Effectiveness of Pentavalent Rotavirus Vaccine in a Large Urban Population in the United States
OBJECTIVE: The goal was to assess the effectiveness of complete (3-dose) or partial (1- or 2-dose) immunization with pentavalent rotavirus vaccine (RV5) against rotavirus acute gastroenteritis (AGE) in US clinical practice.
METHODS: A case-control evaluation was conducted in February through June 2008 at an emergency department in Houston, Texas. Case patients with rotavirus AGE (N = 90) were identified through testing for rotavirus in fecal specimens obtained from 205 children 15 days through 23 months of age presenting with AGE. Control groups included rotavirus-negative AGE patients (N = 115), concurrently enrolled patients with acute respiratory infection (ARI) (N = 228), and up to 10 age- and zip code-matched children sampled from the Houston-Harris County Immunization Registry (HHCIR) for each case patient >8 months of age. Immunization data were obtained from parent records, health care providers, and/or the HHCIR. Vaccine effectiveness was calculated as 1 minus odds of RV5 vaccination for case patients versus control patients, after adjustment for age at presentation and birth date.
RESULTS: The vaccine effectiveness of a complete RV5 series was 89% (95% confidence interval [CI]: 70%–96%) and 85% (95% CI: 55%–95%) with rotavirus-negative AGE and ARI control patients, respectively. Immunization data were available for 44% of case patients (n = 40) from the HHCIR; the estimated 3-dose vaccine effectiveness with these HHCIR control patients was 82% (95% CI: 19%–96%). A complete RV5 series conferred 100% protection (95% CI: 71%–100%) against severe rotavirus disease requiring hospitalization and 96% protection (95% CI: 72%–99%) against disease requiring intravenous hydration. Vaccine effectiveness of 1 and 2 doses against hospitalization and emergency department visits was 69% (95% CI: 13%–89%) and 81% (95% CI: 13%–96%), respectively, using rotavirus-negative AGE and ARI control groups combined.
CONCLUSIONS: In this setting, a complete series of RV5 was highly effective against severe rotavirus AGE. Partial immunization also conferred substantial protection.
WHAT'S KNOWN ON THIS SUBJECT:
RV5 was licensed and recommended for routine immunization of US infants in February 2006. Results from prelicensure trials demonstrated high efficacy against rotavirus-associated physician visits, ED visits, and hospitalizations.
WHAT THIS STUDY ADDS:
We conducted a case-control evaluation of the effectiveness of complete or partial immunization with RV5 against rotavirus AGE in US clinical practice.
Before rotavirus vaccine introduction, nearly every US child suffered an episode of rotavirus acute gastroenteritis (AGE) by the age of 5 years, 1 in 7 visited a physician or an emergency department (ED) because of their illness, and 1 in 70 required hospitalization.1,–,5 Each year, 410 000 physician visits, 205 000 to 272 000 ED visits, and 55 000 to 70 000 hospitalizations because of rotavirus AGE occurred among US children <5 years of age, with total direct and indirect costs of approximately $1 billion.6 A pentavalent rotavirus vaccine (RV5) was licensed in February 2006 and was recommended for routine immunization of US infants, to prevent the health and economic burdens of rotavirus AGE, with 3 doses given at 2, 4, and 6 months of age. In a prelicensure trial, a full 3-dose series of RV5 was 98% efficacious against severe rotavirus AGE, reducing physician visits by 86%, ED visits by 94%, and hospitalizations by 96%.7
Although these data suggest that RV5 will significantly reduce the burden of rotavirus AGE in US children, the effectiveness of vaccination in clinical practice can differ from the efficacy observed in the ideal conditions of a clinical trial.8 Furthermore, only limited information is available on the effectiveness of partial (1- or 2-dose) immunization with RV5. To assess the effectiveness of full and partial immunization with RV5 in clinical practice, we established active surveillance to identify children presenting for treatment of rotavirus AGE at a large children's hospital ED, and we conducted a case-control evaluation. In addition to using traditional approaches of enrolling a comparison group of control patients and conducting extensive follow-up investigation to obtain immunization data, we determined whether data from immunization registries can be used to evaluate vaccine effectiveness.
In February through June 2008, we conducted active surveillance for rotavirus AGE in the ED and inpatient floors at Texas Children's Hospital. Children were enrolled by using convenience sampling including evening and weekend hours, totaling ∼60 hours per week. The ED triage log was used to identify children with a chief complaint of vomiting or diarrhea. Children 15 days through 23 months of age with symptoms of AGE (diarrhea and/or vomiting) were offered participation; if the parents were willing, then consent for enrollment was obtained. Diarrhea was defined as the occurrence of ≥3 loose stools within a 24-hour period, whereas vomiting was defined as ≥1 episode within a 24-hour period. For comparability with other rotavirus surveillance programs, patients with gastrointestinal symptoms for ≥11 days were ineligible.9 In addition, patients who were immunocompromised, patients who did not reside within the greater Houston, Texas, area, and patients whose parents did not speak either English or Spanish were excluded from participation.
We interviewed the parents/guardians to assess demographic characteristics, symptom duration, and illness severity. When available, a parent copy of the child's immunization record was obtained. Permission to search the local immunization registry and to contact the child's immunization providers was obtained. Soiled diapers were collected to obtain fecal specimens. Fecal specimens were stored at 4°C to 8°C until they were tested for rotavirus with a commercial enzyme immunoassay (Premier Rotaclone; Meridian Bioscience, Cincinnati, OH). Rotavirus-positive specimens were characterized at the Centers for Disease Control and Prevention for determination of the G- and P-types by using reverse transcription-polymerase chain reaction and nucleotide sequencing methods, as described previously.10,–,13
Three control groups were selected. The first group included children 15 days through 23 months of age enrolled with AGE symptoms whose fecal specimen tested negative for rotavirus. The use of children who test negative for the disease of interest as control patients was demonstrated in other vaccine effectiveness studies to provide results similar to those obtained with the use of traditional control groups.14,–,16 The second group was composed of children in the same age range who presented to the ED with symptoms of acute respiratory infection (ARI) who did not meet AGE inclusion criteria. Exclusion criteria similar to those of AGE case patients were applied to ARI control patients, and consent to collect immunization information in a manner similar to that used for AGE case patients was obtained. For the third control group, up to 10 children matched to case patients according to date of birth (±30 days) and zip code of residence were selected from the Houston-Harris County Immunization Registry (HHCIR). The HHCIR is a local comprehensive repository of immunization data for children and includes immunizations administered by pubic and private providers.
Immunization Record Collection
Immunization records were requested from several sources. A copy of the child's immunization record was obtained from the parent during enrollment, if available. In addition, records were requested from the immunization providers identified by the parents/guardians during enrollment. Staff members searched the HHCIR for all enrolled patients, and immunization data were obtained for those with established records.
Analyses were performed by using SAS 9.1 (SAS Institute, Cary, NC). Assuming RV5 vaccination coverage of 30% among control patients, we estimated that ≥60 case patients and ≥120 control patients would be necessary to determine vaccine effectiveness of >75% with 90% power.
We used χ2 analyses to compare characteristics of children who were tested for rotavirus with those who were not and characteristics of children with rotavirus-positive specimens with those with rotavirus-negative specimens. Medians were compared by using the Wilcoxon rank-sum test. For analyses using rotavirus-negative AGE and ARI control patients, we used unconditional logistic regression, controlling for age at presentation and month and year of birth, to calculate adjusted odds ratios of vaccination for full and partial vaccination with RV5 (versus no vaccination).17 For case patients and rotavirus-negative AGE and ARI control patients, a dose of vaccine was counted if it was administered ≥14 days before presentation. Both case patients and control patients for whom an immunization record could not be obtained were excluded from analysis. For calculations of vaccine effectiveness, the number of doses of RV5 was defined as the greatest number of doses received from any source.
We examined vaccine effectiveness by using each control group separately and also with the 2 control groups combined. Vaccine effectiveness according to dose was calculated as (1 − odds ratio of vaccination) × 100 for children who were age-eligible to receive the vaccine. To assess protection against illness of varying severity, we examined effectiveness against cases requiring hospitalization and intravenous hydration therapy. Furthermore, we assessed protection against illness of varying severity, classified by using a previously described, 20-point, numerical scoring system18; illnesses with scores of ≥11 were classified as severe.
For the analyses using HHCIR control patients, we used conditional logistic regression to calculate the odds ratio of vaccination with a full series of RV5 (versus no vaccination), because control patients were matched to case patients according to date of birth and zip code. Because delays between administration of RV5 and data entry into the HHCIR might have occurred, analysis was restricted to case patients ≥8 months of age. For control patients enrolled from the HHCIR, a dose of vaccine was counted if it was administered ≥14 days before the date when the control patient was the same age as the matched case patient at the time of clinical presentation.
Because of limited vaccine uptake, we were concerned that receipt of vaccine might be related to factors (eg, health care-seeking behavior or socioeconomic status) that might be related to outcomes of interest, which would bias our findings. To examine the possibility of such bias, we assessed the effectiveness of RV5 against hospitalizations and ED visits for nonrotavirus AGE. We hypothesized that factors associated with increased risk and health care-seeking patterns for nonrotavirus AGE would be generally similar to those for rotavirus AGE; therefore, lack of an association between vaccination and nonrotavirus AGE would indicate an absence of major bias. We assessed the vaccine effectiveness of RV5 against hospitalizations and ED visits for nonrotavirus AGE by using the ARI control patients and matched control patients from the immunization registry.
Approval was obtained from the institutional review boards at Baylor College of Medicine and Texas Department of State Health Services.
We identified 608 potentially eligible children, of whom 403 (66%) were enrolled (Fig 1). Among the 205 nonenrolled patients, the most common reasons were parent refusal (n = 82) and symptom ineligibility (n = 77). Three patients were excluded subsequently, which left a total of 400 children who were included in the evaluation.
Of the 400 children in the evaluation, 57% (n = 288) were male and 69% (n = 277) were Hispanic. Fecal specimens to test for rotavirus were available for 205 patients (51%). Patients who provided a fecal specimen did not differ significantly from those who did not with regard to gender, ethnicity, or age but were significantly more likely to be admitted for hospitalization, to report subjective fever, to have diarrhea and fever at the time of enrollment, to have had a previous visit to a doctor's office (Table 1), and to have a greater number of stools in a 24-hour period.
Ninety (44%) of the 205 children with available fecal specimens tested positive for rotavirus. Compared with children who tested negative for rotavirus, those who tested positive did not differ significantly with regard to gender, ethnicity, prevalence of diarrhea, or need for inpatient admission (Table 2). However, rotavirus-positive children were more likely to report previous treatment for fever, oral treatment for dehydration, and a higher maximal number of vomiting episodes in a 24-hour period, were more likely to be described by their parents as sleepy and less playful, and were more likely to have vomiting, compared with rotavirus-negative children.
Rotavirus-positive children were significantly older than both rotavirus-negative and ARI control patients, with a median age of 17 months, compared with 10 months (P < .001) and 8 months (P < .001), respectively. No significant differences with respect to gender, race, or ethnicity were observed among the 3 groups. Therefore, the 2 control groups were pooled for some analyses. Among all tested patients with AGE, the rotavirus detection rates increased with increasing age, whereas the proportion of children who had received RV5 decreased with increasing age (Fig 2).
Among the 90 rotavirus-positive patients, an adequate amount of fecal specimen for viral strain characterization was available for 62 patients. The most common genotypes were G3P (50%) and G1P (26%); 11% were mixed infections (Fig 3).
Immunization information could be obtained for 79 (88%) of 90 rotavirus-positive children, 206 (90%) of 228 ARI control patients, and 108 (94%) of 115 rotavirus-negative control patients; vaccine effectiveness was calculated by using data for these patients. Although we assessed vaccination by using combined data from several sources, ∼90% of vaccination data for the case patients and rotavirus-negative AGE and ARI control patients were obtained from provider-verified vaccination records.
Age-adjusted vaccine effectiveness against hospitalization and ED visits for a full 3-dose series of RV5 was 89% (95% confidence interval [CI]: 70%–96%) and 85% (95% CI: 55%–95%) with rotavirus-negative AGE and ARI control patients, respectively, and 88% (95% CI: 68%–96%) with the 2 control groups combined (Table 3). With the 2 control groups combined, vaccine effectiveness for 2 doses of RV5 was 81% (95% CI: 13%–96%) and that for 1 dose was 69% (95% CI: 13%–89%). The vaccine effectiveness values for partial immunization were similar when estimated separately with 1 of the control groups, although neither estimate was statistically significant.
Vaccine effectiveness among patients who required hospitalization and intravenous hydration was examined. Of the 90 rotavirus-positive patients, 16 (18%) were hospitalized. Immunization data were available for 15 of those hospitalized patients, and none had received a dose of RV5. In comparison, 6 (46%) of 13 hospitalized, rotavirus-negative patients with AGE with available immunization data (P = .004) and 17 (47%) of 36 hospitalized patients with ARI (P = .001) had received ≥1 dose of RV5. This difference in coverage indicates that vaccine conferred 100% (95% CI: 72%–100%) protection against severe disease requiring hospitalization. Among children who required intravenous hydration, 3-dose vaccine effectiveness was 95% (95% CI: 57%–99%) and 97% (95% CI: 74%–100%) with ARI and rotavirus-negative AGE control patients, respectively, and 96% (95% CI: 72%–99%) with the 2 control groups combined. The total number of patients who required hospitalization and intravenous hydration therapy was insufficient to permit meaningful assessment of the effectiveness of partial immunization.
With the use of a 20-point severity scale, only 16 (18%) of 90 rotavirus-positive cases scored ≤10, which precluded meaningful assessment of vaccine effectiveness against rotavirus AGE of this severity. Against rotavirus AGE with a severity score of ≥11, vaccine effectiveness for a complete 3-dose series was 84% (95% CI: 50%–95%) and 88% (95% CI: 65%–96%) with ARI and AGE control patients, respectively, and 87% (95% CI: 63%–95%) with the 2 control groups combined. With the 2 control groups combined, vaccine effectiveness against rotavirus AGE with a severity score of ≥11 was 89% (95% CI: 15%–99%) for 2 doses of vaccine and 79% (95% CI: 25%–94%) for 1 dose.
Among case patients infected with the G3P rotavirus strain, the 3-dose vaccine effectiveness against all rotavirus AGE was 95% (95% CI: 27%–100%) and 95% (95% CI: 60%–99%) with ARI and rotavirus-negative control patients, respectively, and 95% (95% CI: 57%–99%) with the 2 control groups combined. For all strains other than G3P combined, 3-dose vaccine effectiveness against all rotavirus AGE was 91% (95% CI: 29%–99%) and 93% (95% CI: 41%–99%) with ARI and rotavirus-negative control patients, respectively, and 92% (95% CI: 40%–99%) with the 2 control groups combined.
Vaccination data were available from the HHCIR for 40 (44%) of 90 rotavirus case patients. Of the 50 case patients for whom immunization data were not available in the HHCIR, none had records established in the system and, with the exception of gender, no significant differences were observed between patients with and without HHCIR records. Two (5%) of the 40 case patients included in the registry had been vaccinated with 3 doses of RV5, compared with 68 (18%) of 386 age-matched control patients selected from the HHCIR, which yielded a 3-dose vaccine effectiveness against any rotavirus AGE of 82% (95% CI: 19%–96%). Results from the bias assessment indicated that RV5 did not confer protection against nonrotavirus AGE for the ARI control patients (vaccine effectiveness: −23% [95% CI: −126% to 33%]) or the registry-matched control patients (41% [95% CI: −75% to 80%]).
In this first postlicensure assessment of the performance of RV5 in US clinical practice, a complete 3-dose series of RV5 was 85% to 89% effective against rotavirus AGE requiring an ED visit or inpatient admission. This value is comparable to but slightly lower than the efficacy of 94% to 96% against rotavirus-associated ED visits and hospitalizations noted in the prelicensure trial.7 The effectiveness of vaccination might be slightly lower in our assessment because we evaluated children with less-severe rotavirus disease, compared with the prelicensure trial. Studies have shown that use of the ED as a medical home is sometimes preferred, especially among Hispanic patients, who constituted the vast majority of our population.19,20 When we examined effectiveness against more-severe disease, 3 doses of RV5 were 100% effective against rotavirus-associated AGE hospitalizations and 95% to 97% effective against rotavirus AGE requiring intravenous hydration. It is also possible that variations in vaccine administration practices that commonly occur in office settings, as well as limited misclassification of disease and vaccination status in our assessment, contributed to the slightly reduced effectiveness. Overall, our data confirm the high effectiveness of RV5 against severe rotavirus disease, including disease caused by the G3P rotavirus strain, which was relatively uncommon in the prelicensure trial.
Importantly, our findings indicate that partial immunization with RV5 confers good protection against severe rotavirus disease. When data for our 2 control groups were combined, immunization with 1 or 2 doses conferred protection of 69% or 81%, respectively. Against more-severe disease (severity score of ≥11), 1 or 2 doses of vaccine conferred even greater protection (79% and 89%, respectively). Although significant estimates of vaccine effectiveness for partial immunization with 1 or 2 doses were not obtained in separate analyses with rotavirus-negative and ARI control patients, the point estimates were similar for the 2 comparisons. Approximately 10% to 15% of severe rotavirus disease in US children <5 years of age occurs before the age of 6 months, when a child can be fully immunized,6 and the benefits of partial immunization are key for the prevention of severe rotavirus AGE in young infants and children who do not complete their vaccination series. Furthermore, cost-effectiveness analyses of RV5 conducted before vaccine implementation in the United States empirically assumed protection from 1 or 2 doses of vaccine as 50% of that from a full series21; those analyses may need to be revised on the basis of the greater protective efficacy of partial immunization observed in this assessment.
This evaluation demonstrates the potential utility of an immunization registry for evaluation of vaccine effectiveness. Although immunization data for only 44% of case patients were available from the HHCIR, the effectiveness of a complete 3-dose series of RV5 was estimated to be 82% by using data from the HHCIR alone, which was quite similar to the estimates obtained in analyses with nonrotavirus AGE (89%) and ARI (85%) control patients. The advantages of this novel approach include minimization of staff time for control patient enrollment, reduction in expenses, potential access to robust amounts of patient immunization data, and avoidance of bias resulting from selection of a control group that does not well represent the source population of case patients. Several states have ≥1 immunization registry similar to the HHCIR, and the effectiveness of a vaccine program could be assessed rapidly, with reasonable confidence and substantial economic efficiency, with those data. Because the quality and completeness of information in registries might vary, however, not all registries would be appropriate for vaccine effectiveness evaluations. Efforts to assess potential bias, such as those described here, should be pursued as part of future evaluations.
This observational evaluation has some limitations. First, we were unable to obtain vaccination records for 12% of case patients and 10% and 6% of ARI and rotavirus-negative AGE control patients, respectively. Any potential differences in loss of records might have affected our estimates of vaccine effectiveness. However, reasonably comparable estimates of effectiveness were obtained in analyses using the 3 separate control groups, which argues against any major bias. Second, because a large proportion of our patients with AGE were discharged from the ED within a few hours after admission, we were unable to obtain fecal specimens for approximately one half of all enrolled case patients. Whereas the 2 groups were similar with respect to other characteristics, patients with AGE from whom fecal specimens were obtained had clinically more severe illness than did those from whom specimens could not be obtained. We do not think our vaccine effectiveness estimates are biased by differential enrollment, although they likely reflect effectiveness against rotavirus AGE of above-average severity in the ED setting. Third, the evaluation was conducted at a single center with a predominantly Hispanic population, which may affect the generalizability of our findings. Fourth, permission to review the medical records of patients who declined participation or who were not approached was not obtained; therefore, participants could not be compared with nonparticipants. Fifth, because we did not conduct active surveillance at this hospital before 2008, we were unable to compare the age-specific detection rates and overall numbers of rotavirus and AGE cases in 2008 with the epidemiological features of disease in the prevaccine era.
These data suggest that RV5, as used in US clinical practice, is highly effective against severe rotavirus AGE and that even partial immunization confers substantial health benefits. Because of the limitations of this study, these findings should be confirmed with additional studies conducted in other US clinical practice settings. Because a monovalent rotavirus vaccine was recommended for use in US infants in June 2008, it also will be necessary to determine the effectiveness of monovalent rotavirus vaccine and RV5 when used simultaneously in the community and in combination for individual patients. In addition, as vaccinated children grow older, it will be necessary to document the duration of protection conferred by vaccination. Furthermore, given the wide variety of heterotypic rotavirus genotypes not included in current vaccines and the constant evolution of new strains, it will be important to monitor protection against all rotavirus strains and to determine whether changes in circulating strains occur as a result of vaccine selection pressure in the postvaccination era.
This work was funded by a sole-source grant from the Centers for Disease Control and Prevention that was awarded to the Houston Department of Health and Human Services and then to Texas Children's Hospital. The Centers for Disease Control and Prevention were involved in securing of funding and all aspects of study design, conduct of the study, analysis and interpretation of the data, and preparation and review of the manuscript.
We thank Lizangela Acevado-Gonzalez, Betsy Mayes, Gabrielle Jackson, Chardria Trotter, and Deyanira Verdejo for their capable work in recruiting patients and collecting data; Tara K. Kerin for identification of untypable strains through nucleotide sequencing; Virginia Moyer, MD, for helpful comments regarding the study design; and the Houston Department of Health and Human Services for assistance in procuring funding.
- Accepted August 10, 2009.
- Address correspondence to Julie A. Boom, MD, Baylor College of Medicine, Center for Vaccine Awareness and Research, Texas Children's Hospital, 6701 Fannin St, CC1540, Houston, TX 77030. 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.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- ED =
- emergency department •
- AGE =
- acute gastroenteritis •
- ARI =
- acute respiratory infection •
- HHCIR =
- Houston-Harris County Immunization Registry •
- RV5 =
- pentavalent rotavirus vaccine •
- CI =
- confidence interval
- Fischer TK,
- Viboud C,
- Parashar U,
- et al
- Malek MA,
- Curns AT,
- Holman RC,
- et al
- Parashar UD,
- Holman RC,
- Clarke MJ,
- Bresee JS,
- Glass RI
- Payne DC,
- Staat MA,
- Edwards KM,
- et al
- Das BK,
- Gentsch JR,
- Cicirello HG,
- et al
- Gentsch JR,
- Glass RI,
- Woods P,
- et al
- Gouvea V,
- Glass RI,
- Woods P,
- et al
- Orenstein EW,
- De Serres G,
- Haber MJ,
- et al
- Hosmer DW Jr.,
- Lemeshow S
- Brousseau DC,
- Hoffmann RG,
- Nattinger AB,
- Flores G,
- Zhang Y,
- Gorelick M
- Widdowson MA,
- Meltzer MI,
- Zhang X,
- et al
- Copyright © 2010 by the American Academy of Pediatrics