203 BURDEN OF ROTAVIRUS HOSPITALIZATIONS IN CHILDREN AS ASSESSED BY ACTIVE CASE FINDING

ContextNational disease burden of rotavirus disease requiring hospitalization in the US has been estimated by an epidemiologic case definition (ECD) that combines nonspecific diagnosis codes with age and seasonal patterns of rotavirus illness and that has unknown sensitivity and specificity.ObjectiveTo compare estimates of rotavirus disease requiring hospitalization using the ECD compared with active case finding (ACF = acute gastroenteritis with RV detection).Design/MethodsAge-eligible children were enrolled 5 days a week over 2 years at three pediatric hospitals if acute (< 7 d) diarrhea, vomiting, and/or fever were present; then stools were tested for rotavirus infection. Acute/convalescent serum pairs were collected from children testing rotavirus negative. An age-matched population serum set was collected at one site to estimate age-specific seroconversion during a rotavirus season. Computerized hospital record databases were queried for assignment of ECD-qualifying diagnoses.PatientsOf 73,087 hospitalizations, 14,951 age-eligible patients were screened. Of these, 3,109 had eligible symptoms, 2,254 (72%) were enrolled, and 1,951 (87%) had a stool tested.ResultsOf 1,598 patients with acute gastroenteritis, 208 (13%) met the ECD and 447 (115% more) the ACF. Other children (13% more) with rotavirus illness detected by stool excretion included some with short-stay hospitalizations, F only, or younger age (15-30 d). From the 19% seroresponse rate in 89 serum pairs, 221-296 (106-142% more) rotavirus hospitalizations occurred from children testing rotavirus negative. Rotavirus-positive cases not ECD eligible (n = 94) were younger, less likely to have diarrhea or vomiting, less likely to have acute gastroenteritis attributed to be the cause of hospitalization, and more likely to be given generic diagnoses for "febrile illness " or "viral infection " (each comparison with p < .001). Within ECD age and season limits, ECD was 56% (95 CI: 48, 63%) specific and 44% (42-47%) sensitive compared with ACF.ConclusionsThe epidemiologic case definition did not detect 71-74% of the rotavirus hospitalizations in this multicenter, active surveillance study.