Too Many Blood Donors - Response Bias in the Swiss Health Survey 2012

Background: Data on blood donor status obtained from general surveys and health interview surveys have been widely used. However, the integrity of data on self-reported blood donor status from surveys may be threatened by sampling and non-sampling error. Our study aimed to compare self-reported blood donors (including one-time as well as regular donors) from the Swiss Health Survey 2012 (SHS) with register-based blood donors recorded by blood establishments and evaluate the direction and magnitude of bias in the SHS. Methods: We compared population-weighted SHS point estimates of the number of blood donors with their corresponding 95% confidence intervals to the respective figures from blood donor registries (birth cohorts 1978-1993) and estimates of donors based on period donor tables derived from blood donor registries (birth cohorts 1920-1993). Results: In the birth cohorts 1978-1993, the SHS-predicted number of donors was 1.8 times higher than the respective number of donors based on registry data. Adjusting for foreign and naturalized Swiss nationals that immigrated after their 18th birthday, the SHS overall predicted number of donors was 1.6 times higher. Similarly, SHS estimates for the 1920-1993 birth cohorts were 2.4 and 2.1 times higher as compared to register based estimates. Generally, the differences between SHS and register-based donors were more pronounced in men than in women. Conclusion: Self-reported blood donor status in the SHS is biased. Estimates of blood donors are substantially higher than respective estimates based on blood donor registries. (C) 2016 S. Karger GmbH, Freiburg