Kyaga et al found an intriguing association between creativity and severe mental disorder. ^{Reference Kyaga, Lichtenstein, Boman, Hultman, Långström and Landén1} The study draws its strength from a large sample size. However, the retrospective data collection methodology brings with it certain inherent limitations, which the authors have acknowledged, and causal links have been hinted at in the discussion. We would like to bring to attention two issues. First, the role of potential confounders in selection of occupation has not been taken into consideration. The type of occupation one pursues is governed by multiple factors in addition to personnel interest, including educational qualification, opportunity, awareness, location of the job, financial remuneration, familial and other social commitments. ^{Reference Llena-Nozala, Lindebooma and Portrait2} Many of these variables are likely to be affected by the psychiatric illness, although they are modifiable by many independent factors as well. Hence the occupation choices of both individuals with mental illness and their children (and other family members) are likely to be affected by many variables which need to be taken into consideration when interpreting Kyaga et al's findings.

Another relevant issue for consideration is the way occupation is defined in their study. The definition of occupation used in (mental) health studies has been criticised for being too restrictive. ^{Reference Townsend3} National descriptions of occupation tend to classify only those occupations that have economic relevance. ⁴ Such an approach is likely to miss someone employed as a labourer who paints during their leisure time or to miss certain population groups. For example, in many settings the majority of women are likely to be the primary caregiver (i.e. housewife) and not formally ‘employed’. Future studies could be strengthened by the use of a more comprehensive and inclusive definition of occupation.