Pattern of Risks for Psychiatric and Substance Use Disorders in the Offspring of Parents With Alcohol Use Disorder

Abstract

Objective:

The authors sought to clarify the components of the familial liability to alcohol use disorder (AUD) by examining parent-offspring transmission in a large Swedish population sample.

Methods:

To this end, 1,244,516 offspring in intact families with a mean age at follow-up of 37.7 years (SD=6.8) were examined. Hazard ratios for offspring of parents with AUD were calculated using Cox models for risk of five disorders assessed from Swedish medical and criminal registries: AUD, drug use disorders, attention deficit hyperactivity disorder, major depression, and anxiety disorders.

Results:

The hazard ratio for the offspring was highest for AUD (hazard ratio=2.36), followed by drug use disorder (hazard ratio=2.04), attention deficit hyperactivity disorder (hazard ratio=1.82), major depression (hazard ratio=1.43), and anxiety disorder (hazard ratio=1.43). The risks for AUD were statistically indistinguishable between the children having mothers with AUD compared with those having fathers with AUD and between sons and daughters of a parent with AUD. All risks for offspring having two parents with AUD were higher than those having one parent with AUD, but the increase with two parents with AUD was greatest for AUD, followed by drug use disorder and attention deficit hyperactivity disorder. Age at AUD onset of the parents predicted risk among the offspring more strongly for AUD and drug use disorder, followed by attention deficit hyperactivity disorder, and then major depression and anxiety disorders. Number of recurrences of the parents with AUD predicted risks for all disorders equally. The risk pattern of disorders for the offspring of not-lived-with fathers with AUD was similar to that in the main analysis of intact families. No evidence was found for sex-specific transmission of AUD or a familial female protective effect.

Conclusions:

Familial and likely genetic liability to AUD has three components: a nonspecific risk of common internalizing and externalizing disorders, a moderately specific risk of externalizing disorders, and a highly specific risk of AUD.

The observation that alcohol use disorder (AUD) is transmitted from parents to their offspring is among the oldest in psychiatric genetics, initially noted by the first century Roman writer Plutarch (46–119 A.D.) with the words “Ebrii gignunt ebrios” (one inebriate begets another) (1). The familial aggregation of AUD has been supported by more modern investigations (2–5). Furthermore, parent-offspring transmission of AUD has been examined in both twin-family (6) and adoption studies (7–10). Familial coaggregation has also been studied between AUD and major depression (11, 12), drug use disorders (4, 5), anxiety disorders (4, 13), and attention deficit hyperactivity disorder (ADHD) (4).

Multivariate twin studies (14–16) and molecular studies (17) have identified two clusters of genetically related common psychiatric conditions: internalizing disorders, typically consisting of major depression and anxiety disorder, and externalizing disorders, frequently including AUD, drug use disorders, antisocial personality disorder, and ADHD. Disorders within each of these clusters share a substantial proportion of their genetic risk with each other, and the two sets of disorders are generally positively and more modestly genetically correlated (16, 17). Familial risk of AUD can also be indexed by specific clinical features in affected individuals, particularly age at onset (18–20) and number of episodes (20, 21).

Higher male prevalence of AUD is the most reliable epidemiological feature of AUD (22). As Carter (23) noted decades ago, the most plausible explanation for sex effects in multifactorial familial syndromes is that the sex with the lower prevalence has an endogenous protective factor, thereby requiring higher familial risk for the disorder to develop. Early family studies of AUD generally supported Carter’s hypothesis (2), but a later detailed evaluation of this theory (24) and a family study (4) both produced negative results.

Efforts to clarify the nature of the familial transmission of schizophrenia led to eight classic studies of offspring of two affected parents (25); however, a review of family studies of alcoholism (2) has not revealed a similar effort. The question is what spectrum of disorders would be at particularly increased risk in the offspring of parents both of whom have AUD.

In this study, we analyzed data from a large sample of parents and offspring from the Swedish population to address six specific questions about the risk, in children of parents with AUD, for AUD, two externalizing disorders (drug use disorder and ADHD), and two internalizing disorders (major depression and anxiety disorder). First, what is the relative magnitude of transmission of parental AUD to AUD, drug use disorder, ADHD, major depression, and anxiety disorder in the offspring? Second, are there differences in the magnitude of transmission between paternal and maternal AUD? Third, does parental AUD result in differences in risk between sons and daughters? Fourth, what are the patterns of risk of these disorders in offspring having either one parent or both parents with AUD? Fifth, what is the degree of parent-offspring transmission of these five disorders within families with a parent with AUD, using parental age at onset and the number of recurrences as risk indices? Sixth, are the patterns of risk of these five disorders similar between offspring of parents who contributed only genes to their children (that is, not-lived-with fathers) and those with parents who contributed to both the children’s genes and their rearing?

METHODS

Data for this study were collected from nationwide Swedish registers (see Table S1 in the online supplement). Each person’s unique identification number, replaced with a serial number for confidentiality, was used for registry linkages. Ethical approval was obtained from the regional ethical review board in Lund, Sweden. We included all individuals, here called offspring, born in Sweden from 1970 to 1990 and raised in an intact family (defined as living in the same household as the biological mother and the biological father between ages 0 through 15) (for details of the identification of intact families and families with not-lived-with fathers, see Table S2 in the online supplement).

Diagnoses of AUD in parents and five disorders in offspring (AUD, drug use disorder, ADHD, major depression, and anxiety disorder) were ascertained from the Swedish Hospital Discharge Register, Outpatient Care Register, primary care data (with almost nationwide coverage), the Swedish Prescribed Drug Register, the Swedish Cause of Death Register, the Swedish Criminal Register, and the Swedish Suspicion Register (see Table S3 in the online supplement).

In intact families, we studied the effect of lifetime parental diagnoses of AUD on time to first diagnosis in offspring from the age of 10; these analyses were performed with a multivariable Cox proportional hazards model with age of offspring as the time scale. The offspring were followed until diagnosis, end of study (December 31, 2018), death, or emigration, whichever occurred first. To account for the dependency between siblings, standard errors were obtained using robust variance estimation. The main model included two independent variables, maternal AUD and paternal AUD, and the interaction effect between the two. We also controlled for sex and year of birth of offspring and year of birth of the mother and of the father. To remove problems with nonproportional hazards, we included interaction effects between age of offspring and these controlling variables by using a procedure of splitting data over time. From the main model, we retrieved hazard ratios of having a mother ever diagnosed with AUD and having a father ever diagnosed with AUD, and we determined the hazard ratio of parental AUD on the outcomes by calculating the average of the mother and father regressors. We also calculated contrasts for testing whether the effects of fathers were equal to the effects of mothers. To further analyze the effect of parental AUD on sons and on daughters, we included interaction effects with sex of offspring and the maternal and paternal AUD variables in the model. Furthermore, the effect of having two parents with AUD was retrieved from the sum of the maternal and paternal effects and their interaction. The same modeling approach used with intact families was applied to identify the effect of paternal AUD in families with not-lived-with fathers, while controlling for the potential effect of lived-with mothers with AUD.

To study the effect of age at AUD onset in parents, measured by age at first AUD registration, we used another multivariable Cox proportional hazard model, which included all parent-offspring pairs in intact families that had a parent with AUD (N=102,940). We included a continuous variable for age at AUD onset, divided into 5-year intervals, controlling for birth years of the offspring and the parent and sex of the offspring, along with the relevant interaction effects with offspring age. The same type of model was also applied to study the effect of number of parental AUD recurrences (used as a continuous variable) in the registries. Registrations occurring within 90 days of each other were defined as one recurrence.

In the main model, to account for parental comorbidity of a particular diagnosis, when studying the effect of parental AUD on that same diagnosis in the offspring, we conducted two sensitivity analyses. First, we identified parents with an onset of that diagnosis prior to the onset of AUD. We removed the effect of these parents by including dummy variables for the mothers and fathers with the comorbid condition. Second, we identified parents who ever had both that diagnosis and AUD, and we removed the effect of these parents with the same principle. Furthermore, to account for diagnoses of AUD in offspring, when studying time to diagnosis of drug use disorder, ADHD, major depression, or anxiety disorder, we performed another sensitivity analysis. Offspring with an onset of AUD prior to or on the same day as the diagnosis of interest were excluded from the analyses.

Bonferroni correction procedures for multiple testing, with a significance threshold of 0.05, were applied to all data that included p values. Data analysis was conducted from March 20, 2023, to August 23, 2023. Statistical analyses were performed using R, version 4.2.1 (26) (see Table S4 in the online supplement), and SAS, version 9.4 (27).

RESULTS

A comparison of our sample with the general population is shown in Table 1, and detailed descriptive statistics for the sample are provided in Table 2. The total sample comprised 1,244,516 offspring with a mean age at follow-up of 37.7 years (SD=6.8). Prevalence of the five disorders among the offspring ranged from 2.1% for ADHD to 15.2% for major depression, with a rate of 2.7% for AUD. The offspring demonstrate the expected higher female rates of major depression and anxiety disorder and higher male rates of AUD, drug use disorder, and ADHD. The mean ages at onset for these five disorders in the offspring ranged from 26.2 years for drug use disorder to 31.2 for anxiety disorder. The lifetime prevalence of AUD was 2.1% for mothers and 6.2% for fathers.

TABLE 1.

Comparison of samples in a study of risk for psychopathology among the children of parents with alcohol use disorder with the general Swedish population^a

	Offspring				DS
Measure	Population Sample (N=2,071,848)		Study Sample (N=1,244,516)		Population Sample (N=3,228,061)		Study Sample (N=l,445,225)
	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Year of birth	1980.0	6.3	1979.9	6.1	1949.4	8.6	1949.9	8.1
Age at follow-up (years)	37.4	7.0	37.7	6.8	67.6	8.8	67.6	7.9
	N	%	N	%	N	%	N	%
Male	1,063,922	51.4	645,815	51.9	1,633,359	50.6	722,597	50.0
Highest level of education
Presecondary education	146,346	7.1	57,077	4.6	792,482	24.5	320,358	22.2
Secondary education	887,261	42.8	497,345	40.0	1,405,001	43.5	654,080	45.3
Postsecondary education	1,023,455	49.4	683,633	54.9	1,019,391	31.6	470,627	32.6
Not available	14,567	0.7	6,461	0.5	5,624	0.2	160	0.0
Born in Sweden	2,071,848	100	1,244,516	100	2,967,551	91.9	1,322,247	91.5
Birth region among those born in Swedenb
Stockholm County	396,314	19.1	209,950	16.9	436,527	14.7	167,590	12.7
Northern Sweden	289,603	14.0	178,088	14.3	557,584	18.8	245,056	18.5
Central Sweden	389,959	18.8	230,670	18.5	566,939	19.1	246,574	18.6
Southern Sweden	995,667	48.1	625,804	50.3	1,403,137	47.3	662,968	50.1

^aThe offspring study sample was compared with the sample of all individuals born in Sweden between 1970 and 1990 who were followed up until at least age 15. To produce a comparative sample from the general Swedish population for the parents in the study sample, we included individuals (parents and nonparents) born between the years 1935 and 1964, which covered 95% of the birth years in the parental study sample. To ensure data availability, the individuals had to have been followed up until at least 1995 and did not immigrate to Sweden after 1979.

^bDefinition of birth regions: northern Sweden: Norrbotten, Västerbotten, Jämtland, Västernorrland, and Gävleborg counties; central Sweden: Dalarna, Västmanland, Örebro, Värmland, Uppsala, and Södermanland counties; southern Sweden: Östergötland, Jönköping, Kronoberg, Halland, Västra Götaland, Kalmar, Gotland, Blekinge, and Skåne counties.

TABLE 2.

Characteristics of parents and offspring in a study of risk of psychopathology among the children of parents with alcohol use disorder

Measure	All Offspring (N=l,244,516)		Female Offspring (N=598,701)		Male Offspring (N=645,815)		Biological Mother (1,244,516)		Biological Father (1,244,516)
	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Year of birth	1979.9	6.1	1979.9	6.1	1979.8	6.1	1951.7	7.1	1949.0	7.5
Age at follow-up (years)	37.7	6.8	37.5	6.9	37.8	6.8	66.2	7.2	68.3	7.3
	N	%	N	%	N	%	N	%	N	%
Prevalence
Alcohol use disorder	33,863	2.72	9,520	1.59	24,343	3.77	26,404	2.12	76,536	6.15
Drug use disorder	43,629	3.51	12,089	2.02	31,540	4.88	13,622	1.09	12,621	1.01
ADHD	26,039	2.09	11,544	1.93	14,495	2.24	2,789	0.22	2,772	0.22
Major depression	189,529	15.23	119,755	20.00	69,774	10.80	200,097	16.08	118,250	9.50
Anxiety disorder	185,436	14.90	115,581	19.31	69,855	10.82	144,003	11.57	77,696	6.24
	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Age at first diagnosis (years)
Alcohol use disorder	28.3	8.2	28.7	8.6	28.1	8.1	55.2	10.6	52.3	14.6
Drug use disorder	26.2	6.8	27.3	7.7	25.7	6.4	54.3	10.9	55.3	12.5
ADHD	30.0	7.9	30.7	7.4	29.5	8.2	55.9	7.5	57.4	7.9
Major depression	31.0	7.0	30.7	7.0	31.4	7.0	56.0	10.4	59.3	10.9
Anxiety disorder	31.2	6.9	31.0	7.0	31.4	6.8	57.9	10.0	60.3	10.4

Figure 1A summarizes the hazard ratios for the five disorders in the offspring of parents with AUD. For any parent (mother or father) and male and female offspring combined, the highest hazard ratio was obtained for AUD (hazard ratio=2.36, 95% CI=2.28–2.44), followed by drug use disorder (hazard ratio=2.04, 95% CI=1.97–2.11) and ADHD (hazard ratio=1.82, 95% CI=1.74–1.90). The hazard ratios were appreciably lower and similar for major depression (hazard ratio=1.43, 95% CI=1.41–1.46) and anxiety disorder (hazard ratio=1.43, 95% CI=1.40–1.46).

FIGURE 1.

Effects of parental alcohol use disorder (AUD) on children’s risk of AUD, drug use disorders, attention deficit hyperactivity disorder (ADHD), major depression, and anxiety disorders^a

^a Panel A shows risk of these disorders as hazard ratios and 95% confidence intervals. From left to right, results are shown for five analyses: risk of offspring with a parent with AUD, risk of offspring with a mother with AUD, risk of offspring with a father with AUD, risk of a daughter with a parent with AUD, and risk of a son with a parent with AUD. Panel B shows the risk of these disorders (as hazard ratios and 95% confidence intervals) based on having one parent or both parents with AUD.

For mothers and fathers with AUD, the overall patterns of results were quite similar to those for any parent with AUD, as was seen, for example, for risk for AUD in the offspring of mothers with AUD (hazard ratio=2.36, 95% CI=2.23–2.51) and of fathers with AUD (hazard ratio=2.35, 95% CI=2.28–2.44) (for details, see Table S5 in the online supplement). On the basis of Bonferroni-corrected p values, only risk for major depression and anxiety disorder differed significantly, with risk levels of both being slightly higher in offspring of mothers with AUD.

For the daughters and the sons of parents with AUD, the overall results were quite comparable, with the hazard ratios for AUD being nonsignificantly lower for the daughters of parents with AUD (hazard ratio=2.29, 95% CI=2.16–2.42) than for the sons (hazard ratio=2.39, 95% CI=2.30–2.48). The only significant differences were higher hazard ratios for major depression and anxiety disorder for the sons than for the daughters (for details, see Table S6 in the online supplement).

Analyses of interactions in risk prediction between the sex of the parent and that of the offspring were conducted. For AUD, none of these effects were close to significant, with similar hazard ratios for mother-daughter, mother-son, father-daughter, and father-son combinations (hazard ratios were 2.32 [95% CI=2.11–2.56], 2.38 [95% CI=2.23–2.54], 2.25 [95% CI=2.11–2.39], and 2.40 [95% CI=2.31–2.49], respectively). Five interactions were significant, all involving the prediction of risk for anxiety disorder or major depression (for details, see Table S7 in the online supplement).

Figure 1B shows the hazard ratios for disorders in offspring having both parents with AUD versus only one parent with AUD. The impact of both parents versus one parent with AUD was greatest on AUD in offspring, where the hazard ratios were 4.64 (95% CI=4.26–5.05) and 2.36 (95% CI=2.28–2.44), respectively (the ratio of hazard ratios was 1.97 [95% CI=1.80–2.15]). The increase in risk for offspring of two parents versus one parent with AUD was next greatest for drug use disorder (ratio of hazard ratios was 1.70 [95% CI=1.54–1.87]) and for ADHD (ratio of hazard ratios was 1.38 [95% CI=1.20–1.58]), and the increases in risk were smaller for major depression and anxiety disorder (ratios of hazard ratios were 1.29 [95% CI=1.22–1.37] and 1.26 [95% CI=1.18–1.34], respectively) (for details, see Table S8 in the online supplement).

Next, we determined whether, in offspring of parents with AUD, two clinical features of parental AUD—age at onset and number of recurrences—predicted offspring risk for the five disorders. We first examined early age at onset. As shown in Figure 2A, the impact of early age at onset of parental AUD on risk of the disorders could be divided into three groups, albeit with much smaller effect sizes (compare the y-axes in Figures 1 and 2). The effect was relatively strong for drug use disorder and AUD, intermediate for ADHD, and modest for major depression and anxiety disorder (for details, see Table S9 in the online supplement). We next investigated number of recurrences of parental AUD, based on the number of registrations (Figure 2B). The pattern of results was quite different from that obtained with age at onset, with modest and very similar increases in risk for all five disorders.

FIGURE 2.

Effects of parental alcohol use disorder (AUD) characteristics on children’s risk of AUD, drug use disorders, attention deficit hyperactivity disorder (ADHD), major depression, and anxiety disorders^a

^a Panel A shows the impact, per 5-year units, of parental age at AUD onset on risk of these disorders as hazard ratios and 95% confidence intervals. Panel B shows the impact of parental recurrence of AUD (number of registrations for AUD) on risk of these disorders. Panel C shows the effects of having a not-lived-with father with AUD on risk of these disorders.

The final question was whether the same pattern of results would occur in parent-offspring pairs where the parent played no substantial role in child rearing. We examined the impact of AUD on risk for the five disorders in the 51,003 children who did not live with their fathers. The basic patterns of risk were similar to those seen in our much larger sample of offspring in intact families, although the hazard ratios were modestly lower (Figure 2C).

DISCUSSION

We used large samples of the total population in Sweden to clarify the impact of parental AUD on offspring psychopathology, producing seven noteworthy results. First, consistent with observations dating back to Plutarch, we found substantial cross-generational transmission of AUD risk. The obtained hazard ratio, approximately 2.4, is reassuringly similar to the relative risk for alcohol dependence of 2.0 and 2.3 found in the two most recent family studies of alcohol dependence (4, 5).

Second, congruent with previous twin and family studies (4, 5, 14–16), we found substantial excess risk in offspring of parents with AUD for two other externalizing disorders: drug use disorder and ADHD. However, in accord with several twin and molecular analyses suggesting disorder-specific familial-genetic effects for AUD (28, 29), the hazard ratio for AUD in offspring of parents with AUD was greater than that seen for drug use disorder, which in turn was greater than that for ADHD. This latter finding is consistent with evidence that ADHD is both an externalizing and neurodevelopmental disorder (17).

Third, consistent with previous evidence of moderate familial correlations between externalizing and internalizing disorders (15–17), the hazard ratios of major depression and anxiety disorder for offspring of parents with AUD were clearly greater than unity but lower than those seen for the externalizing disorders.

Fourth, despite the higher male prevalence of AUD compared with the female prevalence rate in both generations, we found no significant differences in AUD risk in offspring based on having either a mother or a father with AUD. These results are inconsistent with a “female protective effect,” which predicts a mean higher familial risk for AUD in women than in men (23). Earlier family studies of AUD reported results suggestive of such an effect, but this has not been consistently seen in more recent investigations (4, 24). Our large population-representative sample was well powered to detect this effect, so our negative findings argue strongly against the likely importance of a female protective effect. Our results are also inconsistent with the notion that intrauterine exposure to excessive blood alcohol levels plays a significant role in the transmission of AUD risk from mothers to children. In our analyses of within- and across-sex parent-offspring transmission, we also saw no significant interactions, suggesting that the familial transmission of AUD is independent of sex, thereby ruling out a substantial role for X-linked variants (which should depress father-son transmission).

Fifth, the pattern of increased risk for offspring having two parents with AUD was qualitatively similar to but quantitatively stronger than that seen for offspring having one parent with AUD. That is, the increase in risk for offspring having two parents with AUD compared with one parent was greatest for AUD, intermediate for drug use disorder and ADHD, and more modest for major depression and anxiety disorder. These results represent an independent replication of our key finding that parents with AUD transmit to their children a pattern of risks that is strongest for AUD and relatively weaker for the other disorders we examined.

Sixth, we expanded our exploration of parent-offspring transmission of AUD from case-control differences to clinical features of parental AUD. Building on earlier analyses (18–21), we examined onset age and recurrences and observed striking differences between the results. Age at AUD onset in parents predicted a pattern of offspring risk similar to that in our main analyses: hazard ratios were highest for the substance use disorders, intermediate for ADHD, and considerably lower for the internalizing disorders. Early age at onset in AUD appears to index general risk for substance use and related externalizing disorders. In contrast, high levels of recurrence in affected parents produced equal and modest increased risks for all five disorders, suggesting that recurrence reflects nonspecific vulnerability to psychopathology.

Finally, because we studied children of intact families, parent-offspring transmission could arise from both genetic and rearing effects. To investigate the relative importance of these forms of transmission, we repeated our analyses for children who did not live with their fathers and with whom the children likely had very limited postnatal contact. Although quantitatively weaker than the findings from our main analyses, the pattern of results was nearly identical to that in our main analyses of intact families, suggesting that the pattern of risk arises predominantly from genetic influences.

Our results suggest that there are three layers to the liability to AUD transmitted from parents to children. The lowest and least specific layer reflects a broad vulnerability to all common forms of psychopathology. This was seen in the increased risk for all of the evaluated disorders, both in our primary analyses (Figure 1A) and in our comparison of risk for offspring having two parents with AUD and one parent with AUD (Figure 1B). Its purest expression emerged in the analysis of the number of registrations of parents with AUD, which indexed an equal but modest risk of AUD, drug use disorders, ADHD, anxiety disorders, and major depression.

The second and moderately more specific layer to the transmitted liability to AUD is risk of the three externalizing syndromes: ADHD, drug use disorders, and AUD itself. We saw the effect of this liability in three analyses: our main results, our analyses of risk patterns in the offspring having two parents with AUD compared with one parent with AUD, and our examination of the impact of parental age at AUD onset on risk in their offspring. All three analyses demonstrated a much stronger transmission of risk to offspring for externalizing disorders than for internalizing disorders.

The third and most specific layer of this familial liability is for AUD itself. We saw this most clearly where the hazard ratio for the transmission of AUD was substantially higher than those seen for drug use disorders or ADHD, typically with nonoverlapping 95% confidence intervals (Figure 1A). We also saw this with the not-lived-with fathers (Figure 2C), suggesting that its origin is likely to be largely genetic. Interestingly, we did not see this effect in analyses of parental age at AUD onset (Figure 2A), where the risks for AUD and drug use disorder in offspring of parents with AUD were very similar, with overlapping confidence intervals. This suggests that early age at AUD onset in parents is actually an index for externalizing or nonspecific substance use disorders rather than specific liability to AUD.

Our results have two potential methodological limitations. The first is the validity of our diagnoses obtained from the Swedish medical and criminal registries. Such diagnoses cannot be expected to replicate those obtained from personal interviews, as they reflect contact between the subject and health care or policing officials. For AUD, in Sweden, we can study only individuals whose alcohol problems affect behavior that is detected in the criminal system (e.g., repeated drunk driving) or alcohol-related medical or psychiatric problems that either cause the individuals to come to medical attention or are detected at medical examination when being seen for other conditions. As suggested by our prevalence figures, when compared with those found for AUD in the United States (30) or nearby Norway (31), we are detecting, on average, more severely affected individuals than those detected by population-based personal interview studies. The validity of our definition of AUD is supported by the high rates of concordance observed across ascertainment methods (10, 32) and the similarity in the pattern of resemblance in relatives for AUD seen in Sweden and in personal interview-based samples (33, 34). The validity of our major depression and anxiety disorder diagnoses are supported by their prevalence, sex ratio, sibling and twin correlations, patterns of comorbidity, and associations with known psychosocial risk factors (35, 36). The validity of ADHD diagnoses in Sweden has been supported by previous studies, including a close association with prescription of stimulants (37).

The second concern is whether our results could arise substantially from parental or child comorbidity (see Tables S10 and S11 in the online supplement). To illustrate the impact of parental comorbid major depression, of the 68,793 parents in our sample with AUD, 17,269 (25.1%) were diagnosed with major depression prior to a diagnosis of AUD. When we censored these cases, the hazard ratio for major depression in the offspring of parents with AUD dropped modestly (from a hazard ratio of 1.43 [95% CI=1.41–1.46] to 1.34 (95% CI=1.31–1.37]). When the other disorders were analyzed in this way, the reductions in hazard ratios were even smaller. We then examined the impact of comorbidity in offspring. Again, taking major depression as an example, of the 189,529 cases of major depression in our offspring sample, 7,887 (4.2%) had an earlier registration for AUD. When these cases were censored in our analysis, the hazard ratio for major depression in offspring having a parent with AUD declined only slightly, from 1.43 (95% CI=1.41–1.46) to 1.39 (95% CI=1.36–1.42). The reductions were uniformly modest for the other disorders. We conclude from these analyses, as illustrated by major depression, that comorbidity pathways likely have, at most, a modest causal role in our observed results.

CONCLUSIONS

Our study explored the nature of the liability to AUD transmitted across generations in a large general-population Swedish cohort. The highest elevation in risk among offspring of parents with AUD was for AUD, followed by increased risks for two other externalizing disorders—drug use disorder and ADHD—and more modest elevations in risk for two internalizing disorders—major depression and anxiety disorder. We demonstrated that a female protective effect played no role in explaining the lower prevalence of AUD in women than men, as the familial risks of the disorder in offspring having mothers with AUD or fathers with AUD were indistinguishable. Offspring having two parents with AUD had a more extreme version of the same pattern of risks compared with offspring having one parent with AUD. We demonstrated the utility of examining, in a family design, clinical features of parental AUD, such that differential patterns of risk of disorders in children were apparent based on parental age at AUD onset and recurrence. Finally, our results suggested a decomposition of the familial risk for AUD into three components: a largely nonspecific risk that included both common internalizing and externalizing disorders, a moderately specific liability to externalizing disorders, and a specific risk of AUD.