Blood Screening for West Nile Virus: The Cost-Effectiveness of a Real-Time, Trigger-Based Strategy

The US Food and Drug Administration has mandated West Nile virus (WNV) screening in the United States since 2003; no guidelines exist, however, for selecting a cost-effective screening platform. Previously, we conducted a cost-effectiveness analysis of regionally imposed screening strategies to prevent transfusion-associated WNV infection [1]. This study demonstrated that nucleic acid testing (NAT) for WNV is not cost-effective for most transmission scenarios in the United States. In areas with low rates of natural WNV transmission, NAT screening provides no clinical benefit to the transfused population while incurring costs. In areas with high rates of natural WNV transmission, seasonal screening of blood that is designated for immunocompromised patients reduces cases of severe disease and may be cost-effective. Because the majority of WNV infections in immunocompetent individuals are asymptomatic, universal screening provides little to no additional clinical benefit, at substantial cost, even in high-transmission areas. Similarly, Custer et al. [2] also concluded that nationally implemented supplemental NAT strategies are not cost-effective.

To designate a screening strategy as cost-effective, the full range of proposed strategies must be evaluated. The emergence of new data on related factors, such as the properties of WNV and characteristics of the screening assay, can also alter decision making. In this report we extend our previous societal perspective cost-effectiveness analysis in response to 2 recent reports. First, we incorporate into our model new data on the infectivity of blood donations found to be positive for WNV by individual NAT, and second, we assess newly proposed screening strategies.

In a recent study, Busch et al. [3] measured the presence of IgM antibodies in blood samples in which WNV had been detected. Among samples that tested positive by minipool NAT followed by confirmatory individual NAT, 9% were positive for the IgM antibody that is thought to neutralize virus infectivity. For samples that tested positive by individual NAT but for which WNV was not detected by minipool NAT, 85% were IgM positive. Busch and colleagues hypothesized that this later group of samples included donations from individuals in the convalescent phase of infection who had become infected with WNV earlier in the season and had mounted an effective immune response, but who had not completely cleared the virus (which was subsequently detected by the more sensitive individual NAT). These data suggest that the viremic samples detected by individual NAT alone are very unlikely to transmit disease and, thus, there is little advantage in withdrawing these individuals' donations from the blood supply.

In another recent report, Custer et al. [4] advocated a real-time monitoring "trigger" screening strategy as an alternative to current national screening strategies: blood collection agencies would conduct routine screening of blood samples for WNV with minipool NAT assays, would switch to individual NAT when a predetermined threshold in the number of positive donations is reached, and then would return to minipool NAT when the rates of WNV infection fall below a specified level. This strategy takes advantage of the higher sensitivity of individual NAT during seasonal peaks of viremic donations. Because IgM-positive donations tend to occur at the end of the WNV season, when infected persons are convalescing, the viremic samples that were missed when minipool testing resumed would be more likely to include low-infectivity samples.

Methods. By extending our previously published model [1], we incorporated WNV real-time monitoring blood donation screening strategies (figure 1) [4] and the new evidence regarding the low infectivity of IgM-positive blood donations [3]. For the time intervals when the real-time monitoring screening strategies indicated minipool testing, we calculated the cost of screening based on the direct costs associated with minipool testing for this period. When the monitoring strategies triggered a switch to individual sample testing, we calculated the cost of screening based on the direct costs for individual sample testing for this duration. Direct costs for screening included the costs of assay kits and reagents, laboratory technician fees, and the costs of discarded false-positive results, donor notification, and retrieval of test-positive samples of WNV-positive samples [1].

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

Real-time monitoring screening strategy. The real-time monitoring screening strategy uses minipool screening until a virus detection threshold is reached, at which time individual sample testing occurs. Minipool testing resumes when there are no West Nile virus–positive donations detected by individual sample testing for 1 week.

In our prior analysis, we used a computer-based mathematical model to estimate cases of neuroinvasive disease in a target population of blood transfusion recipients. We used these predictions and new evidence on the infectivity of viremic donations to estimate lifetime costs and quality-adjusted life expectancy associated with each of the screening strategies. The same model was used to evaluate costs and clinical outcomes associated with the real-time monitoring screening strategies proposed by Custer et al. [4]. A cost-effectiveness analysis was conducted, and incremental cost-effectiveness ratios were calculated for those strategies that were not dominated (i.e., that were more costly and less effective, or that were more costly and less cost-effective, than the next best alternative).

On the basis of data derived from the first 2 years of screening [3], we estimated that 91% of viremic samples that tested positive for WNV by pooled NAT were infectious, and that 15% of positive samples escaping pooled NAT detection were infectious. We modeled the trigger strategies by assuming that minipool screening was conducted until the WNV detection rate exceeded 1 per 1000 donations and 2 positive donations per week, thereby triggering individual sample testing. Individual sample testing continued until no samples with WNV were detected for 1 week by this method, at which point minipool testing was resumed [4]. We also explored the effects of varying the size of the minipool and compared year-round screening for WNV to seasonal screening. table 1 presents the full range of proposed screening strategies for detection of WNV in blood supplies for areas with high transmission intensities.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Predicted quality-adjusted life years (QALYs), costs, and incremental cost-effectiveness ratios associated with nucleic acid test blood screening for West Nile virus in a cohort of 2,000,000 persons.

Results. When we incorporated the recent data on the prevalence and infectivity of IgM-positive donations into our model, we found that the risk of transfusion with a WNV-infected viremic donation decreased, making screening strategies in high-transmission areas even less attractive than had been previously reported. Seasonal, targeted screening for WNV of blood donations designated for immunocompromised patients by individual NAT remained the most cost-effective strategy for high-infection–short-duration transmission areas, even after considering alternative real-time monitoring trigger strategies. Although 1 trigger screening strategy offered clinical benefit beyond restricted screening for WNV in immunocompromised populations, the incremental cost-effectiveness ratio was nearly $4 million per quality-adjusted life year gained—far higher than the commonly accepted willingness to pay threshold cost of $50,000–$100,000 per quality-adjusted life year [5]. For areas with high-infection–long-duration transmission of WNV, the status quo questionnaire strategy was least costly, and seasonal targeted screening for WNV in immunocompromised patients by individual NAT was marginally cost-effective at $91,500 per quality-adjusted life year gained.

Discussion. Real-time monitoring trigger strategies may have initially appeared attractive on the basis of data from 2003 on viremic donations missed by screening pooled donations [3]. However, the number of intercepted viremic donations is a misleading indicator of the benefit of WNV screening, because many of these donations have low infectivity, and the majority of subsequent transfusions occur in immunocompetent individuals who do not experience severe clinical consequences from WNV infection. Therefore, screening is associated with very little, if any, gain in quality-adjusted life expectancy, whereas the cost of screening is substantial.

On the basis of the present analyses, if the number of WNV-positive viremic donations intercepted by NAT were the outcome used in the denominator of the cost-effectiveness ratio, a seasonally implemented, real-time monitoring trigger strategy applied to a high-infection–short-duration transmission area would be approximately $14,500 per viremic sample intercepted. Although the denominator of a cost-effectiveness ratio can certainly be expressed in terms of intermediate outcomes (e.g., viremic donations that were intercepted), results of this kind of analysis are limited: they can only be compared to other analyses using the same intermediate measures of outcome, and only if the consequences (or lack thereof) of the additional averted viremic donation are not reflected anywhere in the analysis. For analyses intended to inform resource allocation and health policy decisions, comparison of cost-effectiveness ratios across different studies and types of interventions is necessary, and therefore, the denominator must be expressed in a common metric (generally life-expectancy gains or quality-adjusted life years) [5]. These cost-effectiveness ratios are generally much higher than those ratios reported that use intermediate outcomes. Calculations that used intermediate health outcomes overestimate the cost-effectiveness of certain screening strategies and may have contributed to earlier reports that advocate blood screening for WNV as a potentially cost-effective intervention to safeguard the blood supply [3, 4, 6].

On the basis of the new estimates of the infectivity of viremic donations and subsequent lower numbers of predicted cases of clinical disease, the total costs associated with transfusion-associated WNV infection decrease substantially with a decrease in treatment costs. For example, the predicted cost for treating transfusion-associated WNV infection in a high-infection, short-duration transmission area when using a status quo questionnaire strategy was 4-fold less than what we had previously calculated. Similarly, the predicted treatment costs associated with all of the supplemental WNV screening strategies decreased. Thus, the cost of screening becomes a larger proportion of the total cost associated with a specific strategy, and the results of the analyses appear to be more sensitive to changes in screening costs and, indirectly at least, assay prices. If the cost of the WNV screening assay doubled, even the seasonal screening of blood for immuocompromised recipients in high-infection, short-duration WNV transmission areas would exceed $100,000 per quality-adjusted life year gained, relative to the status quo questionnaire; it would no longer be considered cost-effective.

This analysis provides further evidence that universal screening for WNV is not a cost-effective intervention, even in areas with high rates of natural transmission of WNV. In areas with high rates of transmission of WNV, seasonal, targeted screening of donations for immunocompromised individuals may be cost-effective, subject to pricing of screening assays.

• Received April 3, 2006.
• Accepted May 22, 2006.

• © 2006 by the Infectious Diseases Society of America