Record linkage research and informed consent: who consents?

BACKGROUND Linking computerized health insurance records with routinely collected survey data is becoming increasingly popular in health services research. However, if consent is not universal, the requirement of written informed consent may introduce a number of research biases. The participants of a national health survey in Taiwan were asked to have their questionnaire results linked to their national health insurance records. This study compares those who consented with those who refused. METHODS A national representative sample (n = 14,611 adults) of the general adult population aged 20 years or older who participated in the Taiwan National Health Interview Survey (NHIS) and who provided complete survey information were used in this study. At the end of the survey, the respondents were asked if they would give permission to access their National Health Insurance records. Information given by the interviewees in the survey was used to analyze who was more likely to consent to linkage and who wasn't. RESULTS Of the 14,611 NHIS participants, 12,911 (88%) gave consent, and 1,700 (12%) denied consent. The elderly, the illiterate, those with a lower income, and the suburban area residents were significantly more likely to deny consent. The aborigines were significantly less likely to refuse. No discrepancy in gender and self-reported health was found between individuals who consented and those who refused. CONCLUSION This study is the first population-based study in assessing the consent pattern in a general Asian population. Consistent with people in Western societies, in Taiwan, a typical Asian society, a high percentage of adults gave consent for their health insurance records and questionnaire results to be linked. Consenters differed significantly from non-consenters in important aspects such as age, ethnicity, and educational background. Consequently, having a high consent rate (88%) may not fully eliminate the possibility of selection bias. Researchers should take this source of bias into consideration in their study design and investigate any potential impact of this source of bias on their results.