Closing the Gap or Widening It? A Decade of Disparities in Mental Health Service Use in Australia

Longstanding inequities persist in Australian mental health service use across socioeconomic and geographic lines. The study used a panel dataset of population-level administrative datasets from the Australian Institute of Health and Welfare, spanning 2015–16 to 2024–25. It examined associations between socioeconomic status, remoteness, age, provider type, and jurisdiction, and two measures of mental health service use. The study design was intended to account for count-data overdispersion and time-invariant unobserved differences across healthcare clusters. The study estimated random-intercept negative binomial specifications. Service-use rates were higher in the Mid and Late periods than in the Early period but not consistently in the Post period. Substantial socioeconomic and geographic disparities, however, remained evident in the adjusted associations. Relative to their respective reference categories, emergency presentation rates were higher among residents of the most disadvantaged SEIFA quintile [main-effect rate ratio, RR = 1.64 (1.55–1.74)] and among people aged ≥85 years [RR = 3.81 (3.61–4.02)]. There were no significant age-by-area interactions, while Medicare-subsidised service rates were lower in the most disadvantaged quintile [RR = 0.62 (0.58–0.67)] and, most markedly, among residents of remote and very remote areas [RR = 0.35 (0.32–0.38)]. SEIFA quintile [main-effect rate ratio, RR = 1.64 (1.55–1.74)] and among people aged ≥85 years [RR = 3.81 (3.61–4.02)]. Age-by-area interactions were significant. Medicare-subsidised service rates were lower in the most disadvantaged quintile [RR = 0.62]. For Medicare-subsidised services, no period-by-category interaction term was statistically significant. However, the concentration index indicated that the socioeconomic gap widened descriptively between the Early (2015/16–2016/17) and Post (2023/24–2024/25) periods. For emergency presentations, a few period-by-category interaction terms were significant while the majority of period-by-category interaction terms for emergency presentations were not statistically significant. Therefore, this study found no consistent evidence that socioeconomic disparities in emergency presentations increased over the index period. Rates in remote and very remote areas rose relative to major cities in the Post period [interaction RR = 1.23 (1.10–1.38)], however. These findings are associational rather than causal, given the study’s design limitations. However, the findings show a persistent inverse pattern, with emergency presentations concentrated in more disadvantaged areas. Medicare-subsidised mental health services use was concentrated in the least disadvantaged areas. This pattern has equity implications and may be relevant to the allocative efficiency of publicly funded mental health systems. Future research should consider incorporating need and cost data for further analysis.