2.4 Discussion

Our analysis shows that universal screening for ZIKV with ID-NAT was not cost-effective at a traditional threshold for the 50 states during the first year of implementation. It was cost-effective in Puerto Rico compared with no screening but not compared with alternative policies. Of note, even the targeted policies we considered were not cost-effective at $1 million per QALY gained for the 50 states and Washington, DC, suggesting a preference (based on cost-effectiveness) for no screening. Although ZIKV has life-threatening sequelae, they were unlikely to manifest because of transfusion transmission during the period of analysis even without screening. At the rate of Zika-infectious donations observed during the 2016-17 epidemic, we would expect transfusion transmission of ZIKV to result in 1 case of Guillain−Barré syndrome every 16 years in Puerto Rico (every 84 years in the 50 states) and 1 case of congenital Zika syndrome every 33 years in Puerto Rico (every 176 years in the 50 states) without screening. Now that the epidemic has ended, the risks of serious adverse events are extremely remote.

The cost-effectiveness of universal Zika screening during this period compares unfavorably with that of other blood safety measures currently in place. For instance, MP-NAT for HIV and hepatitis B and C viruses costs about $1.3 million per QALY gained (Table 2.4) <[9,10,18,85]>.

Because the rate of ZIKV-infectious donations drove screening effectiveness, policies that target donor subpopulations with a higher expected rate had favorable CERs compared with universal screening. However, of the subpopulations we considered, only donations collected in Puerto Rico during high mosquito season had a rate high enough to make ID-NAT or MP-NAT cost-effective.

Our analysis has limited ability to account for all factors salient to the policy decision. The concerted effort of the FDA, blood centers, and manufacturers to rapidly develop and implement ID-NAT is widely regarded as a remarkable achievement [86] that has prevented TTZ in the 50 states and Puerto Rico. Our analysis did not capture the potential benefits to our broader understanding of Zika disease progression and epidemiology conferred by screening the blood supply or the potential effect of removing a blood safeguard on public trust in the safety of the blood supply.

Where data scarcity or model complexity necessitated assumptions, we took care to avoid overestimating testing costs or underestimating the economic and health-related impact of TTZ. For instance, we assumed that the duration of mild febrile illness and its resulting health-state utility loss equaled those of the H1N1 influenza virus in 2009 [76], and our estimate of the percentage of transfusion recipients who are pregnant was derived from a high-risk pregnancy center that may not represent transfusion rates in community settings. Most important, we assumed that all donations positive for ZIKV RNA were potentially infectious regardless of antibody reactivity, which biased interventions to seem more cost-effective. Supplement Table 5 shows key results with donations positive for IgM assumed to be noninfectious.

All prior blood safety models have assumed that recipients who have a TTI resemble the “average” recipient. However, risk for TTI is increased if many components, particularly plasma units, are transfused. Our analysis suggests that high-risk recipients are more likely to be younger than 75 years and to have a shorter expected survival time at baseline. We believe that stakeholders should consider these differences when evaluating the health economic effect of TTIs or other transfusion-related adverse events. We relied on data from Sweden and Denmark to model posttransfusion survival by age and component mix because suitable North American data were not available. We addressed this limitation by calibrating the number of components received and posttransfusion survival to summary statistics derived for American recipients. However, better data on these variables for American recipients would be beneficial.

Since the introduction of HIV antibody testing in 1985, the FDA has never discontinued screening for a TTI once required. Because the infectious donation rate drives the expected benefit and cost-effectiveness of disease marker screening, the FDA might consider identifying infectious rate thresholds for all TTIs, below which screening requirements are downgraded or eliminated. This could help ensure that blood centers allocate resources effectively, cost-efficiently, and in a prioritized manner. For instance, evidence suggests that screening for transfusion-transmitted babesiosis, a widespread TTI in parts of the United States, is a much more effective use of resources than universal ZIKV ID-NAT [64,87,88].

The July 2018 decision to allow MP-NAT as an alternative to ID-NAT will reduce costs with little to no increase in risk to recipients. However, absent a significant increase in the rate of ZIKV-positive donations, any nontargeted screening program will not be cost-effective. Unless the number of ZIKV-positive donations interdicted increases substantially, further reduction of screening requirements should be considered.

In summary, our analysis of the first year of ZIKV screening suggests that ID-NAT was not cost-effective in Puerto Rico compared with alternative policies and that none of the interventions we evaluated were cost-effective for the 50 states. Screening blood donors for ZIKV is currently not cost-effective in any jurisdiction in this study. To date, decision-analytic modeling has played a limited role in informing American blood safety policy. Reports often emphasize cost-effectiveness but include estimation of risk and other policy-relevant outcomes. This report demonstrates that model-based analyses can be tailored to policymakers’ concerns (i.e., risk of serious adverse events). Collaboration could improve understanding between policymakers and modelers the policy relevance and utilization of model-based reports.