Despite over 20 million unrelated donors being listed worldwide, donor attrition at the confirmatory typing (CT) stage of donor acquisition is a key source of delay. Anthony Nolan undertook a study of CT requests from 2010 to 2011 to identify factors associated with attrition. Of 7541 CT requests, 38.2% were cancelled for donor reasons. Of these, 19.4% were personal, 34.1% medical, 36% no contact, 7.9% emigrated and 2.6% others. African (odds ratio (OR) 2.78, P<0.001), African-Caribbean (OR 3.07, P<0.001), Asian (OR 2.65, P<0.001), Jewish (OR 1.54, P=0.009) and Mediterranean (OR=2.38, P<0.001) donors were more likely not to be available compared to Caucasian donors. Female donors were also more likely not to be available (OR=1.32, P<0.001): primarily due to pregnancy. Older donors were less likely to be available in univariate analysis, but this association was not significant after controlling for other factors. Blood donors and those recruited within the past five years had lower rates of attrition. Accumulation of additional attrition-associated characteristics for a given donor was associated with progressively greater odds of attrition (OR 1.99, 2.52, 3.4 and 5.53, respectively, for 1, 2, 3 and 4 risk factors, P<0.001). Donor registries must develop evidence-driven strategies to recruit and retain the most reliable donors.
The use of unrelated donors for haematopoietic progenitor cell (HPC) transplantation has progressed since 1973, which saw the first unrelated donor transplant1 and the foundation of the world’s first BM registry, now known as Anthony Nolan. By 2010, 6309 allografts were performed using unrelated adult donors in Europe alone,2 and almost 80% of the patients of white Northern European (WNE) descent are able to find an appropriately matched unrelated donor on international registries.3 This achievement is made possible through the establishment of large donor registries and global collaborations such as Bone Marrow Donors Worldwide (BMDW)4 and the European Marrow Donor Information System (EMDIS). It is now possible for transplant centers to search online from a panel of over 20 million unrelated donors. Although the outlook for non-white patients remains considerably less positive, enthusiastic recruitment of adult and cord blood donors within ethnic minorities, as well as the foundation of new donor registries in Asia, South America and Africa, are strategies aimed at improving these odds.5
It is known that the time taken to get a patient to transplant influences outcome. In part this relates to disease stage, but a number of studies show an independent effect of time to transplant on transplant outcome measures.6, 7 As a result, registries have increasingly focussed on other ways to improve the time taken to provide a donation.
One major point of delay is the confirmatory typing (CT) stage. Here, registries must make contact with potentially matched donors, some of whom may have been on the register for several decades without regular contact, establish their willingness and fitness to donate, and arrange for further blood samples for confirmatory HLA typing and testing for infectious disease markers. Donor attrition at this stage is a considerable problem: it may take several weeks to trace a donor with obsolete contact details, donors may be medically ineligible to donate8 or they may have personal reasons as to why they do not wish to donate, often related to valid lifestyle issues such as family or travel. This impacts both the donor registries and patients: the registry must bear the financial and reputational impact of having donors typed and listed on the register who will never be available for donation; patients risk a prolonged pretransplant period that may necessitate extra chemotherapy in order to maintain remission. In the first half of 2011, the National Marrow Donor Program (NMDP), the world’s largest unrelated donor registry with almost 10 million donors, estimated that 47% of their donors were not available at ‘first activation’.9 Although other registries have been less forthcoming about attrition rates, it is likely that this problem is widespread.
In order to help improve their attrition rate, the NMDP have attempted to establish which donor characteristics might be associated with donor attrition. Switzer et al.10 published in 1999 on comparative cohorts of donors who decide to, and not to, go ahead at the CT stage. A number of demographic and psychological characteristics were shown to be associated with choosing not to progress at the CT stage. As a natural extension to this study, Switzer recently published on the impact of race and ethnicity, on the decision not to donate, identifying multiple cultural, psychosocial and donation-related factors associated both with donor attrition and ethnicity.3
Although such studies are useful in guiding donor retention strategy, they relate only to that proportion of donors who choose not to donate for personal reasons, and do not address the problems of medical deferral, donor emigration, donor non-contact and other causes. In addition, cultural and racial influences may vary considerably between countries, such that the relevance of these study findings cannot be assumed outside the United States. No other donor register or donor centre has published on donor attrition, although DKMS (Deutsche Knochenmarkspenderdatei) in Germany presented an abstract showing an attrition rate of 18.3% in 53 551 donors requested for CT between 2008 and 2010. Although the causes of donor attrition were not detailed in the abstract, they showed an association between attrition and younger donor age, female sex and non-German nationality.11
Anthony Nolan has undertaken a study of donor attrition at the CT stage. The primary objective was to quantify and characterise that proportion of donors not proceeding at the CT stage and identify donor characteristics that may be associated with increased donor attrition. A further objective was to assess differences in the mode of attrition within certain donor groups, such as ethnic minorities. Ultimately, the findings will inform strategies aimed at reducing donor attrition in the future, and identify areas where further research might be appropriate.
Materials and methods
Data regarding CT-stage requests requiring fresh blood samples from registered donors are prospectively collected at Anthony Nolan. We examined records for a 2-year time period, from 2010 to 2011. Donors were asked for blood samples for one of the three reasons: CT, where high-resolution (allele-level) typing is undertaken at all the key HLA loci deemed essential for optimal donor–recipient matching; extended typing, where clarification of one or more individual HLA loci is required; and DR-typing, where only HLA-DR type is ascertained. Episodes where donors were requested for CT who had previously provided a sample for extended typing or DR typing for the same patient were excluded from the study.
On receiving such a request from either a transplant centre or other registry, Anthony Nolan would immediately attempt to contact the donor through a number of methods, using information stored in the donor database, including telephone, email and postal address. For donors who responded, the implications of the CT-stage process were explained and they were asked if they were still willing to donate. A short medical questionnaire was performed over the telephone, followed by a more detailed medical questionnaire sent by post.
Requests where the donor did not respond were put forward to ‘trace’, where other methods were used to attempt to locate the donor, such as contacting the donor’s general practitioner or searching social media. For this study period, Anthony Nolan did not have access to a national centralized database that might otherwise have provided contemporary donor contact details.
The outcome of each request was defined as completed (blood sample received by Anthony Nolan or shipped to requesting registry/transplant centre), cancelled by transplant centre or cancelled for donor reasons. For those requests cancelled for donor reasons, the cause of cancellation was further defined as cancelled due to: personal reasons (and whether the donor was deleted or kept on the register); medical reasons; failure to contact donor; donor overseas or emigrated, and others. Medical cancellations were further subdivided according to the underlying medical condition.
For each request, donor characteristics including age, sex, ethnicity, prior blood donation and duration of time on the register were documented.
For the purpose of statistical analysis, continuous variables were categorized. These included donor age (18–30, 31–45 and 46–60 years) and duration on the register (less than 1–5, 6–10, 11–15 and 16+ years). When calculating odds ratios (OR), those requests cancelled by the transplant centre were excluded, as it was unknown how many of these donors would have been available to donate. Univariate analyses of donor characteristics influencing donor attrition were performed using a χ2-test for binomial variables and logistic regression for multinomial categorical variables (for example, ethnicity). Multivariate analysis was performed using binary logistic regression. Those characteristics with a statistically significant association with donor attrition (P<0.05) were entered into the multivariate analysis; all variables were entered simultaneously. Associations between variables that potentially predicted donor attrition were assessed for colinearity. Using a variance inflation factor cutoff of >3.0, it was not necessary to remove any variables from the analysis. Differences in the reason for donor attrition between particular groups of donors (for example, male vs female; WNE vs ethnic minorities) were compared using a χ2-test. Those characteristics with at least a trend towards association with donor attrition on multivariate analysis (P<0.1) were then combined to produce a ‘risk score’ model to examine the cumulative effect of possessing more than one attrition-related characteristic on donor attrition. Rates of attrition were calculated for each score level, and ORs derived using donors with a score of 0 as the reference group. Significance values were calculated using a binary logistic regression method. Statistical analysis was performed using PASW Statistics v.18.0.3 (SPSS (Hong Kong) Ltd, Quarry Bay, Hong Kong).
Outcome of CT requests
In all, 7541 requests for CT, extended typing or DR were made during the study period. Of these, 56.3% were completed, 38.2% cancelled for donor reasons and 5.5% cancelled by the transplant centre.
Reasons for donor attrition
Reasons for donor attrition are shown in Table 1. Of requests cancelled for donor reasons, 19.4% were for personal reasons, 34.1% medical, 36% due to failure to contact the donor, 7.9% emigrated/overseas and 2.6% other reasons. For those deferred for medical reasons, the most common cause of deferral was obesity (body mass index >35), which accounted for 20.8% of medical deferrals, and 7.1% of all cancellations for donor reasons. Other common reasons included pregnancy (11.4% of medical deferrals), cancer (10.3%), cardiovascular disease (6.5%) and autoimmune disease (3.5%).
Factors associated with donor attrition
Potential donor characteristics and risk of donor attrition associated with those characteristics are shown in Table 2. In univariate analyses, all donor characteristics were associated with the rate of donor attrition. Factors significantly associated with increased attrition as compared to the reference category include: age 31–45 years (OR 1.23, P<0.001) and 46–60 years (OR 1.58, P<0.001) when compared to 18–30 years; female sex (OR 1.32, P<0.001); African (OR 2.78, P<0.001), African-Caribbean (OR 3.07, P<0.001), Asian (OR 2.65, P<0.001), Jewish (OR 1.54, P=0.009) and Mediterranean (OR=2.38, P<0.001) ethnicities when compared to WNE; and duration on the register of 6–10 years (OR 1.37, P<0.001), 11–15 years (OR 1.65, P<0.001) and 16+ years (OR 1.40, P<0.001) when compared to <1–5 years. However, attrition was lower in those who were also blood donors (OR=0.71, P<0.001).
All variables were entered into the multivariate analysis (Table 3), as all were significantly associated with donor attrition. Interestingly, the effect of donor age was lost for both 31–45 years (P=0.937) and 46–60 years (P=0.34) age groups. Female sex remained associated (P=0.003). African, Asian and Mediterranean ethnicities were associated with higher attrition (P=0.003, P=<0.001 and P=0.001, respectively), but the strength of the association between African-Caribbean ethnicity and attrition was reduced to a marginal significance (P=0.07), and the ethnicity–attrition association for Jewish donors was lost. Duration on the register was also associated with attrition in the multivariate model, and not being a blood donor remained strongly associated (P<0.001).
Patterns of donor attrition between different ethnicities
Figure 1 shows the breakdown of reasons for donor attrition between ethnic groups. The chart includes only those ethnicities within which at least 50 donors were requested for CT, although mixed race donors were not included because of the ethnic heterogeneity within this group. There is a statistically significant effect of ethnicity on the causes of donor attrition (P<0.001). African and African-Caribbean donors had significantly higher rates of no contact (62.2% and 62.0%, respectively, compared to 35.1% for WNE). In contrast, Asians had higher rates of personal deferrals (39.2 vs 19.6%), but fewer medical deferrals (15.6 vs 33.5% for WNE) and similar rates of no contact (39.3 vs 35.1% for WNE).
Patterns of donor attrition between sexes
Figure 2 shows the breakdown of donor attrition between male and female donors. Again there were differences in the underlying cause (P<0.001): females appeared more likely than males to defer for medical reasons (38.9 vs 26.7%), but less likely to be non-contactable (33.7 vs 39.9%). The most common single reason for female medical deferral was pregnancy (23.0%) but, once this is excluded, some differences in the underlying medical deferral remain apparent. Within those deferring for medical reasons, females appeared more likely to be deferred for obesity (26.6 vs 20.6%), autoimmune disease (14.8 vs 8.6%) and cancer (12.9 vs 7.9%), whereas males appeared more likely to be deferred for cardiovascular disease (including hypertension, 10.2 vs 4.3%) and diabetes (7.1 vs 1.7%).
Patterns of donor attrition associated with duration on the register
Figure 3 shows the breakdown of donor attrition by duration on the register. The reasons for donor deferral differ significantly across the categories (P<0.001). Rates of medical deferral increase incrementally through each group, being 19.2%, 24.7%, 32.2% and 41.5%, respectively, for less than 1–5, 6–10, 11–15 and 16+ years on the register. In contrast, rates of personal deferrals decline with duration on the register (31.4, 22.2, 17.8 and 11.1%). Surprisingly, however, rates of no contact increase only marginally with increased duration on the register (35.2, 36.1, 39.9 and 38.4%).
Cumulative ‘risk factors’ for donor deferral
The cumulative effect of possessing more than one attrition-related characteristic on the odds of donor deferral was ascertained by giving each donor in the study a ‘risk score’. Those characteristics assigned a risk score of 1 included: female sex; African, African-Caribbean, Asian or Mediterranean ethnicity; not being a blood donor; and duration on the register >5 years. All other characteristics were assigned a risk score of 0 (for example, male, blood donor). The minimum score for any given donor was zero, and the maximum was four. The results of this analysis are shown in Figure 4, and all ORs are significant to the P<0.001 level. As can be seen, the odds of donor attrition were 1.99 (95% confidence interval 1.53–2.56), 2.52 (2.04–3.41), 3.40 (2.98–5.33) and 5.53 (3.46–9.47) for scores of 1, 2, 3 and 4, respectively.
This investigation is the first comprehensive study of donor attrition outside the United States. Findings indicate that women, those who have been on the register longer, non-blood donors and members of ethnic minority groups are at a higher risk of attrition from the registry than are their counterparts. Furthermore, there is a cumulative risk effect, that is, there is a higher risk of attrition associated with the cumulative addition of each characteristic.
It has been recognized for some time that donor attrition is considerably higher in ethnic minorities,3 and despite detailed studies of factors contributing to donors deciding not to proceed, the reasons for this remain unclear. This finding is not limited to HPC but also to other donations, such as blood12 and solid organs.13 Our study suggests that, in the United Kingdom at least, the mode of donor attrition differs according to ethnicity, and this can guide the registry in their retention strategy.
The effect of sex on donor attrition appears to be predominantly explained by pregnancy. When pregnant donors are excluded from analysis, the difference in donor attrition between sexes disappears (OR=1.03, P=0.68). Differences remain in the mode of attrition, with males more likely to be non-contactable and females still more likely to be medically deferred. However, these differences are small and are unlikely to be useful in guiding retention strategy. In addition, it was apparent that donor aging, rather than donor age, is associated with an increased vulnerability to medical issues. The fact the blood donors were more reliable donors came as no surprise, as this has been well described in previous studies.3, 10, 14
There is no single solution to the problem of donor attrition: registries must employ a number of methods each aimed at reducing an individual mode of donor deferral. Importantly, they must ensure that only committed and non-ambivalent donors are recruited and adequate retention measures are put in place from the start; and that availability of existing donors is regularly re-established. As it is unlikely that any large register will be able to afford to reliably contact every donor, they must be selective in their approach to retention. Of course, registries that are granted access to a national centralized database have potentially a great advantage in tracing those donors with whom initial attempts to make contact fail.
First, the group of donors most likely to be selected (typically younger male donors15 with more common HLA phenotypes) should be identified and become the focus of retention strategy.
Second, a scoring system might be used to further focus retention efforts on those more likely not to be available when requested. Our study shows how a combination of simple and easily obtained demographic characteristics may be combined to produce a risk score that predicts the likelihood of donor attrition. In addition to directing retention strategy, this score might be combined with information on the most recent donor contact and medical suitability, allowing the registry to produce a ‘freshness’ score, which would appear on a search report. By so doing, the transplant team would have a reasonable estimate of the likelihood of a donor being available, which may allow them to select multiple donors simultaneously or consider other stem cell source options immediately. In addition, if such a scoring system could be applied internationally, it might provide a way to benchmark a registry’s success in donor retention.
Finally, minority ethnic donors require particular attention and, rather than focussing on simply recruiting as many such donors as possible, the focus should be on creating a high quality minority ethnic donor panel. Because of the huge diversity of the HLA system, both within populations and between them, several hundred thousand registered donors of a given ethnicity are required to provide an unrelated donor to the majority in need.16, 17 It is traditionally difficult to recruit ethnic minority donors,18, 19, 20 and high rates of attrition further compromise donor availability. Tackling the personal deferral in ethnic minority donors requires a registry-specific study of the underlying decision process behind this behaviour, and this is currently underway at Anthony Nolan. It is hoped that identification of those factors that may underlie the decision not to proceed will aid strategies to pre-empt or modify these behaviours.
Although this study did not address the impact of attrition on ethnic minority patients, it begs the question whether the current approach to providing donors for such patients is the most appropriate. There are other alternatives that may secure the demand of our ethnic minority patients in a way that remains financially viable: these include increasing cord blood provision with focused ethnic minority recruitment (which allows considerably more permissive HLA matching and does not rely on donor availability post-donation), and investment in establishment or expansion of unrelated donor registries in countries that would meet the need of our major ethnic minorities.
There were some limitations to this study. First, the retrospective design means that it is subject to misclassification bias. In order to mitigate this effect, all records where the CT request was documented as not completed on internal spread sheets were double checked against existing donor notes. Second, while useful for donor retention, these findings have limited implications for donor recruitment: donor registries are already focussing their attention on young male donors, as they are most likely to be chosen to donate. This study did not include the impact of donor recruitment methods on attrition; neither did it tackle the socioeconomic and psychosocial aspects of donor attrition, which are targets for intervention at the recruitment stage. Finally, it was impossible to assess the impact that donor attrition has on the potential recipient, and this is a key outcome measure yet to be ascertained in any study.
In conclusion, this large and important study shows a method by which donor registries internationally may assess rates of donor attrition and use their own findings to guide donor retention strategy. The identification of key donor characteristics associated with attrition allows registries to risk-stratify donors and, according to their likelihood of selection, allocate scarce resources for donor retention in an appropriate and efficient manner. Ultimately, such interventions should reduce delays in unrelated donor provision and improve patient outcomes.
Cleaver SA . The Anthony Nolan Research Centre. Bone Marrow Transplant 1993; 11 (Suppl 1): 38–40.
Passweg JR, Baldomero H, Gratwohl A, Bregni M, Cesaro S, Dreger P et al. The EBMT activity survey: 1990-2010. Bone Marrow Transplant 2012; 47: 906–923.
Switzer GE, Bruce JG, Myaskovsky L, Dimartini A, Shellmer D, Confer DL et al. Race and ethnicity in decisions about unrelated hematopoietic stem cell donation. Blood 2012; 121: 1469–1476.
Bone Marrow Donors Worldwide. http://www.bmdw.org (accessed on 1st May, 2013).
Sencer SF, Zhou T, Freedman LS, Ives JA, Chen Z, Wall D et al. Traumeel S in preventing and treating mucositis in young patients undergoing SCT: a report of the Children’s Oncology Group. Bone Marrow Transplant 2012; 47: 1409–1414.
Craddock C, Labopin M, Pillai S, Finke J, Bunjes D, Greinix H et al. Factors predicting outcome after unrelated donor stem cell transplantation in primary refractory acute myeloid leukaemia. Leukemia 2011; 25: 808–813.
Heemskerk MB, van Walraven SM, Cornelissen JJ, Barge RM, Bredius RG, Egeler RM et al. How to improve the search for an unrelated haematopoietic stem cell donor. Faster is better than more!. Bone Marrow Transplant 2005; 35: 645–652.
Lown RN, Shaw BE . 'First do no harm': where do we stand on unrelated hematopoietic cell donor safety? Expert Rev Hematol 2012; 5: 249–252.
Abress L . Retention Strategies. http://www.worldmarrow.org/fileadmin/Committees/EDUC/2011-EDUC/20111103-EDUC-Donor_Retention_NMDP.pdf (accessed on 1st May, 2013).
Switzer GE, Dew MA, Stukas AA, Goycoolea JM, Hegland J, Simmons RG . Factors associated with attrition from a national bone marrow registry. Bone Marrow Transplant 1999; 24: 313–319.
Werzner ASU, Stolze J, Cotta L, Schmidt A, Ehninger G Patient-Related Donor Recruitment Increases Donor Availability On CT Level, http://www.worldmarrow.org/fileadmin/Committees/EDUC/2012-EDUC/20121108-EDUC-Abstracts_Edu_day.pdf (accessed on 16th December, 2013).
Chen AI, Negrin RS, McMillan A, Shizuru JA, Johnston LJ, Lowsky R et al. Tandem chemo-mobilization followed by high-dose melphalan and carmustine with single autologous hematopoietic cell transplantation for multiple myeloma. Bone Marrow Transplant 2012; 47: 516–521.
Direkze S, Mansour M, Rodriguez-Justo M, Kibbler C, Gant V, Peggs KS . Candida kefyr fungal enteritis following autologous BMT. Bone Marrow Transplant 2012; 47: 465–466.
Switzer GE, Myaskovsky L, Goycoolea JM, Dew MA, Confer DL, King R . Factors associated with ambivalence about bone marrow donation among newly recruited unrelated potential donors. Transplantation 2003; 75: 1517–1523.
Schmidt AH, Biesinger L, Baier D, Harf P, Rutt C . Aging of registered stem cell donors: implications for donor recruitment. Bone Marrow Transplant 2008; 41: 605–612.
Muller CR, Ehninger G, Goldmann SF . Gene and haplotype frequencies for the loci hLA-A, hLA-B, and hLA-DR based on over 13 000 German blood donors. Hum Immunol 2003; 64: 137–151.
Kollman C, Abella E, Baitty RL, Beatty PG, Chakraborty R, Christiansen CL et al. Assessment of optimal size and composition of the US National Registry of hematopoietic stem cell donors. Transplantation 2004; 78: 89–95.
Laver JH, Hulsey TC, Jones JP, Gautreaux M, Barredo JC, Abboud MR . Assessment of barriers to bone marrow donation by unrelated African-American potential donors. Biol Blood Marrow Transplant 2001; 7: 45–48.
Ballen KK, Hicks J, Dharan B, Ambruso D, Anderson K, Bianco C et al. Racial and ethnic composition of volunteer cord blood donors: comparison with volunteer unrelated marrow donors. Transfusion 2002; 42: 1279–1284.
Schmidt AH, Solloch UV, Baier D, Yazici B, Ozcan M, Stahr A et al. Criteria for initiation and evaluation of minority donor programs and application to the example of donors of Turkish descent in Germany. Bone Marrow Transplant 2009; 44: 405–412.
The authors declare no conflict of interest.
About this article
Cite this article
Lown, R., Marsh, S., Switzer, G. et al. Ethnicity, length of time on the register and sex predict donor availability at the confirmatory typing stage. Bone Marrow Transplant 49, 525–531 (2014). https://doi.org/10.1038/bmt.2013.206
- marrow and SCT
- allogeneic transplantation
- donor attrition
- unrelated donor
International Journal of Immunogenetics (2021)
Factors associated with registrant availability for unrelated adult donor hematopoietic stem cell donation: Analysis of the stem cell registry at Canadian Blood Services
Multimedia resources to support the recruitment of committed hematopoietic stem cell donors: Perspectives of the most‐needed donors
Development and evaluation of stem cell collection procedure diagrams to support the education and recruitment of committed stem cell donors
Vox Sanguinis (2021)
Biology of Blood and Marrow Transplantation (2020)