The effect of follow-up on limiting non-participation bias in genetic epidemiologic investigations.

The use of a comprehensive follow-up strategy to limit non-participation bias was evaluated in a population-based case-control study of orofacial clefts. Birth parents were requested to provide exposure data, and index children and parents were asked to provide blood specimens. Follow-up included telephone or postal reminders every two weeks for up to three months. Consent to participate was received from 281 (76.6%) case mothers and 246 (72.4%) case fathers. The corresponding totals for controls were 279 (54.7%) and 245 (49.8%). Evaluation of participation rates by intensity of follow-up showed that 23% of case and 18% of control families consented without reminders (first stage); 81% of cases and 83% of controls agreed following one or two reminders (second stage); and the remainder of participants consented following three or more reminders (final stage). Cumulative distributions of sociodemographic characteristics differed little between second and final stage participants. Odds ratios for maternal multivitamin use were similar between second and final stage participants, whereas those for maternal and paternal smoking tended to decline. Although follow-up measures were necessary to enroll most families, use of more than two reminders did not appear to increase the representativeness of the sample; however, termination of recruitment after only two reminders would have led to different conclusions. Future studies require data collection protocols that encourage participation from all population subgroups, and one alternative is presented.