Gastroenteritis hospitalizations in older children and adults in the United States before and after implementation of infant rotavirus vaccination

摘要

Implementation of infant rotavirus vaccination in 2006 has substantially reduced the burden of severe gastroenteritis among US children younger than 5 years. The role of rotavirus in adult gastroenteritis has been less well appreciated. Recent studies report rotavirus detection rates of 18% in emergency departments and 5% from February through May in hospitalized patients, and estimates of 81000 emergency department visits and 18 000 hospitalizations4 in the United States annually. Whether indirect protection (due to reduced transmission of rotavirus) extends to adults remains unclear. Previous studies suggesting such indirect protection were limited to 1 postintroduction season or l hospital setting, so prudent interpretation was warranted. We assessed patterns of gastroenteritis hospitalizations among children aged 5 years or older and among adults before and after implementation of infant rotavirus immunization. Methods | Rotavirus-codedand cause-unspecified gastroenteritis discharges from January 2000 through December 2010 were retrieved from a nationally representative database of hospital inpatient stays, the Nationwide Inpatient Sample, as previously described. Cause-unspecified discharges were examined because testing for rotavirus is infrequently performed in adults. We fitted time series regression models assuming a Pois-son distribution of 2 separate outcomes: monthly counts of ro-tavirus-coded or cause-unspecified discharges. We estimated annual and monthly incidence rate ratios (RR) of the postvac-cine years (2008,2009, and 2010) separately and combined vs the prevaccine years (2000-2006), controlling for month, secular trends, and population size; 2007 was a transition year with limited coverage and was excluded. Separate models were fit for each of the 6 age groups. The study was exempt from institutional review board approval because deidentified aggregated data were used. Significance was assessed as a 2-sided P value of .05 using Stata version 12.0 (StataCorp).