Previous reports suggest that blood donors have a lower mortality risk, which may partially reflect the "healthy donor effect" (HDE). HDE arises in donors due to selection bias and confounding if not appropriately addressed. Using the Sax Institute's 45 and Up Study data linked with blood donation history, we used a "5-year exposure window" method to select donors into regular high-frequency whole blood (WB)donors (at least two donations per exposure year) and low-frequency donors (remaining donors) with an active donation career of 5 years. To further reduce the confounding, we used statistical approaches like the inverse probability weighted (IPW) marginal structural model and the doubly robust targeted minimum loss-based estimator (TMLE), which incorporated machine learning algorithms and time-varying analyses. We selected 4750 (64.7%) low-frequency and 2588 (35.3%) high-frequency donors in the analyses. A total of 69 (1.5%) from the low-frequency and 45 (1.7%) donors from the regular high-frequency group died during the 7-year follow-up period. We did not find any statistically significant association between regular high-frequency blood donation and mortality (IPW RR = 0.98 95% CI 0.68, 1.28). TMLE model also showed similar results to IPW (RR = 0.97 95% CI 0.80, 1.16). Time-varying TMLE did not find any significant association between high-frequency donation and all-cause mortality either (RR = 0.98 95% 0.74, 1.29). We did not find a significant association between regular high-frequency WB donation and all-cause mortality when appropriate methods were employed to minimize the HDE.
Read full abstract