Impact of predictive scoring model and e-mail messages on African American blood donors

Abstract

BACKGROUND: Expanding the African American (AA) donor pool is critical to sustain transfusion support for sickle cell disease patients. STUDY DESIGN AND METHODS: The aims were to: 1) apply cognitive computing on donation related metrics to develop a predictive model that effectively identifies repeat AA donors, 2) determine whether a single e-mail communication could improve AA donor retention and compare retention results on higher versus lower predictive score donors, and 3) evaluate the effect of e-mail marketing on AA donor retention with culturally versus nonculturally tailored message. RESULTS: Between 2011 and 2012, 30,786 AA donors donated blood at least once on whom predictive repeat donor scores (PRDSs) was generated from donor-related metrics (frequency of donations, duration between donations, age, blood type, and sex). In 2013, 28% (8657/30,786) of 2011 to 2012 donors returned to donate on whom PRDS was validated. Returning blood donors had a higher mean PRDS compared to nonreturning donors (0.649 vs. 0.268; p < 0.001). In the e-mail pilot, high PRDS (≥0.6) compared to low PRDS (<0.6) was associated with 89% higher donor presentation rate (p < 0.001), 20% higher e-mail opening rate (p < 0.001), and, specifically among those who opened the e-mail, 159% higher presentation rate (p < 0.001). Finally, blood donation rate did not differ (p = 0.79) as a function of generic (n = 9312, 1.4%) versus culturally tailored (n = 9326, 1.3%) message. CONCLUSION: Computational algorithms utilizing readily available donor metrics can identify highly committed AA donors and in conjunction with targeted e-mail communication has the potential to increase the efficiency of donor marketing.