Health service use in the year before death by suicide among ex-serving ADF members

AIHW conducted modelling to explore differences in health service use between ex-serving members who died by suicide (cases) and other ex-serving members who did not die by suicide (controls). Using a matched case-control design, the modelling estimated the odds of having accessed health services in the 12 months prior to the index date (date of death for cases, matched date for controls). 

The output from the statistical analysis used in a matched case control design (conditional logistic regression modelling) is an odds ratio (OR). An odds ratio greater than one indicates that a particular health service was more commonly accessed by cases compared to controls. An odds ratio less than one suggests relatively lower access among cases. It is important to note that these OR do not measure whether a service caused or prevented suicide, nor do they reflect the effectiveness of services. This information helps identify services with a stronger association with death by suicide, particularly those more likely to have been accessed or accessed closer to the time of death. These findings may support the identification of intervention points, where earlier or different support could be considered. More information on OR is included in Technical notes.

As outlined earlier, ex-serving members who died by suicide were more likely to have comorbidities than other ex-serving members who were alive. To account for this, AIHW performed both crude (unadjusted) and adjusted modelling. The adjusted modelling was performed to control for the confounding impacts of comorbidities using prescription and hospital diagnosis-based indices. While adjustment reduces the magnitude of some observed associations, the overall patterns remain consistent.

Odds of health service use among ex-serving members who died by suicide compared to controls

The odds of cases accessing a health service in the year before the index date was not significantly different to controls using the unadjusted model (OR of 0.92, not statistically significant). However, after adjusting for co-morbidities, cases had 65% lower odds of accessing any health service in the year prior to death compared to controls (adjusted OR of 0.35, statistically significant). This suggests that there is a large group of cases who did not access a health service when compared with controls that had similar health needs.

As the period examined was closer to death, ex-serving members who died by suicide had relatively higher odds of having accessed a health service compared to other ex-serving members. In the week before death, cases had 39% higher odds of accessing a health service than those in the control group (adjusted OR of 1.39). Figure 5 shows the OR across different time periods within the last year.

Conversely, as the period before the index date increased, cases had relatively lower odds of accessing health services, compared with controls. While cases had higher odds of accessing a health service in the week prior to death, cases and controls had similar access to a health service in the last month and last 3 months. Cases were significantly less likely to access a service 6 months and 9 months prior to death by suicide (adjusted OR of 0.74 and 0.49, respectively).

These patterns may reflect a shift from routine or ongoing care earlier in the year to more crisis-driven help seeking health service contact closer to death. However, these figures should be interpreted with caution as it cannot be determined whether the care accessed was directly related to suicide, nor whether risk was identifiable at the time of contact.

The trend of increasing health service access near death and diminishing associations at longer time intervals was consistent across all health service types. In the week before death, the highest odds of accessing health services were observed for ED presentations and hospital admissions. Access of MBS, PBS and DVA-funded (MBS equivalent) services showed smaller differences, with odds close to or below 1 throughout the year. 

This report also analysed ex-serving members access of mental health and non-mental health services. Mental health service access was significantly more common among cases. Across the last year, cases were four times more likely to have accessed mental health services than controls (adjusted OR of 4.04). The relatively higher odds of accessing health services was sustained across all time periods, peaking at nine months before death (adjusted OR of 4.23) and remaining elevated in the week before death (adjusted OR of 2.47). The associations were strongest for hospital-based (ED and admitted care) mental health services, with adjusted odds ratios ranging from 3.1 to 6.3. For more information see Supplementary tables S5.1 to S5.7.

The markedly increased healthcare access in the week before death by suicide, with consistently higher odds of access, particularly for mental health services across all settings, indicates multiple potential opportunities for intervention. While the strongest associations were seen in hospital services (ED presentation and admitted care), the substantial associations found in primary care services suggest the potential for intervention exists across the healthcare system. This pattern indicates that while hospital care settings may show the strongest signals, primary care settings could offer additional opportunities for risk identification and engagement, potentially at an earlier stage. In particular, the peak in mental health services at nine months appears to be one potential ‘moment’ that deserves further investigation.

Characteristics associated with health service use differences among ex-serving members who died by suicide compared with other ex-serving members

AIHW analysed whether there were common characteristics of ex-serving members who died by suicide compared to other ex-serving members who remained alive based on their health service use in the last year. In this section AIHW has focussed on mental health and overall hospital services (either ED or admitted care) as these were the services that showed strongest differences between cases and controls.  For more information on other health services, see Supplementary tables S6.1 and S6.2.

Cases had four times the odds of having accessed mental health services in the year before death compared to controls (adjusted OR of 4.04). Cases also were more likely to have accessed hospital services (either ED or admitted care) compared to controls (adjusted OR of 1.89). These associations were stronger among females, people who were Officer ranks and people who had separated from the ADF between 5 years (inclusive) and 10 years earlier, with adjusted odds ratios (adjusted for comorbidities) ranging from 2.9 to 17.9.

Figure 6 outlines characteristics that were associated with higher odds of health service use between cases and controls, for both mental health and overall hospital services.