This project, which collected and analyzed WNV screening data for 95% of US blood donations during the 2003 epidemic, identified 944 viremic donations among 4.6 million donations screened from July 1 to October 31. The number of viremic donors identified in 2003 is a slight underestimate since viremic donors identified by participating ABC and ARC centers outside the study time frame and viremic donors identified at nonparticipating collection centers and the military blood program were not included. The ABC and ARC data indicate that ≈1,000 West Nile viremic donors were identified in the United States in 2003 by prospective NAT screening, and consequently ≈1,500 potentially infectious blood components were interdicted before transfusion (24
). This yield is particularly remarkable when compared with NAT screening for HIV-1 and HCV, which identified only 12 HIV and 170 HCV-infected antibody-negative donations among ≈39 million donations screened in the first 4 years of testing (13
One goal of this project was to monitor the geographic and temporal distribution of WNV in the US blood donor population. We documented rates of viremic blood donors exceeding 3 per 1,000 donations in some states during the peak of the 2003 epidemic. The proportion of confirmed positive donations identified by minipool-NAT paralleled the neuroinvasive case reports in each state. Blood donor NAT screening data are useful for population surveillance because the testing has a rapid turnaround time, infections are identified soon after WNV acquisition, many of these infections remain asymptomatic, and typically those in whom symptoms develop are identified before illness onset. Communication of WNV donor screening data from blood centers to state and county health departments thus augments national surveillance and facilitates more complete national reporting of human WNV infections to CDC's ArboNET program (24
Our estimate of an average 6.9-day period of viremia detectable by minipool-NAT correlates well with the duration of viremia that was documented after intentional WNV inoculation of human cancer patients in the 1950s (25
). In those studies, the duration of viremia (detected by intracerebral inoculation in mice, which is less sensitive than minipool-NAT) correlated with underlying disease severity and averaged 6.2 days in a subset of relatively healthy patients.
Our results have limitations. We assumed that WNV incidence in blood donors reflects incidence in the general population. Blood donors differ from the general population with respect to age; however, serologic surveys indicate that age is not associated with the likelihood of WNV infection acquisition but is associated with severity of disease (1,2,6,7
). Some racial, ethnic, and socioeconomic groups are also underrepresented in the blood donor population. Because WNV is a mosquitoborne arbovirus, incidence may vary among these demographic subgroups, which could bias extrapolations based on donor data. Moreover, potential donors with fever or headache are deferred from donation because the combined symptoms may indicate WNV infection; thus, blood donor screening data would underestimate infection incidence in the general population. However, we believe an underestimate is unlikely since the primary viremia phase of infection detected by minipool-NAT tends to precede development of WNV-related symptoms (1,10,26
Although projections of seasonal incidence estimates based on donor data have limitations, they represent a source of data independent from national disease reporting. Completeness of reporting of WNV neuroinvasive cases to ArboNET is unknown and likely varies among states. The ratio of total infections to neuroinvasive cases is also not precisely known, thus adding uncertainty to incidence data extrapolated from such cases. Using blood donor screening data, we project that ≈256 people are infected with WNV for each person in whom neuroinvasive disease develops (95% CI 112–401). This ratio is similar to that observed in a serologic survey in Romania, which estimated that 1 in 140–320 infections results in neuroinvasive disease (6
). Previous estimates of the total number of persons infected in the United States are based on a serologic survey in New York City that indicated that 1 in 140 infections (95% CI 61–217) results in neuroinvasive disease (2,7
). Although CIs around the New York City estimate and our ratio overlap, the blood donor screening data suggest that previous projections may have underestimated the total number of persons infected. Similar analyses to determine the proportion of infections that result in febrile illness or other clinical manifestation of WNV would be of interest. However, reporting of these illnesses to ArboNet is incomplete and highly variable by state and over time and hence not appropriate for this purpose. Follow-up studies of viremic donors have demonstrated that febrile syndromes develop in 20% to 30% of patients (26
), consistent with reports from other studies (1–3
Our approach of using NAT yield data to project WNV infections has advantages over serologic strategies. Performing large-scale, community-based serologic surveys to estimate infection incidence is prohibitively expensive, is subject to participation bias, and can be biased by previous exposures to WNV or infections by other flaviviruses that cross-react on WNV IgM and IgG assays (9,27–31
). Given the extent of recent WNV spread in the United States, interpretation of future serologic surveys will require determination of baseline prevalence before each epidemic year, evaluation of serial samples throughout the epidemic to accurately estimate infection incidence, or both.
In conclusion, our study demonstrates that in addition to preventing many transfusion-transmitted WNV infections, routine donor NAT screening has valuable public health applications, both as an early indicator of human epidemic activity regionally and as a surveillance tool to help monitor national infection incidence. In addition, this study highlights the value of establishing a national system for compiling blood donor data, which would enable ongoing and timely surveillance of WNV and other established and emerging infectious diseases.