Reasons for non-response in observational pharmacogenetic research

PURPOSE: In epidemiological studies, non-response may introduce bias and limit generalizability. In genetic pharmacoepidemiological research, collection of DNA might be a major reason for non-response. We determined reasons for non-response and compared characteristics of non-responders and responders in a pharmacogenetic case-control study.

METHODS: Myocardial infarction (MI) cases and controls, who were antihypertensive drug users, were recruited through community pharmacies that participate in the Pharmaco-Morbidity-Record-Linkage-System (PHARMO). The PHARMO database comprises drug dispensing histories of about 2 000 000 subjects from a representative sample of Dutch community pharmacies linked to the national registry of hospital discharges. Independent samples t-test and ANOVA-statistics were used to analyse the differences in continuous variables between responders and non-responders. chi2 statistics and logistic regression were used to compare categorical variables.

RESULTS: We approached 1871 cases and 14 102 controls of whom 794 MI cases (42.4%) and 4997 controls (35.4%) responded. We could not approach 2194 patients of whom 63.1% had died and 31.2% moved to another pharmacy. Main reasons for non-response were health problems or hospital stays (16.2%, OR 1.47; 95%CI: 1.00-2.16). Other reasons were old age or dementia (16.9%, OR 1.82; 95%CI: 1.24-2.65). Only a small percentage (1.1%, OR 1.43; 95%CI: 0.41-5.03) mentioned DNA sampling as a reason. About 30% of the non-responders did not give a reason. Women were statistically significantly (p < 0.0005) less willing to participate than men (38.8% versus 31.3%). An association with age was also found (mean age 64.6 versus 66.5 yrs) (p < 0.0005).

CONCLUSION: In a pharmacogenetic case-control study fear for genetic screening was not a major reported reason for non-response. Females were less willing to participate than males.