ABSTRACT
PURPOSE: To study whether the difference in the demographic characteristics of participants and non participants could result in biased prevalence estimates and associations. AIM: To compare the non-participant and participant characteristics, and to ascertain if non-response bias is present in the rural population of the Chennai Glaucoma Study (CGS). METHODS: Rural participants and non-participants were compared with regard to socio-demographic variables (age, gender, religion, mother tongue, literacy and employment). RESULTS: 4800 subjects aged 40 years or over were enumerated, 82% (3934: 45% male and 55% female) responded. Gender did not influence participation (adjusted OR-1.11, CI: .91-1.36). Subjects in the 70-79 year age group were more likely to respond (OR-1.76; CI-1.31-2.38). Hindus had a higher participation rate than Christians or Muslims (adjusted OR-2.63, CI: 1.80-3.84). The other predictors of participation were illiteracy (adjusted OR-1.44, CI: 1.22-1.70), unemployment (OR-1.28, CI: 1.04-1.58), place of residence (main villages) (OR-6.66, 95% CI: 4.6-9.64). CONCLUSION: Based on our study findings, it does not seem likely that participation bias will affect the study results.
