Investigating the effect of demographic and socioeconomic variables on misclassification by the SRQ-20 compared with a psychiatric interview

Background Previous research among primary care attenders in Latin America has noted that females and those with less education may over-report psychiatric complaints on the SRQ-20, compared with responses to a standardized psychiatric interview administered by a clinician. In this paper, the association between demographic and socioeconomic variables and misclassification by the SRQ-20 were investigated and the size of misclassification estimated in a populationbased survey. Method A cross-sectional survey of a random sample of private households included 683 adults aged 15 years and over living in Olinda, Recife Metropolitan Region, Pernambuco, Brazil. The SRQ-20 results were compared with an interview administered by a psychiatrist. The effect of demographic and socioeconomic variables on misclassification by the SRQ-20 was assessed by calculating the odds ratio (OR) for being a case on the SRQ-20 after adjustment for being a case on the psychiatric interview. Logistic regression was used to investigate the size of misclassification, adjusting the association between common mental disorders, defined by the SRQ-20,and different variables for the psychiatric interview results. Results In the univariate analysis, females, the elderly, the less educated, manual workers, housewives and migrants did tend to over-report complaints in the absence of symptoms.However, the apparent influence of age, education, occupation and migration on misclassification by the SRQ-20 was markedly reduced and became statistically non-significant after adjustment for sex and for the other variables in the table. In contrast, the gender effect was not altered after adjustment. Conclusions Our results suggest that misclassification on the SRQ-20 was mostly related to being female, though this did not entirely explain the increased prevalence of CMD in women in the sample. Further research is needed to understand why different ways of measuring CMD can lead to different results