Appendix 1 for Busch et al. manuscript in March 2006 EID
Serological Testing of Samples from Bismarck and Minot, North Dakota
To estimate TMP-NAT, we needed to obtain minipool-NAT and IgM data from a region with high West Nile virus activity in 2003 but no previous activity. We used archived plasma samples from donations to blood centers based in Bismark and Minot, North Dakota since this area experienced epidemic activity for the first time in 2003. The rationale for this approach is based on the observation that IgM antibodies appear within two to three weeks following infection (28) and remain in circulation for a minimum half-life of about two months (29-31). Donors who live in a region without West Nile virus activity before 2003 (such as North Dakota) should not have formed West Nile virus antibodies (IgM or IgG) unless they were exposed elsewhere. Owing to the serological dynamics of West Nile virus infection, if the regional epidemic is of short duration (< 2 months), one can assume that IgM prevalence measured about three weeks after the end of the epidemic will be a good estimate of the seasonal West Nile virus incidence because virtually all people who acquired infection during the epidemic would have developed IgM, and most would still have elevated IgM in circulation.
We accessed the Blood Systems Laboratory (BSL) repository (Tempe, AZ) to obtain all archived plasma preparation tubes with sufficient residual volume from Bismarck and Minot. In 2003, BSL routinely saved and froze at -70oC plasma preparation tube specimens derived from donations collected in areas with elevated West Nile virus activity; this repository contained only NAT non-reactive specimens since NAT reactive plasma aliquots had been diverted for confirmatory testing using multiple West Nile virus assays including IgM. All frozen plasma specimens were shipped frozen to the Viral Reference Laboratory and Repository Core at Blood Systems Research Institute (BSRI, San Francisco, CA) where they were scanned into electronic box maps and stored at -80oC. After comparing the electronic box maps with an independently compiled master list to verify that all tubes originated from Bismarck and Minot, specimens were withdrawn anonymized (i.e., donation center location and dates were retained and linked to a new study bar code identifier, and the blood unit identification code that links to personal identifiers was destroyed). Anonymous IgM testing was performed on 3,922 of 7,012 minipool-NAT non-reactive donations collected between July 1 and September 27, 2003. We also prospectively obtained all plasma preparation tube specimens collected from Bismarck between June 7 and June 24, 2004 to assess IgM and IgG prevalence. For operational and budgetary reasons, only 1,000 samples from one region could be collected prospectively; since 75% of the 2003 archived specimens originated from Bismarck, we chose that region for testing antibody prevalence in June 2004.
IgM screening was performed at the Viral Reference Laboratory and Repository Core at BSRI on the 2003 (n=3,922) and 2004 (n=1,000) samples using the Focus Diagnostics, Cypress, CA, Flavivirus (West Nile) IgM Capture enzyme-linked immunosorbent assay (ELISA) per package insert directions (32). All initially IgM-reactive specimens were retested with the inclusion of a background subtraction step that excludes possible false-positive IgM results. All initially IgM-reactive or equivocal samples were then shipped frozen to Focus Diagnositics where the IgM assay results were confirmed in a licensed laboratory that had developed the screening assay. We defined IgM positivity based on the results from Focus Technologies. Similar procedures without a background subtraction step were used to evaluate IgG positivity in June 2004 samples using the Focus Technologies Flavivirus (West Nile) IgG ELISA (33).
We estimated IgM prevalence for each calendar week between July 1 and September 27, 2003 by combining the weighted average of 1) the weekly IgM prevalence observed in NAT screened-reactive donations and 2) the weekly IgM prevalence observed in NAT screened non-reactive donations. Weighting was performed because IgM testing was performed on a sample of NAT screened non-reactive donations but virtually all NAT screened reactive donations. We calculated approximate 95% confidence intervals (CIs) assuming the weekly prevalence estimates were normally distributed with variance equal to the weighted variances of the NAT reactive prevalence and of the NAT non-reactive prevalence (21). Prevalence of IgM and IgG in June 2004 were calculated by dividing the number of IgM or IgG positive donations by the number of donations screened in that two-week period. We also obtained approximate 95% confidence intervals around these prevalence estimates (20).
|Funded by the Blood Systems Foundation|