Spousal concordance for substance use and anxiety disorders

JOURNAL OF

PSYCHIATRIC

RESEARCHJournal of Psychiatric Research 41 (2007) 942–951

www.elsevier.com/locate/jpsychires

Spousal concordance for substance use and anxiety disorders

Nancy Low *, Lihong Cui, Kathleen R. Merikangas

Section on Developmental Genetic Epidemiology, Mood and Anxiety Disorders Program, Intramural Research Program, National Institute of

Mental Health, National Institutes of Health, Department of Health and Human Services, 35 Convent Drive (MSC 3720), Bethesda, MD 20892-3720, USA

Received 18 March 2006; received in revised form 27 October 2006; accepted 2 November 2006

Abstract

Assortative mating – the tendency for mate selection to occur on the basis of similar traits – plays an essential role in understandingthe genetic contribution to psychiatric illness. It also carries significant impact on clinical prognosis and is an important mechanismexplaining spousal concordance. This study uses a family study design ascertaining 225 probands with substance abuse/dependence, anx-iety disorders, and controls to address: (1) Is there spousal concordance or cross-concordance for substance use and/or anxiety disorders?(2) Is the spousal concordance or cross-concordance associated with worse clinical outcomes? (3) What is the mechanism of the concor-dance or cross-concordance? Results show a high magnitude of spousal concordance for substance use disorders with a third of the sub-stance probands’ spouses also substance dependent. In contrast, there was no spousal concordance for anxiety disorders. Couples werealso concordant for having ‘‘no disorders.’’ Both substance use and anxiety disorder concordance were associated with poorer globalfunctioning and persistent illness. Assortative mating is a likely mechanism for spousal concordance given the elevated rate of substanceuse disorders among the relatives of spouses’ of substance probands. Implications for family/genetic studies and the transmission of sub-stance use disorders and ‘‘no disorders’’ include: (1) at the individual level, spousal concordance influences probands’ course of illness,couples’ marital functioning, and offspring’s genetic and environmental context; and (2) at the population level, it shifts the general dis-tribution of substance use disorders and ‘‘no disorders’’ by reducing the ‘‘average’’ couple concordance and increasing the number con-cordant and discordant couples at extremes of the distribution.� 2006 Elsevier Ltd. All rights reserved.

Keywords: Spousal concordance; Assortative mating; Psychiatric genetics; Substance abuse/dependence disorders; Anxiety disorders

1. Introduction

Spousal concordance, the tendency for couples toappear to share the same specific traits, has been of longstanding interest to researchers and clinicians alike fromdiverse fields of study. The traits investigated among cou-ples have ranged from physical/somatic characteristics(such as, blood pressure, height, weight, eye color, race)(Wolanski, 1974, 1994; Speers et al., 1986; Knuimanet al., 1996; Maes et al., 1997), to medical disorders (forexample, multiple sclerosis, cancer, hypertension, infertil-ity) (Mascie-Taylor, 1986; Mascie-Taylor and Boldsen,

0022-3956/$ - see front matter � 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.jpsychires.2006.11.003

* Corresponding author. Tel.: +1 301 496 7840; fax: +1 301 480 2915.E-mail address: [email protected] (N. Low).

1988; Ebers et al., 2000), to sociocultural and behavioralpatterns (such as, religious affiliation/attendance, educa-tion, diet, smoking, exercise, other health habits) (Koloneland Lee, 1981; Reynolds et al., 2000; Bloch et al., 2003),and personality and cognition (such as, extroversion/intro-version, IQ) (Farley and Davis, 1977; Farley and Mueller,1978; Mascie-Taylor and Gibson, 1979; Merikangas, 1982;Mascie-Taylor and Boldsen, 1984; Mascie-Taylor, 1989;Heun and Maier, 1993; Feng and Baker, 1994; Merikangasand Brunetto, 1996; Dubuis-Stadelmann et al., 2001).

Spousal concordance can be attributed to several differ-ent mechanisms which include: (1) tendency to marry thosewho already have similar traits (assortative mating);(2) development of the trait in the second partner, inresponse to the first partner who already has it (contagion);

mailto:[email protected]

N. Low et al. / Journal of Psychiatric Research 41 (2007) 942–951 943

(3) mutual influence among partners to develop the trait;and (4) sharing of common pathogenic factors (for exam-ple, environment, diet, economic status) that predisposeto the trait (Nielsen, 1964; Merikangas, 1982; McLeod,1995). Though several of these mechanisms have beenaddressed in previous literature (Hagnell and Kreitman,1974; Merikangas, 1982; Merikangas et al., 1983; McLeod,1994, 1995), few studies have considered a full range ofmechanisms adequately due to the multiple types of data(e.g. longitudinal, family, measured exposures, etc.) thatwould need to be collected.

Assortative mating, or spouse selection based on similartraits, has been found for numerous psychiatric disordersincluding depression, substance abuse, alcoholism, schizo-phrenia and anxiety disorders (Merikangas and Spiker,1982; Parnas, 1988; McLeod, 1993, 1995; Mathews andReus, 2001). Spousal concordance for anxiety and sub-stance use disorders was first examined in the 1960s withthe investigation of phobic states (Kreitman, 1962, 1964;Agulnik, 1970; Hagnell and Kreitman, 1974; Buglasset al., 1977; Hafner, 1977) and heavy drinking (Cisin andCahalan, 1968). Yet to date there are only a few studiesof alcohol/substance use disorders among couples fromclinical samples (Rimmer and Winokur, 1972; Hall et al.,1983a,b; Moskalenko et al., 1992). Moreover, these priorstudies have methodologic limitations including selectionof probands from strictly inpatient alcoholic treatmentwards, and a lack of clinician-administered structured diag-nostic instrument, blinded direct assessment of spouses,control/comparison group, or application of modern diag-nostic nomenclature. On the other hand, numerous popu-lation-based studies of alcohol/substance use disorders doreport spousal concordance (Cisin and Cahalan, 1968;Hagnell and Kreitman, 1974; Kolonel and Lee, 1981; Glei-berman et al., 1992; McLeod, 1993, 1995; Demers et al.,1997; Windle, 1997; Galbaud du Fort et al., 1998; Maeset al., 1998; Graham and Braun, 1999; Leonard and DasEiden, 1999).

There are few studies of spousal concordance of anxietydisorders using modern psychiatric assessments. Earlierstudies of clinical samples (Kreitman, 1964; Agulnik,1970; Buglass et al., 1977; Hafner, 1977) employed mea-sures of current state rather than lifetime disorders, there-fore the low couple correlations observed would notnecessarily rule out spousal concordance. Evaluations ofcommunity samples using checklists or lay-intervieweradministered standardized diagnostic assessments (Kreit-man, 1962; Eagles et al., 1987; Tambs, 1991; Zimmer-man-Tansella and Lattanzi, 1991; McLeod, 1995;Galbaud du Fort et al., 1998; Maes et al., 1998; Dubuis-Stadelmann et al., 2001) have been inconsistent – in partdue to variable sampling techniques and anxiety measures– but not provide convincing support for spouse concor-dance for anxiety disorders.

At the conceptual level, assortative mating has impor-tant implications on society and the individual. At the pop-ulation level, if non-random mating occurs for similar

traits, the distribution and constitution of genetic risks fac-tor for psychiatric illness (the traits of interest) is affectedbecause the Hardy-Weinberg equilibrium cannot beassumed. Assortative mating will increase the populationvariance of a given trait and the proportion of individualswho are homozygous for the trait. At the individual level,assortative mating can influence the course and outcomeof marriage, in addition to determining the genetic and bio-logical makeup and parenting of offspring.

Additionally, cross-concordance – the tendency for per-sons with one trait to marry persons with another traitmore frequently than would be expected if they were cho-sen at random from the population – can lead to erroneousinferences regarding the co-segregation of two traits infamilies. Few studies, however, have investigated thecross-concordance of anxiety and substance use disorders,which may be even more common than concordance forthese conditions.

In this study, we address the following questions:

1. Is there spousal concordance or cross-concordance forsubstance use and/or anxiety disorders?

2. Is spousal concordance or cross-concordance for sub-stance use and/or anxiety disorders associated withworse clinical outcomes among probands?

3. What is the mechanism of the observed spousal concor-dance or cross-concordance?

This study distinguishes itself from previous literaturebased on the strengths of its methodological approachesand substantive questions. We apply a contemporary fam-ily study design to a sample ascertained from both outpa-tient treatment clinics and the same local community,using standardized psychiatric assessments with modern

diagnostic nomenclature administered by clinically trained

interviewers, and direct interviews with spouses to address:(a) concordance (including for an absence of psychiatricdisorders) and cross-concordance patterns and (b) theirclinical outcomes as established by the best-estimate pro-cess conducted by a panel of experienced clinicians; (c)the effect of psychiatric comorbidity in spouses and pro-bands, and spousal family history of psychiatric disorders,and (d) quantitative exploration of several mechanisms ofspousal concordance.

2. Methods and materials

2.1. Sample and measures

2.1.1. Probands

Two-hundred fifty-five probands participated in thestudy. They were recruited from clinics for substance useand/or anxiety disorders, and from the community (thesame geographic area) through a random digit dialing pro-cedure. All of the controls and a proportion (29.7% =58/195) of the affected probands were recruited from thecommunity. The rationale for including a combination of

944 N. Low et al. / Journal of Psychiatric Research 41 (2007) 942–951

probands from the clinic and community was to minimizeBerkson’s bias (treatment-seeking bias). The diagnosticinterview for adults was the semi-structured Schedule forAffective Disorders and Schizophrenia (Endicott and Spit-zer, 1978), current and lifetime versions, which was modi-fied to incorporate DSM-III and DSM-III-R criteria andto include more detailed information on substance useincluding patterns and the interrelations of substance useand anxiety symptoms and disorders.

The major proband disorder groups used to addressthe major goals of the study were: (1) substance abuse/dependence (i.e., alcohol and/or marijuana and/or seda-tives) with and without anxiety disorders (i.e., panic disor-der and/or social anxiety disorder) (N = 121); (2) anxietydisorders alone (i.e., panic disorder and/or social anxietydisorder without a lifetime history of a substance use dis-order) (N = 74); and (3) controls (no Axis I disorders)(N = 60).

Probands were excluded if they had evidence of schizo-phrenia, schizoaffective disorder, mental retardation, or asignificant medical disorder causing their psychiatric symp-toms. Probands with opioid and cocaine abuse/dependencewere excluded. Comorbid affective and other anxiety disor-ders were not excluded among the affected proband groups.Substance abuse/dependent probands with and withoutanxiety disorders were combined in the analyses becausepreliminary examination found no substantial differencesbetween the groups.

2.1.2. SpousesTwo hundred fifty-five spouses were included in the

study. Fourteen percent (35/255) of the probands hadmultiple spouses. If more than one spouse was available,either the current or the most recent spouse was included.Diagnostic information on all of the spouses was obtainedby direct interview with SADS (see Section 2.1.1) and astructured family history interview using a modified ver-sion of the Family History-Research Diagnostic Criteria(Andreasen et al., 1977) to collect diagnostic spectruminformation and to obtain both DSM-III and DSM-III-R diagnoses. This modified version also enabled the col-lection of more extensive information on patterns andsequelae of substance and alcohol use and anxiety symp-toms and disorders and as well as the extent of the infor-mant’s knowledge on the index subject. Independentinterviewers, blind to the identity and diagnoses of theproband, directly interviewed the spouses either in personor by telephone.

2.1.3. Relatives

Four hundred thirty-nine first-degree relatives wereincluded in the relatives of the spouses. All relatives were18 years or older, and mainly included siblings and parentsof the spouses. Diagnostic procedures for the relatives,including direct interviews and family history collection,were the same as for the probands and spouses.

2.1.4. Mating types

Probands and spouses were classified into six mating typesto examine patterns of concordance and cross-concordanceamong couples. Based on frequencies of proband disordergroups described above, the following six hierarchical mutu-ally exclusive mating types were created for the analyses:

1. Both members of the couple affected with substanceabuse/dependence with or without an anxiety disorder;

2. Both members of the couple affected with an anxietydisorder;

3. One member affected with substance abuse/dependenceand the other with an anxiety disorder;

4. One member affected with substance abuse/dependenceand the other with no Axis I disorder;

5. One member affected with an anxiety disorder and theother with no Axis I disorder; and,

6. Both members have no Axis I disorder.

2.2. Procedures

All procedures complied with strict ethical standards forthe treatment of human subjects and were approved by theYale University School of Medicine Institutional ReviewBoard. After obtaining consent, probands were directlyinterviewed with the structured interview on health, behav-ior, mental disorders and substance use disorders.

Interviewers with experience in clinical psychiatry andsubstance abuse administered the interview, and an exten-sive effort was made to establish reliability. Reliability wasestablished at the study initiation and was tested periodi-cally throughout the study. Kappas derived from jointratings of individual interviews were generally higher forsubstance abuse (0.72–0.94) than for anxiety or affectivedisorders (0.54–0.78) across the first three series of train-ing sessions. Comparison of diagnosis obtained by in per-son interview versus telephone interview showed highlevels of agreement across all diagnostic categories. Addi-tional methodologic details concerning this family studyare provided in a previous publication (Merikangaset al., 1998).

2.3. Best estimate diagnostic procedures

Diagnoses of anxiety and substance use disorders werebased upon all available information, including the diag-nostic interview, family history reports, and medicalrecords using the best estimate method of diagnoses (Leck-man et al., 1982). Each substance of use, abuse and depen-dence was characterized by: age of onset, quantity,frequency, chronicity, substance of choice, number ofsymptoms, and severity. Best estimate diagnoses weremade by clinicians with extensive experience in either theevaluation and treatment of substance abuse or moodand anxiety disorders. We did not include specific algo-rithms for the best estimate diagnostic process because of

Table 1Characteristics of probands and spouses

Characteristic Proband (N = 255) Spouse (N = 255)

Age in years (mean, s.d., range) 39.7 ± 6.1(24–56)

40.3 ± 6.5(22–65)

Male (%) 49.4 50.6Socioeconomic status

(mean, range)37.9 (11–80) 37.3 (11–72)

Percent employed (%) 82.5 94.6Marriage duration

(mean, years, range)13.9 (1–34)

Couple’s household income(in thousands of dollars) (%)

629 28.130–59 45.560+ 26.4

Religion (%)

Catholic 57.1 53.2Protestant 18.9 29.7Jewish 4.3 4.5Other 7.5 3.6Not affiliated 12.2 9.0

Education (%)

Partial High School or lower 10.6 3.6High School 31.5 32.4Partial university or higher 57.9 64.0


the variability in the information as well as the descriptivenature of the narrative summaries provided in the directassessments and family history interviews. The clinicianswere made blind with respect to the diagnostic status ofthe probands when making best estimates of the spousesand relatives. Direct interview information was prioritizedamong all the available diagnostic data because of the well-established under-estimation of diagnoses in non-inter-viewed relatives. When direct assessment was unavailable,family history information gathered from multiple, reliableinformants was used. Direct interviews and family historyinformation are far more concordant for observable disor-ders such as substance abuse and behavior disorders, thanfor less readily observable disorders such as mood and anx-iety disorders (Orvaschel et al., 1982; Thompson et al.,1982).

2.4. Statistical analyses

Pearson and Spearman correlations were calculated toinvestigate the associations between probands and spousesfor continuous and ordinal variables, respectively. Chi-square tests were calculated to evaluate the associationbetween proband and spouse diagnostic groups.

Odds ratios were estimated from logistic regressionmodels conducted separately for each of the major disorderoutcomes (substance use disorders, anxiety disorders andno Axis I disorders) in spouses controlling for correspond-ing proband diagnoses, source of proband ascertainment(clinic versus community), spouse comorbidity (substanceuse, anxiety and affective disorders), spouse family psychi-atric history (substance use, anxiety and affective disor-ders), spouse age and sex.

Given the non-normal distribution of probands’ globalassessment of functioning ratings, mean number of currentdisorders and rates of substance use disorders amongspouses’ relatives, non-parametric Kruskal–Wallis testswere used to examine the differences for each of these vari-ables among the different mating types. If significant differ-ences were found among the groups, pair-wise comparisonswere made using Wilcoxon Rank Sum tests. Bonferronicorrection was used for multiple comparisons to adjustthe p-values.

3. Results

Sociodemographic characteristics of the probands andspouses are shown in Table 1. Couples were correlatedfor age (r = 0.73) and socioeconomic status levels(r = 0.63), and concordant for levels of education(p < 0.000) and religious affiliation (p < 0.000). All coupleswere white.

3.1. Spousal concordance and associations

Table 2 shows patterns of concordance among couples.Of the 255 probands, 121 had substance abuse/dependence,

74 had anxiety alone, and 60 had no Axis I disorders (con-trols). Thirty-eight percent of probands with substance usedisorders had spouses with substance use disorders, com-pared to a rate of 11.7% among spouses of controls – athreefold elevation. In contrast, the rate of anxiety disor-ders among spouses of probands with anxiety disorderswas 24.3%, compared to 20.0% among control spouses.There was also spouse concordance for the absence of psy-chiatric disorders: 68.3% of control spouses had no disor-der, compared to 56.8% of spouses of anxious probands,versus 38.0% of spouses substance probands. Finally, nosystematic patterns of cross-concordance were observed.Chi-squared testing on this concordance table yielded ap-value of 0.0006, and the specific effects of the probanddisorder groups were modelled in Table 3.

The associations between disorders in probands andspouses are examined in Table 3. The results of three logisticregression models are presented. The three groups of spousedisorders – substance use, anxiety, and no disorders – are themajor outcomes of interest. The measure of association(odds ratio) and p-values for each disorder are displayedin the columns of the table. The main effects and all poten-tial confounders are represented in the rows of the table.

Findings in the first column are in agreement with Table2. There is a strong and significant association betweenproband and spouse substance use disorders as shown byan odds ratio (OR) of 7.6 (95% CI 1.9–30.4), indicatingthat the risk of substance abuse/dependence amongspouses of probands with substance abuse/dependenceis nearly 8 times that of spouses of probands withoutsubstance use disorders. Male spouses (displayed as spouse

Table 2Rates of disorders among spouses by proband diagnostic group

Proband diagnostic group Spouse diagnostic group Total (N)

Substance use disorder (%) (n) Anxiety disorder (%) (n) No disorder (%) (n)

Substance use disorder 38.0 (46) 24.0 (29) 38.0 (46) 121Anxiety disorder 18.9 (14) 24.3 (18) 56.8 (42) 74Control 11.7 (7) 20.0 (12) 68.3 (41) 60

Total (N) 67 59 129 255

Chi-square test p-value = 0.0006.

Table 3Associations between spouse and proband disorders for substance use, anxiety, and no disorders

Spouse disorders

Substance use Anxiety No disorder

Odds ratio (95%confidence interval)

p-Value Odds ratio (95%confidence interval)

p-Value Odds ratio (95%confidence interval)

p-Value

Main effects: proband disorders

Substance use disorder 7.6 (1.9, 30.4) 0.004 0.8 (0.3, 2.3) ns NA NAAnxiety disorder 1.0 (0.3, 3.9) ns 1.2 (0.4, 3.1) ns NA NANo disorder NA NA NA NA 2.5 (1.4, 4.5) 0.003

Covariates in spouses:

Spouse comorbidity

Substance use disorder NA NA 0.7 (0.2, 2.2) ns NA NAAnxiety disorder 0.9 (0.3, 2.8) ns NA NA NA NAAffective disordera 2.4 (0.7, 8.2) ns 3.4 (1.3, 8.7) 0.0115 NA NASpouse family history

Substance use disorder 3.2 (1.0, 10.3) 0.0457 b ns NA NAAnxiety disorder b ns 1.2 (0.5, 3.0) ns NA NAAffective disordera b ns b ns NA NA

Other spouse characteristics

Sex (0 = female,1 = male) 16.1 (3.4, 77.1) 0.0005 0.4 (0.2, 1.2) ns 1.1 (0.7, 1.9) nsAge 1.0 (0.9, 1.1) ns 1.0 (0.9, 1.0) ns 1.0 (1.0, 1.0) ns

Covariates in probands

Comorbid affective disordera 1.0 (0.2, 3.9) ns 1.0 (0.3, 3.0) ns NA NASource of ascertainment

(0 = community,1 = clinic)

1.4 (0.3, 6.3) ns 0.6 (0.2, 1.9) ns NA NA

NA = not applicable.ns = not significant (p > 0.05).a Includes major depression and bipolar disorder.b Fit into model separately because of dependent nature and removed because not significant.


sex in Table 3) have 16 (95% CI 3.4–77.1) times the odds tohave substance abuse/dependence, compared to femalespouses. Spouses with a family history of substance use dis-order have 3 times the odds (95% CI 1.0–10.3) of having asubstance use disorder than a spouse with no family his-tory. Comorbid proband and spouse anxiety and affectivedisorders, spousal age and family history of mood and anx-iety disorders did not contribute to spousal concordancefor substance abuse/dependence.

As shown in the second column, no evidence was foundfor an association between proband and spouse anxiety dis-orders (OR = 1.2, 95% CI 0.4–3.1). Comorbid affective dis-order in the spouse was the only significant predictor ofspousal anxiety disorders (OR = 3.4, 95% CI 1.3–8.7).Finally, the third column demonstrates that probandsand spouses with no disorders also assort together(OR = 2.5, 95% CI 1.4–4.5). Source of proband ascertain-

ment (from clinic or community) was not a significant pre-dictor in either the substance abuse/dependence (OR = 1.4,95% CI 0.3, 6.3) or anxiety disorder models (OR = 0.6,95% CI 0.2, 1.9).

3.2. Clinical outcomes by mating types

Differences in the clinical outcomes of probands by theirmating types are examined in Table 4. Significant differ-ences among probands’ lifetime global assessment of func-tioning (GAF) scores were observed. The lowest globalassessment of functioning ratings were found among: cou-ples concordant for substance abuse/dependence (64.0) (2SUB) and for anxiety (66.1) (2 ANX), couples where onlyone member had substance abuse/dependence (63.4) (1SUB), and couples where one member has substance

Table 4Proband global assessment of functioning (lifetime GAF) and current psychiatric disorders by mating type

Mating type Number of couples(N = 255)

GAF(mean)

Average number of currentdisorders

Both members with substance use disorder (2 SUB) 51 64.0a1 9.8d1

Both members with anxiety disorder (2 ANX) 35 66.1a1 9.5d1

One member with substance use disorder (1 SUB) 63 63.4a1 6.9d2

One member with substance use disorder and the other with anxiety disorder (1SUB/1 ANX)

23 69.3a2 4.7e

One member with anxiety disorder (1 ANX) 46 75.5b 5.0e

Both members have no disorder (0 ILL) 37 83.3c 0f

GAF p-valuesa Current disorders p-valuesa

Between a1 and a2: ns Between d1 and d2: p = 0.02Between [a1 and a2] and b: p < 0.0001 Between d2 and e: nsBetween a2 and b: ns Between d1 and e: p = 0.002Between b and c: p < 0.0001 Between [d1 and d2] and f: p < 0.0001Between [a1 and a2] and c: p < 0.0001 Between e and f: p < 0.0001

a Adjusted significant level (Bonferroni method): 0.0033.


abuse/dependence and the other has anxiety (69.3) (1 SUB/1 ANX). There was no significant difference between any ofthe above global assessment of functioning ratings amongthese mating types. The next increment in global assess-ment of functioning ratings was among couples with onlyone anxious member (75.5) (1 ANX). Between 1 ANXand 1 SUB/1 ANX couples, there was no difference. Finallyas would be expected, the highest global assessment offunctioning ratings were from couples with no disorders(83.3).

The other proband clinical outcome of interest – persis-tence (as measured by the average number of current diag-noses) of substance use, anxiety, mood and any otherpsychiatric disorder – classified by mating types is also pro-vided in Table 4. Concordant (same disorder) mating typesfor substance abuse/dependence and for anxiety hadgreater numbers of current diagnoses (9.8 and 9.5, respec-tively), compared to all other mating types (6.9–1 SUB,4.7–1 SUB/1 ANX, 5.0–1 ANX, 0–0 ILL). Though afterBonferroni correction, concordant mating types (2 SUBand 2 ANX) were not different then 1 SUB mating type(p-value = 0.02).

3.3. Mechanisms of concordance

In order to address the mechanism of spousal concor-dance, the temporal relation between onset of psychiatric

Table 5Proportion of spouses’ relatives with substance use disorders by number of m

Number of members withsubstance use disorder

Number ofcouples

Number of spouses’relatives

0 13 46

1 39 145

2 18 74

a Adjusted significance level (Bonferroni method): 0.0167.

illness and marriage date was investigated. Seventy-onepercent of probands had their substance use disorder beginprior to marriage, compared to 63.6% among spouses. Foranxiety disorders, 78.0% of probands had their onsetbefore marriage and among spouses, the rate was 75.5%.The onset of any psychiatric disorder occurred prior tomarriage in 87.3% of probands and 80.0% of spouses.

To address assortative mating as a mechanism ofspousal concordance, we examined whether ‘‘substanceabusing/dependent families’’ assorted together. To thisend, we examined the proportion of spouses’ relativeswith substance use disorders based on the number ofmembers in the couple with a substance use disorder(i.e., 0, 1, 2 members of the couple). We inferred assorta-tive mating as the operating mechanism if rates of sub-stance abuse/dependence among the spouses’ relativesincreased with more members of the couple affected withsubstance abuse/dependence. Indeed, Table 5 demon-strates that couples where both members had substanceuse disorders had a spousal relative rate of 35.1%, whichwas substantially and significantly greater the correspond-ing rate in couples where one member had substanceabuse/dependence (21.4%); and greater than the spousalrelative rate in couples with no substance use disorders(17.4%). The rates between couples with one and nomembers with substance abuse/dependence were notsignificantly different.

embers with substance use disorders

Proportion of spouses’ relatives withsubstance use disorders (%)

p-Valuea

17.4 0 versus1 member:ns

21.4 1 versus 2 members:0.0001

35.1 0 versus 2 members:0.0001


4. Discussion

This study advances previous research on spousal con-cordance for substance use and anxiety disorders both withrespect to its scientific questions and methodologicalapproach. Substantively, it addresses: (1) both concor-dance and cross-concordance for substance use and anxietydisorders; (2) clinical outcomes associated with specificmating types; and (3) mechanisms for spousal concor-dance. Methodologically, this study is among few thatemploy: (1) clinically trained interviewers to administer astandardized, semi-structured diagnostic instrument forpsychiatric disorders; (2) a family study approach withdirect interviews of spouses and first-degree relatives inaddition to the family history method; (3) analytic modelsthat adjust for psychiatric comorbidity in spouses and pro-bands, and spousal family psychiatric history simulta-neously as well as other relevant predictors.

Several major findings are demonstrated. First, there is ahigh magnitude of spousal concordance for substance usedisorders – more than a third of the spouses of probandswith substance abuse/dependence also have a lifetime his-tory of substance abuse/dependence. In contrast, there isno evidence for spousal concordance for anxiety disorders.There is also concordance for ‘‘no psychiatric disorders’’(that is, absence of any Axis I disorder), as 53% of thespouses of controls had no lifetime disorder themselves.Second, concordant mating types (both members with thesame disorder) are associated with worse clinical outcomesamong probands. Third, assortative mating appears to be alikely mechanism to explain spousal concordance forsubstance use disorders.

4.1. Patterns of spousal concordance across disorders

We observed a large and significant association betweenprobands and spouses for substance use disorders(OR = 7.6, 95% CI 1.9-30.4). This is in agreement withother clinical (Winokur et al., 1970; Rimmer and Winokur,1972; Hall et al., 1983a,b) and convenience (Schuckit et al.,1994) samples using standardized psychiatric assessmenttools. Only one of these earlier studies included a controlgroup and considered comorbid psychiatric diagnoses(Rimmer and Winokur, 1972).

Likewise, this finding is also confirmed in population-based studies which yielded odds ratios ranging from 2.7to 12.4 for alcohol and/or drug use disorders concordance(Cisin and Cahalan, 1968; Hagnell and Kreitman, 1974;Kolonel and Lee, 1981; Gleiberman et al., 1992; McLeod,1993, 1995; Demers et al., 1997; Windle, 1997; Galbauddu Fort et al., 1998; Maes et al., 1998; Graham and Braun,1999; Leonard and Das Eiden, 1999) (McLeod, 1993, 1995;Galbaud du Fort et al., 1998). Our analyses were morecomprehensive because we included proband and spousepsychiatric comorbidity, and spousal family history (ofsubstance use, affective, and anxiety disorders) in the finalassociation models.

In contrast, there was no evidence for spousal concor-dance for anxiety disorders. These findings confirm thelack of spousal concordance for anxiety disordersobserved in the prior sparse clinical (Kreitman, 1964;Agulnik, 1970; Buglass et al., 1977; Haber and Jacob,1997) and abundant community studies (Hagnell andKreitman, 1974; Eagles et al., 1987; Tambs, 1991;Zimmerman-Tansella and Lattanzi, 1991; Galbaud duFort et al., 1998; Dubuis-Stadelmann et al., 2001) usingeither diagnostic or dimensional ratings of anxiety. Oneprior study did report an association between anxietydisorders in couples (McLeod, 1995).

Cross-concordance was not observed between sub-stance use disorders and anxiety in our study. It has how-ever been reported in previous literature in the typicalpattern of males with alcohol and/or substance use disor-ders assorting with anxious (panic, phobias, generalizedanxiety disorder) and/or depressed females (Merikangaset al., 1985; McLeod, 1995; Galbaud du Fort et al.,1998), and can lead to transmissible associations betweendisorders in offspring when the two disorders are dis-tinctly transmitted (Weissman et al., 1978a,b; Merikangaset al., 1988a,b, 1992).

4.2. Mechanisms of concordance

This study addresses the first two mechanisms of spousalconcordance as listed in Introduction – assortative matingand contagion. Couples in our sample demonstrated bothprimary and secondary assortative mating. Primary assor-tative mating (defined here as assortment on the basis ofdisorders in the families of origin of the couple members)was examined by comparing the rates of substance use dis-orders among relatives of spouses married to substanceprobands and controls. We observed a twofold elevationin spouses’ relative rates in dually affected couples com-pared to well couples, and no difference in only singlyaffected couples over control couples. These findings areconsistent with other samples from clinics (Rimmer andWinokur, 1972; Hall et al., 1983a,b), but not from the com-munity (McLeod, 1995).

Secondary assortative mating (that is, non-randomselection of a martial partner based on correlates or riskfactors for the trait/disorder of interest) was also evidentin our sample because our couples were highly correlatedon sociodemographic characteristics, such as race, age,SES levels, religion, and educational attainment (Meri-kangas, 1982). Previous literature (Merikangas, 1982;McLeod, 1995; Qian, 1998; Lewis and Oppenheimer,2000) addresses this mechanism which is more likely toplace couples at risk for psychopathology in general,rather than substance use or anxiety disorders specifi-cally. Our couples, however, were all derived from thesame geographic community.

We tested the ‘‘contagion’’ mechanism of spousal con-cordance by examining the temporal relationship betweenthe onset of psychiatric disorders and marriage. The vast


majority of psychiatric illness began in both probandsand spouses prior to marriage, implying that when indi-viduals married, both already had illness. This lack oftemporal association points away from the contagionmechanism explanation; however, the natural onset ofmany anxiety disorders tends to be prior (e.g. teens) tothe typical age of marriage (e.g. 20s or later) and sub-stance use disorders’ onsets (Kessler et al., 2005) canoften coincide with it.

4.3. Implications for family/genetic studies and impact on

familial transmission

Assortative mating has important implications at thelevels of the individual, the couple, potential offspring,and society at large. Previous research on assortativelymated couples found poorer psychiatric prognoses for bothmembers of the couple, lower marital and social function-ing, higher divorce rates, and reduced coping skills (Wood-ruff et al., 1972; Weissman et al., 1978a,b; Merikangas,1982, 1984; Merikangas et al., 1983, 1992; McLeod,1994). This study observed a dose effect between probandglobal functioning and anxiety mating types: couples with0, 1, and 2 anxious members had a significant, progressivedecrement in global assessment of functioning scores (83.3,75.5 and 66.1, respectively); whereas among couples withsubstance misuse, no difference in global functioning wasobserved between mating types with 1 and 2 membersaffected. Notably, couples where both were anxious didnot differ from: (1) those with only one member who withsubstance abuse/dependence (66.1 versus 63.4, respec-tively), or (2) those with one substance abuser/dependentand the other anxious (66.1 versus 69.3, respectively). Cou-ples with only one anxiety member seem to fall at an inter-mediate level between well (no disorders) couples and allother mating types.

Illness persistence patterns among specific matingtypes showed a stepwise pattern. Overall, three groupsemerged: (1) Concordant mating types (2 ANX and 2SUB) had the highest number of persist psychiatric dis-orders (approximately 10); (2) couples with singlyaffected members (1 ANX or 1 SUB) and cross-concor-dant couples (1 ANX/1 SUB) formed an intermediategroup with approximately 5-7 current disorders; andfinally (3) the well couples with no disorders. These clin-ical observations highlight the need to encourage the psy-chiatric assessment and treatment of spouses, along withthe recognition that their typical role in a treatment planas a reliable caregiver/supporter/advocate and symptomreporter of the proband cannot be assumed. In addition,more intensive treatment of both the proband andspouse will be required.

Consequences to the offspring of assortatively matedcouples have been reported on widely, as the social andfamily structure of offspring is negatively impacted as cou-ples are not able to perform their parental duties asexpected. In addition, offspring are exposed to increased

genetic and other biological risk factors transmitted bythe dually affected parents. Studies have demonstratedincreased prevalence of substance misuse and anxiety dis-orders among offspring of couples with depression and sub-stance use disorders (Merikangas et al., 1988a,b; Maeset al., 1998). Spousal concordance for substance misuseshould be identified as a major risk factor the developmentof psychiatric disorders among offspring. The goal in mindis to improve the prognosis among those offspring who arealready affected, and to predict and prevent illness for thenext generation.

In the application of these findings to a population level,we did observe a strong association among couples to haveno disorders (OR = 2.5, 95% CI 1.4–4.5), consistent withother studies (Hall et al., 1983a,b). Thus assortative matingfor substance use disorders and no disorders may result in aclustering of ill and well families at opposite extremes ofthe population distribution. As a consequence of this shift,failure to incorporate assortative mating (for example, col-lecting probands with and without psychiatrically affectedspouses – from clinics and the community) will result inbiased findings relevant only to a subset of the populationthat is either substance abusing or well.

5. Limitations

These findings should be interpreted with these limita-tions. First, selection of some probands from clinics maylead to overestimation of the magnitude of spousal con-cordance, since the proband’s treatment may facilitatetreatment of the spouse. Second, the clinical correlatesof probands were derived from those seeking treatment,and therefore, may be more representative severe illness.We did, however, attempt to minimize clinical ascertain-ment as source of bias through correction for sampling(proband ascertainment in the logistic models werefound not to be significant) in the statistical analyses.Third, the study was cross-sectional and retrospective,as such, then limited in the power to make definitiveconclusions about temporal associations and causality.Fourth, we were unable to fully explore and test othermechanisms of spousal concordance (such as, mutualinfluence among partners and sharing common patho-genic factors) due lack of this type of testable data inthe sample. Finally, all our participants were Caucasianand therefore our results cannot be extended to otherethnic/racial groups.

References

Agulnik PL. The spouse of the phobic patient. British Journal ofPsychiatry 1970;117:59–67.

Andreasen NC, Endicott J, Spitzer RL, Winokur G. The family historymethod using diagnostic criteria. Reliability and validity. Archives ofGeneral Psychiatry 1977;34:1229–35.

Bloch KV, Klein CH, de Souza e Silva NA, Nogueira Ada R, Salis LH.Socioeconomic aspects of spousal concordance for hypertension,


obesity, and smoking in a community of Rio de Janeiro, Brazil.Arquivos Brasileiros de Cardiologia 2003;80:179–86, 171–8.

Buglass D, Clarke J, Henderson AS, Kreitman N. A study of agoraphobichousewives. Psychological Medicine 1977;7:73–86.

Cisin IH, Cahalan D. Comparison of abstainers and heavy drinkers in anational survey. Psychiatric Research Reports of the AmericanPsychiatric Association 1968;24:10–21.

Demers A, Bisson J, Palluy J. Wives’ convergence with their husbands’alcohol use: social conditions as mediators. Journal of Studies onAlcohol 1997:368–77.

Dubuis-Stadelmann E, Fenton BT, Ferrero F, Preisig M. Spousesimilarity for temperament, personality and psychiatric symptomatol-ogy. Personality and Individual Differences 2001:1095–112.

Eagles JM, Walker LG, Blackwood GW, Beattie JA, Restall DB.The mental health of elderly couples. II. Concordance forpsychiatric morbidity in spouses. British Journal of Psychiatry1987;150:303–8.

Ebers GC, Yee IM, Sadovnick AD, Duquette P. Conjugal multiplesclerosis: population-based prevalence and recurrence risks in off-spring. Canadian Collaborative Study Group. Annals of Neurology2000;48:927–31.

Endicott J, Spitzer RL. A diagnostic interview: the schedule for affectivedisorders and schizophrenia. Archives of General Psychiatry1978;35:837–44.

Farley FH, Davis SA. Arousal, personality, and assortative mating inmarriage. Journal of Sex and Marital Therapy 1977;3:122–7.

Farley FH, Mueller CB. Arousal, personality, and assortative mating inmarriage: generalizability and cross-cultural factors. Journal of Sexand Marital Therapy 1978;4:50–3.

Feng D, Baker L. Spouse similarity in attitudes, personality, andpsychological well-being. Behavior Genetics 1994;24:357–64.

Galbaud du Fort G, Bland RC, Newman SC, Boothroyd LJ. Spousesimilarity for lifetime psychiatric history in the general population.Psychological Medicine 1998;28:789–802.

Gleiberman L, Harburg E, DiFranceisco W, Schork A. Familial trans-mission of alcohol use: V. Drinking patterns among spouses,Tecumseh, Michigan. Behavior Genetics 1992;22:63–79.

Graham K, Braun K. Concordance of use of alcohol and other substancesamong older adult couples. Addictive Behaviors 1999;24:839–56.

Haber JR, Jacob T. Marital interactions of male versus female alcoholics.Family Process 1997;36:385–402.

Hafner RJ. The husbands of agoraphobic women: assortative mating orpathogenic interaction? British Journal of Psychiatry 1977;130:233–9.

Hagnell O, Kreitman N. Mental illness in married pairs in a totalpopulation. British Journal of Psychiatry 1974;125:293–302.

Hall RL, Hesselbrock VM, Stabenau JR. Familial distribution of alcoholuse: I. Assortative mating in the parents of alcoholics. BehaviorGenetics 1983a;13:361–72.

Hall RL, Hesselbrock VM, Stabenau JR. Familial distribution of alcoholuse: II. Assortative mating of alcoholic probands. Behavior Genetics1983b;13:373–82.

Heun R, Maier W. Morbid risks for major disorders and frequencies ofpersonality disorders among spouses of psychiatric inpatients andcontrols. Comprehensive Psychiatry 1993;34:137–43.

Kessler RC, Berglund P, Demler O, Jin R, Merikangas KR, Walters EE.Lifetime prevalence and age-of-onset distributions of DSM-IV disor-ders in the National Comorbidity Survey Replication. Archives ofGeneral Psychiatry 2005;62:593–602.

Knuiman MW, Divitini ML, Bartholomew HC, Welborn TA. Spousecorrelations in cardiovascular risk factors and the effect of marriageduration. American Journal of Epidemiology 1996;143:48–53.

Kolonel LN, Lee J. Husband-wife correspondence in smoking, drinking,and dietary habits. American Journal of Clinical Nutrition 1981;34:99–104.

Kreitman N. Mental disorder in married couples. Journal of MentalScience 1962;108:438–46.

Kreitman N. The patient’s spouse. British Journal of Psychiatry 1964;110:159–73.

Leckman JF, Sholomskas D, Thompson WD, Belanger A, WeissmanMM. Best estimate of lifetime psychiatric diagnosis: a methodologicalstudy. Archives of General Psychiatry 1982;39:879–83.

Leonard KE, Das Eiden R. Husband’s and wife’s drinking: unilateral orbilateral influences among newlyweds in a general population sample.Journal of Studies on Alcohol 1999;13(Suppl):130–8.

Lewis SK, Oppenheimer VK. Educational assortative mating acrossmarriage markets: non-Hispanic whites in the United States. Demog-raphy 2000;37:29–40.

Maes HH, Neale MC, Eaves LJ. Genetic and environmental factors inrelative body weight and human adiposity. Behavior Genetics1997;27:325–51.

Maes HH, Neale MC, Kendler KS, Hewitt JK, Silberg JL, Foley DL,et al. Assortative mating for major psychiatric diagnoses in twopopulation-based samples. Psychological Medicine 1998;28:1389–401.

Mascie-Taylor CG. Assortative mating and differential fertility. Biologyand Society 1986;3:167–70.

Mascie-Taylor CG. Spouse similarity for IQ and personality andconvergence. Behavior Genetics 1989;19:223–7.

Mascie-Taylor CG, Boldsen JL. Assortative mating for IQ: a multivariateapproach. Journal of Biosocial Science 1984;16:109–17.

Mascie-Taylor CG, Boldsen JL. Assortative mating, differential fertilityand abnormal pregnancy outcome. Annals of Human Biology1988;15:223–8.

Mascie-Taylor CG, Gibson JB. A biological survey of a Cambridgesuburb: assortative marriage for IQ and personality traits. Annals ofHuman Biology 1979;6:1–16.

Mathews CA, Reus VI. Assortative mating in the affective disorders: asystematic review and meta-analysis. Comprehensive Psychiatry2001;42:257–62.

McLeod JD. Spouse concordance for alcohol dependence and heavydrinking: evidence from a community sample. Alcoholism Clinical andExperimental Research 1993;17:1146–55.

McLeod JD. Anxiety disorders and marital quality. Journal of AbnormalPsychology 1994;103:767–76.

McLeod DR. Social and psychological bases of homogamy for commonpsychiatric disorders. Journal of Marriage and the Family1995;57:201–14.

Merikangas KR. Assortative mating for psychiatric disorders andpsychological traits. Archives of General Psychiatry 1982;39:1173–80.

Merikangas KR. Divorce and assortative mating among depressedpatients. American Journal of Psychiatry 1984;141:74–6.

Merikangas KR, Brunetto W. Assortative mating and psychiatric disor-ders. Balliere’s Clinical Psychiatry International Practice and Research1996;2:175–85.

Merikangas KR, Spiker DG. Assortative mating among in-patientswith primary affective disorder. Psychological Medicine 1982;12:753–764.

Merikangas KR, Bromet EJ, Spiker DG. Assortative mating, socialadjustment, and course of illness in primary affective disorder.Archives of General Psychiatry 1983;40:795–800.

Merikangas KR, Leckman JF, Prusoff BA, Pauls DL, Weissman MM.Familial transmission of depression and alcoholism. Archives ofGeneral Psychiatry 1985;42:367–72.

Merikangas K, Weissman M, Prusoff B, John K. Assortative mating andaffective disorders: psychopathology in offspring. In: Kellam S, editor.The family and psychopathology. New Brunswick, NJ: RutgersUniversity Press; 1988.

Merikangas KR, Weissman MM, Prusoff BA, John K. Assortative matingand affective disorders: psychopathology in offspring. Psychiatry1988b;51:48–57.

Merikangas K, Rounsaville B, Prusoff B. Familial factors in vulnerabilityto substance abuse. In: Glantz M, Pickens R, editors. Vulnerability todrug abuse. Washington, DC: American Psychological Association;1992. p. 75–98.

Merikangas KR, Stevens DE, Fenton B, Stolar M, O’Malley S, WoodsSW, et al. Co-morbidity and familial aggregation of alcoholism andanxiety disorders. Psychological Medicine 1998;28:773–88.


Moskalenko VD, Vanyukov MM, Solovyova ZV, Rakhmanova TV, Vladi-mirsky MM. A genetic study of alcoholism in the Moscow population:preliminary findings. Journal of Studies on Alcohol 1992;53:218–24.

Nielsen J. Mental disorders in married couples (assortative mating).British Journal of Psychiatry 1964;110:683–97.

Orvaschel H, Thompson WD, Belanger A, Prusoff BA, Kidd KK.Comparison of the family history method to direct interview. Factorsaffecting the diagnosis of depression. Journal of Affective Disorders1982;4:49–59.

Parnas J. Assortative mating in schizophrenia: results from the Copen-hagen High-Risk Study. Psychiatry 1988;51:58–64.

Qian Z. Changes in assortative mating: the impact of age and education,1970–1990. Demography 1998;35:279–92.

Reynolds CA, Baker LA, Pedersen NL. Multivariate models of mixedassortment: phenotypic assortment and social homogamy for educa-tion and fluid ability. Behavior Genetics 2000;30:455–76.

Rimmer J, Winokur G. The spouses of alcoholics: an example ofassortative mating. Diseases of the Nervous System 1972;33:509–11.

Schuckit MA, Tipp JE, Kelner E. Are daughters of alcoholics more likelyto marry alcoholics? American Journal of Drug and Alcohol Abuse1994;20:237–45.

Speers MA, Kasl SV, Freeman DH, Ostfeld Jr AM. Blood pressureconcordance between spouses. American Journal of Epidemiology1986;123:818–29.

Tambs K. Transmission of symptoms of anxiety and depression in nuclearfamilies. Journal of Affective Disorders 1991;21:117–26.

Thompson WD, Orvaschel H, Prusoff BA, Kidd KK. An evaluation of thefamily history method for ascertaining psychiatric disorders. Archivesof General Psychiatry 1982;39:53–8.

Weissman MM, Myers JK, Harding PS. Psychiatric disorders in a USurban community: 1975–1976. American Journal of Psychiatry1978a;135:459–62.

Weissman MM, Prusoff BA, Thompson WD, Harding PS, Myers JK.Social adjustment by self-report in a community sample and inpsychiatric outpatients. Journal of Nervous and Mental Disease1978b;166:317–26.

Windle M. Mate similarity, heavy substance use and family history ofproblem drinking among young adult women. Journal of Studies onAlcohol 1997;58:573–80.

Winokur G, Reich T, Rimmer J, Pitts Jr FN. Alcoholism. 3. Diagnosisand familial psychiatric illness in 259 alcoholic probands. Archives ofGeneral Psychiatry 1970;23:104–11.

Wolanski N. The stature of offspring and the assortative mating ofparents. Human Biology 1974;46:613–9.

Wolanski N. Assortative mating in somatic traits and its consequences.Studies in Human Ecology 1994;11:73–111.

Woodruff Jr RA, Guze SB, Clayton PJ. Divorce among psychiatric out-patients. British Journal of Psychiatry 1972;121:289–92.

Zimmerman-Tansella C, Lattanzi M. The Ryle marital patterns test aspredictor of symptoms of anxiety and depression in couples in thecommunity. Social Psychiatry and Psychiatric Epidemiology 1991:221–9.

Spousal concordance for substance use and anxiety disorders

Documents

Transcript of Spousal concordance for substance use and anxiety disorders