Secular trends in the age at onset of bipolar I disorder - Support for birth cohort effects from interational, multi-centre clinical observational studies.

OBJECTIVE: To examine any association of birth decade, sex and exposure to alcohol and/or substance use disorders (ASUD) with age at onset (AAO) of bipolar I disorder (BD-I).

METHODS: Using data from a representative clinical sample of 3896 BD-I cases recruited from 14 European countries, we examined AAO distributions in individuals born in consecutive birth decades. Cumulative probabilities with Mantel-Cox log-rank tests, pairwise comparisons and Odds Ratios (OR) with 95% confidence intervals (95% CI) were employed to analyze AAO according to birth decade, sex, and presence or absence of an ASUD.

RESULTS: In the total sample, median AAO of BD-I decreased from about 41 years for those born in the 1930s to about 26 years for those born in the 1960s. In a sub-sample of 1247 individuals (selected to minimize confounding), AAO significantly decreased for males and females born in each consecutive decade between 1930 and 50 (OR: 0.65; 95% CI: 0.51, 0.81), and for cases with an ASUD as compared to without (OR: 0.77, 95% CI: 0.69, 0.87). The best fitting regression model identified an independent effect for each birth decade and an interaction between ASUD status and sex, with a consistently earlier AAO in males with an ASUD (OR: 0.79: 95% CI: 0.70, 0.91).

CONCLUSIONS: In BD-I cases diagnosed according to internationally recognized criteria and recruited to pan-European clinical observational studies, the AAO distributions are compatible with a birth cohort effect. A potentially modifiable risk factor, namely ASUD status, was associated with the observed reduction in AAO, especially in males.