Overcoming Standardization: Revealing Hidden Age Patterns of Suicide with Spatiotemporal Models

TL;DR

This paper introduces age-structured hierarchical Bayesian models to better understand and estimate complex age patterns in suicide data, overcoming limitations of traditional standardization methods.

Contribution

It proposes a novel modeling approach that incorporates space-time, space-age, and time-age interactions for more accurate analysis of suicide patterns.

Findings

Improved model fit with age effects included

Detected rising suicide risk from 2017 to 2022

Identified nonlinear age-related risk patterns

Abstract

Indirect standardization is widely used in disease mapping to control for confounding, but relies on restrictive assumptions that may bias estimates if violated. Using data on suicide-related emergency calls, this study highlights such limitations and proposes age-structured hierarchical Bayesian models as an alternative. These models incorporate space-time, space-age, and time-age interactions, allowing for more accurate estimation without strong assumptions. The results show improved model fit, especially when including age effects. The best model reveals a rising temporal trend (2017--2022), a nonlinear age pattern, and stronger risk increases among younger individuals compared to older ones.