Career

21 views

Endorsement frequencies and factor structure of DSM‐III‐R and DSM‐IV Generalized Anxiety Disorder symptoms in women: implications for future research, classification, clinical practice and comorbidity

Abstract We investigated dimensions of liability to Generalized Anxiety Disorder (GAD) and whether evidence exists for distinct pathological versus normal clusters in the population. Structured interviews were administered to a general population
of 13
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Share
Tags
Transcript
  International Journal of Methods in Psychiatric Research, 2005, Volume 14, Number 2, pages 69–81 69 Endorsement freq u encies and factor str u ct u reof DSM-III-R and DSM-IV Generali z edAn x iet y Disorder s y mptoms in w omen:implications for f  u t u re research, classification,clinical practice and comorbidit y THOMAS S. KUBARYCH, 1 STEVEN H. AGGEN, 1  JOHN M. HETTEMA, 1 KENNETH S. KENDLER, 1, 2 MICHAEL C. NEALE 1, 2 1 Department of Ps y chiatr y , Virginia Common w ealth Uni v ersit y , Richmond, Virginia, USA 2 Department of H u man Genetics, Virginia Common w ealth Uni v ersit y , Richmond, Virginia, USA Abstract We investigated dimensions of liability to Generalized Anxiety Disorder (GAD) and whether evidence exists for distinct pathological versus normal clusters in the population. Structured interviews were administered to a general populationsample of 2,163 female twins in a cross-sectional design. Endorsement rates were estimated using full information maxi-mum likelihood factor analyses of the DSM-III-R and DSM-IV GAD symptoms, which provides appropriate treatment of the stem-probe structure of the clinical interview.Endorsement rates were highest for symptoms retained in DSM-IV. For both DSM-III-R and DSM-IV, a two-factormodel fit the data better than a single-factor model. There was no evidence for non-normality in the liability to GAD. ForDSM-III-R, autonomic symptoms loaded on a factor with panic disorder, while fatiguability, difficulty concentrating andhypervigilance loaded on a factor with major depression. For DSM-IV, all items loaded on one factor, and muscle tensionalso loaded on a second. Major depression, panic, phobias and alcohol dependence diagnoses also loaded on the first factor.Conclusions: future research involving structured interviews should take into account the stem-and-probe format and focuson common factors rather than separate disorders; GAD is not a unidimensional construct and pathological anxiety maydiffer only quantitatively from normal anxiety. Key words: Generali z ed An x iet y Disorder, factor anal y sis, missing data, str u ct u red inter v ie w , discriminant v alidit y normal and pathological w orr y . DSM-III-R (AmericanPs y chiatric Association, 1987) criteria for a diagnosisof GAD req u ired at least 6 months of  u nrealistic ore x cessi v e w orr y abo u t a v ariet y of circ u mstances u nre-lated to another A x is I disorder. Generali z ed An x iet y Disorder w as assessed in DSM-III-R b y 18 s y mptomsdi v ided into three cl u sters: motor tension, a u tonomich y peracti v it y and v igilance and scanning. Si x or mores y mptoms, e x perienced conc u rrentl y , w ere req u ired forthe diagnosis. Introduction O v erlap bet w een the s y mptoms of Generali z ed An x iet y Disorder (GAD), the other an x iet y disorders anddepression ca u ses considerable diffic u lt y in differentialdiagnosis. Since GAD’s introd u ction in DSM-III(American Ps y chiatric Association, 198 0 ), re v isions inassociated s y mptom criteria, req u ired d u ration, anddiagnostic hierarch y of the disorder ha v e largel y beenmoti v ated b y the need to address iss u es of pre v alence,reliabilit y , discriminant v alidit y and differencesbet w een  Kubarych et al. 7 0 We are not a w are of an y st u dies that address thestr u ct u re of GAD s y mptoms in the general pop u lation.Kenard y , E v ans and Oei (199 2 ) cond u cted a principalcomponents anal y sis of  3 9 0 o u tpatients w ith an x iet y disorders on patients’ ratings on the An x iet y S y mptoms and Beliefs Scale. A fo u r-component sol u -tion emerged. The components w ere interpreted asrespirator y s y mptoms, v estib u lar s y mptoms, a u tonomicaro u sal and ps y chological threat. This anal y sis w ascond u cted on a self-selected sample of patients w ith a v ariet y of an x iet y disorders seeking treatment.Merikangas, Zhang, A v ene v oli, Achar yy a, Ne u ensch w ander and Angst ( 2003 ) arg u ed that persis-tence of s u bthreshold-le v el an x iet y and depressionfrom earl y ad u lthood to mid-ad u lthood ill u strates theimportance of st u d y ing the contin uu m of an x iet y anddepression, not j u st diagnostic cases. These in v estiga-tors fo u nd that, in a comm u nit y -based cohort,comorbid an x iet y and depression w ere more persistentthan either s y ndrome alone. An x iet y alone tended tode v elop into either depression alone or comorbid an x i-et y and depression o v er time. Depression alone anddepression comorbid w ith an x iet y tended to be morestable than an x iet y alone o v er time. The patterns w eresimilar for s u bthreshold and threshold disorders.Whether or not a diagnosis is w arranted is a s u btl y different q u estion from that of  w hether or not a personis s u ffering from s y mptoms. Merikangas et al. ( 2003 )emphasi z e that contemporar y diagnostic s y stems fail toco v er depressi v e and an x iet y states among s u bjects w ho do not meet d u ration and impairment criteria y ete x hibit rec u rrent distress and ha v e a histor y of treat-ment. There are fl u ct u ations across threshold ands u bthreshold le v els o v er time, so classification onl y b y means of threshold le v el criteria at each e v al u ation w ill fail to capt u re the majorit y of cases w ith persistentan x iet y thro u gh the lifespan.The red u ction to si x s y mptom criteria in DSM-IV(American Ps y chiatric Association, 1994) w as primar-il y based on a single m u lti-site st u d y that assessed thefreq u enc y and u tilit y of the 18 DSM-III-R GAD s y mp-toms (Marten, Bro w n, Barlo w , Borko v ec, Shear, andL y diard, 199 3 ). This st u d y of  20 4 GAD patients at fo u rsites identified se v en s y mptoms as ha v ing satisfactor y reliabilit y and endorsement freq u enc y : irritabilit y , rest-lessness, m u scle tension, diffic u lt y concentrating, sleepdiffic u lties, feeling ke y ed u p and eas y fatig u abilit y . T w oof these s y mptoms, restlessness and feeling ke y ed u p, w ere combined into one criterion in DSM-IV. None of these s y mptoms belonged to the a u tonomic h y perac-ti v it y cl u ster.One of the goals in red u cing the n u mber of criterias y mptoms for GAD diagnosis w as to impro v e discrimi-nant v alidit y . T u r v e y , Ste v ens and Merikangas (1999)compared the discriminant v alidit y of DSM-III-R v ers u s DSM-IV criteria for GAD u sing data from ast u d y of the familial aggregation of an x iet y disordersand alcoholism. No differences w ere fo u nd in the dis-criminant v alidit y of the t w o definitions. The a u thorsspec u lated that the lack of difference in v alidit y bet w een the t w o definitions w as d u e to lo w pre v alenceof a u tonomic h y peracti v it y s y mptoms in both theirsample and earlier st u dies.In a non-clinical sample of 18 3 st u dents, Joormanand Stöber (1999) fo u nd that, among the si x DSM-IVGAD s y mptoms, onl y m u scle tension w as u niq u el y related to pathological w orr y . Diffic u lt y concentrating w as related to depression. The a u thors concl u ded thatm u scle tension is a specific aspect of pathological w orr y and that discriminant v alidit y bet w een GADand major depression co u ld be impro v ed b y criteriathat emphasi z e m u scle tension and de-emphasi z e diffi-c u lt y concentrating. R u scio, Borko v ec and R u scio( 200 1), ho w e v er, s u ggested that pathological w orr y differs q u antitati v el y rather than q u alitati v el y fromnormal w orr y .The DSM is designed to classif  y s u bjects as eithercases or non-cases. The q u estion of  w hether ps y -chopatholog y is better v ie w ed as discrete classes or as acontin u o u s dimension of liabilit y is m u ch disc u ssed(Waller and Meehl, 1998; Loranger, 1999; Clark,1999; Costa and Widiger, 2002 ; Pickles and Angold, 2003 ). There is a s u bstantial loss of information, andconseq u entl y statistical po w er, w hen a contin u o u s v ariable is treated as discrete. For e x ample, e v en u nderoptimal conditions of 5 0 % pre v alence, appro x imatel y three times the sample si z e is needed to assess familialresemblance for a dichotomi z ed rather than contin u -o u s v ariable (Neale, Ea v es and Kendler, 1994).In the present st u d y , w e in v estigate the item proper-ties and factor str u ct u re of the 18 DSM-III-R and 6DSM-IV GAD s y mptomatic criteria in a pop u lationsample. O u r aims are:•to cond u ct factor anal y ses of the SCID-basedDSM-III-R and IV stem items and probe s y mptomsin a general pop u lation sample taking into consid-eration the stem-and-probe format;  Generalized Anxiety Disorder symptoms in women 71 •to in v estigate the dimensionalit y of the DSM-IVs y mptoms;•to test for non-normalit y in these dimensions, w hich w o u ld indicate that s u bjects’ liabilities onthat factor w ere not distrib u ted along a contin uu mand s u ggest different categories; and•to identif  y factors in GAD s y mptoms that ma y bedifferentiall y associated w ith v ario u s patterns of comorbidit y , treatment efficac y and o u tcomes. Method S u bjects w ere 2 ,16 3 Ca u casian same-se x female t w insfrom the first w a v e of the pop u lation-based VirginiaT w in Registr y . The response rate w as 9 2 %. The agerange w as 17 to 54 y ears (mean 30 .1, SD 7.6), andmedian income w as in the $ 30 , 000 to $ 3 4,999 range.Details of these samples are gi v en b y Kendler andPrescott (1999). The assessment of GAD s y mptomso v er the pre v io u s y ear w as based on the Str u ct u redClinical Inter v ie w for DSM (SCID) inter v ie w forDSM-III-R (Spit z er and Williams, 1985). T w o ‘stem’q u estions w ere u sed:•‘In the last y ear, ha v e y o u had a time lasting at least5 da y s, w hen y o u felt an x io u s or w orried most of the time?’•‘In the last y ear, ha v e y o u had a time lasting at least5 da y s, w hen most of the time y o u r m u scles felttense, or y o u felt j u mp y or shak y inside?’If the response to either of these items w as ‘ y es’, the 18DSM-III-R indi v id u al GAD s y mptoms (‘probes’) w ereassessed. Note that the d u ration req u irement for thesestem q u estions w as deliberatel y set m u ch lo w er thanthat req u ired for GAD diagnosis in either DSM-III-Ror DSM-IV. Empirical data s u pport the h y pothesis thatshorter d u rations of GAD s y mptoms reflect the samecontin uu m of liabilit y as the f  u ll y s y ndromal disorder(Kendler, Neale, Kessler, Heath and Ea v es, 199 2 ). Nine h u ndred of  2 ,16 3 s u bjects recei v ed the 18 probes.Factor anal y sis of item-le v el data obtained from aninstr u ment that has a stem-probe format presents non-tri v ial complications. Most cases (1, 2 6 3 of  2 ,16 3 ) inthe present sample contain missing data, and w o u ld beeliminated b y methods that u se list w ise deletion. Thiselimination w o u ld restrict the anal y sis to those w ith atleast one positi v e screen and co u ld bias the res u lts.The probe s y mptom data missing d u e to v al u es of the stem items are ass u med to be ‘missing at random’(MAR) according to the Little and R u bin (1987) defi-nition. Therefore, w e u sed f  u ll information ma x im u mlikelihood anal y sis of the item data to reco v er as y mp-toticall y u nbiased estimates of the pop u lationparameters (Enders, 200 1). The model u sed incorpo-rates the ass u mption that for each item there is anormal distrib u tion of liabilit y u nderl y ing the response, w ith a threshold be y ond w hich the item is responded toin the affirmati v e. This liabilit y -threshold model is w idel y u sed in factor anal y sis, item response theor y (at w o-parameter normal ogi v e model) and genetic anal y -ses (Neale and Cardon, 199 2 ). In principle, the anal y sisin v ol v es n u merical integration o v er as man y non-missing items as e x ist in the response pattern of thes u bject. In practice, this method is comp u tationall y demanding (especiall y for the t w o-stem and 18-itemDSM-III-R diagnostic criteria) so w e u sed a marginalma x im u m likelihood approach (Bock, Gibbons andM u raki, 1988) implemented as a mi x t u re distrib u tionin M x (Neale, Boker, Xie and Maes, 2002 ) u sing ordi-nal data anal y sis as described b y Neale, Aggen,K u bar y ch, Fole y and Kendler (in preparation) andMehta, Neale and Fla y (in press). The res u lts w o u ldprobabl y differ if the data w ere not MAR.Factor anal y sis pro v ides a method to in v estigateso u rces of co v ariation bet w een different meas u res. Itallo w s e x amination of ho w w ell the items in a scaleeach assess a latent trait. The eq u i v alence of the factormodel to the normal ogi v e item response theor y (IRT)model has been noted b y others (Takane and DeLee uw ,1987). Another reason for the pop u larit y of factoranal y sis is that factor scores can be comp u ted andeither correlated w ith meas u res of other constr u cts tohelp clarif  y the nat u re of the factors or u sed as predictor v ariables in m u ltiple regression or dependent v ariablesin ANOVA. There are se v eral different methods of comp u ting factor scores, ho w e v er. Each of these meth-ods has shortcomings and the choice of method canlead to different concl u sions (Grice, 200 1). Therefore,instead of  u sing factor scores to predict other diagnosesfor e x ternal v alidation, w e e x plicitl y entered v ariablesrepresenting lifetime major depression, phobias, alco-hol dependence and panic disorder in f  u rther factoranal y ses. We ha v e pre v io u sl y reported comorbidit y bet w een GAD and these disorders in w omen in termsof both tetrachoric correlations and odds ratios(Kendler, Walters, Neale, Kessler, Heath and Ea v es,1995). Tetrachoric correlations w ere 0 .677 (GAD andmajor depression), 0 . 3 8 2 (GAD and phobias), 0 . 2 81  Kubarych et al. 7 2 (GAD and alcoholism) and 0 .484 (GAD and panicdisorder).Lifetime major depression w as diagnosed u sing DSM-III-R criteria (American Ps y chiatric Association,1987)) and req u ired fi v e or more s y mptoms. Lifetimealcohol dependence w as also based on DSM-III-R crite-ria and req u ired three or more of nine s y mptoms for atleast 2 8 da y s. Phobia w as diagnosed u sing an adaptationof DSM-III criteria (American Ps y chiatric Association,198 0 ), w hich req u ired the presence of one or more of  22 fears that the respondent recogni z ed as u nreasonableand that, in the j u dgment of the inter v ie w er, objecti v el y interfered w ith the respondent’s life (Kendler, M y ers,Prescott and Neale, 200 1). The diagnosis of panic w asbased on DSM-III-R and req u ired fo u r s y mptoms thatpeak w ithin 1 0 min u tes and either fo u r attacks w ithinone month or that the patient w orries abo u t f  u t u reattacks (American Ps y chiatric Association, 1987). Weha v e sho w n that these approaches reflect the same con-tin uu m of liabilit y as the f  u ll y s y ndromal disorders(Kendler, Gardner and Prescott, 200 1). Altho u gh these v ariables are based on s y mptom criteria, not formaldiagnoses, w e adopt the shorthand of referring to themas ‘diagnoses’ belo w . All ‘diagnoses’ are lifetime. Results Endorsement frequency The first t w o col u mns of Table 1 gi v e the ma x im u mlikelihood estimates of the thresholds for the 18 DSM-III-R and 6 DSM-IV GAD s y mptoms in z -scores(col u mn 1) and the probabilit y of being abo v e thisthreshold in a normal distrib u tion (col u mn 2 ). Fore x ample, w ith a z -score of 1. 00 , the threshold at w hich as u bject w o u ld be e x pected to endorse the second stem isone standard de v iation abo v e the mean, so abo u t 16%of s u bjects w o u ld be e x pected to endorse this item. Thethird col u mn gi v es the obser v ed endorsement ratesamong the 9 00 s u bjects w ho ans w ered ‘ y es’ to one orboth of the stems. For e v er y s y mptom, the endorsementrates of the 9 00 s u bjects w ho ans w ered ‘ y es’ to a stemq u estion e x ceed or eq u al the proportion that w o u ld bee x pected to endorse the item in a normal pop u lation.Th u s, the general pattern is as e x pected: s u bjectsendorsing either of the stems ha v e liabilit y higher thanthe pop u lation mean. These res u lts s u bstantiall y repli-cate the Marten et al. (199 3 ) rates and v erif  y that thes y mptoms w ith the highest endorsement freq u encies w ere retained in DSM-IV. In onl y one case did as y mptom that w as dropped from DSM-IV ha v e anendorsement freq u enc y (col u mn 3 ) that e x ceeded thatof an y of the s y mptoms retained: ‘Was y o u r stomachoften u pset, or did y o u ha v e na u sea or diarrhea?’( 0 .581) w as higher than both ‘Did y o u r m u scles oftenfeel tense, sore or ach y ?’ ( 0 .567) and ‘Were y o u so ner- v o u s y o u had tro u ble concentrating?’ ( 0 .5 2 9). Factor analyses Confirmator y factor anal y ses w ere u ndertaken for bothone and t w o factors. For the DSM-III-R s y mptoms, thechi-sq u are difference bet w een the one- and t w o-factormodels of 144.5 on 19 degrees of freedom w as highl y significant, rejecting the h y pothesis that DSM-III-RGAD is a u nidimensional constr u ct. For the DSM-IVs y mptoms, the chi-sq u are difference of 68 on 9 degreesof freedom w as also highl y significant, rejecting theh y pothesis that DSM-IV GAD is a u nidimensionalconstr u ct.A series of model-fitting anal y ses w ere performed u sing M x (Neale et al., 2002 ) to address comorbidit yw ith other ps y chiatric disorders. Initiall y , w e sepa-ratel y factor anal y sed the GAD s y mptomatic criteriafor DSM-III-R and DSM-IV along w ith the t w ostems. In the DSM-IV case, it w as necessar y for iden-tification p u rposes to fi x one loading on one of thefactors. We decided to fi x the loading of the first stemto z ero on the second factor ( w hich w e did for bothDSM-III-R and DSM-IV). We reasoned that the firststem, abo u t feeling an x io u s or w orried most of thetime, w as the most general indicator of GAD; fi x ing itto z ero on one factor sho w s its relationships to theother factor. Ne x t, w e added diagnoses for lifetimemajor depression, phobias, alcohol dependence andpanic as v ariables to the factor anal y ses, and per-formed a series of rotations to e x amine theirrelationships to the factors.Table 2 combines the res u lts of the factor anal y sesfor DSM-III-R and DSM-IV, w ith and w itho u t theadded v ariables for major depression, alcohol depen-dence, an y phobia and panic. In the anal y sis w ith theadded v ariables, the loadings for the stems and probes w ere fi x ed to the v al u es obtained in o u r anal y ses tosho w the relationships of major depression, alcoholdependence, phobias and panic to these factors. DSM-III-R criteria Both stem items loaded highl y on the first factor ( 0 .75and 0 .86). Si x probes had significant (> 0 .4) loadings  Generalized Anxiety Disorder symptoms in women 7 3 on both factors. These incl u ded cardio v asc u lar s y mp-toms: ‘Did y o u often feel short of breath?’ ( 0 .46 and 0 .48) and ‘Did y o u r heart often po u nd or race?’ ( 0 .47and 0 .59). Also loading on both factors w ere ‘Did y o u often tremble, t w itch or feel shak y ?’ ( 0 .61 and 0 .5 0 );‘Did y o u often feel di zzy or lightheaded?’ ( 0 .55 and 0 .4 3 ); ‘Did y o u often ha v e fl u shes (hot flashes) orchills?’ ( 0 .5 2 and 0 .5 3 ); and ‘Did s u dden noises oftenstartle y o u ?’ ( 0 .47 and 0 .44).Factor 1 also has s u bstantial loadings for ‘Did y o u rm u scles often feel tense, sore or ach y ?’ ( 0 .8 3 ); ‘Did y o u often feel ph y sicall y restless – co u ldn’t sit still?’ ( 0 .5 2 );‘Did y o u often tire easil y ?’ ( 0 .6 0 ), ‘Did y o u often feelke y ed u p or on edge?’ ( 0 .57), ‘Were y o u so ner v o u s y o u had tro u ble concentrating?’ ( 0 .59) and ‘Were y o u often irritable or especiall y impatient?’ ( 0 .55). Thisfactor contains items that o v erlap w ith s y mptoms of Major Depressi v e Disorder (MDD), s u ch as tiring Table 1. Comparison of theoretical probabilit y of being abo v e threshold w ith obser v ed endorsement rates for DSM-III-Rand DSM-IV GAD s y mptomsItemThreshold p > thresholdObs 1 or more probe( z score)= 1 (N = 9 00 ) Stems In the last y ear, ha v e y o u had a time lasting at least 0 . 2 5 0 .4 0 1 0 .95 0 5 da y s, w hen y o u felt an x io u s, ner v o u s or w orriedmost of the time?In the last y ear, ha v e y o u had a time lasting at least 1. 00 0 .159 0 . 3 5 2 5 da y s, w hen y o u r m u scles felt tense, or y o u feltj u mp y or shak y inside? DSM-III-R Did y o u often tremble, t w itch or feel shak y ? 0 .79 0 . 2 16 0 . 3 8 2 * Did y o u r m u scles often feel tense, sore or ach y ? 0 .47 0 . 3 18 0 .567† Did y o u often feel ph y sicall y restless –co u ldn’t sit still?– 0 . 0 8 0 .5 32 0 .7 0 6* Did y o u often tire easil y ? 0 .17 0 .4 3 1 0 .61 0 Did y o u often feel short of breath?1. 3 5 0 . 0 88 0 .151Did y o u r heart often po u nd or race? 0 .76 0 . 22 5 0 . 3 66Did y o u s w eat a lot? Were y o u r hands often cold and clamm y ? 0 .94 0 .174 0 . 2 76Did y o u r mo u th often feel dr y ? 0 .81 0 . 2 1 0 0 . 320 Did y o u often feel di zzy or lightheaded?1.1 3 0 .1 2 8 0 . 232 Did y o u often ha v e fl u shes (hot flashes) or chills?1.1 0 0 .1 3 7 0 . 2 4 3 Was y o u r stomach often u pset, or did y o u ha v e na u sea or diarrhea? 0 . 0 8 0 .466 0 .581Did y o u u rinate more often than u s u al? 0 .98 0 .164 0 . 23 1Did y o u ha v e tro u ble s w allo w ing, or get a l u mp in y o u r throat? 0 .9 3 0 .177 0 . 2 46†Did y o u often feel ke y ed u p or on edge?– 0 .79 0 .784 0 .9 03 Did s u dden noises often startle y o u ? 0 .7 0 0 . 2 41 0 . 3 7 3 * Were y o u so ner v o u s y o u had tro u ble concentrating? 0 . 3 1 0 . 3 5 2 0 .5 2 9* Did y o u often ha v e tro u ble falling asleep?– 0 .11 0 .54 3 0 .659* Were y o u often irritable or especiall y impatient? – 0 . 2 9 0 .61 3 0 .77 3 DSM-IV Did y o u r m u scles often feel tense, sore or ach y ? 0 .44 0 . 332 0 .567Did y o u often tire easil y ? 0 .17 0 .4 33 0 .61 0 Were y o u so ner v o u s y o u had tro u ble concentrating? 0 . 3 8 0 . 3 5 3 0 .5 2 9Did y o u often ha v e tro u ble falling asleep?– 0 .11 0 .544 0 .659Were y o u often irritable or especiall y impatient? – 0 . 2 8 0 .61 0 0 .77 3 Did y o u feel restless or ke y ed u p and on edge?– 0 .96 0 .8 3 1 0 .946* = S y mptom retained in DSM-IV. † = Combined into one s y mptom in DSM-IV.
Advertisement
Related Documents
View more
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks