Religious service attendance is protective against the diseases of despair: evidence from regression, sibling-fixed effects, and instrumental variables analyses.

It is unclear whether the large secular decline in religiosity has contributed to the dramatic rise in the "deaths of despair." We contribute to the recent epidemiologic literature estimating more rigorous effects of religiosity on health by examining the association between religiosity and the diseases of despair via regression, sibling fixed effects (SFE) analyses, instrumental variable (IV), and cross-lag analyses. We used the US Add Health sample when respondents were in Waves (W) 3-5 (ages: 18-43). We measured religious service attendance and a composite outcome consisting of painkiller abuse, past-year suicidal ideation, and weekly binge drinking. We estimated linear probability models, SFE, IV, and cross-lag models. Confounders included parental socio-demographics, community/school characteristics, and individual socio-demographics. Greater religious service attendance was negatively associated with the composite outcome in the pooled sample (β =-0.031; p < .5) and at each wave (W3 β=-0.025; W4 β=-0.040; W5 β=-0.028; all p < .5). Conclusions were similar in SFE models (W3-5 pooled β=-0.013), IV models (W4 β=-0.081; W3-5 pooled β=-0.064, all p < .5, F>100, and overidentification p > .10) and cross-lag models (W3-5 pooled β=-0.023, p < .5). The consistent results across models suggests that the large decline in religious service attendance likely contributed to the rise in the deaths of despair.