West Nile Virus Infection among Humans, Texas, USA, 2002–2011

Melissa S. Nolan, Jim Schuermann, and Kristy O. Murray

Author affiliations: Author affiliations: Baylor College of Medicine, Houston, Texas, USA (M.S. Nolan, K.O. Murray); Texas Department of State Health Services, Austin, Texas, USA (J. Schuermann)

Abstract

We conducted an epidemiologic analysis to document West Nile virus infections among humans in Texas, USA, during 2002–2011. West Nile virus has become endemic to Texas; the number of reported cases increased every 3 years. Risk for infection was greatest in rural northwestern Texas, where Culex tarsalis mosquitoes are the predominant mosquito species.

The first documented case of West Nile virus (WNV) infection in North America occurred during an outbreak of encephalitis in 1999 in New York, New York, USA (1). Within a few years, WNV had rapidly spread across the United States and was being transmitted throughout most of the country (2). In Texas, the first cases of human infection with WNV were reported in 2002, and cases have been reported every year since. Our objective was to epidemiologically describe WNV infections among humans over the first decade of virus transmission in Texas.

The Study

We analyzed deidentified surveillance data for all cases reported to the Texas Department of State Health Services (TxDSHS) during 2002–2011. Reporting of West Nile neuroinvasive disease (WNND) to the TxDSHS was made mandatory in 2000, and West Nile fever (WNF) was added to the list of reportable conditions in 2005. Onset dates were categorized according to work week for each year and were used to create an epidemic curve and yearly incidence graph. Attack rates (no. cases/100,000 population) were stratified by demographics, and a 2-tailed t test was used to detect differences between the overall attack rate for the state and each demographic variable. Stratified calculations were based on the number of variable-specific reported cases over the variable-specific population. We estimated the seroprevalence by using published ratios (no. infected: no. WNND cases) and state population estimates for 2011 (3–5). By using the number of reported WNND cases, defining WNF as 26% of seroprevalence estimates, and using previously published cost estimates for WNND and WNF cases, we calculated the total economic cost of infections over the past decade (6,7).
Relative risks (RRs) for each county were calculated for each year. Expected number of cases per county (e) were calculated by using a validated equation e = (r × n) (c), where r is the state incidence rate, n is the state population, and c is the county population (5,8). To assess potential associations between counties and increased RRs and to examine urban–rural characteristics of counties and increased RRs, we performed Kruskal-Wallis 1-way analysis of variance by ranks. All calculations were run by using Stata version 12.0 software (StataCorp, College Station, TX, USA).