Coronary heart disease in mothers and fathers of adult children with alcohol use disorders

Abstract

Background and aim

Having a family member with an alcohol use disorder (AUD) may negatively affect a person’s health. Our aim was to study the long-term risk of coronary heart disease (CHD) in parents who have an offspring with AUD.

Design

Cohort study with Cox regression models and co-sibling analyses.

Setting

Sweden.

Participants

From population registers, we selected all parent-offspring pairs in which the parent was born in Sweden between 1945 and 1965.

Measurements

Baseline was set when the offspring was 15 years old and AUD was assessed from medical and criminal registers. The parents were followed for CHD during a mean follow-up of 18 years. Hazard ratios (HRs) in mothers and fathers were calculated and adjusted for potential confounders (year of birth, age at childbirth, sex of the child, parent’ AUD, educational level, and marital status).

Findings

In mothers, the adjusted HR for CHD was 1.24 (95% CI = 1.19–1.28) in relation to having a child with AUD. In fathers, the HR for CHD was lower than in mothers but still increased; the adjusted HR was 1.08 (95% CI = 1.05–1.12). In the co-sibling analyses, the HRs for mothers were similar to the HRs estimated from the population-based sample, but in fathers the association did not remain significant (HR = 0.98 [0.90–1.06]).

Conclusions

In Sweden, there appears to be an association between having an offspring with alcohol use disorder and increased risk of developing coronary heart disease. For fathers, the association did not remain in co-sibling analyses.

INTRODUCTION

Globally, excessive alcohol use is one of the leading risk factors for premature mortality and disability [1,2]. In addition to the direct effects on the heavy drinker, alcohol use disorders (AUD) may be harmful for the personone’ family members, (e.g. by increased stress, family strain, and failure to fulfill role obligations at work and at home [3,4].) These effects may negatively impact the mental and physical health of family members, but this topic has mostly represented a neglected research field [4].

An overview of the burden of ill-health in family members affected by addiction (e.g. partners, parents, siblings, and extended family members) concluded that having a relative with problematic drinking behavior is associated with impaired health in numerous ways [4]. Cross-sectional population survey studies from Australia, New Zealand, and Chile found negative effects on personal wellbeing and quality of life in individuals with heavy drinkers in their lives within a range of family and other relationships, (i.e. partners, parents, children, and siblings but also among friends and work colleagues [5–7].) In addition, an American study compared healthcare costs and found significantly higher costs in spouses and children of patients treated for alcohol problems and higher prevalence of medical conditions such as congestive heart failure, ischemic heart disease, and diabetes [8]. Another study suggested that there is a causal relationship between presence of heavy drinkers in the respondents’ social circle of family, friends, and co-workers and lower levels of self-reported mental wellbeing as well as anxiety [9]. A recent Swedish study, which was based on self-report questionnaires to parents with children with drug problems, reported negative consequences from offsprings’ drug problems on parental social life, economy, and mental health [10]. However, few studies have examined the somatic health in parents with an offspring with AUD; previous literature has mainly focused on spouses and children to persons with AUD [4] and/or examined mental health outcomes in parents to individuals with AUD [4,10,11]. Parents usually have a strong bond to their children, and an important role concerning protection and guidance, which may result in negative health consequences if the child is affected by AUD.

Coronary heart disease (CHD) is the leading cause of death in the Western world [12]. Psychosocial distress may contribute to a higher risk of CHD, in addition to traditional cardiovascular risk factors like hypertension and diabetes type 2 [13,14]. Low socioeconomic status constitutes another well-known risk factor for CHD [15], and low familial socioeconomic status in childhood, (e.g. low parental educational attainment, is also related to increased risks of AUD in adulthood [16,17].) Whether the potentially stressful effects of having a teenager or adult child with AUD is associated with increased long-term risks of CHD is, however, not known. In such a study, it is important to keep in mind that a potential association may be confounded by several factors, such as socioeconomic status and shared genetic and environmental factors [18].

The first aim of this cohort study was to assess the risk of CHD in parents to individuals with AUD, adjusting for potential confounders, and by using data from Swedish medical, criminal, and population registers during a mean follow-up of 18 years. The second aim was to investigate whether this potential association remained in a co-sibling analysis, by comparing siblings with and without an offspring with AUD, to control for shared genetic and environmental factors [18].

METHODS

In Sweden, each person has a unique personal identification number that can be linked to several population-based registers with national coverage after replacing the personal identification number with a pseudonymized serial number to preserve confidentiality. In this study, data was retrieved from the following national registers: Multi-Generation Register (parents-offspring data), Total Population Register (family data), National Patient Register (in- and outpatient data), The Swedish Suspicion Register, The Swedish Crime Register (crime data), and The Swedish Prescribed Drug Register (data on drugs against alcohol addiction). Detailed information about the registers is described in Supporting information Data S1. The study complies with the Declaration of Helsinki and we secured ethical approval for this study from the Regional Ethical Review Board in Lund (No. 2008/409). The analysis was not pre-registered, and the results should be considered exploratory. From the Swedish population register, we selected all possible biological parent-offspring pairs (n = 3 984 167) where the parent was born in Sweden between 1945 and 1965. Each parent could be included several times in the analyses, with more than one offspring. The offspring in these pairs were born between 1959 and 2001 (mean year of birth 1983). The parents in this study constituted the study participants at risk for CHD, and AUD in their offspring (yes/no) was regarded as a predictor variable. Baseline was set when the offspring was 15 years old. From baseline until the end of follow-up (December 2017), AUD was assessed in the offspring and the parents were followed for CHD. The follow-up started somewhere between 1974 and 2016. An overview/timeline of the study period is shown in Fig. 1.

Figure 1.

Timeline of the study

Covariates

We included several potential confounders at baseline. The rationale for measuring these variables at baseline was because we intended to assess them before the occurrence of the main predictor, AUD in the child. The choice of potential confounders was based on previous studies about risk for AUD and CHD [12,13,15–17], as previous research on CHD risk in parents with AUD offspring is limited. In the parent generation we identified sex, year of birth, age at childbirth, date of first registration of AUD, number of years of education, marital status, and cohabitation with their offspring. AUD in the parents was not assessed after baseline. In the offspring generation, we identified sex, year of birth, first registration of AUD after baseline, age at AUD, and age at the end of follow-up.

Number of years of education was measured as highest achieved education and was standardized (with mean 0 and SD 1) separately for each combination of year of birth and sex, where approximately 12 years was the mean value (0). Marital status was categorized as married versus not married. The studied parent was coded as married regardless of whether the spouse was the biological parent of the child or not. Marital status was used as a proxy for a stable family situation. Cohabitation with their offspring was measured at baseline and categorized into a dichotomous variable (parent-offspring residing in the same household vs not residing in the same household at baseline). The reason for measuring cohabitation status at baseline was based on the hypothesis that cohabiting with the child during his/her childhood creates a stronger bond between parent and child, compared with not residing in the same household. Household was defined as follows: from 1960 to 1985 (every 5th year) we used HouseholdID from the Population and Housing Census. The HouseholdID includes data on all individuals living in the same dwelling. For the years that we did not have this information, we approximated the HouseholdID with the information from the year closest in time. From 1986 and onward (every year) we used the FamilyID from the Total Population Register. The FamilyID is defined by individuals that are related and who are registered at the same property (a person can only be registered at one property in the population database).

CHD and AUD

In the parent generation, date of first registration for CHD was assessed by identification of the following International Classification of Disease (ICD) codes in the Swedish medical registers; ICD8/9: 410–414 and ICD10: I20–I25. Death records were not used to identify CHD. Those who had a CHD before baseline were excluded.

AUD for both offspring (first registration in 1974) and parents (first registration in 1969) was identified in the Swedish medical and mortality registries by the following ICD codes: ICD8: 291, 303, 980, 571.0; ICD9: V79B, 305A, 357F, 571A-D, 425F, 535D, 291, 303, 980; ICD10: E244, G312, G621, G721, I426, K292, K70, K852, K860, O354, T51, F10. AUD was also identified in the Crime Register by codes 3005, 3201, which reflect convicted crimes related to alcohol abuse, that is, drunk driving or being drunk in charge of a maritime vessel and in the Suspicion Register by codes 0004, 0005 (only those individuals with at least two alcohol-related crimes or suspicion of crimes were included). Finally, the Prescribed Drug Register was used to identify the drugs disulfiram (Anatomical Therapeutic Chemical (ATC) Classification System N07BB01), acamprosate (N07BB03), and naltrexone (N07BB04).

Statistics

To investigate the effect of AUD in a child and future risk for CHD in a mother or father, we used Cox proportional hazards models where the follow-up time in number of months was measured from age 15 in the child until first registration for CHD in a parent, death or emigration in a parent, or end of follow-up (December 2017), whichever came first. In model A, we included AUD in the child as a time-dependent covariate suggesting that the exposure starts at the date of AUD registration in the child. Model B also included year of birth of the parent, age at childbirth, and sex of the child. Model C additionally included highest achieved educational level and marital status in the parent, and in model D, AUD in the parent was included as a time-dependent variable. We tested if the effect of AUD in a child varied across follow-up time, but we found no significant differences, (i.e. the proportional hazards assumption was not violated.) In the final model (model E), we investigated the interaction between AUD in a child and cohabitation parent–child. This was done by including the covariate cohabitation parent–child as well as an interaction term between cohabitation parent–child and AUD in the child. A significant interaction term suggests that the effect of AUD in the child varies as a function of cohabitation status. In the results section we, therefore, present two hazard ratios (HRs), one for the effect of AUD in a child for parents who cohabited with their offspring and one for parents who did not. In all models, we adjusted for the possibility that a parent could be included several times in the analysis, using a robust sandwich variance estimator. In a sensitivity analysis, we replicated our final model using only the first-born offspring for each parent. All models were estimated separately for fathers and mothers and we tested for equality of the regression coefficient “AUD in child” between mothers and fathers [19].

In an additional approach to control for confounding, we compared full siblings in the parental generation, where one had an offspring with AUD and the other did not. For this co-sibling analysis, we used stratified Cox proportional hazards models, with a separate stratum for each sibling pair. This means that factors, both known and unknown, shared by the siblings, will not confound the estimates [20]. In an additional model, we adjusted the co-sibling analyses for parental year of birth, age at childbirth, sex of the child, educational level, marital status, and AUD. All statistical analyses were performed using SAS 9.4.

The study is reported using the RECORD checklist (in Supporting information Data S1).

RESULTS

At baseline, the mean age in fathers and mothers was 45.2 years and 42.6 years, respectively (Table 1). During the follow-up (mean 18 years), a total of 158 750 (8.3%) fathers and 74 427 (3.6%) mothers were diagnosed with CHD (Table 1), which occurred at an average age of 57.9 years (fathers) and 57.6 years (mothers). The rate of having a child with AUD was 3.7% (n = 70108) in fathers and 4.0% (n = 83 109) in mothers. Out of the AUD registrations in children, 2.4% were medical/somatic complications to AUD and the rest were related to alcohol abuse based on ICD codes (ICD10: T51, F10) and AUD from the crime or prescribed drug registers (data not shown in tables). The corresponding proportions in fathers and mothers were 1.7% and 2.5%, respectively. When the child was 15 years old, 28% of the fathers and 9% of the mothers were not residing in the same household as their child.

Table 1.

Basic characteristics of parents born in Sweden between 1945 and 1965 and their offspring

	Fathers	Mothers
Parents characteristics
N (parent-children pairs)	1 919 549	2 064 618
Age at baseline, mean (SD)	45.2 (5.6)	42.6 (5.4)
CHD, n (%)	158 750 (8.3%)	74 427 (3.6%)
AUD in child, n (%)	70 108 (3.7%)	83 109 (4.0%)
Year of birth, mean (SD)	1954 (6.1)	1955 (6.2)
Age at childbirth, mean (SD)	30.2 (5.6)	27.6 (5.4)
Education (years), mean (SD)a	12.0 (2.9)	12.4 (2.8)
Married at baseline, n (%)	1 259 313 (65.6%)	1 354 623 (65.6%)
AUD in parent, n (%)	195 870 (10.2%)	81 263 (3.9%)
Same household at baseline, n (%)	1 375 889 (71.7%)	1 877 660 (90.9%)
Offspring characteristics	Offspring to fathers	Offspring to mothers
Sex of the child (males), n (%)	986 890 (51.4%)	1 060 703 (51.4%)
Year of birth, mean (SD)	1984 (8.4)	1982 (8.7)
Age at AUD, mean (SD)	24.3 (7.5)	25.6 (8.4)
Age at end of follow-up, mean (SD)	31.1 (8.3)	33.9 (8.6)

CHD = coronary heart disease; AUD = alcohol use disorders.

^aNumber of years of education were measured as highest achieved education, presented as unstandardized mean values. In the regression models the variable was standardized (with mean 0 and SD 1) per year of birth and sex.

In mothers, the crude HR for CHD was 1.31 (95% CI = 1.26–1.36) in relation to having a child with AUD (Table 2). The increased HR declined only modestly after adjustments for parental year of birth, age at childbirth, sex of the child, educational level, marital status, and AUD in parent; the HR was 1.24 (1.19–1.28). In fathers, the corresponding HRs for CHD were lower than in mothers but still increased in relation to having a child with AUD; the unadjusted HR was 1.12 (95% CI = 1.08–1.15) and the adjusted HR was 1.08 (95% CI = 1.05–1.12) (Table2b, model D). In the sensitivity analysis, using only the first-born offspring for each parent, the HRs only changed marginally (HR in mothers was 1.21 [95% CI = 1.15–1.28] and in fathers 1.08 [95% CI = 1.03–1.13]).

Table 2.

Adjusted HR with 95% CI in relation to AUD in child (models A–E)

	Model A	Model B	Model C	Model D	Model E
Mothers
AUD in child (yes vs no)	1.31 (1.26; 1.36)	1.36 (1.31; 1.42)	1.28 (1.23; 1.33)	1.24 (1.19; 1.28)	1.06 (0.95; 1.18)
Year of birtha		0.99 (0.99; 0.99)	0.99 (0.99; 0.99)	0.99 (0.99; 0.99)	0.99 (0.98; 0.99)
Age at childbirtha		1.05 (1.05; 1.05)	1.06 (1.06; 1.06)	1.06 (1.06; 1.06)	1.06 (1.06; 1.06)
Sex of the child (male vs female)		1.01 (0.99; 1.02)	1.01 (0.99; 1.02)	1.01 (1.00; 1.02)	1.01 (0.99; 1.02)
Educationb			0.80 (0.79; 0.81)	0.81 (0.80; 0.81)	0.81 (0.80; 0.82)
Married at baseline (vs unmarried)			0.84 (0.82; 0.86)	0.86 (0.84; 0.87)	0.87 (0.85; 0.89)
AUD in parent				1.64 (1.59; 1.70)	1.60 (1.51; 1.70)
Same household at baseline (yes vs no)					0.84 (0.81; 0.86)
Same household at baseline aAUD in child c					1.20 (1.07; 1.34)
Fathers
AUD in child (yes vs no)	1.12 (1.08; 1.15)	1.17 (1.13; 1.21)	1.12 (1.08; 1.15)	1.08 (1.05; 1.12)	1.00 (0.96; 1.05)
Year of birtha		0.98 (0.98; 0.99)	0.98 (0.98; 0.98)	0.98 (0.98; 0.98)	0.98 (0.98; 0.98)
Age at childbirtha		1.06 (1.06; 1.06)	1.06 (1.06; 1.07)	1.06 (1.06; 1.07)	1.06 (1.06; 1.07)
Sex of the child (male vs female)		0.99 (0.98; 1.00)	1.00 (0.99; 1.01)	1.00 (0.99; 1.01)	1.00 (0.99; 1.01)
Educationb			0.87 (0.86; 0.88)	0.88 (0.87; 0.88)	0.88 (0.87; 0.88)
Married at baseline (vs unmarried)			0.88 (0.86; 0.89)	0.90 (0.89; 0.92)	0.96 (0.94; 0.98)
AUD in parent				1.31 (1.27; 1.35)	1.27 (1.24; 1.31)
Same household at baseline (yes vs no)					0.86 (0.85; 0.88)
Same household at baseline aAUD in childa					1.12 (1.06; 1.19)

AUD = alcohol use disorders; HR = hazard ratios.

^aHR per increasing year, continuous variable.

^bHR per increasing step of education standardized per year of birth and sex, continuous variable.

^cInteraction between same household at baseline and AUD in child.

Because the HRs seemed higher in mothers, we tested for equality of the regression coefficient “AUD in Child” between mothers and fathers. Equality between mothers and fathers could be rejected in model A, B, C, and D (P < 0.0001). However, in model E, only the effect for living in the same household at baseline could be rejected (P < 0.0001 for living in the same household vs P = 0.26 for not living in the same household). These results indicate that the effect of AUD in a child on parental CHD is different for mothers and fathers.

The E-value is defined as the minimum strength of association that an unmeasured confounder would need to have with both the exposure and the outcome, conditional on the measured covariates, to fully explain away a specific exposure-outcome association [21,22]. We calculated the E-value based on model D, and for mothers it was 1.79 (1.67 for the lower limit of the 95% CI) and for fathers it was 1.37 (1.28 for the lower limit of the 95% CI).

Cohabitation parent-offspring

The interaction between AUD in child and cohabitation was positive (Table 2a and 2b), which means that the effect on CHD of having a child with AUD was stronger if the parents cohabited with the child when he or she was 15 years old than if they did not. The HR for CHD in relation to AUD in a child was 1.26 (95% CI = 1.21–1.32) in mothers in the same household as the child, and 1.06 (95% CI = 0.95–1.18) when the household was not shared. The corresponding HRs for fathers were 1.12 (95% CI = 1.08–1.17) and 1.00 (95% CI = 0.96–1.05), respectively.

Co-sibling analyses

The association was also explored in co-sibling analyses, that is, in full same-sex siblings where one had an offspring with AUD and the other did not. The HR for CHD in relation to AUD in a child was 1.25 (95% CI = 1.13–1.38) in mothers, adjusted for parental year of birth, and 1.10 (95% CI = 1.01–1.19) in fathers (Table 3). After full adjustment, the increased HR remained in mothers 1.27 (95% CI = 1.14–1.41) but disappeared in fathers 0.98 (95% CI = 0.90–1.06).

Table 3.

Co-sibling analysis

Co-sibling analysis	Mothers	Fathers
Sample size, n	21 276	17 904
Model A	1.25 (1.13; 1.38)	1.10 (1.01; 1.19)
Model D	1.27 (1.14; 1.41)	0.98 (0.90; 1.06)

AUD = alcohol use disorders. Sample size = female respective male pairs that are discordant for the exposure variable AUD in child. Model A: unadjusted. Model D: adjusted for parental year of birth, age at childbirth, sex of the child, educational level, marital status, AUD in parent. Hazard ratios with 95% CI in parental full siblings where one had an offspring with AUD and the other did not.

DISCUSSION

According to this nationwide study, parents of children who were registered for AUD had an increased risk of developing CHD during a mean follow-up time of 18 years. The results indicate a potentially negative effect on cardiovascular health of the distress of having a child with AUD. The increased risk remained in mothers in the co-sibling analyses but not in fathers, indicating that the relationship in fathers was a result of shared genetic and environmental factors.

To our knowledge, an association between having a child with AUD and the long-term risk of CHD has previously neither been studied in the scientific literature, nor been considered in clinical practice. A potential underlying mechanism behind our findings is psychological distress that is associated with an increased risk of CHD [13,14,23]. Psychological distress is likely to be increased among relatives to individuals with both substance use disorders and/or mental illness. For example, a study of care-givers of young adults with first episode psychosis reported high levels of psychological distress and increased rates of being overweight, and having diabetes and hypertension [24]. Another study showed that cortisol patterns were disturbed by chronic stress in parents of adult children with serious mental illness [25]. Several studies have reported that parents and family members of individuals with AUD or drug problems suffer from anxiety, stress, feelings of guilt and shame, and subsequent deteriorated mental and physical health [4,10,11]. Addiction in the family may lead to conflicts and problematic relationships, which may disturb the parents’ social network and opportunities for relaxation, a healthy lifestyle, and taking care of one’ own health. Reduction of stress and a healthy lifestyle, such as physical activity and a healthy diet, are important factors in the prevention of CHD [13,14], and may be limited by the time, energy, and money spent on a problematic relationship with a heavy drinking child. Earlier studies have shown that family members of substance misusing individuals have low socioeconomic status and psychological and physical symptoms [26].

Other strenuous life circumstances have also been related to cardiovascular disease. For example, having a spouse with cancer has been found to be associated with increased risk of acute myocardial infarction [27]. According to a meta-analysis of caregivers of family members with dementia, caregivers had higher levels of stress hormones and poorer global self-reported health compared with non-caregivers [28]. Potential explanatory factors may be poorer health habits and physiological responses to psychological distress. The association between psychological distress and CHD may be mediated by biological as well as behavioral mechanisms. Earlier studies have suggested that psychological distress influences atherosclerotic mechanisms and cardiovascular disease through inflammatory markers, regulation of the hypothalamic–pituitary–adrenal axis and impaired vascular function [29–31]. A Swedish study found that females with several stress symptoms (e.g. tiredness, sleeping problems, and nervousness) have an increased risk of myocardial infarction, irrespective of traditional cardiovascular risk factors [32]. To a certain degree, cardiovascular diseases aggregate in families [33], as well as other cardiovascular risk factors in addition to alcohol consumption (e.g. hypertension, obesity, smoking habits and dietary patterns [34].) In the present study, the association between CHD and having an adult child with AUD remained in discordant female siblings where one of the siblings had an AUD child and the other had not, which supports that the association is not solely explained by shared genetic and environmental factors.

After adjustments for potential confounding factors, the increased HR of CHD in parents with AUD children was attenuated in fathers, and in the co-sibling analyses the association remained only in the mothers but not in the fathers. This may be explained by the higher absolute risk of CHD in males compared with females [35], therefore resulting in a relatively lower relative risk in fathers. Another contributing explanation may be that mothers may be more affected by their children’ circumstances and health than fathers. Psychological distress has been reported as a cardiovascular risk factor in both men and women, but women may be more likely to suffer from different types of stress [14]. It has been shown that mothers’ health related quality of life is more affected than fathers’, by caregiving of adult children suffering from mental disorder [36]. However, another study found no sex differences in parents to adult children with developmental problems [37]. The positive interaction between cohabitation with the child and CHD indicates that the stronger effect of AUD in a child on mothers’ CHD may be because children are more likely to live with their mothers than with their fathers.

Strengths and limitations

This study has several strengths. First, to our knowledge, this is the first long-term cohort study that has examined the association between AUD in children and CHD risk in parents. The study was based on a large nationwide study population with almost complete coverage in the registers. Second, AUD in the child was measured as a time-varying variable before the onset of CHD in the parent, to determine the start of the exposure at the date of AUD registration in the child. Third, adjustments were made for several potential confounding factors and, finally, a co-sibling analysis was conducted in an attempt to adjust for shared familial factors, both genetic and environmental.

The study also has some limitations. The study was done in a Swedish setting and the generalizability to other geographic and ethnic populations is not known. Like many long-term cohort studies, we were not able to assess all changes during follow-up. However, we adjusted for parental AUD as a time-dependent variable. Another potential limitation is residual confounding because we, for example, did not have data on lifestyle, such as smoking and physical inactivity. Although we adjusted for the parents’ educational level, residual confounding from undetected familial socioeconomic factors is possible. Previous studies have concluded that childhood circumstances, such as low familial socioeconomic status and educational attainment, are related to increased risk of AUD in adulthood [16,17]. Traditional CHD risk factors like hypertension and diabetes mellitus were not included, because these could rather be considered as potential mediators than “true” confounders. Further studies are, therefore, highly warranted.

In conclusion, this unique cohort study of almost 4 million Swedish individuals shows that mothers with an offspring who is registered for AUD are at increased risk of developing CHD. The association was also found in fathers but did not remain in the co-sibling analyses. Having an AUD offspring seems to be a CHD risk factor that has previously not been considered in clinical practice and preventive efforts may be needed in women with a child with AUD. Further studies are needed about potential mechanisms and factors explaining the association between child AUD and CHD in parents.

Supplementary Material

Supplement 1

Data S1 Supporting Information.