A target trial emulation to estimate the effect of blood donation on mortality among blood donors in Australia

Background and Objectives The healthy donor effect (HDE) is a bias evident when comparing the effect of blood donation on health, often leading to conclusions such as ‘donors with a high frequency of donation have better health outcomes than non‐donors or donors with a lower frequency of donation’. To overcome this bias, we used a target trial emulation method while examining the relationship between blood donation and mortality among blood donors in Australia. Materials and Methods We emulated 60 target trials from July 2006 to June 2011 using the Sax Institute's 45 and Up Study data, which was linked with other electronic health databases, including blood donation data in New South Wales, Australia. We conducted observational analogues of intention‐to‐treat (ITT) analyses comparing donors with non‐donors, adjusting for variables that impact mortality. Hazard ratios (HRs) were approximated by the pooled logistic regression model. Results The 60 trials generated 263,300 person‐trials with 121,967 unique participants included in the analysis. The unadjusted ITT mortality HR (95% confidence interval [CI]) between donors and non‐donors was 0.57 (95% CI: 0.42–0.77). The HR was 0.70 (95% CI: 0.52–0.95) and 1.0 (95% CI: 0.73–1.35) after adjusting for age and sex, and then all the baseline covariates, respectively. Conclusion Blood donation did not significantly affect mortality during the follow‐up when appropriate methods to mitigate the HDE were used, although blood donors were observed to have a lower crude mortality rate than non‐donors.