In this analysis of forty years of Australian mortality data examining deaths due to substance poisoning, alcohol-related disease, and suicide, the most common of the three causes of death over the study period was suicide. Between 1980 and 1999, combined mortality rates were relatively stable, reflecting the combined effect of both a reduction in alcohol-related disease deaths and increase in substance poisoning deaths. This was followed by a decline until 2006 and a subsequent increase to the end of the study period, primarily attributable to corresponding changes in the rates of substance poisoning and suicide deaths among males. Over the study period, combined age-specific mortality rates remained relatively stable, but patterns varied between cause of deaths, as well as by sex, age, birth cohort, and period.

Increases in substance poisoning deaths were of considerable concern in Australia in the late 1990s, largely attributed to increased availability of cheap, higher purity heroin [21]. After the turn of the century, an unprecedented reduction in heroin supply across all Australian jurisdictions was accompanied by a reduction in fatal overdoses [21, 22]. The findings from the present study reflect these period effects of substance poisoning deaths, highlighting the importance of considering changes in the local context—whether they be environmental, social, economic, behavioural, or healthcare-related—when interpreting temporal patterns of death.

In the US, Scotland, and Canada [12], recent and substantial increases in substance poisoning deaths have been attributed in part to increasing prescription medicines use (particularly opioids) [23,24,25], novel benzodiazepines [26], and increasing use of heroin and illicit fentanyl in North America [23]. This trend is somewhat reflected in the Australian setting, which so far lacks the proliferation of illicit fentanyl, with the increase in substance poisoning deaths since 2006 coinciding with substantial increases in opioid analgesic prescribing [27, 28]. Importantly, changes in administrative coding around this time may have influenced when the observed increase in substance poisonings occurred [15]. Although examining specific drugs involved was beyond the study’s scope, a 2021 analysis showed opioids were the most commonly involved drug class in Australian overdose deaths, with over two-thirds of opioid overdose deaths involving pharmaceutical opioids [29]. Harm-reduction efforts, including drug treatment policies, are important strategies to reduce the scale of these substance-related harms.

Alcohol consumption in Australia increased post-World War II and peaked in the 1970s, attributed to increasing affluence, reduced restrictions, and declines in “puritanism” around this time [30]. In the 1980s, consumption declined and has remained relatively stable since the 1990s. Previous analyses have shown alcohol use tends to peak between the ages of 40–60-years [31], but that alcohol use and risky drinking behaviours have declined among recent birth cohorts [31, 32]. In the present study, these findings are reflected in the older ages and strong cohort trends of alcohol-related disease deaths, whereby rates were highest among 50–80-year-olds across the study period and absolute rates have declined among successive birth cohorts; similar cohort and age trends have been observed internationally [6, 12]. It is unsurprising these deaths appear to occur at older ages: alcohol-related disease mortality captures a range of health effects and chronic diseases resulting from long periods of heavy alcohol consumption [33]. Accordingly, a higher proportion of alcohol-related disease deaths will occur after long induction periods, compared to the more acute poisoning and suicide deaths. Also of note is the difference between sexes, with more substantial declines in alcohol-related disease deaths observed among male compared to female cohorts born after the 1950s. This convergence in male–female alcohol-related disease deaths may reflect the increase in women’s drinking over this time, as well as the narrowing of the gender gap in risky drinking behaviours [34, 35]. Future research will be required to assess whether changes in alcohol consumption among later birth cohorts translate into a greater convergence of alcohol-related disease mortality as these individuals reach mid-life.

Of the cause-specific deaths considered, suicide remained the dominant cause over the study period, with patterns of deaths varying by sex and age. Suicide deaths among older-aged men were consistently higher than among middle-aged men, concordant with prior global research [36]. In the 1990s, the increased rate of suicide deaths among young (20–35-years) males may have related to increasing unemployment in this age group [37], while reduced suicide mortality rates among middle-aged (40–60-years) females may reflect disproportionate antidepressant uptake in this age group [38]. As the circumstances surrounding substance-related deaths can be complex, it is possible that some suicide deaths involving drugs may have been misclassifications (whereby intent was incorrectly deemed intentional), especially considering these deaths coincided with the previously mentioned increase in heroin availability. However, hanging was the most common method of suicide among young men in the 1990s [39], limiting misclassification of poisoning deaths as suicides as an explanation for observed increases in young men. Reasons for the dynamic changes in female suicide deaths over time, particularly the increased rate since 2000, is unclear, and requires further exploration. Regardless of the contributing factors, that suicide deaths remained high throughout the study period demonstrates a critical need to review the effectiveness of current suicide prevention strategies.

Overall, patterns of deaths due to substance poisoning, alcohol-related disease, and suicide differed, providing further evidence that caution is needed when considering a narrative that groups these deaths as driven by shared phenomena [40]. In US and UK studies, as in this study, drug-related deaths appear to be the main driver of increases in overall rates of the collective group [5, 7, 9, 12]. This observation has led researchers to question whether these deaths are indicative of a “despair” problem or a drug overdose problem (or more specifically, an opioid overdose problem) [41, 42]. Given the overlapping and intersectional drivers of these deaths, including socioeconomic disadvantage, declining social support structures, increasing global insecurity—all contributing to rising levels of psychological distress, there remains value in considering these deaths as having shared influences and associated risks. However, our findings add to the growing body of evidence demonstrating the importance for researchers, policymakers, and other stakeholders to remain aware of the heterogeneity between these causes of death, and the potential need for the development of differing strategies to address and prevent these deaths. Our study, and others examining the age, period, and/or cohort effects of these deaths [6, 43, 44], also demonstrate how causes of death differentially affect sub-populations, as well as the effects of local phenomena, such as the impact of disruption to Australian drug markets on substance poisoning deaths. Although strategies targeting underlying drivers and psychological distress may be universal, knowledge of these differences and contexts is essential for the development of targeted and context-specific interventions, as well as the accurate reporting of temporal changes in mortality rates.

APC modelling provides a powerful framework for inferring meaningful structure from complex data. Use of an expanded definition of these deaths allowed the capture of drug and other alcohol-related disease deaths which have the same underlying socio-structural drivers, but which may be excluded by analyses limited to drug poisonings and alcohol-related liver disease. The availability of forty years of mortality data provided scope for examining trends across a period of considerable social and economic change, allowing cohort effects to be examined.

Several limitations must be noted. This study examined cause of death records and may be limited by incorrect or unclear coding. Since 2006, the Australian Bureau of Statistics has implemented a revisions process for improving the quality of cause of death data and, in particular, the coding of intent [15]. This may influence comparisons across calendar time; however, as fewer substance poisoning deaths with undetermined intent (ICD codes Y10-Y15) were observed in the years before compared to after implementation of the revision process (data not shown), the impact of this administrative change on observed patterns is expected to be minimal. Nevertheless, intent may be difficult to ascertain and the underreporting of suicide deaths is a known problem [45]. Although international research has demonstrated associations between these causes of death and socioeconomic factors (e.g., ethnicity, education, employment) [1, 2, 11], detailed data on these were unavailable, and these relationships require further exploration in the Australian context.