Mahsa M Yazdy1, Martha M Werler2. 1. Slone Epidemiology Center at Boston University, Boston, MA. Electronic address: mahsa.yazdy@channing.harvard.edu. 2. Slone Epidemiology Center at Boston University, Boston, MA.
Abstract
PURPOSE: To evaluate whether selection bias might be introduced because of differential self-selection into an internet-based case-control study. METHODS: We compared sociodemographic factors and behavioral factors for cases and controls who opted for their first entrée into the study be via the internet (Web group) versus the telephone (Phone group), using data from a study on clubfoot. RESULTS: Overall, 200 of 1,825 mothers were in the Web group and were more likely to be non-Hispanic White, nonsmokers, more educated, wealthier, and took ≥7 medications/vitamins in pregnancy. These differences were greater for case mothers, introducing the potential for selection bias. Maternal smoking is an established risk factor for clubfoot and was confirmed in the Phone group (OR = 1.61) but was not observed in the Web group (OR = 1.15). CONCLUSION: Our results suggest there is potential for selection bias if enrollment is solely internet based for a case-control study.
PURPOSE: To evaluate whether selection bias might be introduced because of differential self-selection into an internet-based case-control study. METHODS: We compared sociodemographic factors and behavioral factors for cases and controls who opted for their first entrée into the study be via the internet (Web group) versus the telephone (Phone group), using data from a study on clubfoot. RESULTS: Overall, 200 of 1,825 mothers were in the Web group and were more likely to be non-Hispanic White, nonsmokers, more educated, wealthier, and took ≥7 medications/vitamins in pregnancy. These differences were greater for case mothers, introducing the potential for selection bias. Maternal smoking is an established risk factor for clubfoot and was confirmed in the Phone group (OR = 1.61) but was not observed in the Web group (OR = 1.15). CONCLUSION: Our results suggest there is potential for selection bias if enrollment is solely internet based for a case-control study.
Authors: Samantha E Parker; Cara T Mai; Matthew J Strickland; Richard S Olney; Russel Rickard; Lisa Marengo; Ying Wang; S Shahrukh Hashmi; Robert E Meyer Journal: Birth Defects Res A Clin Mol Teratol Date: 2009-11