Unexpected lower prevalence of depression in patients with diabetes: selection bias in a waiting room population.

Background: We initially studied the association between diabetes and depression in general practice attendees. Subsequently, we investigated whether our findings were influenced by selection bias.

Methods: Unexpectedly, the prevalence of depression was lower in diabetes patients (4%) than in subjects without diabetes (14%). To explore the possibility of selection bias, we first described the selection process from the general population to our study population and provided prevalence data of diabetes and depression in the different populations based on literature and our data. Second, we performed a sensitivity analysis and described possible reasons for selection. Third, we studied the association between other chronic diseases and depression.

Results: These analyses suggested that the lower prevalence of depression in diabetes patients was due to selection bias. Visiting the general practitioner for a control visit seemed to play a role in this selection process.

Conclusions: This study illustrated the potential for selection bias in a waiting room population. The degree of bias depended on the exposure under study.