Modelling health service of ex-serving members admitted for self-harm
Odds of using a health service for ex-serving members admitted for intentional self-harm
AIHW conducted modelling to further understand the relationship between use of health service and self-harm (leading to hospital admission). The output from the modelling is an odds ratio (OR) which compares the odds of accessing a health service among ex-serving members in the self-harm cohort to the odds among ex-serving members in the comparison cohort.
This report presents OR for three different periods to provide distinct insight into the differences in health service use between the cohorts. The first OR is for health service use in the year before admission, the second is for health service use in the year after. The third OR represents the difference between the two OR, which represents the change in the OR in the year before and after the index admission. The change over time is of interest because it potentially distinguishes health service use associated with the self-harm event, rather than reflecting the baseline health differences between the cohorts. Box 3 outlines the interpretation of the OR.
Box 3: Interpreting odds ratios (OR) in this report
This report presents multiple OR by different health service use and by different periods of analysis relative to the index hospital admission. In each instance, the OR compares the odds of accessing health service in the self-harm cohort compared to the comparison cohort.
The results of each OR by period of analysis could be interpreted as:
- The first OR is for the year before hospital admission. If OR>1, those in the self-harm cohort had higher odds of accessing health services before the event compared to the comparison cohort. If OR<1, the self-harm cohort had lower odds of accessing health services before the event.
- The second OR is for the year after hospital admission. If OR>1, those in the self-harm cohort had higher odds of accessing health services after the event compared to the comparison cohort. If OR<1, the self-harm cohort had lower odds of accessing health services after the event.
- The third OR is the difference between the first two OR. If OR>1, those in the self-harm cohort showed a relative increase in health service use after the event compared with the comparison cohort. If OR<1, those in the self-harm cohort showed a relative decrease in health service use after the event compared with the comparison cohort (which could be due to both cohorts increasing, but the comparison cohort had a greater relative increase, or could be due to a decrease in the self-harm cohort).
The OR do not provide insight on the effectiveness of services such as whether a service is more or less likely to lead to self-harm but describe differences in the use of health service for those who were admitted for self-harm compared to those who were admitted for other reasons. More information on OR is included in the Technical notes.
As outlined earlier, ex-serving members admitted for self-harm were more likely to have comorbidities than ex-serving members admitted for non-injury reasons, this report therefore includes both crude (unadjusted) and adjusted modelling. The adjusted modelling was performed to control for the confounding impacts of comorbidities as well as matching variables with residual confounding effects. Caution is advised when interpreting the fully adjusted results for PBS services, hospital services and any health service use, as the comorbidity indices (RxRisk Index from PBS/RPBS data and MACSS index from hospital data) were derived from these same data sources. While adjustment reduces the scale of observed associations, the overall patterns remain consistent.
Ex-serving members admitted for intentional self-harm were more likely to access health services
In the year before admission, those in the self-harm cohort had significantly higher odds of accessing any health service (OR=5.57) compared with the comparison cohort. The OR after adjusting for comorbidities was 2.03, meaning those in the self-harm cohort were more likely to have accessed health services compared to the comparison cohort during that period. The rest of this section will report adjusted OR, unless expressed otherwise.
In the year after admission, the self-harm cohort had significantly higher odds of accessing a health service (OR=1.73), suggesting continued greater engagement with health services compared to the comparison cohort. However, the OR for the change in health service use from the year before to the year after self-harm admission was 0.85 (not significant). This indicates a reduction in the relative difference over time (Figure 4).
The highest OR by health service types were observed for ED presentations, PBS services and admitted care, in both the year before and after self-harm admission. For example, in the year after admission, the odds of health service use in the self-harm cohort compared to the comparison cohort were highest for ED (OR=2.69), followed by PBS service (OR=1.79), and admitted care (OR=1.64).
When health services accessed were disaggregated by mental health and non-mental health categories, differences were especially pronounced for mental health services among the self-harm cohort compared to the comparison cohort. In the year before admission, odds of accessing mental health services were substantially higher in the self-harm cohort, with OR of 6.97 (before) and 12.64 (after), compared to the comparison cohort.
Of the mental health services, in the year after self-harm admission, the self-harm cohort had significantly higher odds of accessing each service type compared with the comparison cohort. The self-harm cohort were more likely to access PBS services (OR=9.09), ED services (OR=7.64), admitted care (OR=7.60), MBS services (OR=6.28) and DVA-funded MBS equivalent services (OR=2.75). The odds of the self-harm cohort accessing each mental health service increased in the year after relative to the year before admission (except for ED presentations which remained stable).
Figure 4: Odds ratios of accessing health services among ex-serving members admitted for intentional self-harm in the year before and after admission compared to those admitted for non-injury related reasons
Chart showing odds ratios of health service use among ex-serving members in admitted for self-harm compared to those admitted for non-injury reasons in the year before or after admission.
Rates of using a health service for ex-serving admitted for intentional self-harm
In addition to OR, this report includes rate ratios (RR) to assess differences in the frequency of service use between ex-serving members in the self-harm and comparison cohorts. Unlike OR, which compare whether a person accessed a service at least once, RR compare the frequency of health service contacts per person year between the self-harm and the comparison cohort, providing insight into how intensively health services were used across the cohort.
An RR greater than 1 means that ex-serving members in the self-harm cohort had a higher rate of health service use than those in the comparison cohort. As with the OR analysis, RR results are presented for the year before admission, the year after admission, and the relative change over time between the two cohorts.
The RR do not provide insight on the effectiveness of health services such as whether a health service is more or less likely to lead to self-harm but describe the patterns of use between those who were admitted for self-harm compared to those who were admitted for reasons other than injury. More information on RR is included in the Technical notes.
AIHW performed both crude and adjusted modelling to account for comorbidities. Caution is advised when interpreting the fully adjusted results for PBS services, hospital services and any health service use, as the comorbidity indices (RxRisk Index from PBS/RPBS data and MACSS index from hospital data) were derived from these same data sources. While adjustment reduces the observed associations, the overall patterns remain consistent.
Ex-serving members admitted for intentional self-harm had higher rates of health service use than ex-serving members admitted for non-injury reasons
In the year before admission, the self-harm cohort had significantly higher rates of health service use than the comparison cohort (RR=2.09). The RR after adjusting for comorbidities was 1.31, meaning that rates of health service use were 31% higher in the self-harm cohort compared to the comparison cohort (Figure 5). In the year after admission, this difference increased further, with an RR of 1.41 (or 41% higher rate of health service use). When comparing changes over time, the increase in service use was 9% greater in the self-harm cohort than in the comparison cohort. The rest of this section will report adjusted RR, unless expressed otherwise.
The self-harm cohort had higher rates of health service use for each in-scope service in the year before and after admission, though the size of these differences varied. In the year before admission, the self-harm cohort had higher rates of ED presentations (RR=2.77), PBS services (RR=1.39), MBS services (RR=1.34) and admitted care (RR=1.20) compared to the comparison cohort.
In the year after admission, these differences increased with the self-harm cohort having higher rates of ED presentations (RR=2.61), admitted care (RR=1.46), MBS service (RR=1.43) and PBS services (RR=1.41) than the comparison cohort.
Of the MBS subservices, the self-harm cohort had a higher rate of services than the comparison cohort in the year before and after, especially for allied health and specialist services.
Mental health service use was also higher among the self-harm cohort in both time periods, with rate ratios of 2.55 (before) and 3.07 (after), compared to the comparison cohort. This represents a 21% relative increase in mental health service use over time in the self-harm cohort compared to the comparison cohort.
Of the mental health services, in the year after self-harm admission, the self-harm cohort had significantly higher rates of using each health service type compared with the comparison cohort, with RR ranging from 2.58 (PBS services) to 7.77 (ED presentations).
Figure 5: Rate ratios of using health services among ex-serving members admitted for intentional self-harm in the year before and after admission compared to those admitted for non-injury related reasons
Chart showing rate ratios of health service use among ex-serving members admitted for self-harm compared to those admitted for non-injury reasons in the year before or after admission.
Characteristics of ex-serving members admitted for intentional self-harm who used health services
Common characteristics of ex-serving members who accessed health services are presented in this report. This section of the report discusses the results for mental health services as it showed the highest OR and a RR of health service use in the year before and after self-harm admission.
Ex-serving members in the self-harm cohort more likely to access mental health services compared to the comparison cohort both before and after admission were: females, those who had served in the RAAF and those who held office ranks (Figure 6).
Ex-serving members in the self-harm cohort more likely to access mental health services compared to the comparison cohort after the admission were those: aged 35 to 44 and 55 to 64 years, separated due to contractual/administrative reasons, served for 10 or more years, non-DVA clients, DVA White card holder, and without comorbidities (Figure 6).
Figure 6: Characteristics associated with differences in health service use among ex-serving members admitted for intentional self-harm in the year before and after admission compared to those admitted for non-injury related reasons
Chart showing odds or rate ratios of health service use among ex-serving members admitted for self-harm compared to those admitted for non-injury reasons in the year before or after admission by demographic and ADF-service characteristics.
Supplementary tables S4.2 and S4.3 (see data) presents the characteristics associated with higher odds of accessing health services and tables S5.2 and S5.3 presents the characteristics associated with higher rates of using health services.