Disentangling Social-Genetic From Rearing-Environment Effects for Alcohol Use Disorder Using Swedish National Data

Abstract

Investigations of social-genetic effects, whereby a social partner’s genotype affects another’s outcomes, can be confounded by the influence of the social partner’s rearing environment. We used marital information on more than 300,000 couples from Swedish national data to disentangle social-genetic from rearing-environment effects for alcohol use disorder (AUD). Using observational and extended-family designs, we found that (a) marriage to a spouse with a predisposition toward AUD (as indexed by a parental history of AUD) increased risk for developing AUD; (b) this increased risk was not explained by socioeconomic status, the spouse’s AUD status, or contact with the spouse’s parents; and (c) this increased risk reflected the psychological consequences of the spouse having grown up with an AUD-affected parent (i.e., a rearing-environment effect) rather than a social-genetic effect. Findings illustrate that a spouse’s rearing-environment exposures may confer risk for AUD.

Keywords

social-genetic effects social-rearing-environment effects marriage alcohol use disorder

Spouses have profound influences on behavioral outcomes (Kiecolt-Glaser & Wilson, 2017) and on alcohol use in particular. A robust literature documents spousal influences on alcohol consumption (Leonard & Rothbard, 1999) and, in a series of studies that we conducted in the Swedish population, that marriage to a spouse without an alcohol use disorder (AUD) reduces the proband’s risk of developing AUD (Kendler, Larsson Lönn, Salvatore, Sundquist, & Sundquist, 2018; Kendler, Larsson Lönn, Salvatore, Sundquist, & Sundquist, 2017). These spousal socialization effects are interesting from a genetic perspective because the spouse’s drinking behavior is also genetically influenced (Verhulst, Neale, & Kendler, 2015). Thus, evidence for spousal influence at the behavioral level raises the interesting yet relatively unexplored possibility that there are social-genetic effects (Wolf, Brodie, Cheverud, Moore, & Wade, 1998) for alcohol outcomes, whereby a spouse’s genotype influences a partner’s alcohol use and likelihood of developing an AUD.

The possibility that a spouse’s genotype could influence a partner’s drinking behavior is consistent with emerging findings from a growing literature on social genomics (Baud et al., 2017; Domingue & Belsky, 2017) that demonstrate associations between a social partner’s genotype and another’s outcomes. Although the field of social genomics is relatively new, the concept of social-genetic effects has a long history in evolutionary and human genetics (Ashbrook & Hager, 2017; Wolf et al., 1998) and is related to the idea that genes influence phenotypes beyond the body as part of what Dawkins (1982) termed “the extended phenotype.” Historically, social-genetic effects have been studied in the context of biologically related individuals, such as the effect of maternal genotype on offspring outcomes such as crown-heel length at birth.

Only recently have researchers begun to investigate social-genetic effects among biologically unrelated individuals. In findings from an animal (rodent) model, Baud et al. (2017) found that the genotype of a cage mate accounted for 9% to 18% of the variation in a series of health-related outcomes. In human samples, findings from the National Longitudinal Study of Adolescent to Adult Health (Add Health) indicated that friends’ and schoolmates’ genotypes predicted educational attainment (Domingue et al., 2018) and smoking behavior (Sotoudeh, Harris, & Conley, 2019). In the UK BioBank sample, individuals whose spouses carried the protective A allele of the ADH1B single-nucleotide polymorphism rs1229984 drank less compared with those whose spouses were not carriers (Clarke et al., 2019).

A challenge when studying social-genetic effects in observational studies of humans is that a social partner’s genotype is also correlated with his or her rearing environment through passive gene–environment correlation (Plomin, DeFries, & Loehlin, 1977). This makes it difficult to disentangle the effects of a social partner’s genotype from the effects of the social partner’s rearing environment. Our goal in this study was to examine the plausibility of social-genetic effects for AUD in marital dyads and to use an extended-family design to examine whether the risk associated with a social partner’s AUD predisposition reflects social-genetic as opposed to rearing-environment effects. We examined three questions. First, does a spouse’s AUD predisposition, as indexed by spousal parental history, increase risk of developing AUD during marriage, above and beyond the proband’s own predisposition? Second, are the effects of a spouse’s AUD predisposition explained by (a) socioeconomic assortment, (b) whether the spouse is AUD affected, or (c) contact with the spouse’s parents? And finally, using an extended-family design, can we determine whether associations between a spouse’s AUD predisposition and likelihood of developing AUD reflect the effects of the spouse having a genetic predisposition to AUD (i.e., a social-genetic effect) as opposed to the psychological consequences of growing up with an AUD-affected parent (i.e., a rearing-environment effect)?

Method

Swedish national registries

We linked nationwide Swedish registers via the unique 10-digit identification number assigned at birth or immigration to all Swedish residents. The identification number was replaced by a serial number to ensure anonymity. Data came from a series of population registries: the Total Population Register, containing information about year of birth, sex, and family and marital status; the Multi-Generation Register, linking individuals born after 1932 to their parents; the Hospital Discharge Register, containing hospitalizations for Swedish inhabitants from 1964 to 2015; the Prescribed Drug Register, containing all prescriptions in Sweden picked up by patients from 2005 to 2015; the Outpatient Care Register, containing information from all outpatient clinics from 2001 to 2015; the Crime Register, which included national complete data on all convictions in lower court from 1973 to 2015; the Swedish Suspicion Register, which included national data on individuals strongly suspected of crime from 1998 to 2015; and the Mortality Register, which contains dates and causes of death from 1952 to 2016.

We examined two study groups. Our first study group included men and women born between 1955 and 1990 and in opposite-sex first marriages before age 35. Our second study group, which we term the extended-family sample, included couples in opposite-sex first marriages before age 35 for whom we had information on whether or not the partners grew up with their biological parents. Individuals were defined as having lived-with parents (LWPs) if they lived with their biological parents in the same household for more than 80% of the years before age 15. Individuals were defined as having not-lived-with parents (NLWPs) if they were registered in the same household as their biological parents for fewer than 10% of the years before age 15. Neither spouse was AUD affected prior to marriage in either study group.

Measures

The outcome variable was AUD within 10 years of marriage, censored at divorce or widowhood. AUD was defined from Swedish medical registries by the International Classification of Diseases (ICD) codes (ICD8: 571.0, 291, 303, 980; ICD9: V79B, 305A, 357F, 571A, 571B, 571C, 571D, 425F, 535D, 291, 303, 980; and ICD 10: E244, G312, G621, G721, I426, K292, K700, K701, K702, K703, K704, K709, K852, K860, O354, T510, T512, T511, T513, T518, T519, F101, F102, F103, F104, F105, F106, F107, F108, F109) and from the Prescribed Drug Register if the person retrieved disulfiram (Anatomical Therapeutic Chemical [ATC] Classification System N07BB01), acamprosate (N07BB03), or naltrexone (N07BB04). In addition, we identified individuals as AUD cases if they were convicted or suspected of at least two alcohol-related crimes according to Law 1951:649, Paragraph 4 and 4A and Law 1994:1009, Chapter 20, Paragraph 4 and 5 from the Swedish Crime Register, as well as Code 3005 and 3201 in the Suspicion register.

To assess AUD predispositions, we assessed lifetime AUD in biological parents. Yearly information on marital status and household (from the years 1960, 1965, 1970, 1975, 1980, 1985) or family ID (from 1990 and onward) were obtained from the Total Population Register. Husbands and wives were identified as opposite-sex pairs who were married and had the same family identification number.

Statistical analysis

We first ran an actor-partner interdependence model (Kenny, Kashy, & Cook, 2006) to examine the association between a spouse’s AUD predisposition and the proband’s (i.e., the focal individual’s) risk of developing AUD during marriage while controlling for the proband’s AUD predisposition. Because AUD was defined as a dichotomous outcome and parental AUD as a dichotomous explanatory variable, we followed the advice of Loeys and Molenberghs (2013) and used a generalized-estimating-equation model in which the clustering within couples is handled by a correlation coefficient. We started with a full model in which the proband’s AUD predisposition and the spouse’s AUD predisposition were allowed to vary by sex, which was captured with interaction terms between sex and the proband’s AUD predisposition and sex and the spouse’s AUD predisposition, respectively. We then compared how well the models fitted the data using the quasi-information criterion (QIC; Pan, 2001). Preliminary analyses indicated that a model that included an interaction term between sex and proband’s AUD predisposition, but not spouse’s AUD predisposition, provided the best fit for the data according to the QIC.

We then ran a series of analyses to probe alternative explanations for the association between a spouse’s AUD predisposition and proband AUD. In robustness analyses examining whether the risk associated with marriage to a spouse with an AUD predisposition could be more simply explained by the proband’s socioeconomic background, we included proband parental education as a covariate. This was measured as the highest level of education attained by either parent and categorized into three levels (1 = compulsory school, 2 = high school, and 3 = university). Preliminary analyses indicated that a model that included the interaction between parental education and sex, in addition to the interaction between proband AUD predisposition and sex, fitted the data best (lowest QIC).

Next, to investigate whether the effect of a spouse’s AUD predisposition was independent of whether the spouse had AUD, we ran a logistic regression analysis, stratified by sex, that included the proband’s predisposition to AUD, the spouse’s predisposition to AUD, the spouse’s AUD phenotype (absent or present), and the interaction between the two latter terms. An interaction differing from unity would indicate that the association between a spouse’s AUD predisposition and proband AUD varies as a function of whether the spouse is AUD affected. If the spouse’s AUD phenotype explained the association between the spouse’s AUD predisposition and proband AUD, we would expect the effect of the spouse’s AUD predisposition to be dramatically reduced when the spouse is AUD affected. If the interaction term can be represented by unity, this would indicate that the association between the spouse’s AUD predisposition and proband AUD does not vary when the spouse is AUD affected versus AUD unaffected.

In the final set of robustness analyses examining the possibility that contact with a spouse’s parents explained the association between a spouse’s AUD predisposition and proband AUD, we ran a logistic regression analysis including the spouse’s AUD predisposition, distance to the spouse’s parents (measured in kilometers and logarithm transformed), and the interaction between the two. Detailed measures regarding interpersonal contact were not available in the context of this registry-based study, so we used distance to the spouse’s parents as a reasonable proxy. An interaction differing from unity would indicate that the association between the spouse’s AUD predisposition and proband AUD varies as a function of proximity to the spouse’s parents. In contrast, an interaction that can be represented by unity would indicate that the association between a spouse’s AUD predisposition and proband AUD does not differ as a function of proximity to the spouse’s parents. These analyses were limited to couples who lived less than 100 km away from their own and their spouse’s parents so we could examine this question in a more homogenous set of couples.

We then conducted analyses using the extended-family sample to disentangle the genetic and environmental mechanisms underlying the effects of a spouse’s AUD predisposition. We ran a logistic regression analysis including the spouse’s AUD predisposition, whether the spouse came from an NLWP or an LWP family, and the interaction between the two. The contrast of interest was whether AUD resemblance between probands and their spouse’s parents differed when spouses did not grow up with their biological parents (i.e., came from an NLWP family) versus when they did (i.e., came from an LWP family). A spouse’s NLWPs provided genes but not a rearing environment to the spouse; thus, AUD resemblance between probands and their spouse’s NLWPs would be consistent with a social-genetic effect. In contrast, a spouse’s LWPs provided both genes and a rearing environment to the spouse. Thus, AUD resemblance between probands and a spouse’s LWPs would reflect either the psychological consequences of the spouse having grown up with an AUD-affected parent (i.e., a rearing-environment effect) or the combined effect of a spouse’s AUD genetic predisposition and having grown up with an affected parent. In preliminary analyses of the extended-family sample, we found that a model that combined parameter estimates across men and women provided a better fit for the data (according to the QIC) than a model allowing for different parameter estimates in men and women. We thus present results from the model in which the estimates for men and women were combined and that controlled for sex.

All analyses were conducted using SAS software Version 9.4 for Windows (SAS Institute, Cary, NC).

Results

We first examined whether a spouse’s AUD predisposition predicted probands’ likelihood of developing AUD during the first 10 years of marriage in a sample of opposite-sex spouses in first marriages in which neither had AUD prior to marriage (N = 367,398 couples). Descriptive statistics for this sample are shown in Table 1.

Table 1.

Descriptive Statistics Regarding Alcohol Use Disorder (AUD) Predispositions (Indexed by Parental History of AUD) and AUD Registration Within the First 10 Years of Marriage for Husbands and Wives in Opposite-Sex First Marriages

Group and variable	Total	No parental history of AUD in husband or wife	Parental history of AUD in husband but not wife	No parental history of AUD in husband but in wife	Parental history of AUD in husband and wife
Husbands	367,398	300,578 (81.8%)	30,299 (8.20%)	32,074 (9.6%)	4,447 (1.20%)
Husband AUD in 10 years	1,912 (0.52%)	1,257 (0.42%)	357 (1.18%)	211 (0.66%)	86 (1.93%)
Criminal behavior (prior to marriage)	43,182 (11.75%)	30,831 (10.26%)	6,154 (20.31%)	4,979 (15.52%)	1,218 (27.39%)
Low parental education	88,902 (24.20%)	72,927 (24.26%)	6,822 (22.52%)	8,079 (25.19%)	1,074 (24.15%)
Mid parental education	156,883 (42.70%)	126,371 (42.04%)	14,086 (22.58%)	14,288 (44.55%)	2,138 (48.08%)
High parental education	121,613 (33.10%)	10,128 (33.70%)	9,391 (30.99%)	9,707 (30.26%)	1,235 (27.77%)
Age at marriage (years)	28.53 (2.43)	28.55 (3.41)	28.49 (3.47)	28.46 (3.48)	28.24 (3.58)
Birth year	1966.53 (7.64)	1966.45 (7.67)	1966.92 (7.41)	1966.76 (7.61)	1967.15 (7.36)
Wives	367,398	300,578 (81.80%)	30,299 (8.20%)	32,074 (9.6%)	4,447 (1.2%)
Wife AUD in 10 years	871 (0.24%)	530 (0.18%)	103 (0.34%)	197 (0.61%)	41 (0.92%)
Criminal behavior (prior to marriage)	11,146 (3.03%)	7,800 (2.60%)	1,186 (3.91%)	1,797 (5.60%)	363 (8.16%)
Low parental education	80,644 (21.95%)	66,197 (11.01%)	6,753 (22.29%)	6,716 (20.94%)	978 (21.99%)
Mid parental education	162,891 (44.23%)	13,120 (21.82%)	13,975 (46.12%)	15,485 (24.83%)	2,230 (50.15%)
High parental education	123,863 (33.71%)	103,180 (34.33%)	9,571 (31.59%)	9,873 (30.78%)	1,239 (27.86%)
Age at marriage (years)	26.92 (3.54)	26.94 (3.52)	26.95 (3.95)	26.76 (3.46)	26.56 (3.72)
Birth year	1968.14 (7.59)	1968.07 (7.61)	1968.48 (7.47)	1968.45 (7.51)	1968.83 (7.38)

Note: The table shows ns for all variables except age at marriage and birth year, for which means are given. For the two rows showing ns for husbands and wives, values in parentheses are percentages of the total sample. For the next five rows below each heading, values in parentheses are either percentages of the total sample (Column 2) or percentages of the subsample in that column (Columns 3–6). For age at marriage and birth year, values in parentheses are standard deviations.

Using an actor-partner interdependence model, we examined whether a spouse’s AUD predisposition was associated with the proband’s AUD registration and whether this spousal effect held after controlling for the proband’s AUD predisposition. As shown in Table 2, marriage to a spouse with an AUD predisposition was associated with increased risk for the proband developing AUD during marriage, odds ratio (OR) = 1.66, 95% confidence interval (CI) = [1.50, 1.84]. Although the best-fitting model included an interaction between sex and the proband’s AUD predisposition, this term was not significant, OR = 0.85, 95% CI = [0.71, 1.02]. The OR for an interaction term represents the excess (or reduction) in risk associated with the moderator’s non–reference group. Thus, this nonsignificant interaction term indicates that the AUD risk associated with proband predispositions (OR = 3.36, 95% CI = [2.90, 3.90] for women) was not significantly different for men (i.e., OR = 3.36 × 0.85 = 2.85, 95% CI = [2.56, 3.17]).

Table 2.

Results of the Logistic Regression Model Predicting Alcohol Use Disorder (AUD) as a Function of Proband and Spouse AUD Predispositions (Indexed by Parental History of AUD)

Parameter	Odds ratio	95% CI	p
Proband sex (reference = women)	2.31	[2.10, 2.53]	< .0001
Proband AUD predisposition	3.36	[2.90, 3.90]	< .0001
Spouse AUD predisposition	1.66	[1.50, 1.84]	< .0001
Proband AUD Predisposition × Proband Sex	0.85	[0.71, 1.02]	.0781
Proband AUD predisposition (men; derived association)	2.85	[2.56, 3.17]	< .0001

Note: Significant results are indicated by boldface (p < .05). CI = confidence interval.

We then conducted a series of robustness analyses aimed at probing whether family socioeconomic status, spouse’s AUD status, or contact with the spouse’s AUD-affected parents (as indexed with a measure of distance to the spouse’s parents) explained the risk associated with a spouse’s AUD predisposition. In the first set of robustness analyses, we included parental education as a covariate. As shown in Table 3, there was no reduction in the risk associated with marriage to a spouse with an AUD predisposition even after we controlled for parental education, OR = 1.66, 95% CI = [1.49, 1.84].

Table 3.

Results of the Logistic Regression Model Predicting Alcohol Use Disorder (AUD) as a Function of Proband and Spouse AUD Predispositions, Controlling for Parental Education

Parameter	Odds ratio	95% CI	p
Proband sex (reference = women)	2.78	[2.30, 3.36]	< .0001
Proband AUD predisposition	3.34	[2.88, 3.88]	< .0001
Spouse AUD predisposition	1.66	[1.49, 1.84]	< .0001
Proband AUD Predisposition × Proband Sex	0.85	[0.71, 1.02]	.0821
Proband parental education (mid vs. low)	1.39	[1.15, 1.67]	.0005
Proband parental education (high vs. low)	1.24	[1.02, 1.52]	.0296
Proband Parental Education (Mid vs. Low) × Proband Sex	1.11	[0.99, 1.25]	.0732
Proband Parental Education (High vs. Low) × Proband Sex	0.98	[0.87, 1.11]	.8083

Note: Significant results are indicated by boldface (p < .05). CI = confidence interval.

In the second set of robustness analyses, we probed whether the spouse’s AUD predisposition provided unique predictive power above and beyond the spouse’s AUD phenotype. Descriptive statistics regarding AUD in husbands and wives as a function of spouses’ AUD predispositions and spouse AUD are shown in Table S1 in the Supplemental Material available online. As shown in Table 4, the interaction between spouse’s AUD predisposition and the spouse’s AUD status when predicting husbands’ or wives’ AUD did not significantly differ from unity (husbands: OR = 1.79, 95% CI = [0.96, 3.33]; wives: OR = 1.03, 95% CI = [0.54, 1.97]). This suggests that the association between a spouse’s AUD predisposition and proband AUD did not vary as a function of the spouse’s AUD status. Although these effects were not statistically significant at a conventional level, we note that the interaction effect for husbands approached significance (p = .0677) and suggested that the association between a spouse’s AUD predisposition and a husband’s likelihood of AUD registration during marriage was stronger when the spouse was AUD affected, OR = 2.70, 95% CI = [1.47, 4.97] compared with when the spouse was AUD unaffected, OR = 1.51, 95% CI = [1.33, 1.72].

Table 4.

Results of the Logistic Regression Model Predicting Alcohol Use Disorder (AUD) as a Function a Spouse’s AUD Predisposition (Indexed by Parental History of AUD) and the Spouse’s AUD Phenotype, Separately for Husbands and Wives

Parameter	Husbands			Wives
Parameter	Odds ratio	95% CI	p	Odds ratio	95% CI	p
Proband AUD predisposition (reference = proband without AUD predisposition)	2.83	[2.54, 3.15]	< .0001	3.28	[2.82, 3.81]	< .0001
Spouse AUD predisposition (reference = spouse without AUD predisposition)	1.51	[1.33, 1.72]	< .0001	1.68	[1.39, 2.03]	< .0001
Spouse AUD (reference = spouse without AUD)	7.07	[4.64, 10.77]	< .0001	8.81	[6.04, 12.83]	< .0001
Spouse AUD Predisposition × Spouse AUD	1.79	[0.96, 3.33]	.0677	1.03	[0.54, 1.97]	.9318
Spouse AUD predisposition when spouse AUD affected (derived association)	2.70	[1.47, 4.97]	.0014	1.73	[0.93, 3.22]	.0861

Note: Significant results are indicated by boldface (p < .05). CI = confidence interval.

In the third set of robustness analyses, we aimed to examine whether contact with a spouse’s AUD-affected parents explained the increased risk associated with a spouse’s AUD predisposition. We tested whether there was an interaction between the spouse’s AUD predisposition (as indexed by parental AUD) and distance to the spouse’s parents. We identified 245,024 couples who lived less than 100 km away from their own and their spouse’s parents for these analyses. Descriptive statistics regarding AUD predispositions, distance to parents, and AUD registration for this subsample are shown in Table S2 in the Supplemental Material. As shown in Table 5, the interaction between a spouse’s AUD predisposition and distance to the spouse’s parents did not significantly differ from unity, OR = 1.04, 95% CI = [0.95, 1.14]. Thus, the association between a spouse’s AUD predisposition and proband AUD did not vary as a function of proximity to the spouse’s parents.

Table 5.

Results of the Logistic Regression Model Predicting Alcohol Use Disorder (AUD) as a Function of Proband and Spouse AUD Predispositions (Indexed by Parental History of AUD) and Distance to the Spouse’s Parents

Parameter	Odds ratio	95% CI	p
Proband sex (reference = women)	2.25	[2.03, 2.49]	< .0001
Parental history of AUD: proband (reference = proband without parental history of AUD)	3.18	[2.61, 3.88]	< .0001
Parental history of AUD: spouse (reference = spouse without parental history of AUD)	1.56	[1.23, 1.99]	.0003
Distance to proband’s parents (log km)	1.04	[1.00, 1.08]	.0341
Distance to spouse’s parents (log km)	1.01	[0.98, 1.05]	.5808
Parental History of AUD × Distance to Parents: Proband	0.96	[0.89, 1.04]	.3618
Parental history of AUD × Distance to Parents: Spouse	1.04	[0.95, 1.14]	.4050
Distance to spouse’s parents with history of AUD (log km; derived association)	1.05	[0.96, 1.14]	.2593

Note: Significant results are indicated by boldface (p < .05). CI = confidence interval.

Having established that marriage to a spouse with an AUD predisposition is associated with increased risk for developing AUD during marriage, we next used an extended-family design to evaluate whether these associations reflected social-genetic or rearing-environment effects. The study group for these analyses (the extended-family sample) consisted of 318,084 couples in opposite-sex first marriages for whom we had information on whether or not both partners grew up with their biological parents. Descriptive statistics for this sample are shown in Table S3 in the Supplemental Material.

We used logistic regression to examine whether the risk associated with a spouse’s AUD predisposition differed as a function of whether the spouse came from an NLWP or an LWP family, as tested with an interaction. As shown in Table 6, marriage to a spouse from an NLWP family was associated with increased risk for AUD, OR = 1.68, 95% CI = [1.43, 1.97]. Above and beyond this, there was some evidence of a significant interaction between a spouse’s AUD predisposition and whether the spouse came from an NLWP or an LWP family, OR = 0.71, 95% CI = [0.52, 0.97]. Consequently, a spouse’s AUD predisposition significantly predicted the proband’s AUD registration during marriage only when the spouse came from an LWP family, OR = 1.67, 95% CI = [1.41, 3.17]. In contrast, the association between a spouse’s AUD predisposition and proband AUD was not significant when the spouse came from an NLWP family, OR = 1.19, 95% CI = [0.91, 1.56].

Table 6.

Results of the Logistic Regression Predicting Alcohol Use Disorder (AUD) as a Function of Proband and Spouse AUD Predispositions (Indexed by Parental History of AUD), Living Arrangements While Growing Up, and Their Interactions

Parameter	Odds ratio	95% CI	p
Proband sex (reference = women)	2.29	[2.08, 2.53]	< .0001
Proband AUD predisposition (reference = proband without AUD predisposition)	2.77	[2.41, 3.17]	< .0001
Spouse AUD predisposition (reference = spouse without AUD predisposition)	1.67	[1.41, 3.17]	< .0001
Proband from NLWP family (reference = proband from LWP family)	2.01	[1.71, 2.37]	< .0001
Spouse from NLWP family (reference = spouse from LWP family)	1.68	[1.43, 1.97]	< .0001
Proband AUD Predisposition Parental AUD × Proband From NLWP Family	0.85	[0.65, 1.11]	.2365
Spouse AUD Predisposition × Spouse From NLWP Family	0.71	[0.52, 0.97]	.0334
Spouse AUD predisposition when spouse from NLWP family (derived association)	1.19	[0.91, 1.56]	.2099

Note: Significant results are indicated by boldface (p < .05). CI = confidence interval; LWP = lived-with parent; NLWP = not-lived-with parent.

Discussion

We used information about spouses’ AUD predispositions and living arrangements while growing up to investigate social-genetic effects for AUD. Using measures of parental history of AUD to index spouses’ AUD predispositions, we found that a spouse’s AUD predisposition was associated with an increased risk of developing AUD during marriage. As expected, the magnitude of the effect of the spouse’s AUD predisposition was smaller than the effect of the proband’s AUD predisposition. The spousal effect was of the same magnitude and remained significant even after we controlled for the proband’s parental education, suggesting that the risk associated with a spouse’s AUD predisposition did not simply reflect an effect of the proband’s broader socioeconomic background that may have constrained his or her spousal selection.

We probed two alternative explanations for our finding that a spouse’s AUD predisposition increased risk for developing AUD. First, we tested whether the effect of the spouse’s AUD predisposition simply reflected the spouse’s AUD phenotype. We found that a spouse’s AUD predisposition and a spouse’s AUD phenotype had largely independent effects, and we found no evidence that the risk associated with the spouse’s AUD predisposition was simply a recapitulation of the spouse’s AUD phenotype. In direct contrast to the alternative hypothesis that a spouse’s AUD phenotype explains the effect of a spouse’s AUD predisposition, the trend observed in men indicated the opposite. The association between a spouse’s AUD predisposition and a husband’s likelihood of AUD registration during marriage became stronger when the spouse was AUD affected compared with when the spouse with AUD unaffected. Second, because we used parental history to index AUD predispositions, we examined whether the risk associated with having an AUD-affected in-law was stronger when probands had more contact with their in-laws. Using geographical distance to index probable contact with in-laws, we found no evidence that living closer increased the risk associated with marriage to a spouse who has an AUD-affected parent. Taken together, these analyses suggest that the risk associated with marriage to a spouse who has an AUD predisposition is not explained by the spouse’s AUD phenotype or by the proband’s direct exposure to the spouse’s AUD-affected parents.

We next used an extended-family design to delineate whether the risk associated with marriage to a spouse with an AUD predisposition reflected social-genetic effects as opposed to rearing-environment effects. Parental history of AUD captures both genetic and environmental components of AUD risk for offspring; accordingly, the spousal AUD-predisposition effects documented above only begin to establish the plausibility of social-genetic effects for AUD. We used information about whether the partners grew up with their biological parents and examined the magnitude of the spousal AUD-predisposition effect both when the spouse’s parents provided genes as well as a rearing environment to the spouse and when the spouse’s parents provided genes but not a rearing environment.

We found that a spouse’s AUD predisposition increased the risk of developing AUD to a significant extent when the spouse was raised by his or her biological parents. In contrast, the spouse’s AUD predisposition did not significantly increase risk for AUD when the spouse was raised in another home. Thus, the risk associated with marriage to a spouse with an AUD predisposition likely reflects the psychological consequences of the spouse having grown up with an AUD-affected parent rather than a social-genetic effect. Importantly, we observed these effects after controlling for the risk associated with marriage to a spouse who was raised in another home.

The results from this study are informative for the broader literature on spousal influences for alcohol outcomes, which has typically focused on alcohol-specific contagion models in which drinking behaviors in one partner are socially transmitted to the other (Leonard & Mudar, 2003). Our findings go beyond this to demonstrate that marrying a spouse with an affected parent increases AUD risk, even if the spouse is unaffected. This underscores the far-reaching impact of growing up with an AUD-affected parent, which extends even to the spouses of those adult children. Although we are unable to delineate the specific mechanisms underlying these spousal effects, one hypothesis is that growing up with an AUD-affected parent may socialize individuals to act in ways that reinforce a spouse’s problem drinking behavior, such as engaging in rewarding leisure activities that involve alcohol or caretaking for a spouse during a drinking episode.

Another hypothesis is that given that parental AUD is a risk factor for lower dyadic functioning (Harter, 2000), couples in which one or both partners grow up with an affected parent may be at particular risk of using alcohol as a tool to improve their marital interactions (Fairbairn & Testa, 2016). More broadly, spouses who grew up with an AUD-affected parent may have other psychosocial liabilities that limit their ability to engage in the types of health-promoting behaviors typically associated with marriage. Consistent with this possibility, studies of the intergenerational transmission of antisocial behavior and substance use disorders indicate that parents pass a general vulnerability to these disorders to their offspring, in whom they can manifest as impulsivity and behavioral disinhibition (Hicks, Krueger, Iacono, McGue, & Patrick, 2004).

The results from our study also contribute to ongoing discussions regarding sociogenomics (Braudt, 2018) and highlight the need for designs in which it is possible to disentangle purported social-genetic effects from a correlated rearing environment. In view of the pervasive evidence for passive gene–environment correlation for many behaviors (Jaffee & Price, 2007), caution is warranted when interpreting evidence for social-genetic effects. Indeed, our findings suggest that the risk associated with a spouse’s AUD predisposition reflects rearing-environment effects rather than social-genetic effects.

Limitations

These results should be interpreted in view of the study’s limitations. First, AUD diagnoses came from national registries and for this reason generally capture the more severely affected individuals. The underdiagnosis of AUD may be the greatest weakness of our study and may have led us to underestimate the effects of a spouse’s AUD phenotype in our robustness analyses. However, we note that, somewhat tempering this concern, the rates of AUD observed here closely parallel the rates of alcohol dependence observed in an epidemiological study in nearby Norway (Kringlen, Torgersen, & Cramer, 2001). Second, geographical distance is an imperfect proxy for our analyses of whether contact with a spouse’s AUD-affected parents was likely to explain the risk associated with marrying a spouse with an AUD predisposition. However, more detailed measures regarding contact with in-laws were not available in our sample. Third, we included only legally married opposite-sex couples in our analyses, and it is unknown whether the pattern of findings generalizes to nonmarital cohabiting relationships, which are relatively common in Sweden (Organisation for Economic Co-operation and Development, 2016), or to same-sex couples.

Fourth, follow-up was censored by divorce, which raises potential concerns about bias resulting from censoring. We believe that these concerns are minimized in view of prior evidence that parental AUD increases risk for offspring divorce (Salvatore, Larsson Lönn, Sundquist, Sundquist, & Kendler, 2018). Thus, to the extent that marriage to a spouse with an AUD predisposition (as indexed by parental AUD) increases risk for both AUD and marital dissolution, censoring because of divorce would attenuate the association between spousal AUD predispositions and AUD. Fifth, the three-level measure of parental education used in robustness analyses did not fully capture all aspects of family socioeconomic status (e.g., parental occupation and income). Sixth, the critical test in our extended-family design was of differences in relative risks (ORs), which are known to be sensitive to base rates. We ran a series of analyses using methods less sensitive to base rates, and as summarized in the Supplemental Material, the pattern of effects was similar.

Conclusions

In the Swedish population, marriage to a spouse with an AUD predisposition increased the risk of developing AUD during marriage, even when the spouse was personally unaffected. Furthermore, the risk associated with marriage to a spouse with an AUD predisposition is not explained by socioeconomic status nor by contact with the spouse’s parents. By probing the nature of this effect using an extended-family design, we found evidence for a rearing-environment effect: Marriage to a spouse who grew up with an AUD-affected parent, but not marriage to a spouse who inherited only a genetic predisposition to AUD, increased risk for AUD. Thus, although there are strong genetic and biological components to an individual’s risk for AUD, these findings extend the robust literature on marriage and alcohol to demonstrate the complex pathways through which a spouse’s environmental exposures may also impact one’s own risk for AUD.

Supplemental Material

Salvatore_Supplemental_Material_rev – Supplemental material for Disentangling Social-Genetic From Rearing-Environment Effects for Alcohol Use Disorder Using Swedish National Data

Supplemental material, Salvatore_Supplemental_Material_rev for Disentangling Social-Genetic From Rearing-Environment Effects for Alcohol Use Disorder Using Swedish National Data by Jessica E. Salvatore, Sara Larsson Lönn, Jan Sundquist, Kristina Sundquist and Kenneth S. Kendler in Psychological Science

Footnotes

Transparency

Action Editor: Steven W. Gangestad

Editor: D. Stephen Lindsay

Author Contributions

J. E. Salvatore and K. S. Kendler developed the study concept and design. S. Larsson Lönn analyzed the data. J. E. Salvatore drafted the manuscript, and S. Larsson Lönn, J. Sundquist, K. Sundquist, and K. S. Kendler provided critical revisions. All authors approved the final version of the manuscript for submission. K. Sundquist and K. S. Kendler share last authorship of this article.

ORCID iD

Jessica E. Salvatore

Supplemental Material

Additional supporting information can be found at

References

Ashbrook

D. G.

Hager

(2017). Social interactions and indirect genetic effects on complex juvenile and adult traits. Methods in Molecular Biology, 1488, 499–517. doi:10.1007/978-1-4939-6427-7_24

Baud

Mulligan

M. K.

Casale

F. P.

Ingels

J. F.

Bohl

C. J.

Callebert

. . . Stegle

(2017). Genetic variation in the social environment contributes to health and disease. PLOS Genetics, 13(1), Article e1006498. doi:10.1371/journal.pgen.1006498

Braudt

D. B.

(2018). Sociogenomics in the 21st century: An introduction to the history and potential of genetically informed social science. Sociology Compass, 12(10), Article e12626. doi:10.1111/soc4.12626

Clarke

T.-K.

Adams

Howard

D. M.

Xia

Davies

Hayward

. . . McIntosh

A. M.

(2019). Genetic and shared couple environmental contributions to smoking and alcohol use in the UK population. Molecular Psychiatry. Advance online publication. doi:10.1038/s41380-019-0607-x

Dawkins

(1982). The extended phenotype. Oxford, England: Oxford University Press.

Domingue

B. W.

Belsky

D. W.

(2017). The social genome: Current findings and implications for the study of human genetics. PLOS Genetics, 13(3), Article e1006615. doi:10.1371/journal.pgen.1006615

Domingue

B. W.

Belsky

D. W.

Fletcher

J. M.

Conley

Boardman

J. D.

Harris

K. M.

(2018). The social genome of friends and schoolmates in the National Longitudinal Study of Adolescent to Adult Health. Proceedings of the National Academies of Sciences, USA, 115, 702–707. doi:10.1073/pnas.1711803115

Fairbairn

C. E.

Testa

(2016). Relationship quality and alcohol-related social reinforcement during couples interaction. Clinical Psychological Science, 5, 74–84. doi:10.1177/2167702616649365

Harter

S. L.

(2000). Psychosocial adjustment of adult children of alcoholics: A review of the recent empirical literature. Clinical Psychology Review, 20, 311–337. doi:10.1016/s0272-7358(98)00084-1

10.

Hicks

B. M.

Krueger

R. F.

Iacono

W. G.

McGue

Patrick

C. J.

(2004). Family transmission and heritability of externalizing disorders: A twin-family study. Archives of General Psychiatry, 61, 922–928. doi:10.1001/archpsyc.61.9.922

11.

Jaffee

S. R.

Price

T. S.

(2007). Gene-environment correlations: A review of the evidence and implications for prevention of mental illness. Molecular Psychiatry, 12, 432–442. doi:10.1038/sj.mp.4001950

12.

Kendler

K. S.

Larsson Lönn

Salvatore

J. E.

Sundquist

(2018). The origin of spousal resemblance for alcohol use disorder: Causal inference utilizing longitudinal and within-person designs. JAMA Psychiatry, 75, 280–286. doi:10.1001/jamapsychiatry.2017.4457

13.

Kendler

K. S.

Lönn

Salvatore

Sundquist

(2017). Divorce and the onset of alcohol use disorder: A Swedish population-based longitudinal cohort and co-relative study. American Journal of Psychiatry, 174, 451–458. doi:10.1176/appi.ajp.2016.16050589

14.

Kenny

D. A.

Kashy

D. A.

Cook

W. L.

(2006). The analysis of dyadic data. New York, NY: Guilford.

15.

Kiecolt-Glaser

J. K.

Wilson

S. J.

(2017). Lovesick: How couples’ relationships influence health. Annual Review of Clinical Psychology, 13, 421–443. doi:10.1146/annurev-clinpsy-032816-045111

16.

Kringlen

Torgersen

Cramer

(2001). A Norwegian psychiatric epidemiological study. American Journal of Psychiatry, 158, 1091–1098. doi:10.1016/0924-9338(96)88727-x

17.

Leonard

K. E.

Mudar

(2003). Peer and partner drinking and the transition to marriage: A longitudinal examination of selection and influence processes. Psychology of Addictive Behaviors, 17, 115–125. doi:10.1037/0893-164x.17.2.115

18.

Leonard

K. E.

Rothbard

J. C.

(1999). Alcohol and the marriage effect. Journal of Studies on Alcohol, 13, 139–146. doi:10.15288/jsas.1999.s13.139

19.

Loeys

Molenberghs

(2013). Modeling actor and partner effects in dyadic data when outcomes are categorical. Psychological Methods, 18, 220–236. doi:10.1037/a0030640

20.

Organisation for Economic Co-Operation and Development. (2016). SF3.3. Cohabitation rate and prevalence of other forms of partnership. Retrieved from http://www.oecd.org/els/family/SF_3-3-Cohabitation-forms-partnership.pdf

21.

Pan

(2001). Akaike’s information criterion in generalized estimating equations. Biometrics, 57, 120–125. doi:10.1111/j.0006-341x.2001.00120.x

22.

Plomin

DeFries

J. C.

Loehlin

J. C.

(1977). Genotype-environment interaction and correlation in the analysis of human behavior. Psychological Bulletin, 84, 309–322. doi:10.1037/0033-2909.84.2.309

23.

Salvatore

J. E.

Larsson Lönn

Sundquist

Kendler

K. S.

(2018). Genetics, the rearing environment, and the intergenerational transmission of divorce: A Swedish national adoption study. Psychological Science, 29, 370–378. doi:10.1177/0956797617734864

24.

Sotoudeh

Harris

K. M.

Conley

(2019). Effects of the peer metagenomic environment on smoking behavior. Proceedings of the National Academies of Sciences, USA. 116, 16302–16307. doi:10.1073/pnas.1806901116

25.

Verhulst

Neale

M. C.

Kendler

K. S.

(2015). The heritability of alcohol use disorders: A meta-analysis of twin and adoption studies. Psychological Medicine, 45, 1061–1072. doi:10.1017/S0033291714002165

26.

Wolf

J. B.

Brodie

E. D.

Cheverud

J. M.

Moore

A. J.

Wade

M. J.

(1998). Evolutionary consequences of indirect genetic effects. Trends in Ecology & Evolution, 13(2), 64–69. doi:10.1016/s0169-5347(97)01233-0

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB