Age-period-cohort effects on suicide mortality in Andalusia, Spain (2000-2024): demographic masking and sustained pandemic excess.
To analyse suicide mortality trends in Andalusia, Spain (2000-2024) through age-period-cohort (APC) decomposition, identify trend change-points, and quantify pandemic-associated excess mortality.
This population-based study used Spanish National Statistics Institute data (INE; heading 098, ICD-10 × 60-X84). Age-standardised rates (European Standard Population 2013) were computed by sex. Methods comprised joinpoint regression, Prais-Winsten modelling, an APC model (intrinsic estimator), and a log-linear counterfactual projecting the 2000-2019 trend onto 2020-2024, with quasi-Poisson and age-range sensitivity analyses.
18,350 suicide deaths were recorded (77.4% male). Crude rates remained stable (EAPC = - 0.12%, p = 0.615) whereas age-standardised rates declined significantly (EAPC = - 0.96%, p < 0.001), revealing demographic masking. Joinpoint regression identified a single change-point reversing the long-term decline, at 2018 for both sexes (EAPC - 1.67% to + 2.65%) and 2019 for men (- 2.01% to + 3.84%); neither post-reversal segment reached significance. The APC model (adults aged 40-84; deviance R² = 0.983) showed an ascending age gradient, a period nadir in 2015-2019, and declining cohort risk. The counterfactual estimated 783 excess deaths during 2020-2024 (O/E = 1.23, 95% CI 1.20-1.27; p < 0.001), concentrated among those aged 15-64.
The age-standardised decline reached its nadir around 2018-2019 and reversed thereafter; relative to the projected pre-pandemic trend, the pandemic period (2020-2024) showed a sustained excess without attenuation over the five years observed. These findings support age-targeted prevention; provincial heterogeneity warrants formal subnational analyses before recommending geographically differentiated strategies.
This population-based study used Spanish National Statistics Institute data (INE; heading 098, ICD-10 × 60-X84). Age-standardised rates (European Standard Population 2013) were computed by sex. Methods comprised joinpoint regression, Prais-Winsten modelling, an APC model (intrinsic estimator), and a log-linear counterfactual projecting the 2000-2019 trend onto 2020-2024, with quasi-Poisson and age-range sensitivity analyses.
18,350 suicide deaths were recorded (77.4% male). Crude rates remained stable (EAPC = - 0.12%, p = 0.615) whereas age-standardised rates declined significantly (EAPC = - 0.96%, p < 0.001), revealing demographic masking. Joinpoint regression identified a single change-point reversing the long-term decline, at 2018 for both sexes (EAPC - 1.67% to + 2.65%) and 2019 for men (- 2.01% to + 3.84%); neither post-reversal segment reached significance. The APC model (adults aged 40-84; deviance R² = 0.983) showed an ascending age gradient, a period nadir in 2015-2019, and declining cohort risk. The counterfactual estimated 783 excess deaths during 2020-2024 (O/E = 1.23, 95% CI 1.20-1.27; p < 0.001), concentrated among those aged 15-64.
The age-standardised decline reached its nadir around 2018-2019 and reversed thereafter; relative to the projected pre-pandemic trend, the pandemic period (2020-2024) showed a sustained excess without attenuation over the five years observed. These findings support age-targeted prevention; provincial heterogeneity warrants formal subnational analyses before recommending geographically differentiated strategies.