Divorce has been the subject of many a Hollywood movie, epic novel and heart-wrenching country ballad. Perhaps, this is not so surprising considering that divorce is rated only after “Death of a spouse” as the second most stressful life event (Holmes and Rahe, 1967). Divorce often represents a psychologically distressing period, requiring significant adjustment.

Whist in the UK, divorce rates have steadily declined over the past 10 years across all age groups, Sweden has seen a record increase in divorce in recent years. In 2013, there were over 25,100 divorces in Sweden, the highest figure for divorce since 1975.

Epidemiological research has consistently demonstrated an association between divorce, alcohol consumption and risk for alcohol use disorder (AUD) (Grant et al, 2015; Kessler et al, 1998). Compared with married individuals, divorcees:

• consume larger amounts of alcohol (Power et al, 1999),
• have a greater incidence of drinking to harmful levels (Halme et al,2008)
• and are more likely to be diagnosed with a AUD during their lifetime (Grant et al, 2015).

However, the mechanisms underlying the association between alcohol-related problems and divorce are poorly understood. Possible explanations include:

• Shared confounding factors (e.g. social class, genetic vulnerability etc.) that contribute to both divorce and problematic alcohol use
• A causal pathway between divorce and alcohol-related problems.

The present study aimed to explore the nature of the association between divorce and AUD, using a longitudinal cohort design by examining;

1. Divorce and subsequent risk for AUD
2. The influence of confounding factors and spouse AUD status
3. Risk for AUD in pairs of relatives concordant for marriage and discordant for divorce
4. The temporal (timeframe) association in married individuals, between divorce and AUD diagnosis
5. Association between divorce and risk of AUD relapse in individuals registered with AUD prior to marriage
6. Association between re-marriage after divorce and subsequence risk for AUD
7. Association between widowhood and risk for AUD

Methods

Participants

Participants were Swedish nationals born between 1960-1990, married and living with their spouse in or after 1990, with no AUD diagnosis prior to marriage.

The risk for AUD in pairs of relatives was assessed by recruiting sibling and cousin pairs (born within 3 years of each other) and monozygotic twins.

Measures

AUD was identified from Swedish medical registries and the prescribed drug register for pharmacotherapies for AUDs. AUD was also identified using convictions for at least 2 alcohol-related crimes.

Cofounding measures included socioemotional status (parents’ highest education), early externalising behaviour (criminal behaviour or drug abuse before age 19) and familiar risk (one or more parents, siblings or cousins with AUD).

Divorce and widowhood were determined using data from the population register. Re-marriage was defined as first marriage registered after first registration of divorce or widowhood.

Data analysis

A Cox proportional hazards model was used to estimate the risk of AUD as a function of divorce or widowhood. Analyses were adjusted for year or birth, externalising behaviour, parental education and familial risk for AUD. AUD in spouse was also included to examine the association with divorce.

A co-relative design was used to examine the effect of divorce when adjusting for unmeasured familial confounding. Only pairs concordant for marriage and discordant for both divorce and AUD contributed to estimates.

Estimates from an Aalen’s Additive Regression Model were used to test whether a family history of AUD or a history of externalising behaviour prior to age 18 modified the impact of divorce on AUD risk. These analyses were adjusted for birth year and parental education.

Results

The main cohort included 942,366 individuals, with the average age at marriage 30 years and AUD onset around 8-9 years later.

At follow-up, 16% males and 17% females were divorced and 1.1% of males and 0.5% were registered with a AUD.

Divorce and subsequent risk for AUD

• Divorce was strongly associated with subsequent onset of AUD for both
  • Males (hazard ratio=5.98, 95% CI=5.65 to 6.33) and
  • Females (hazard ratio=7.29, 95% CI=6.72 to 7.91).

Influence of confounding factors and spouse AUD status

• Low parental education, prior deviant behaviour and family history all substantially predicted AUD risk
• These three confounders also modestly decreased the association of AUD with divorce for:
  • Males (hazard ratio=5.09, 95% CI=4.81 to 5.39) and
  • Females (hazard ratio=6.31, 95% CI=5.82 to 6.86)
• Divorce had a stronger association with risk for future AUD for males and females if the spouse did not have a history of AUD.

Risk for AUD in pairs of relatives concordant for marriage and discordant for divorce

• The co-relative analysis revealed a moderate decline in the association between divorce and AUD risk in discordant relative pairs compared with that observed in the general population, with a stronger decline in siblings than cousins for males
• For females, the association was similar in cousins compared with the general population, but considerably lower in full siblings
• The hazard ratio estimated for AUD onset given divorce in married monozygotic twins discordant for divorces was:
  • Males 3.45 (95% CI=1.70 to 7.03) and
  • Females 3.62 (95% CI=1.29 to 10.18).

The temporal (timeframe) association in married individuals, between divorce and AUD diagnosis

• For males married between 18-25 years, rate of AUD registration rose in individuals with a future divorce and peaked in the year of divorce. Additionally, the rate of AUD onset remained elevated for the next 15 years in those individuals did not remarry. A similar pattern was observed for males married between 26-32 years
• For females married between 18-25 years, increased risk for AUD occurred only 3 years prior to divorce, with a peak in the year of divorce and risk remaining elevated for the next 15 years in those individuals did not remarry. A similar pattern was observed for females married between 26-32 years
• Together these figures show an increased rate of first AUD registration following divorce, with a proportion of the total AUD-divorce association arising from AUD predisposing to divorce.

Association between divorce and risk of AUD relapse in individuals registered with AUD prior to marriage

• In participants registered with an AUD prior to marriage, divorce was associated with an increased risk of AUD relapse for
  • Males (hazard ratio=3.20, 95% CI=2.86 to 3.59) and
  • Females (hazard ratio=3.56, 95% CI=2.75 to 4.60).

Association between re-marriage after divorce and subsequence risk for AUD

• A Cox proportional model, with remarriage as a time-dependent covariate showed a substantial decline in risk for first AUD registration for
  • Males (hazard ratio=0.56, 95% CI=0.52 to 0.64) and
  • Females (hazard ratio=0.61, 95% CI=0.55 to 0.69).

Association between widowhood and risk for AUD

• A cox proportional model, with death of a spouse as a time-dependent covariate showed an association of widowhood with risk for onset of AUD in
  • Males (hazard ratio=3.85, 95% CI=2.81 to 5.28) and
  • Females (hazard ratio=4.10, 95% CI=2.98 to 5.64)
• Adding confounders of low parental education, prior deviant behaviour and family history produced a small reduction in these associations.

Conclusions

The present study provides evidence for:

• A strong association between divorce and subsequent onset of AUD, with rates of first onset of AUD increasing after divorce being 6-fold for males and over 7-fold for females
• Low parental education, prior deviant behaviour and family history for AUD were predictive risk factors for AUD. Additionally, these factors reduced the association between divorce and onset of AUD by 15%
• The association between divorce and AUD was part causal
• Increased risk of AUD onset beginning a few years prior to divorce. The risk for AUD also increased substantially in the year of divorce and remained high in those who did not remarry
• Individuals registered with AUD prior to marriage were at increased risk for AUD relapse following divorce
• Protective effects of re-marriage on risk for AUD, indicative of a causal association between marital status and AUD risk
• A strong association between widowhood and increased risk for AUD

Strengths and limitations

The findings from this study should be interpreted within the context of the following methodological limitations:

• Firstly, the study did not directly ask individuals about their AUD diagnosis, instead national records were used to establish AUD registration. It is possible that the rates of AUD reported in this paper would not be consistent with self-report of AUD
• Secondly, these data only reflect a “snapshot” of marriages between 1960-1990. With the divorce rate in Sweden reported to be at a record high in recent years, it would be timely to follow up this study with an exploration of the association between divorce and AUD risk in the current climate of increased divorce rates.

Implications

The results of this study contribute to our understanding of the association underling divorce and AUD. The findings suggest that the nature of the relationship between spousal loss through divorce or widowhood and increased risk for AUD may be both causal and non-causal.

These findings have implications for prevention and interventions for problem drinking in individuals approaching and going through divorce. For example, results of the study indicate that increased risk for onset of AUD began in the few years leading to divorce and increased in the year of the divorce. These findings suggest that early detection of risk for AUD may be possible in individuals expressing marital problems.

Similarly, the results of the study indicate that the association between increased risk for AUD and divorce was enduring and remained for several years following divorce. This suggests that any intervention aimed at addressing AUD during divorce may be required to extend beyond the divorce period itself.

Findings of the present study also indicate that divorce had a stronger association with risk for AUD if the spouse did not have a history of AUD. This is consistent with the notion that a non-AUD partner may be supportive in monitoring and helping control the drinking of an individual with AUD. Intervention programs may therefore benefit from exploring the inclusion of social support for individuals with AUD going through divorce or widowhood.

Links

Primary paper

Kendler et al (2017) Divorce and the Onset of Alcohol Use Disorder: A Swedish Population-Based Longitudinal Cohort and Co-Relative Study http://ajp.psychiatryonline.org/doi/pdf/10.1176/appi.ajp.2016.16050589

Other references

Holmes and Rahe Stress Inventory (1967)

Grant et al (2015) Epidemiology of DSM-5 Alcohol Use Disorder. JAMA Psychiatry. 2015 Aug; 72(8): 757–766. doi: 10.1001/jamapsychiatry.2015.0584

Kessler et al (1998) The social consequences of psychiatric disorders, III: probability of marital stability. American Journal of Psychiatry Published online: August 01, 1998 [Abstract]

Power et al (1999) Heavy alcohol consumption and marital status: disentangling the relationship in a national study of young adults. Addiction First published: October 1999 DOI: 10.1046/j.1360-0443.1999.941014774.x [Abstract]

Halme et al (2008) Hazardous drinking: prevalence and associations in the Finnish general population. Alcoholism Clinical and Experimental Research First published: 8 July 2008 DOI: 10.1111/j.1530-0277.2008.00740.x [Abstract]