Confounders in studies of suicide by occupation.

Milner et al 1 make a commendable analysis of the effect of occupation on suicide risk, drawing on an imperfect field of research. Their work does not, however, allow for the effects of the differential demographic profiles (particularly gender) of those employed in each occupational category. This is particularly important when, in the UK in 2011, there was an 18.2 per 100 000 suicide rate among males compared with a 5.6 per 100 000 rate among females. 2 If an occupation were almost entirely filled with men aged 30 to 44, with their suicide rate of 22.2 per 100 000, 2 it would not be surprising that its rate of suicides was significantly higher against all reference groups. The United Nations Statistics Division figures show a striking difference in the gender balance of the ISCO categories in the UK from 2009 census data. In ISCO-9, with its high suicide rate ratio of 1.8 (95% CI 1.5–2.3) in Milner et al, 60.4% were male. 3 In ISCO-4, with its rate ratio of 0.8 (95% CI 0.6–0.9), only 22.5% are male. 3 This relationship does not correlate across the ISCO categories, but it is enough of a confounder to be of interest. Despite the advantages of the rate ratio, it does not correct for gender, whereas the proportionate mortality ratio does. It may be premature to dismiss its utility, until we have better data-sets that are more amenable to correction for demographic factors. The paper by Milner et al 1 is an excellent meta-analysis, but there are a few lacunae in the interpretation of results. First, suicide rates differ from country to country. Second, causes differ in age groups and in different countries. For example, in India a 2012 analysis 2 revealed that the self-employed category accounted for 38.7% of victims, of whom 11.4% worked in farming/agriculture, 4.7% in business and 2.9% in professional occupations. Students and unemployed victims accounted for 5.5% and 7.4% respectively; 18% of those who died were housewives. Third, educational status also affects suicide rates. In India, the majority of suicides (46%) are by people with a middle-or primary-level education. These categories have been relatively constant for a long time. 2,3 Fourth, the causes differ from culture to culture: in India, family problems constitute the majority (26%), followed by illness (21%). So, drawing and applying conclusions from meta-analysis to different cultures and countries is difficult. Author's reply: We …