← Back to GodExamined.com
Measured Efficacy · Peer-Reviewed
Religion & Depression Risk
The studies behind the claim that active faith lowers depression — a large cohort showing reduced incident depression, and a striking ~90% reduction among those at the highest inherited risk.
Li et al. (2016) — lower depression in a large cohort
ANNALS OF BEHAVIORAL MEDICINE · 2016;50(6):876–884
48,984
U.S. women followed
1996–2008 (Nurses’ Health Study)
Lower risk
of incident depression with
higher service attendance
In a prospective cohort of 48,984 women from the Nurses’ Health Study, more frequent religious service attendance was associated with a significantly lower risk of developing depression. The protective association held after adjusting for a wide range of health and social factors. This is representative of a broad cohort literature placing the reduction in depression risk for active attenders in roughly the 22–30% range.
Design: prospective cohort, ~12-year follow-up; depression defined as physician-diagnosed clinical depression, antidepressant use, or severe depressive symptoms.
Miller, Weissman et al. (2012) — ~90% lower risk at highest inherited risk
AMERICAN JOURNAL OF PSYCHIATRY · 2012;169(1):89–94
OR 0.09
~one-tenth the risk (~90% lower)
95% CI 0.01–0.82
10 yrs
prospective follow-up,
high familial-risk offspring
This ten-year prospective study followed the adult offspring of depressed and non-depressed parents. Among those at high familial (inherited) risk of depression, participants who reported that religion or spirituality was highly important to them had about one-tenth the risk of experiencing major depression over the following decade (odds ratio 0.09) — and, strikingly, the protective effect was strongest precisely in those at the highest risk.
Read honestly: the ~90% figure comes from the high-risk subgroup (n = 72), so the confidence interval is wide (0.01–0.82) — a real and statistically significant effect, but from a modest subsample. It is best stated as “up to ~90% lower among those at highest inherited risk,” not a population-wide average. Li et al. (above) gives the large-cohort, general-population figure.