Blood donors' long-term health : implications for transfusion safety

Abstract

Continuous attention to transfusion safety through improvement of disease screening and donor selection has succeeded in reducing the risks of transfusion transmitted disease to practically immeasurable levels. Despite this progress, surprisingly little is known about disease occurrence among blood donors and whether there are possible long-term effects of repeated whole-blood or apheresis donation. Several investigations have addressed possible inadvertent non-infectious health effects of blood transfusions per se, but to date there is no conclusive evidence regarding whether blood transfusions from donors with preclinical cancer can result in cancer development in the recipient. To address such research questions, we assembled the Scandinavian Donations and Transfusions (SCANDAT) database, containing detailed data on blood donors and their donations, with transfusions and transfusion recipients encompassing more than three decades. All studies described in this thesis were based on this database.

In the first study, we describe the creation of the SCANDAT database and its contents, as well as the results from our analyses of its quality. In total, the database contains 1,134,290 donors with 15,091,280 donations, and 1,311,079 recipients with 11,693,844 recorded transfusions. Although direct evaluations of data quality were not possible, we assessed quality by various indirect methods and judged the database to be of a sufficiently high standard for epidemiological research investigations.

In the second study, we compared the mortality and cancer incidence in a cohort of 1,110,329 blood donors to the rates in background population. The relative risks were expressed as standardized mortality ratios (SMR) and standardized incidence ratios (SIR). Blood donors had an overall mortality 30% lower (99% confidence interval [CI] 29%-31%) and cancer incidence 4% lower (99% CI 2%-5%) than the background population. Furthermore, blood donors recruited in more recent years exhibited a lower relative mortality than those who started earlier.

Within the cohort of blood donors from the second study, the third study was conducted using a nested case-control design. Relative risks of cancer in relation to number of donations made, or iron loss endured, was estimated with conditional logistic regression. We identified a total of 10,866 donors who were diagnosed with a malignancy between their first recorded blood donation and study termination and selected 107,140 individually matched controls. We found no clear association between number of donations and risk of cancer overall. The risk of non-Hodgkin lymphoma was increased among frequent plasma donors, the odds ratio among donors with ≥20 plasma donations relative to those with <3 donations in the period from 3-12 years before diagnosis, was 2.00 (95% CI 1.15-3.46). Further, among male donors only, we found that the risk for selected cancers decreased with increasing iron loss in the period 3-7 years before diagnosis of the case (p<0.001).

Of the 354,094 transfusion recipients eligible for analysis in the fourth study, 12,012 (3.4%) were exposed to blood products from donors who developed cancer within 5 years. The relative risk of cancer overall comparing recipients of blood from precancerous donors to recipients of blood from non-cancerous donors was 1.00 (95% CI, 0.94-1.07). We also did not find any excess risk when we considered the site and severity of the cancer in the donor, nor when we assessed site-specific cancer risks among the recipients.