In accordance with previous research, we found that marital dissolution was related to an increase in mental health problems for all divorced groups. The worsening in mental health for the divorced compared to people remaining married occurred even though people who were going to divorce reported relatively poor mental health also before the divorce. Our results may be taken as support both for the health selection (poor baseline mental health) and social causation models (the divorce causes a further worsening in mental health) [6–8]. To our knowledge, the present study was the first to investigate the effects of divorce on change in mental health in couples where at least one of the spouses had a risky consumption of alcohol prior to the divorce. We hypothesized that the mental health of high risk users would worsen after the divorce, whereas the mental health of the individuals divorcing high risk users could either improve or worsen after the divorce. The results showed that female high-risk users and men divorcing female high-risk users had a significantly higher increase in mental health problems post divorce, compared to female divorced low risk users and their spouses. A similar trend was seen for men with a risky consumption of alcohol. The worsening of mental health after a divorce seemed to last longer in male ex-spouses of female high-risk users than in other divorced men, and the same but not significant trend was observed for divorced men who themselves had a high pre-divorce alcohol consumption. Although the effects were only significant in couples where the woman drank heavily, low statistical power makes us unable to conclude in regards to a true gender difference. The observed similar tendencies for men who drink heavily in fact suggest that there may be similar increased, and perhaps long-lasting effects in both genders.
The hypothesis that divorcing a high-risk user would lead to a decrease in mental health problems was not supported – divorcing a high-risk user appears to be related to increasing mental health problems. Men divorcing high-risk users even showed a stronger increase in problems than did other divorced men. This suggests that the adversities related to divorcing a high-risk user, in addition to adversities related to divorce in general, outweigh the benefits of getting rid of problems associated with spousal high-risk drinking. There could be several reasons for this. The strain related to living with a high-risk using spouse may continue to affect life after the divorce, and in many cases, the contact between the spouses does not end once the divorce is final. For instance, custodial disagreements or limited control over the relationship between ex-spouse and children may affect the other spouses’ mental health more in cases when the ex-spouse drinks heavily. Sometimes alcohol abusing persons left alone may act rather vindictive toward their ex-spouse, perhaps especially when drunk. Also, leaving a person with alcohol problems may undoubtedly sometimes impose feelings of guilt.
Despite common beliefs and previous research, high-risk alcohol use may not in all circumstances be negative. A recent study showed that a high consumption was related to less spousal mental distress, once the variation caused by problems directely associated with alcohol abuse (like being critisiced for drinking too much) was accounted for . Thus, a high consumption of alcohol per se may not necessarily be related to mental health problems for the spouse during the marriage, which in turn may make divorcing a high-risk user just as hard, or even harder than divorcing a normal consumer. Our somewhat broad definition of high-risk alcohol use in the present study (top 13.1% of men and 9.3% of women) may entail that not all in fact have developed a problematic relationship to alcohol, but rather are at risk of it. This may to a certain degree have affected our results, in line with the findings in the aforementioned study.
Female high-risk users experienced more of an increase in mental health problems after the divorce compared to the female normal consumers. The same, but non-significant trend was observed in high-risk drinking men. Previous research has found a higher risk of increasing alcohol consumption and a higher risk of developing or relapsing into alcohol abuse for divorced couples [22, 23]. Mental distress after a divorce has been found to return to baseline levels only after remarriage , and people who drink excessively may have a lower chance of remarrying than people who drink less.
The analyses of time since divorce showed that in general, most of the detrimental effect of divorce takes place during the first years after divorce (1–3 years) and that the effect after 4–7 years is not much worse than after 8–10 years. Apparently most of the healing occurs within the first few years. This finding is in line with results of several previous studies – that have shown increased mental health problems the first few years after the divorce that recedes as time goes by, but stays at an elevated level compared to baseline levels [38–40]. There is, however, a great deal of variability in recovery from mental health problems after the divorce. In one study, about half of the respondents improved with time, whereas about one fourth of the sample got worse .
The significant interaction effect between time since divorce and wives’ high-risk use on the husbands’ change in mental health, and the same, but non-significant trend seen for male high-risk users, indicate that the healing process may be particularly slow for these men. During the first years after the divorce, the mental health problems increased dramatically both in the high-risk groups and the reference groups. The mental health problems declined with time, but were significantly higher for men divorcing high-risk users than for the reference groups even 8–10 years after the divorce. The observed gender differences may reflect true variation or random fluctuations, where a true tendency of delayed healing only could be observed in men.
The previous studies that have reported subsiding mental health problems a few years after the divorce, indicate that a divorce may be seen as a temporary crisis in which most people manage to get through on the positive side as time goes by [38, 39]. This finding corresponds well with our results for the reference groups, as opposed to those for men from couples with high pre-divorce alcohol consumption. These seemed to experience more of a chronic strain. The same may be true for women. Interestingly, Hetherington  found that although most people tend to adjust well to a divorce in time, about 10% report elevated scores on depression more than 10 years post-divorce. Typically, members of this group were troubled by alcohol or drug abuse, other health problems, low self-esteem, low social support, poverty and anti-social behavior. This may indicate that in some cases negative effects of divorce may be cumulative – problems cause more problems to arise. The trends for men from pre-divorce high-risk couples essentially showed the same pattern. These men (and possibly women) may be particularly vulnerable for the adverse effects of divorce on mental health, causing the healing process to take longer time.
The present study is to our knowledge the first to investigate how spouses of high-risk users or the high-risk users themselves reacts to divorce in terms of change in mental health problems. We were able to investigate this due to our large and representative sample of approximately 12,000 couples with two-wave questionnaire data with an 11-year interval, and registry based data on marital status and time of change in marital status.
However, there are methodological limitations to our study. Only 38.0% of the invited couples had returned all four questionnaires in both HUNT1 and 2. This may have caused a selection bias. However, an attrition study of the HUNT 2 sample showed that high alcohol consumption and mental distress in HUNT 1 only predicted non-participation in HUNT 2 moderately well (alcohol consumption: OR = 1.27 for the top 3% consumption; mental distress: OR = 1.84 for the top 1%) . Divorce was a stronger prediction of non-participation in HUNT 2 (OR = 1.98). In general, even highly selective non-participation are not usually expected to influence associations between variables dramatically , giving reason to believe that our estimates have not been severely affected by non-participation. However, if people with the strongest loss of mental health from T1 to T2 tend to drop out of the study, and if this selection effect is stronger among divorced people than among people who are still married, the observed association between divorce and mental health problems may be somewhat attenuated. Correspondingly, if such a selection effect is stronger among alcohol burdened divorcing couples than among other divorcing couples, the observed difference between divorcees from alcohol burdened couples and other divorced people will be attenuated. However, we believe that the risk of such a differential selection effect is small.
Due to few divorced couples where both spouses had participated at T1 and T2, we were compelled to shift the focus from alcohol abuse to heavy risk use in order to maximize the number of divorced cases with a (likely) alcohol problem. The top 13% of the men and top 9% of the women were categorized as high-risk users in our study. The male heavy risk use falls within the 12-month prevalence rates for alcohol use disorders among men (16%), whereas the female group is somewhat larger than the 12-month prevalence rates for women (6%) . Compared to the affirmative response percentages on the alcohol frequency, influence of alcohol and excessive drinking items of the low-risk groups, the high-risk men and women clearly exhibit drinking behaviors representing a risky consumption. However, the broadly defined high-risk categories may have somewhat attenuated our effect estimates, especially for women.
The 11 year time gap between T1 and T2 makes us able to investigate long term effects of divorce. However, a lot of events may have taken place during this long period. People may remarry, start new families, even both remarry and divorce, get somatically ill or experience other kinds of events that may impact the mental health in a positive or negative way at T2. Remarriage is related to improvements in mental health , and as some of the respondents divorced between T1 and T2 are bound to have remarried before T2, the inclusion of these may have underestimated the effects of divorce on change in mental health.
The mental health index used in the present study was based on nine items asking about a variety of symptoms indicative of general mental health problems. As the measure has not been broadly validated, there is a risk of unsatisfactory construct validity of the outcome measure. However, the high correlation with the Hopkins Symptom Checklist  - a well validated measure of mental distress – speaks in support of the validity of the measure. Rather than using the term “mental distress”, we chose the more general term “mental health problems” due to the wide variety of symptoms tapped by the nine items.
Perhaps the largest methodological weakness of our study is the small groups in the interaction analyses of time since divorce and heavy risk alcohol use. The small number of observations and the wide confidence intervals imply that the results from the analyses of time since divorce should be interpreted with caution. Especially a significant observed effect in one gender only cannot be taken as evidence that the same effect does not exist in the other gender as well. The results need to be replicated in a larger sample.
Besides of random errors, our results could be systematically biased due to confounding between alcohol consumption and mental health problems from sources not adjusted for. For instance heavy drinkers might tend to marry persons with mental health problems. We would expect such confounding to primarily affect the baseline mental health problems, not the estimates of change in mental health. But we also cannot rule out the possibility of confounding from sources of vulnerability that make people from alcohol risk couples respond more strongly to divorce than do other people. Such confounding would inflate the observed differences between divorced subjects from alcohol exposed couples and other divorced subjects.