Extended Data Fig. 1: Results of age–period–cohort analysis from the first Poisson log-linear regression model.
From: Age and generational patterns of overdose death risk from opioids and other drugs
Black horizontal line represents no effect, the black curves represent the estimated effect and the orange and green shaded area represent the associated confidence interval. a, b and c show the age, period and cohort deviations which reveal significant statistical deviation for most ages, periods and cohorts (since the confidence interval does not cover 0). The net drift from this analysis was 8.23% per year CI = (8.17, 8.30) and is also statistically significant since the confidence interval does not cover 0 (D). d also shows statistically significant differences in the age-specific local drifts against the net drift which range between 4% to 11%. e shows the cross-sectional and longitudinal age curves for overdose mortality. The cross-sectional curve reveals the age-specific mortality rates for the reference period of p0 = 2000, and the longitudinal age curve reveals the age-specific mortality rate for the reference cohort of c0 = 1960. The clear divergence of these two age curves indicates important cohort effects that can also be revealed from the net drift. (F) and (G) reveal the period and cohort rate ratios of mortality rate against the reference period (p0 = 2000) and reference cohort (c0 = 1960), respectively.
