Willingness Is Not a Donation: Behavioral Evidence on Digital Donor Recruitment in LMICs

Question: How have digital donor-recruitment and emergency blood-donor notification platforms in low- and middle-income countries been evaluated for actual donation behavior rather than self-reported willingness?

Answer

Across these papers, the honest answer is that they largely have not — and the papers that come closest treat the discrepancy as their headline finding rather than as a solved measurement problem. The dominant LMIC evidence base is cross-sectional willingness and "mobilisability" survey work, whose own authors conclude that stated willingness overestimates the population actually available for emergency donation [1]. The single retrieved study that measures realized behavior — a 400-episode comparison of social-media-sourced versus conventionally recruited donors — records delay-to-transfusion and donor irregularities rather than willingness, and finds digital recruitment substantially worse on both [3]. Everywhere else, behavioral measurement appears only where a registry or routine administrative dataset already exists, and those demonstrations sit in high-income or registry-mature settings [8] [11]. Taken together, the literature supports a structural position: the binding constraint on behavioral evaluation is not analytic ambition but the absence of donor registries in precisely the settings where digital recruitment has proliferated, so willingness has been substituted for donation by default rather than by design.

Willingness surveys dominate, and their own authors report that they overstate the donor pool

In Ghana, 91.8% of 1,067 digitally reachable respondents reported high future-donation willingness, but only 54.1% met all four criteria for urgent mobilisability, leading the authors to conclude that digital reachability and stated willingness may overestimate the population pool available for emergency donation [1]. Even within self-report, the binding constraint is capacity rather than intent: willingness to install a trusted donation app (83.2%) and to respond to a trusted urgent request (82.7%) both exceeded practical flexibility to leave current activities (66.6%) [1]. In Qingdao, 37.0% of 1,058 respondents reported a lifetime donation while 39.2% intended to donate in the next 12 months, and the determinants of past behavior and future intention were asymmetric — with higher education positively associated with past donation but adversely associated with future intention [6]. That asymmetry matters for evaluation design because it removes any justification for treating intention as a validated proxy for the same construct.

The one LMIC study measuring realized behavior finds digital recruitment slower and less reliable, not merely unmeasured

A prospective analysis of 400 transfusion episodes across tertiary hospitals in Bangladesh recorded delay-to-transfusion as its primary outcome and found social-media-sourced donors associated with 5.98 versus 2.97 hours (p<0.001), with delay of four hours or more in 83.6% versus 17.6% of episodes (OR 23.78) [3]. Donor-related irregularities occurred in 85% of social-media episodes and in none of the conventional-donor episodes, and social-media sourcing remained the strongest independent predictor of delay (adjusted OR 18.09), while safety outcomes did not differ significantly between groups [3]. This is the only retrieved study in which an LMIC digital recruitment channel is judged on what happened to patients and to the blood supply rather than on what respondents said they would do.

Absent registries, digital recruitment cannot be evaluated behaviorally — and registry linkage is difficult even where it exists

The Bangladesh authors state the precondition directly: in the absence of structured donor registries, social media platforms have become a dominant mechanism for blood donor recruitment in many low-resource settings [3]. The Ghana mobilisability work reaches the same conclusion from the demand side, recommending that digital blood-donor solutions rest on verifiable health-system requests [1]. Linkage is not a solved problem even in Europe: of 17 blood establishments across 16 countries, 76% could in principle link donor to product data and 59% donor to recipient data, but only 29% had automated donor-to-recipient linkage, with legal constraints and lack of consent cited by 11 establishments and resources by 10 to 14 [7]. Where donation activity is register-captured, as in the Danish Blood Donor Study, year-specific probabilities of donation activity can be estimated for 169,893 participants from national registers, showing that donation behavior itself is a tractable administrative construct — but only given the register [11].

Registry-linked behavioral evaluation of donor outreach exists, and shows it can work — in high-income, registry-mature settings

Using the Portuguese blood institute registry for invitation planning in the Lisbon region, an optimization framework was evaluated on actual attendance and found demand fulfillment of 86.1% for a greedy heuristic versus 90.0% for a binary integer linear program, with the trade-offs including larger donor-session distance and greater invitation burden per non-high-frequency donor [8]. The framework explicitly targets mobilization of eligible inactive and lapsing donors and integrates travel convenience and blood-type demand targets into invitation scheduling [8]. This is the strongest available demonstration that invitation strategies can be judged on realized attendance — yet it evaluates a planning algorithm against a mature registry, not a deployed notification platform, and reports no cost per additional donation.

Analogous LMIC trials show the correct design — randomized digital notification with routine-data endpoints — has already been executed for other behaviors

A trial protocol randomizes 342,818 national health insurance members in Kumasi to SMS reminders, autorenewal, or control, with the effect on renewals evaluated from routine NHIS data rather than from participant report [9]. In India, a collaborative bandit algorithm deployed with roughly 6,500 Kilkari participants to personalize automated call timing produced a statistically significant improvement in call pick-up rates compared with random scheduling [10]. Both measure an objectively observed response to a digital notification at scale in an LMIC; neither concerns blood donation, and neither has been replicated on a donor registry.

Emergency-mobilization studies in blood deserts evaluate system safety and feasibility, not whether notified donors arrive

In Lodwar, Kenya, chart review found stockouts on 40 of 126 days (32%), donor transfusion-transmission infection prevalence of 5.4%, and a 99.2% negative predictive value for rapid testing against laboratory-based assays [4]. The follow-on community walking blood bank protocol plans to assess effectiveness through prospective laboratory-based data on changes in blood ordering practices before and after protocol initiation, alongside key informant interviews [12]. Risk modeling for the same model estimates 12 to 56 transfusion-transmitted infections per 100,000 donations depending on rapid-test sensitivity, and that 15,351 donations would be needed under a high-sensitivity rapid strategy to incur one additional infection relative to laboratory-based testing [14]. These studies measure whether the emergency supply is safe and whether stockouts abate, not the donor-level conversion from notification to venepuncture.

The donor-side barriers LMIC studies identify are logistical and experiential, which notification platforms do not address

Among 100 Bangladeshi whole-blood donors, each one-point increase in procedural pain score reduced the odds of greater willingness to donate again by 54% (adjusted OR 0.46), while perceived needle quality showed no association [2]. In Ghana, women (AOR 0.60), those reporting work-schedule barriers (AOR 0.43), and those more than 30 minutes' night-time travel from the nearest healthcare facility (AOR 0.66) had lower adjusted odds of high mobilisability [1]. Qualitative work in Faisalabad attributes donation willingness to religion, personal beliefs, family and friends, and socioeconomic status, alongside low understanding of benefits and risks [5]. A platform that optimizes message delivery therefore addresses none of the measured binding constraints, and the Ethiopian readiness assessment shows why supply-side context matters too — perioperative transfusion at 2.4% was interpreted as consistent with supply rationing rather than optimized clinical practice [13].

Conflicts and gaps

The willing-vs-behavioral literatures point in different directions. The Ghana survey frames willingness as a predictable overestimate but never observes a donation [1], whereas the Bangladesh study observes behavior and attributes the problem to unregulated social media rather than to digital recruitment as such, concluding that integration into regulated donor systems is what is required [3]. The Lisbon scheduling work cuts against any general claim that digital outreach is ineffective, since constraint-aware invitation achieved 86–90% demand fulfillment [8] — but that result may reflect registry maturity rather than transferable intervention content, and no retrieved paper tests whether it transfers. The Bangladesh comparison is observational, so confounding by indication is unaddressed: no retrieved paper reports a randomized or quasi-experimental evaluation of a donor-notification platform with donation as the endpoint in an LMIC, and none reports cost per additional unit collected through digital recruitment.

What would change the answer

A registry-linked randomized trial of notification timing and framing, powered on units collected per 1,000 invitations and modeled on the 342,818-member NHIS reminder design with routine-data endpoints [9], would settle whether the Lisbon attendance gains survive in an LMIC [8]. Adaptive allocation of contact times, as deployed in Kilkari against call pick-up [10], offers a low-cost proximal behavioral endpoint that could be measured before a full donation trial. A calibration study that administers the four-criteria mobilisability instrument and then follows respondents into blood bank records would quantify the willingness-to-donation conversion ratio and tell the field how far existing willingness estimates must be discounted [1]. Any such protocol needs the obstacle taxonomy of the European linkage survey addressed up front — legal basis, consent, and resources for automated linkage [7] — and should carry delay-to-transfusion and donor-irregularity rates as co-primary outcomes rather than post-hoc secondary measures [3].

References

[1] Shen, H., Agorinya, I. A., Ayanore, M. A., Brede, M., Chapman, A., Head, M.. "Beyond willingness to donate: a cross-sectional online survey to understand urgent blood-donor mobilisability among digitally reachable adults in Ghana". medRxiv. 2026-10-19T00:00:00.000Z. URL: https://doi.org/10.64898/2026.08.27.26361538

[2] hoque, a., Rahman, M., Basak, S. K., Mamun, A. A.. "Procedural Pain, Rather Than Donor-Perceived Needle Quality, Is Associated With Willingness to Donate Again: A Multicenter Cross-Sectional Study in Bangladesh". medRxiv. 2026-07-15T00:00:00.000Z. URL: https://doi.org/10.64898/2026.07.13.26357726

[3] Hoque, A., Rahman, M., Basak, S. K., Mamun, A. A.. "Association of social media-sourced blood donors with transfusion delay and donor-related irregularities: A multicentre study in Bangladesh". medRxiv. 2026-04-17T00:00:00.000Z. URL: https://doi.org/10.64898/2026.04.08.26350439

[4] Raykar, N., Chelagat, T., Barnes, L. S., Kumar, N., Raguveer, V., Makanga, C., Musa, A., Williams, W., Lokoel, G., Omondi Onyango, S., Mutebi, E., Esekon, E., Cap, A. P., Delaney, M., Del Signore, M., Dey, T., Gupta, S., Kumar, P., Kundu, S., Mathew, F., Puyana, J. C., Roy, N., Sawhney, R., Stoeger, E., Thivalapill, N., Thompson, A. A., Valencia, A. M., Wesonga Wangamati, C.. "The Local Initiative For Emergency Blood (LIFE-Blood) Study: A mixed-methods, single center exploration of civilian walking blood bank need, feasibility, and safety in a low-resource blood desert". medRxiv. 2025-05-23T00:00:00.000Z. URL: https://doi.org/10.1101/2025.05.23.25327506

[5] Aslam, N., Shafiq, H., Shahid, A. N., Baber, M., Ayub, A.. "Perceptions of public about Blood Transfusion in Allied hospital Faisalabad: A qualitative study.". medRxiv. 2025-06-24T00:00:00.000Z. URL: https://doi.org/10.1101/2025.06.24.25330083

[6] cheng, f., zhang, l., Wang, B., Dai, Z.. "Differential Determinants of Past Behavior and Future Intention Regarding Voluntary Blood Donation: A Cross-Sectional Study of Knowledge, Attitudes, and Practices in Qingdao, China". medRxiv. 2026-06-17T00:00:00.000Z. URL: https://doi.org/10.64898/2026.06.16.26355761

[7] Wehrens, S. M., Arvas, M., Fustolo-Gunnink, S. F., Vinkovic Vlah, M., Waters, A., Erikstrup, C., Drechsler, L. O., Stanworth, S. J., van den Hurk, K.. "Opportunities for Vein-to-Vein Datasets from a Blood Establishment Perspective: towards a 'Pan-European Transfusion Research InfrAstructure' (PETRA)". medRxiv. 2026-03-26T00:00:00.000Z. URL: https://doi.org/10.64898/2026.03.24.26348611

[8] André Carneiro, Pedro T. Monteiro, Rui Henriques. "Optimizing Donor Outreach for Blood Collection Sessions: A Scalable Decision Support Framework". arxiv. 2026-04-01T00:00:00.000Z. URL: https://arxiv.org/abs/2603.29643

[9] Nübler, L., Kwawukume, M. A., Ibrahim, F., Neumann, A., Okoe-Boye, B., Addo-Cobbiah, V., Owusu-Dabo, E., Akohene Mensah, K., Struckmann, V., Knauss, S., Emmrich, J. V., Waitzberg, R., Affanyi, E., Afflu, O., Annor-Darkwah, J., Blankson, D., Asenso-Boadi, F., Cazier, J., Bencivenga, J., Pioch, C., Siegel, M., Quentin, W., Opoku, D.. "ReachUHC: A randomized controlled trial study protocol of a mobile phone-based reminder and automatic renewal intervention to increase health insurance renewal rates in Kumasi, Ghana". medRxiv. 2024-11-21T00:00:00.000Z. URL: https://doi.org/10.1101/2024.11.21.24317685

[10] Arpan Dasgupta, Mizhaan Maniyar, Awadhesh Srivastava, Sanat Kumar, Amrita Mahale, Aparna Hegde, Arun Suggala, Karthikeyan Shanmugam, Aparna Taneja, Milind Tambe. "Learning to Call: A Field Trial of a Collaborative Bandit Algorithm for Improved Message Delivery in Mobile Maternal Health". arxiv. 2025-11-25T00:00:00.000Z. URL: https://arxiv.org/abs/2507.16356

[11] Traeholt, J., Didriksen, M., Helenius, D., Christoffersen, L. A. N., Dinh, K. M., Dowsett, J., Mikkelsen, C., Hindhede, L., Quinn, L. J. E., Bruun, M. T., Aagaard, B., Hansen, T. F., Hjalgrim, H., Rostgaard, K., Sorensen, E., Erikstrup, C., Pedersen, O. B. V., Hansen, T., Schork, A. J., Markussen, B., Ostrowski, S. R.. "Multi-stage reweighting to correct for participation bias in a nationwide biobank with nested recruitment". medRxiv. 2026-04-02T00:00:00.000Z. URL: https://doi.org/10.64898/2026.04.01.26349852

[12] Kumar, N., Mathew, F., Sawhney, R., Thompson, A., Makanga, C., Williams, W., Wesonga, C., Musa, A., Barmasai, H., Wakwabubi, M., Wangia, E., Opiyo, C., Esekon, E., Lockhoel, G., Puyana, J. C., Kumar, P., Delaney, M., Roy, N., Cap, A., Dey, T., Onyango, S., Mutebi, E., Nyanamba, N. O. M., Lokuruka, K. K., Hongo, K., Raykar, N., Chelagat, T., Barnes, L.. "A Protocol to Implement and Evaluate a Community Walking Blood Bank for Hemorrhagic Shock When Banked Blood is Unavailable - The Local Initiative For Emergency Blood (LIFE-Blood) Study, Phase 2.". medRxiv. 2025-06-11T00:00:00.000Z. URL: https://doi.org/10.1101/2025.05.27.25328015

[13] Belachew, F. K., Ambese, T. Y., Dulla, P. K., Soboka, T. S., Sisay, A., Dawit, A., Ariza, F., Moore, J.. "Perioperative Patient Blood Management (PBM) in Africa: a multicentre mixed-methods assessment of health system readiness and clinical practice in Ethiopia.". medRxiv. 2026-07-31T00:00:00.000Z. URL: https://doi.org/10.64898/2026.07.30.26359311

[14] Thivalapill, N., Geng, Z., Kumar, N., Roy, N., Bidanda, B., Raykar, N.. "The risk of transfusion transmissible infection in a civilian walking blood bank using rapid diagnostic tests: A modeling study". medRxiv. 2025-10-23T00:00:00.000Z. URL: https://doi.org/10.1101/2025.10.21.25336343

How this survey was made

This survey answers one question: “How have digital donor-recruitment and emergency blood-donor notification platforms in low- and middle-income countries been evaluated for actual donation behavior rather than self-reported willingness?”

We keep our own collection of recent preprints and refresh it every day from bioRxiv, medRxiv, arXiv, PsyArXiv, SocArXiv and ChemRxiv. For each survey, we search that collection for recent work and keep the strongest matches: papers that use the same terms as the question, and papers on the same topic in different words.

The text was written from the titles and abstracts of those papers; the full texts were not read. Statements that draw on a particular paper are marked with a number in brackets. The reference list just above is the 14 papers whose work this text draws on, chosen from the 130 papers those searches returned; the rest of the shortlist is not printed. It is not a complete review of everything published on the question.

Treat this as a starting point and a reading list, not a peer-reviewed review. The searches behind this survey ran on October 21, 2026 and covered work from the previous two years.