Access to means of suicide, occupation and the risk of suicide: a national study over 12 years of coronial data

At a population level reducing access to lethal means is one of few suicide prevention strategies for which there is convincing evidence [1, 2]. Specifically, there have been a number of studies showing that reducing the availability and accessibility of firearms [3], charcoal [4], as well as installing physical barriers and/or safety nets around bridges and tall buildings [5], have been associated with a reduction in suicide deaths, with little evidence of means substitution [5, 6].

The availability of potentially lethal suicide methods has also been cited as an explanation for higher rates of suicide in specific occupational groups [7‐12]. Certainly, this may be the case for those working in jobs with ready access to potentially lethal suicide methods, for example, firearms in the case of police and members of the defence force [13], and medicinal drugs in the case of medical and veterinary professionals [8]. However, this explanation is insufficient to explain the elevated suicide rates in other occupations, such as construction workers [14, 15] who would have the same ability to access suicide methods as the general population.

There has been limited research on the relationship between access to means and suicide within the employed population, and those that have been conducted have been based in specific occupational settings (e.g., police, military, doctors and veterinarians) [8, 9, 16]. One of the few studies in this area [17], found that several occupations with access to means actually had lower suicide rates than the employed population. However, this research concluded that access to means played a role in influencing the choice of suicide method within specific occupational groups, rather than being related to an overall greater suicide risk.

The present research seeks to further progress research in this area. At a national level, we seek to assess whether occupations with greater access to potentially lethal suicide methods have elevated suicide rates compared to those that do not have such access. Related to this, we will also explore whether persons employed in occupations with access to specific suicide methods use these methods to end their life. We also seek to assess whether the relationship between occupational access to lethal means and suicide varies for females compared to males. We hypothesise gender as an effect modifier based on past research showing that males and females have different preferences for suicide method, with males tending to use more lethal and violent suicide methods than females [18, 19].

As can be seen in Table 2, over 30% of males in OAM with access to firearms and medicine and drugs used these methods to end their lives, while over 20% used carbon monoxide poisoning as compared to 7.2, 4.5, and 16.9% respectively of ONAM. However, hanging was the most prevalent method for the majority of suicides.

Table 3 indicates that around one-half (46%) of females employed in OAM used medications and 50% of OAM with access to carbon monoxide (albeit based on only 12 cases) used these methods to end their lives, compared to 18 and 14.7% of those employed in ONAM. A greater proportion of females in OAM used firearms (17.4%) compared to 2.4% of ONAM.

Table 4 describes age-standardised rates of suicide per 100,000 for OAM and ONAM by gender. As can be seen, those persons working in OAM had higher rates of suicide (both males and females). However, the magnitude of these differences varied by gender. Females employed in OAM had about twice the suicide rate of those employed in ONAM. Males employed in OAM, on the other hand, had only slightly elevated rates compared to ONAM.

We next conducted negative binomial regression testing whether the relationship between access to means and suicide varied by gender. The interaction term was significant (females in OAM had suicide rates 2.15 higher than males in ONAM, 95% CI 1.84 to 2.52, p < 0.001). The model with the interaction term was also provided significantly better fit compared to a model without the interaction term (LR χ² = 84.60, p < 0.001). On the basis of this, we stratified analyses by gender (Table 5). Table 5 shows that access to means was associated with a 3.02 RR of suicide for females (95% CI 2.60, 3.50, p < 0.001) and a 1.24 RR of suicide for males (95% CI 1.16, 1.33, p < 0.001), after controlling for year of death, occupational skill level and age (only the adjusted models are presented here for the sake of brevity). We would note that the male RR associated with some categories of occupational skill level was substantially higher than the RR associated with OAM.

Our results suggest that occupational access to means is an important risk factor for suicide. Those occupations characterized by greater access and familiarity with potentially lethal suicide methods had overall higher suicide rates than those without. Females with access were at particularly elevated risk compared to men. At the same time, we would acknowledge that, within the employed population, males and females tend to use similar suicide methods, with hanging being the most common method choice for most groups, followed by carbon monoxide poisoning.

In stating these results, our paper has a number of limitations that need to be taken into consideration. First, the underreporting of suicide (because of misclassification of cause of death) can be expected in NCIS, as in any official record of death. Second, occupation may have been misreported by police collecting information on occupation or in the coding process, which may have occurred despite independent coding by two researchers and the use of a structured approach to classification. Third, data on suicide methods is incomplete as initial data collection from NCIS did not include information on suicide methods. We have subsequently obtained this data for most cases but were unable to match suicide method for a small proportion of cases (approximately 10%). It is also worth note that NCIS is not an open data source. Fourth, there are several confounders that we were unable to control for due to restrictions in the data set and lack of exposure information (e.g., psychosocial job stressors, other stressors outside the workplace). Last, we have taken an overly inclusive approach to the classification of OAM. Thus we believe, if anything, we have over-represented occupations with access and knowledge about suicide methods. This would have the effect of biasing our results towards the null. We would also note that the descriptive analysis reported in Table 2 was based on a small number of cases for females, and thus we would caution against over-interpretation of results (e.g., occupations with access to carbon monoxide poisons being based on 12 cases only). The major strengths of our study include the use of the best available individual-level data on suicide and coverage across an entire national population over a twelve-year period.

Factors influencing an individual’s choice of suicide method are believed to include availability, knowledge about how to use a potential suicide method, and the overall perceived cultural acceptability of the method [28‐31]. For example, there was a substantial reduction in firearm suicides in Australia after implementation of national policy restricting access to guns [32]. At the same time, there has been an increase in hanging, which is possibly because this method is more difficult to regulate [32]. It is also possible that stigma about hanging and criminal behaviour have reduced followed the cessation of capital punishment [33]. These issues are important from a prevention perspective, considering hanging is the most commonly used method in Australia. A previous study conducted with a group of suicide survivors who used carbon monoxide poisoning indicated several other risk factors, including awareness of the method through media sources [34]. This aligns with other research which has shown that media portrayal of certain methods is an important factor in the decision to use that specific method [35]. It is important to note that most past research about choice of suicide means has been conducted at the general population level or within specific clinical samples. There has been comparatively limited research on the use of specific methods within occupational groups. Despite this, we would argue that similar explanations (e.g., ready access to potentially lethal suicide methods and greater knowledge about how to use these) are likely to comprise part of the explanation for our results. However, factors influencing choice of one method over another may be particularly pertinent as a risk within occupational settings compared to the general population, given that workers in OAM may use these potential suicide methods in their everyday work. For example, doctors have substantial knowledge about the toxicity of prescription medications and the amount that would be required to result in serious harm or death.

Choice of suicide method is also likely reflects a complex number of gender-specific factors. Previous research suggests that a preference for a quick and painless death and the use of method that entails no risk of disfigurement have been found to be important influences on method choice for females [28]. Our results suggest that, where they have ready access, employed females will use methods such as carbon monoxide poisoning and drug overdose rather than hanging. They may have chosen these methods because, as suggested above, they are associated with less pain and disfigurement than other methods. In saying this, a greater proportion of females who had occupational access to firearms used these to end their lives, despite the fact that this method is potentially associated with considerable disfigurement. This latter finding, and the fact there is a limited amount of research on gender differences in method choice within the employed population, highlights the need for future research on this topic.

OAM have been identified as being “at risk” in several previous studies [8, 9, 16, 17, 36]. The reasons for this have been linked to a variety of factors, including adverse working contexts, selection issues into the occupation, as well as access to lethal means. We would highlight the fact that we found elevated suicide rates among several occupational groups without access to lethal means, such as labourers (males) and machinery worker (females), aligning with our past research [37, 38]. We controlled for occupational skill level in the present analysis to account for potential confounding by socio-economic position as well as by other work-related factors that pattern by occupational status in Australia and internationally, such as exposure to poor psychosocial working conditions, or job stressors [39]. We acknowledge there is undoubtedly a number of complex factors underpinning suicide in the employed population aside from access to lethal means. Our results, nevertheless, suggest that means access appears to play a role for some occupational groups more than others.

The findings of this study suggest the importance of controlling access to lethal methods in occupations where these are readily available, and where there is evidence that these are particularly utilised by those who die by suicide. These prevention initiatives need to be combined with strategies that address other risk factors within these occupations, including adverse working conditions [40, 41]. Risk factors are likely to vary between different occupations. This highlights the importance of prevention initiatives being tailored for specific occupational groups. Our results also indicate that even within a specific OAM, there may be a need for different interventions for males and females.