West Nile Virus Epidemic, Northeast Ohio, 2002

Serum samples and sociodemographic data were obtained from 1,209 Ohio residents. West Nile virus immunoglobulin M (IgM) and IgG antibodies were detected by enzyme-linked immunosorbent assay and confirmed. Children were 4.5 times more likely to become infected yet 110× less likely to have neuroinvasive disease develop.


The Study
The survey was conducted December 5-12, 2002. Stratified multistage cluster sampling was used to estimate countywide and subpopulation prevalence rates. The county was divided into 3 risk strata (Table 1). Census tracts were sampled within strata with probability proportional to population. Within each census tract, clusters of ≈50 households were formed. At random points, residents were approached for recruitment until 10 participating households were enrolled from each cluster.
Residents >5 years of age who had lived in the household since July 1, 2002, were asked to participate by providing a blood sample and responding to a questionnaire. One person from each household completed a questionnaire about the home environment. Questionnaires developed by the Centers for Disease Control and Prevention (CDC) were used (10). Informed consent was obtained from all participants or their legal guardian. Assent was obtained from minors >8 years of age. Residents were offered a US $10 gift certificate and test results as compensation. Persons who were pregnant, mentally handicapped, or taking anticoagulants were not enrolled. Institutional review board approval was obtained from University Hospitals of Cleveland.
Serum samples were screened with a WNV-specific immunoglobulin M (IgM) antibody-capture (MAC) enzyme-linked immunosorbent assay (ELISA) (11) and indirect IgG ELISA at Focus Laboratories (Cypress, CA, USA). Positive IgM and IgG were defined as an antibody index >2.0 and >0.9, respectively. All IgM-and IgG-positive samples were sent to the Viral and Rickettsial Laboratory, California Department of Health Services (Richmond, CA, USA) for confirmatory plaque reduction neutralization tests to identify WNV and St. Louis encephalitis virus (SLEV)-specific neutralizing antibody. At the second laboratory, WNV MAC-ELISAs (12) were repeated and IgG ELISAs for WNV, SLEV, and dengue were performed (13). Laboratory-based case definitions were developed (Table 2). SPSS version 11.5 (SPSS Inc, Chicago, IL, USA) and SUDAAN version 8.0 (Research Triangle Institute, Research Triangle Park, NC, USA) were used for preliminary analyses and to assess differences in demographics, behavior, and clinical characteristics between seropositive and seronegative persons. Since SUDAAN variance estimation did not accommodate our complex sample design, we developed formulas to provide better estimates of variance and confidence intervals (CIs) using an α = 0.05. Unless noted, all analyses were weighted. Individual weights were derived by taking the inverse of the probability of selection.
The standard Horvitz-Thompson estimator was used for point estimation (14). For variance estimates, all sources of variation that resulted from the selection process were included by using standard Taylor series approximations. To calculate the confidence interval for the true prevalence ratio (PR), we approximated the variance of the logarithm of the sample PR by using standard Taylor series method. The end points of this interval were exponentiated to obtain the interval for PR.

Conclusions
Participants were recruited from 13 Cuyahoga County municipalities and 9 Cleveland neighborhoods. Of  Seroprevalence varied significantly between age groups (p<0.05) ( Table 1). Based on reported WNND cases and age-stratified seroprevalence rates, we estimate that 1 case of WNND occurred per 4,167 infected children 5-17 years of age, per 154 infected adults 18-64 years of age, and per 38 infected persons >65 years of age (Figure). Strata-specific seroprevalence values ranged from 1.5% to 3.3% but were not statistically different (Table 1).
In 2002, Cuyahoga County experienced its largest epidemic of arboviral encephalitis and meningitis, yet only 1.9% of the county's population became infected during this first WNV transmission season. In the 733-km 2 area of Cuyahoga County, 155 cases of encephalitis and meningitis (WNND incidence: 111 cases/million population) occurred; the seroprevalence was 1.9% countywide and 2.5% in the selected highest risk survey stratum.
Little is known about WNV infection rates in children (15). In contrast to a previous study (8), our study demonstrated an age-dependent risk for WNV infection. The antibody prevalence in the 5-to 17-year age group was significantly greater than in older age groups. These data suggest that children were 4.5 times more likely to be infected than older persons. In this study, children reported spending more time outdoors and using less personal protective measures, which likely contributed to their higher seroprevalence rate. In 2002, only 4 cases of WNND were reported in the 5-to 17-year age group, resulting in a WNND:infection ratio of 1:4,200 compared to a 1:38 ratio among persons >65 years of age. Thus, the risk for WNND after infection may be as much as 110× greater in adults >65 years of age, as compared to children. Inclusion of a larger number of children in this study compared to previous studies allowed these age-stratified analyses to be completed.
Although WNV seroprevalence was similar to those measured in previous outbreaks (7,8), our study was the first to demonstrate that the risk for WNV infection can be age-dependent. Children had a higher rate of infection than adults, but serious neurologic disease developed in few of them. This finding has implications for public health practice and emphasizes the need for children to use protective measures to prevent mosquito bites to further lower their risk for infection with WNV and other mosquitoborne viruses. Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 11, No. 11, November 2005 Figure. Comparison of age-stratified seroprevalence rates (gray bars) to the age-stratified incidence of West Nile neuroinvasive disease (WNND) (black line). Seroprevalence rates were measured in the 2002 seroprevalence study. The incidence of WNND was based on cases reported through the local disease reporting system during the 2002 transmission season.