Rotavirus (RV) in children has decreased since the introduction of RV vaccines. With use of an enzyme immunoassay (EIA) for RV detection, U.S. RV vaccine effectiveness (VE) calculations using a prospective, case-controlled, study design has ranged from 70 – 91%. Prior data using rotavirus PCR has underestimated RV VE compared to EIA. We used RV rapid antigen test results to determine VE in children presenting to the emergency department (ED) or requiring hospitalization.
Stool samples from subjects with ED or hospital encounters from July 2007 – June 2014 at 2 Atlanta-area children’s hospitals were tested by rapid antigen detection for RV (Xpect®RV, Thermo Scientific [Remel], Lenexa, KS). Subjects were excluded if duplicate, <8 mo. of age, born before January 2006, partially vaccinated for RV, had unknown vaccination status, or had a non-ED outpatient encounter. Controls were RV-negative children, and cases were matched if both date of birth and test date were +/- 30 days. Cases and controls were matched according to inpatient or ED status and analyzed separately. VE was calculated among children with available controls as 1 – (vaccination rate of cases/vaccination rate of controls). Subgroup analysis included community-acquired RV (AGE symptoms or RV test sent <3d after hospitalization) and the non-immunocompromised.
RV was identified in 723/5,839 (12%). Of these 723, 337 (47%) were not eligible (106 with age <8 mo., 143 with birth before 2006, 41 with partial vaccination, 31 with unavailable vaccine information, 16 non-ED outpatient). Among the 386 eligible children, 269 (70%) were matched with 630 controls. Overall VE was estimated at 51% (95% CI: 46-57%), and was lower among children >3 yrs. of age (VE 21%; 95% CI: 16-36%). Estimated ED-only VE was 64% (95% CI: 53-84%). Inpatient VE was 49% (95% CI: 41-58%), and was similar when limited to community-acquired, non-immunocompromised patients (50%; 95% CI: 44-57%).
The RV VE estimated by this retrospective, case-controlled study design was lower than that of other prospective studies using EIA, and particularly low in older children. These data suggest that prospectively enrolled VE studies using EIA may be needed to obtain accurate VE estimates.
S. E. Gillespie, None
R. Jerris, None
L. Westblade, None
A. L. Shane, None
C. Kraft, None
E. J. Anderson, Abbvie: Consultant , Consulting fee
MedImmune: Investigator , Research grant
Roche: Editorial assistance in writing a manuscript , Editorial assistance in writing a manuscript