Age, period and cohort analysis of suicide trends in Australia, 1907–2020

• Published in: The Lancet regional health. Western Pacific Vol. 51; p. 101171
• Main Authors: Spittal, Matthew J., Mitchell, Rachel, Clapperton, Angela, Laughlin, Adrian, Sinyor, Mark, Page, Andrew
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01-10-2024
• Subjects: Age-period-cohort; Australia; Suicide; Suicide trends; Suicide trends; Suicide; Australia; Age-period-cohort
• ISSN: 2666-6065; 2666-6065
• DOI: 10.1016/j.lanwpc.2024.101171

Suicide rates have been increasing in Australia since the mid-2000s, especially for women aged ≤25 years. We conducted an age-period-cohort study to investigate these recent trends in the context of historical Australian suicide rates.
Data on annual suicides in Australia from 1907 to 2020 were extracted from the General Record of Incidence of Mortality. We modelled age-specific effects for a reference cohort, after adjustment for period effects.
We found evidence of age, cohort and period effects. For males, compared to the cohort born in 1946–1950, rates were higher for all cohorts born after this year. The period effect showed peaks in the risk of male suicide in the mid 1960s and the early 1990s, followed by a decline in risk until early 2010, after which the risk began to rise again. For females, compared to the cohort born in 1946–1950, the risk of suicide was higher for all cohorts born after this, with the highest risk for those born in 2006–2010. The period effect for females showed an elevated risk of suicide in the mid 1960s followed by a sharp decline, and an increase in risk after 2009.
Suicide rates in Australia have fluctuated substantially over time and appear to be related to age trends as well as period and cohort trends. Advocacy and policy making tends to focus on contemporaneous changes in suicide rates. However, this study shows that focusing only on year-on-year changes in suicide rates ignores underlying trends for specific population birth-cohorts.
None.