Burden of Acute Gastroenteritis Hospitalizations and Emergency Department Visits in US Children That Is Potentially Preventable by Rotavirus Vaccination: A Probe Study Using the Now-Withdrawn RotaShield Vaccine
BACKGROUND. With the implementation of a new rotavirus immunization program in the United States in 2006, determining the potential health benefits of rotavirus vaccination is important. We estimated the burden of acute gastroenteritis hospitalizations and emergency department visits in US children that are potentially preventable by rotavirus vaccination.
METHODS. We conducted a retrospective cohort analysis of children who in 1998–1999 were eligible to receive a now-withdrawn rotavirus vaccine (RotaShield) and were continuously enrolled in 1 of 6 managed care organizations in the Vaccine Safety Datalink. Estimates of vaccine effectiveness against all-cause gastroenteritis hospitalizations and emergency department visits adjusted according to month of birth, gender, and managed care organizations were calculated as 1 minus the risk ratio of outcomes among children in different dose groups. The burden of acute gastroenteritis prevented by vaccination was compared with the rotavirus burden estimated by 2 previously used indirect methods.
RESULTS. The effectiveness of a full 3-dose RotaShield series over a 1-year follow-up period was 83% against all-cause gastroenteritis hospitalizations and 43% against all-cause gastroenteritis emergency department visits. An increasing number of doses improved the effectiveness in preventing gastroenteritis hospitalizations, but no clear trend was observed between number of doses and effectiveness in prevention of gastroenteritis emergency department visits. The proportion of gastroenteritis hospitalizations and emergency department visits prevented by vaccination was substantially greater than the 48% to 53% of year-round hospitalizations and 33% of emergency department visits estimated to result from rotavirus by indirect methods.
CONCLUSIONS. The withdrawn rotavirus vaccine was highly effective in preventing hospitalizations and emergency department visits for all-cause acute gastroenteritis and the health benefits of vaccination against rotavirus may be greater than previously estimated.
In February 2006, a new rotavirus vaccine (RotaTeq, Merck Vaccines, Whitehouse Station, NJ) was licensed in the United States and recommended for routine immunization of all US infants.1 To understand the potential health benefits of the US rotavirus vaccination program, knowledge of the burden of severe acute gastroenteritis in US children that is potentially preventable by vaccination is required.
Previous studies have estimated that rotavirus accounts for 30% to 50% of hospitalizations and 20% to 50% of emergency department (ED) visits for acute gastroenteritis among US children.2–8 Because testing for rotavirus is not routinely performed in children with acute gastroenteritis, these estimates of rotavirus disease burden have been derived primarily by using indirect approaches.9–11 For example, rotavirus hospitalizations in US children have been estimated by examining hospitalizations for acute gastroenteritis of all causes and then estimating the proportion attributable to rotavirus by either 1) applying rotavirus detection rates from active surveillance for laboratory-confirmed rotavirus infection among children hospitalized at a sentinel hospital (Brandt method)2 or 2) measuring excess hospitalizations during the winter months when rotavirus is prevalent relative to a summer baseline (residual method).10
In 1998, a previous rotavirus vaccine (RotaShield [Wyeth Lederle Vaccines, Philadelphia, PA]) was licensed and recommended for routine immunization of all US infants.12 Approximately 1 million doses of this vaccine were administered to 500 000 infants13 before it was withdrawn from the market in 1999 because of an association with intussusception, an uncommon but potentially serious adverse event.14,15 In prelicensure trials in industrialized countries, RotaShield was shown to be 80% to 100% efficacious against severe rotavirus disease16,17 and 2 postlicensure vaccine effectiveness studies using case-control methodology showed that the vaccine was 100% effective against severe rotavirus disease requiring hospitalization.18,19 Given the high effectiveness of RotaShield in preventing severe acute gastroenteritis caused by rotavirus, measuring its effectiveness in preventing severe acute gastroenteritis from all causes could provide a more clinically meaningful estimate of the potential health benefits of vaccination.
In this study, we retrospectively examined data from a large cohort of children enrolled in 6 US managed care organizations (MCOs) to assess the effectiveness of RotaShield vaccination in preventing hospitalizations and ED visits for acute gastroenteritis from all causes. Our findings suggest that the potential health benefits of rotavirus vaccination for US infants could be greater than previously estimated from disease-burden figures derived by using indirect approaches.
The Vaccine Safety Datalink (VSD) project was established in 1991 as a collaboration between the Centers for Disease Control and Prevention and several MCOs to monitor vaccine safety in the United States.20 Information on vaccination, medical outcomes, and covariates is gathered prospectively from clinical and administrative databases for the entire population enrolled in the VSD MCOs and enables calculation of rates of medical events in vaccinated and unvaccinated individuals. Most of the RotaShield vaccine administration in the US occurred between October 1998 and July 1999. We conducted a retrospective cohort analysis of children who were potentially age-eligible to have received RotaShield vaccine (ie, were born April 1, 1998 through May 31, 1999) and who were continuously enrolled from November 1, 1998 (when RotaShield vaccine was introduced) until the end of the 2 follow-up periods defined below in 1 of 6 MCOs (Fig 1). The continuous enrollment criterion was used to ensure all RotaShield vaccinations and acute gastroenteritis hospitalizations and ED visits among the eligible cohort were captured.
Two follow-up periods were examined. The first follow-up period was a full calendar year from August 1, 1999, through July 31, 2000. Because rotavirus infection peaks during winter months, a second follow-up period from November 1, 1999, through April 30, 2000, was also examined, because it was the first winter rotavirus season during which all cohort members would have been eligible to be vaccinated.
Ascertainment of Vaccination Status
Dates of vaccination for all children included in the cohort were used to determine the vaccination status as of August 1, 1999, as 0 (unvaccinated), 1, 2, or 3 doses. Because vaccination was suspended in July 1999, vaccination status of each eligible child did not change throughout the 2 follow-up periods.
Ascertainment of Outcomes
Outcomes of interest included hospital and ED discharges for all-cause acute gastroenteritis,21 including rotavirus, and were identified by using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. These ICD-9-CM codes included diarrhea of determined etiology (bacterial [001–005, excluding 003.2, and 008.0–008.5], parasitic [006–007, excluding 006.2–006.6], and viral [008.6 and 008.8, including rotavirus (008.61)]) and diarrhea of undetermined etiology (009.0–009.3, 558.9, and 787.91).
Because person-time for all children included in the cohort was equal as a result of the continuous enrollment criteria during a given follow-up period, the proportion of children with at least 1 acute gastroenteritis visit among all children for each dose group was used to calculate vaccine effectiveness (VE). We calculated VE as 1 minus the risk ratio (RR), where RR equals the RR of outcomes among children in different dose groups. Because the vaccine was not uniformly administered during the cohort eligibility period and because the risk of hospitalization and ED visit for acute gastroenteritis varies according to age of the child, logistic regression was used to calculate RRs adjusted for month of birth, gender, and MCO. VE was calculated for all-cause acute gastroenteritis hospitalizations and ED visits. VE was not calculated separately for hospitalizations and ED visits that were specifically coded as rotavirus because the rotavirus-specific ICD-9-CM code was assigned too infrequently at these MCOs to permit a meaningful assessment.
Assessment of Bias
Because of limited vaccine uptake, we were concerned that distribution of vaccine may be related to factors that may also be related to study outcomes of interest (eg, health-seeking behavior, socioeconomic status) that would bias our findings. Two approaches were used to assess the possibility of such bias. First, we assessed the effectiveness of RotaShield vaccine against hospitalizations and ED visits for acute respiratory illness (ARI) that were identified by using ICD-9-CM codes 480–487 and 466.1. We hypothesized that factors associated with increased risk and health care-seeking patterns of severe ARI are generally similar to those of acute gastroenteritis; thus, the lack of an association between vaccination and ARI would indicate absence of major bias. Second, we assessed vaccine effectiveness against acute gastroenteritis hospitalizations and ED visits that occurred during the nonrotavirus season from May 2000 through October 2000 among children continuously enrolled from November 1, 1998, through October 31, 2000. During the nonrotavirus season, very few acute gastroenteritis events are attributable to rotavirus, and thus only limited vaccine effectiveness would be expected in the absence of bias.22
Comparison of Rotavirus Burden Estimates
To investigate different methods of estimating the disease burden of rotavirus, we compared the number of rotavirus hospitalizations and ED visits for the unvaccinated study cohort determined by 2 previously used indirect approaches with the number based on the proportion of acute gastroenteritis hospitalizations and ED visits prevented by RotaShield vaccination. For the first estimate of rotavirus hospitalizations based on the indirect Brandt method, the age-specific numbers of all-cause acute gastroenteritis hospitalizations were multiplied by the age-specific proportion of rotavirus infections among all acute gastroenteritis admissions identified during a surveillance study at Children's Hospital National Medical Center from 1974 to 1982.2 For the second estimate of rotavirus hospitalizations and an estimate of rotavirus ED visits based on the indirect residual method, the number of summer (May through October) acute gastroenteritis hospitalizations and ED visits were subtracted from the number of winter (November through April) acute gastroenteritis hospitalizations and ED visits. The excess hospitalizations and ED visits during the winter season were attributed to rotavirus.10
Hospitalizations and Emergency Department Visits for Acute Gastroenteritis During the Entire Year
From August 1, 1999, through July 31, 2000, 64599 children who were age-eligible to receive 1 or more doses of RotaShield vaccine were continuously enrolled. These children were aged 2 to 16 months at the start of the study period. Approximately half (51%) of the study population was male. Overall, 13399 (21%) of the 64 599 children in the cohort received at least 1 dose of RotaShield vaccine, including 12% who received 1 dose only, 5% who received 2 doses only, and 3% who received all 3 doses. However, vaccine coverage varied substantially according to MCO and age. Vaccinated children were younger than unvaccinated children at the start of the study period. The median age of fully vaccinated (3 doses) children was 8 months, of children who received 2 doses was 6 months, of children who received 1 dose was 4 months, and of unvaccinated children was 10 months.
From August 1, 1999, through July 31, 2000, 371 (0.6%) children were hospitalized for all-cause acute gastroenteritis (Table 1). Among unvaccinated children, the rate was 6.5 per 1000 person-years. Rates of all-cause acute gastroenteritis hospitalization decreased incrementally among children vaccinated with increasing number of doses. VE against all-cause acute gastroenteritis hospitalization adjusted by month of birth, gender, and MCO for children with 1 dose of vaccine was 52% (95% confidence interval [CI]: 28% to 68%), for 2 doses was 70% (95% CI: 40% to 85%), and for fully vaccinated children with 3 doses was 83% (95% CI: 45% to 95%).
From August 1, 1999, through July 31, 2000, 1268 (2.0%) children visited an ED for all-cause acute gastroenteritis (Table 1). The rate was 20.5 per 1000 person-years among unvaccinated children and almost half that for children who received 3 doses of rotavirus vaccine. VE against all-cause acute gastroenteritis ED visit adjusted by month of birth, gender, and MCO for children with 1 dose of vaccine was 26% (95% CI: 10% to 40%), for 2 doses was 14% (95% CI: −13% to 34%), and for fully vaccinated children with 3 doses was 43% (95% CI: 15% to 61%).
Hospitalizations and Emergency Department Visits for Acute Gastroenteritis During the Winter Rotavirus Season
A total of 68 149 children aged 5 months to 19 months that were age-eligible to receive 1 or more doses of RotaShield vaccine were continuously enrolled from November 1, 1999, through April 30, 2000, the estimated 1999–2000 winter rotavirus season. During this season, 291 (0.4%) children in the cohort were hospitalized for all-cause acute gastroenteritis (Table 2). Among unvaccinated children, the seasonal rate was 9.8 per 1000 person-years. Seasonal rates of all-cause gastroenteritis decreased incrementally among children vaccinated with increasing doses of rotavirus vaccine. VE against all-cause acute gastroenteritis hospitalization during the rotavirus season adjusted by month of birth, gender, and MCO for children with 1 dose of vaccine was 54% (95% CI: 26% to 72%), for 2 doses was 76% (95% CI: 42% to 90%), and for fully vaccinated children with 3 doses was 85% (95% CI: 40% to 96%).
During the 1999–2000 winter season, 886 (1.3%) children visited an ED for all-cause acute gastroenteritis (Table 2). Among unvaccinated children, the seasonal rate was 27.3 per 1000 person-years; this rate decreased by almost half in children who received 3 doses of rotavirus vaccine. Compared with unvaccinated children, VE against all-cause acute gastroenteritis ED visits during the winter season adjusted by month of birth, gender, and MCO for children with 1 dose of vaccine was 36% (95% CI: 19% to 50%), for 2 doses was 36% (95% CI: 9% to 55%), and for fully vaccinated children with 3 doses was 49% (95% CI: 17% to 69%).
Assessment of Bias
RotaShield vaccine was not effective in preventing ARI hospitalizations and ED visits or significantly effective in preventing all-cause acute gastroenteritis hospitalizations and ED visits during the nonrotavirus season (Table 3). In addition, no dose response trend between the number of RotaShield doses and effectiveness against either of these 2 outcomes was observed.
Comparison of Disease-Burden Estimates
By using the indirect Brandt method, an estimated 53% (177 of 332) all-cause acute gastroenteritis hospitalizations from August 1999 through July 2000 among unvaccinated children would be attributed to rotavirus. By using the indirect winter residual method, an estimated 48% (160 of 332) all-cause acute gastroenteritis hospitalizations and 33% (349 of 1049) all-cause acute gastroenteritis ED visits between August 1999 and July 2000 would be attributed to rotavirus. By comparison, from our VE estimates, up to 83% (276 of 332) of all-cause acute gastroenteritis hospitalizations and up to 43% (451 of 1049) of all-cause acute gastroenteritis ED visits among unvaccinated children were prevented by a 3-dose vaccination series.
Our findings clearly demonstrate the potential impact of an effective rotavirus vaccine in preventing acute gastroenteritis hospitalizations and ED visits among US children. In the VSD study cohort, the effectiveness of a full 3-dose RotaShield series over an entire year of follow-up was 83% against all-cause acute gastroenteritis hospitalizations and 43% against all-cause acute gastroenteritis ED visits. Even less than a full 3-dose series of RotaShield vaccination seemed quite effective, with 2 doses of vaccine preventing 70% and 1 dose preventing 52% of all-cause acute gastroenteritis hospitalizations over the entire year. In comparison, by applying indirect methods used previously to estimate rotavirus burden, only 48% to 53% of year-round hospitalizations and 33% of ED visits for acute gastroenteritis in unvaccinated children in the VSD study would have been attributed to rotavirus and deemed potentially preventable by vaccination. Thus, the potential health benefits of the new US rotavirus vaccination program could be substantially greater than previously thought based on epidemiologic data alone.
The lack of VE against ARI hospitalizations and ED visits and against acute gastroenteritis events outside the rotavirus season argue against bias in our study. Furthermore, our figure of 83% VE of RotaShield against year-round acute gastroenteritis hospitalizations is consistent with the fact that the new rotavirus vaccines prevented 59% to 72% of all-cause acute gastroenteritis hospitalizations in recently completed clinical trials.23,24 Several factors might explain the greater observed effectiveness of rotavirus vaccines against acute gastroenteritis hospitalization and ED visit than that expected based on epidemiologic studies. First, active surveillance of children with acute gastroenteritis could underestimate rotavirus burden because of factors such as failure to obtain an adequate stool specimen from patients in a timely manner, improper handling or storage of specimens, or inadequate sensitivity of the enzyme immune assay commonly used for detection of rotavirus.25 For example, in the Brandt et al study, some tests were performed by using less-sensitive rectal swabs rather than bulk stool and some testing was performed by the less-specific electron microscopy technique, which likely resulted in an underestimate of the true rotavirus burden. In addition, the rotavirus season and disease burden many have changed during the 25 years since the Brandt et al study was performed. Second, in prelicensure trials, RotaShield seemed to protect to some extent against adenovirus gastroenteritis and also reduced the severity of gastroenteritis caused by sappoviruses and adenoviruses,26,27 which might account in part for the greater than anticipated VE, although the mechanism of effect is unknown. During the nonrotavirus season, the vaccine seemed to demonstrate some effectiveness against all-cause acute gastroenteritis, although it was not significant. This nonrotavirus season VE could be due to rotavirus cases that occurred outside of the November to April rotavirus season or perhaps because of cross protection against other enteric viruses. Finally, it is possible that vaccinated and nonvaccinated children were not randomly distributed in each MCO population. If vaccinated children were more likely to interact with other vaccinated children, they would be less frequently exposed to rotavirus than nonvaccinated children and this would exaggerate the measured effectiveness of vaccination.
Some limitations should be considered in the interpretation of our data. First, because the rotavirus-specific ICD-9-CM code was infrequently used, we were unable to specifically assess VE against outcomes coded as rotavirus gastroenteritis. Second, rotavirus testing was not routinely performed at these MCOs, so we were not able to verify the rotavirus season at these MCOs. However, available national surveillance data from a sentinel laboratory system indicate that the 1999–2000 rotavirus season was fairly typical, with onset in the months of November through December and decline in activity from April through May. Third, our estimates of the number of hospitalizations and ED visits prevented by vaccination will only equal the true burden of disease if the 3-dose vaccine efficacy is 100%. If this assumption is incorrect or if the vaccine indeed cross-protects against other pathogens, we may incorrectly estimate the burden of rotavirus disease; however, our estimates of the health benefits of vaccination would still be valid. Fourth, because the vaccine was available for only a short period of time, vaccine effectiveness could only be examined in children aged 5 to 19 months at the beginning of the winter season because these were the only children eligible to have received the vaccine. Rates of rotavirus disease are highest among children aged 4 to 23 months,28,29 and so the large impact observed in all-cause acute gastroenteritis hospitalizations and ED visits likely reflects a high proportion of rotavirus infection in the age-restricted cohort in our study and would likely be lower among children outside this age range. Finally, although we did not find evidence of bias, it is possible that some unaccounted differences in characteristics of vaccinated and unvaccinated children could have influenced our VE estimates.
Our findings show that the now-withdrawn RotaShield vaccine was highly effective in preventing hospitalizations and ED visits for all-cause acute gastroenteritis in US children. More importantly, the directly measured reduction in acute gastroenteritis hospitalizations and ED visits from RotaShield vaccination seems to be greater than what was anticipated based on previous disease-burden estimates. Thus, the potential health benefits of the new US rotavirus vaccination program may be greater than previously anticipated. The VSD should provide an excellent platform for monitoring hospitalizations and ED visits to assess the impact of new rotavirus vaccines in reducing the burden of severe acute gastroenteritis in US children.
We thank the following Vaccine Safety Datalink sites for participation in this study: Kaiser Permanente Northwest (Portland, OR); Kaiser Permanente Medical Care Program of Northern California (Oakland, CA); Southern California Kaiser Permanente Health Care Program (Los Angeles, CA); HealthPartners Research Foundation (Minneapolis, MN); Marshfield Clinic Research Foundation (Marshfield, WI); and Kaiser Permanente Colorado (Denver, CO).
- Accepted June 19, 2008.
- Address correspondence to Jacqueline E. Tate, PhD, Centers for Disease Control and Prevention, 1600 Clifton Rd, NE Mail Stop A47, Atlanta, GA 30333. E-mail:
The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
To understand the potential health benefits of the US rotavirus vaccination program, knowledge of the burden of severe acute gastroenteritis in US children that is potentially vaccine preventable is required. Previous rotavirus disease-burden estimates were based on indirect approaches.
What This Study Adds
Our findings demonstrate the potential impact of an effective rotavirus vaccine in preventing acute gastroenteritis hospitalizations among US children. The health benefits of rotavirus vaccination could be greater than previously estimated from disease-burden figures derived by using indirect approaches.
- ↵Brandt CD, Kim HW, Rodriguez WJ, et al. Pediatric viral gastroenteritis during eight years of study. J Clin Microbiol.1983;18 (1):71– 78
- Koopman JS, Turkish VJ, Monto AS, et al. Patterns and etiology of diarrhea in three clinical settings. Am J Epidemiol.1984;119 (1):114– 123
- Matson DO, Estes MK. Impact of rotavirus infection at a large pediatric hospital. J Infect Dis.1990;162 (3):598– 604
- Waters V, Ford-Jones EL, Petric M, et al. Etiology of community-acquired pediatric viral diarrhea: a prospective longitudinal study in hospitals, emergency departments, pediatric practices and child care centers during the winter rotavirus outbreak, 1997 to 1998. The Pediatric Rotavirus Epidemiology Study for Immunization Study Group. Pediatr Infect Dis J.2000;19 (9):843– 848
- Widdowson MA, Meltzer MI, Zhang X, et al. Cost-effectiveness and potential impact of rotavirus vaccination in the United States. Pediatrics.2007;119 (4):684– 697
- ↵Klein EJ, Boster DR, Stapp JR, et al. Diarrhea etiology in a children's hospital emergency department: a prospective cohort study. Clin Infect Dis.2006;43 (7):807– 813
- ↵Ho MS, Glass RI, Pinsky PF, Anderson LJ. Rotavirus as a cause of diarrheal morbidity and mortality in the United States. J Infect Dis.1988;158 (5):1112– 1116
- ↵Rennels MB, Glass RI, Dennehy PH, et al. Safety and efficacy of high-dose rhesus-human reassortant rotavirus vaccines: report of the National Multicenter Trial. United States Rotavirus Vaccine Efficacy Group. Pediatrics.1996;97 (1):7– 13
- ↵Perez Mato S, Perrin K, Scardino D, Begue RE. Evaluation of rotavirus vaccine effectiveness in a pediatric group practice. Am J Epidemiol.2002;156 (11):1049– 1055
- ↵Parashar UD, Holman RC, Clarke MJ, Bresee JS, Glass RI. Hospitalizations associated with rotavirus diarrhea in the United States, 1993 through 1995: surveillance based on the new ICD-9-CM rotavirus-specific diagnostic code. J Infect Dis.1998;177 (1):13– 17
- ↵Jackson LA, Jackson ML, Nelson JC, Neuzil KM, Weiss NS. Evidence of bias in estimates of influenza vaccine effectiveness in seniors. Int J Epidemiol.2006;35 (2):337– 344
- ↵Vesikari T, Karvonen A, Prymula R, et al. Efficacy of human rotavirus vaccine against rotavirus gastroenteritis during the first 2 years of life in European infants: randomised, double-blind controlled study. Lancet.2007;370 (9601):1757–1763
- ↵Stockman LJ, Staat MA, Holloway M, et al. Optimum diagnostic assay and clinical specimen for routine rotavirus surveillance. J Clin Microbiol.2008;46 (5):1842– 1843
- Copyright © 2009 by the American Academy of Pediatrics