Males, and especially males aged between 40 and 70, are at increased risk of suicide. It is therefore of the utmost importance to gain better insight into the risk factors among the group of men between 40 and 70.
Using data from Statistics Netherlands, which has extensive demographic, financial, health insurance, and mortality data, we looked at all suicides of men aged 40-70 during the years 2012-2020. We included data on age, migration background, personal income, household income, household net wealth, marital status, household position, employment status, unemployment benefits, disability benefits, education level, region, physical healthcare costs, and whether they were treated in mental healthcare.
We used a machine learning model introduced in Berkelmans et al. to not only identify individual risk factors, but also interactions of risk factors that together increase the risk of suicide even further.
We found groups where the suicide rate was as high as 289 per 100,000 compared to the 20 per 100,000 among the general population of men between 40 and 70. Based on single risk factors, groups were the suicide rate exceeded 40 per 100,000 were those with a household income in the bottom 25% (41 per 100,000), those with high healthcare costs (up to50 per 100,000), those on long term unemployment or disability benefits (51 and 56 per 100,000 respectively), those living alone (52 per 100,000), and those receiving mental healthcare (115 per 100,000).
As for interacting factors, we found many interactions related to income and mental healthcare. These included both instances where the risk was lower than if the risk factors acted independently as well as instances where the risk was higher. For instance, both low personal income and low household net wealth increased the risk of suicide, while the compounded risk was lower compared to when they were to act independently. On the other hand, we found that though having an age between 60 and 70 appeared protective within men of middle age in general, it was a risk factor among those receiving mental healthcare.

We were limited to existing demographic data and could therefore not include psychological factors, but nonetheless managed to find high risk groups. The results from this study have practical implications for targeting campaigns and interventions in a more effective manner.