessler  RCKendler  KS Latent class analysis of lifetime depressive symptoms in the NationalComorbidity Survey. Am J Psychiatry. 1998;1551398- 1406
PubMed
Cardno  AGJones  LAMurphy  KCSanders  RDAsherson  POwen  MJMcGuffin  P Dimensions of psychosis in affected sibling pairs. Schizophr Bull. 1999;25841- 850
Link to Article
Wickham  HWalsh  CAsherson  PTaylor  CSigmundson  TGill  MOwen  MJMcGuffin  PMurray  RSham  P Familiality of symptom dimensions in schizophrenia. Schizophr Res. 2001;47223- 232
PubMed Link to Article
Leckman  JFWeissman  MMPrusoff  BACaruso  KAMerikangas  KRPauls  DLKidd  KK Subtypes of depression. Arch Gen Psychiatry. 1984;41833- 838
PubMed Link to Article
Sullivan  PFPrescott  CAKendler  KS The subtypes of major depression in a twin registry. J Affect Disord. 2002;68273- 284
PubMed Link to Article
Wing  JKBabor  TBrugha  TBurke  JCooper  JEGiel  RJablenski  ARegier  DSartorius  N SCAN: Schedules for Clinical Assessment in Neuropsychiatry. Arch Gen Psychiatry. 1990;47589- 593
PubMed Link to Article
Everitt  BS Making Sense of Statistics in Psychology.  Oxford, England Oxford University Press1996;291
Joreskog  KGSorbom  D LISREL VI: Analysis of Linear Structural Relationshipsby Maximum Likelihood, Instrumental Variables, and Least Squares Methods. 4th Uppsula, Sweden Dept of Statistics, University of Uppsula1986;
Bentler  PM Comparative fit indexes in structural models. Psychol Bull. 1990;107238- 246
PubMed Link to Article
Bentler  PMBonett  DG Significance tests and goodness of fit in the analysis of covariancestructures. Psychol Bull. 1980;88588- 606
Link to Article
Steiger  JH Structural model evaluation and modification: an interval estimationapproach. Multivariate Behav Res. 1990;25173- 180
Link to Article
Marsh  HWBalla  JWHau  K-T An evaluation of incremental fit indices: a clarification of mathematicaland empirical properties. Marcoulides  GASchumacker  REedsAdvancedStructural Equation Modeling Issues and Techniques. Mahwah, NJ LawrenceErlbaum Associates Inc1996;315- 354
Browne  MWCudeck  R Alternative ways of assessing model fit. Bollen  KALong  JSedsTesting StructuralEquation Models. Newbury Park, Calif Sage Publications1993;136- 161
Byrne  BM Structural Equation Modeling With LISREL, PRELIS,and SIMPLIS: Basic Concepts, Applications, and Programming.  Mahwah, NJ Lawrence Erlbaum Associates Inc1998;
Bollen  KALong  JS Testing Structural Equation Models.  Newbury Park, Calif Sage Publications1993;
Johnson  JHorwath  EWeissman  MM The validity of major depression with psychotic features based on acommunity study. Arch Gen Psychiatry. 1991;481075- 1081
PubMed Link to Article
Kendler  KSNeale  MCKessler  RCHeath  ACEaves  LJ The lifetime history of major depression in women. Arch Gen Psychiatry. 1993;50863- 870
PubMed Link to Article
McGuffin  PKatz  RAldrich  J Past and present state examination: the assessment of "lifetime ever"psychopathology. Psychol Med. 1986;16461- 465
PubMed Link to Article
Rice  JPRochberg  NEndicott  JLavori  PWMiller  C Stability of psychiatric diagnoses. Arch Gen Psychiatry. 1992;49824- 830
PubMed Link to Article
Kessler  RCMcGonagle  KASwartz  MBlazer  DGNelson  CB Sex and depression in the National Comorbidity Survey, I: lifetimeprevalence, chronicity and recurrence. J Affect Disord. 1993;2985- 96
PubMed Link to Article
Ninan  PTBerger  J Symptomatic and syndromal anxiety and depression. Depress Anxiety. 2001;1479- 85
PubMed Link to Article
Kessler  RCNelson  CBMcGonagle  KALiu  JSwartz  MBlazer  DG Comorbidity of DSM-III-R major depressivedisorder in the general population: results from the US National ComorbiditySurvey. Br J Psychiatry Suppl. June1996;17- 30
PubMed
Posternak  MAZimmerman  M Partial validation of the atypical features subtype of major depressivedisorder. Arch Gen Psychiatry. 2002;5970- 76
PubMed Link to Article
Farmer  AMcGuffin  P The classification of the depressions: contemporary confusion revisited. Br J Psychiatry. 1989;155437- 443
PubMed Link to Article
McGuffin  PKatz  RWatkins  SRutherford  J A hospital-based twin register of the heritability of DSM-IV unipolar depression. Arch Gen Psychiatry. 1996;53129- 136
Link to Article
Hyman  SEFenton  WS Medicine: what are the right targets for psychopharmacology? Science. 2003;299350- 351
PubMed Link to Article

Correspondence

CME
Also Meets CME requirements for:
Browse CME for all U.S. States
Accreditation Information
The American Medical Association is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The AMA designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 CreditTM per course. Physicians should claim only the credit commensurate with the extent of their participation in the activity. Physicians who complete the CME course and score at least 80% correct on the quiz are eligible for AMA PRA Category 1 CreditTM.
Note: You must get at least of the answers correct to pass this quiz.
Your answers have been saved for later.
You have not filled in all the answers to complete this quiz
The following questions were not answered:
Sorry, you have unsuccessfully completed this CME quiz with a score of
The following questions were not answered correctly:
Commitment to Change (optional):
Indicate what change(s) you will implement in your practice, if any, based on this CME course.
Your quiz results:
The filled radio buttons indicate your responses. The preferred responses are highlighted
For CME Course: A Proposed Model for Initial Assessment and Management of Acute Heart Failure Syndromes
Indicate what changes(s) you will implement in your practice, if any, based on this CME course.
Submit a Comment

Multimedia

Some tools below are only available to our subscribers or users with an online account.

Web of Science® Times Cited: 60

Related Content

Customize your page view by dragging & repositioning the boxes below.

Articles Related By Topic
Related Collections
PubMed Articles
JAMAevidence.com

The Rational Clinical Examination
Make the Diagnosis: Depression

The Rational Clinical Examination
Original Article: Is This Patient Clinically Depressed?