Personalized medicine comes to a blood center near you

While personalized medicine has transformed healthcare over the last two decades, the blood donation process has remained basically unchanged. In this issue of TRANSFUSION, the article by Kim et al.1 shows how discrete event simulation (DES) modeling2 can be used to select the right donors and minimize deferrals, thus ensuring that the donor's experience is optimized to encourage repeat donation. Three important developments in blood banking preceded this study. First, until the last decade, it was very difficult to extract a large amount of data from donor records, as most were on paper. This made the process arduous and feasible for auditing only a random sample of donors to look for trends that might affect both donor eligibility status and rate of return. With the Food and Drug Administration's requirement for use of 510 (k) cleared Blood Establishment Computer Systems,3 large amounts of donor data are now available.4 The second development was the advent of big data, in which extremely large data sets may be analyzed computationally to reveal patterns, trends, and associations, especially relating to human behavior and interactions.5 Management of the blood supply is inherently challenging, with a complexity that is likely to increase in the years ahead, as donation rates wane yet populations age and may require more transfusions.6, 7 The planning of future collection activities should be based on appropriate forecasts of blood demand, which is influenced by several factors including geographical region, type of hospital and services provided, patient population, and hospital activity level. Predictive models could prove instrumental in optimizing donor and inventory management. For instance, predictive models can expand on initiatives that have become routine in-patient blood management.8-10 In China, a group employed a series of algorithm-based models to guide preoperative red blood cell (RBC) transfusion.11 In general, their models proved to be superior to clinical decision-making when predicting intra-operative RBC use. In England, a group described the use of time-series methods, drawing on historical data to predict RBC demand up to 24 weeks ahead.12 This becomes important to keep pace with changing patterns of usage.13 In Canada, a modeling framework was devised to optimize regional blood supply chains.14 DES modeling arose from a need to find an optimal method for extrapolating outcomes from clinical trials.2 DES comprises a type of microsimulation method that can be applied to healthcare decision problems. Individual instantiations of a system are generated by using a random process to draw a large number of times from probability distributions. (This approach is also known as Monte Carlo or probabilistic simulation).2 In DES modeling, sequences of events are simulated by drawing directly from probability distributions of event times.2 The third development was that mounting evidence indicates frequent donors require iron supplementation to prevent significant iron depletion, which may have deleterious health consequences.15 There were a number of studies performed to support this finding. HEmoglobin Iron and Recovery Study (HEIRS) found that only donors on iron supplementation replaced donated iron.16 Eight weeks of iron supplementation provided nearly all the measured improvement in total body iron. The Strategies To Reduce Iron DEficiency (STRIDE) study demonstrated that 19 or 38 mg of daily iron supplementation and iron status information after ferritin testing was effective in mitigating post-donation iron deficiency.17 The Comparison of the History of Donation and Iron LeveLs in Teen Blood Donors (CHILL) study found that being a donor aged 16–18 years old is an independent risk factor for iron deficiency in blood donors of any donation frequency.18 An Association for the Advancement of Blood and Biotherapies Association Bulletin at the time recommended educational materials, iron supplementation, and consideration of increasing the interdonation interval as mitigation strategies.15 None of these are optimal. Donors often do not read educational materials. Many US blood centers were not comfortable prescribing or even recommending iron supplementation as they did not have good mechanisms to monitor compliance and potential side effects. There were medical and legal concerns with blood centers practicing outside their normal scope of operations with the added liability of patient care. Finally, the US blood supply is tenuous at best, to the point that in 2020 a report from the US Department of Health and Human Services on the Adequacy of the National Blood Supply was submitted to Congress.19 Lengthening the inter-donation interval is simply not feasible to maintain an adequate blood supply at this time. Kim et al. used a simulation model to assess four different invitation strategies together with the current donation strategy.1 Under the current policy in the UK (“strategy A"), donors are invited to return a minimum of 12 weeks (men) or 16 weeks (women) after each donation, up to four times per year. Three alternative strategies (“strategies B, C, D") based on post-donation hemoglobin testing were modeled and compared with current practice/strategy A. While often used with plateletpheresis donors to determine if their platelet count qualifies them for their next donation,20 post-donation testing is uncommon in the whole blood donation arena in most countries. In the strategies evaluated by Kim and colleagues, a previous single hemoglobin measurement was used to calculate the probability of a donor's hemoglobin being above the UK threshold level (13.5 g/dL for men, 12.5 g/dL for women) over time, thus allowing the investigators to develop a personalized inter-donation interval. The three modeled strategies differed in the level of certainty that a returning donor would have acceptable hemoglobin and/or the minimum post-donation interval. The final strategy (E) took somewhat of a belt-and-suspenders approach, as it also used post-donation testing to inform inter-donation intervals with medium or high certainty of being over the hemoglobin threshold, combined with limited on-site testing using hemoglobinometry (HemoCue AB, Ängelholm, Sweden) for donors with medium certainty. The earlier Comparison of Four Methods to Measure Hemoglobin Concentrations in Whole Blood Donors (COMPARE)21 study formed the groundwork for this study. Data from 9360 female and 7948 male donors in COMPARE were used to inform the estimation of parameters in the DES model. Fortunately for the present study, 8441 women (90%) and 7421 men (93%) in COMPARE returned to donate within a year. Of these, 78% of women and 80% of men had known hemoglobin levels at the second visit. These data were used to model the donors' post-donation hemoglobin recovery over time. For the current study, adverse events were defined as deferrals for low hemoglobin levels and “inappropriate bleeds”—blood collections that should not have been performed but were anyway due to falsely high point-of-care hemoglobin test result. Among the five strategies evaluated, donations per adverse event were maximized under strategy C (high [90%] certainty of hemoglobin above the threshold, with a minimum 16-week return for women/12 weeks for men). Strategy C yielded an estimated 14.8 donations/adverse events in women (95% uncertainty interval [UI] 11.6, 19.2) and 26.9 donations/adverse events in men (95% UI 20.8, 42.6). Strategy C also minimized costs in women at £27.37/donation. For both men and women, strategy D (high [90%] certainty of hemoglobin above threshold with minimum 12-week return for women/8 weeks for men) maximized donations while also minimizing costs in women at £27.37/donation. In men, while costs were minimized at £26.93/donation under the current strategy (A), this was least favorable in terms of adverse events, at 7.14 donations per adverse event (95% UI 6.1, 8.5). The current strategy also had the least favorable donations per adverse event in women. The authors concluded that a strategy of re-inviting donors with a ≥ 90% probability of future hemoglobin level (based on the last previous measurement) being above the donation threshold (strategies C and D) yielded the highest number of donations per adverse event in both sexes and also had the lowest cost per donation in women. More importantly, they noted that this approach would be predicted to prevent 111,000 low hemoglobin deferrals and 128,000 inappropriate bleeds annually in the UK. Adopting strategy C/D would represent a paradigm shift in that only post-donation testing for hemoglobin would be used, thus improving operational efficiency and potentially donor satisfaction. Further, since adverse events disproportionately occur in women, strategy C/D also goes some way to redressing this inequality as inappropriate bleeds are minimized, and low hemoglobin deferrals are eliminated. Of note, post-donation testing of whole blood donors is already in place in several European counties, including France, Belgium, and Denmark.22 There were limitations of the DES model. In the internal validation, the number of inappropriate bleeds was underestimated in both women (114 v. 125, −9%) and men (51 v. 68, −25%). Low hemoglobin deferrals were also under-predicted, particularly in men (21 v. 27, −22%). These discrepancies arose in part because the underlying model for hemoglobin fits less well at the most extreme values of hemoglobin at recall and because the threshold for donation in men is further into the tail of the hemoglobin distribution than in women. This could be an issue for those donors with hemoglobin levels near the required thresholds, potentially allowing for collections that should not occur and potentially placing the donor at risk for subclinical anemia. On a positive note, the results showed pronounced benefits of personalized inter-donation intervals in women. For example, while the vast majority (98%) of men are re-invited at the minimum time under all strategies, in women, 12%–16% (depending on strategy) are re-invited later, including 7%–10% at 24+ weeks post-donation. This reflects those women in whom a low hemoglobin deferral is averted under a personalized strategy. This is very good news for blood centers that want to avoid subclinical anemia by over-collecting this population. As donors may have concerns about how frequent donation affects their health, evidence that allows for a better understanding of how post-donation testing can identify the optimal interval window to protect donor health would be valuable to include in donor communications and policy. The importance of moving toward a personalized blood donation qualification process cannot be overstated. A 2021 study from the Austrian Red Cross used data from more than 123,000 whole blood donors over a period of 5.5 years.23 The results confirmed that temporary deferrals hurt future donation behavior. The effect is strongest with donors early in their donation history.23 Similarly, a large-scale registry study of whole blood donors in the Netherlands conducted between 2013 and 2015 that included 343,825 donors found that first-time donors deferred for low hemoglobin were less likely to return and slower to return than other donors.24 In conclusion, the study by Kim et al. nicely demonstrates how DES modeling could be used to adjust standard deferral lengths, often set many years ago by regulatory agencies, challenging the norm of pre-donation hemoglobin testing. Personalized medicine is coming to a blood center near you. There is a saying in blood banking that to get a donor to present, you have to ask them and make it convenient. I would add that the donor needs to have a good blood donation experience to want to return. If blood centers can more accurately predict when a donor should present based on previous hemoglobin results, this would reduce the opportunity for deferral. A donor who presents is eligible, and is able to help the community is a happy donor!