Mental distress predicts divorce over 16 years: the HUNT study

Background: The association between mental distress and divorce is well established in the literature. Explanations are commonly classified within two different frameworks; social selection (mentally distressed people are selected out of marriage) and social causation (divorce causes mental distress). Despite a relatively large body of literature on this subject, selection effects are somewhat less studied, and research based on data from both spouses is scarce. The purpose of the present study is to investigate selection effects both at the individual level and the couple level. Methods: The current study is based on couple-level data from a Norwegian representative sample including 20,233 couples. Long-term selection effects were tested for by means of Cox proportional hazard models, using mental distress in both partners at baseline as predictors of divorce the next 16 years. Three identical sets of analyses were run. The first included the total sample, whereas the second and third excluded couples who divorced within the first 4 or 8 years after baseline, respectively. An interaction term between mental distress in husband and in wife was specified and tested. Results: Hazard of divorce was significantly higher in couples with one mentally distressed partner than in couples with no mental distress in all analyses. There was also a significant interaction effect showing that the hazard of divorce for couples with two mentally distressed partners was higher than for couples with one mentally distressed partner, but lower than what could be expected from the combined main effects of two mentally distressed partners. Conclusions: Our results suggest that mentally distressed individuals are selected out of marriage. We also found support for a couple-level effect in which spouse similarity in mental distress to a certain degree seems to protect against divorce.