Over the 2007 and 2008 rotavirus seasons, 1 dose of RV5 was associated with an 88% VE against RGE hospitalizations and ED visits combined (95% CI: 45%–99%; Table 3). The magnitudes of VE were similar when the analysis was restricted to infants who received only 1 dose of their respective vaccines (ie, did not receive a subsequent dose) (VE: 84%, 95% CI: <0%–99%). In the outpatient setting, VE against RGE was 100% (95% CI: 54%–100%). Results of analyses restricted to infants who received only 1 dose of their respective vaccines were again similar. In a supplementary analysis, we expanded the outcome definition to diagnostic codes recorded in all positions of the claims; VE was somewhat lower for both the combined hospitalization and ED setting (VE: 78%, 95% CI: 27%–95%) and the outpatient setting (VE: 73%, 95% CI: 8%–94%). Adjusted estimates of VE for RGE were not calculated because of the small number of events among the infants receiving RV5.
One dose of RV5 was associated with a 44% VE against AGE identified in the combined hospitalization and ED setting (95% CI: 18%–62%; Table 2). The VE was somewhat higher among those who received only 1 dose of their respective vaccines (VE: 67%, 95% CI: 34%–84%). The regression result adjusted for age was similar (VE: 42%, 95% CI: 16%–60%), as was the result adjusted for sex and calendar year. In the outpatient setting, VE against AGE was 17% (95% CI: 6%–23%). The sex-adjusted and calendar-adjusted VE estimate was also similar, as was the adjusted estimate accounting for age, sex and rotavirus season (VE: 13%, 95% CI: 2%–23%), as were the results of analyses restricted to infants who received only 1 dose of their respective vaccines.
In the same period, 2 doses of RV5 was associated with a 94% VE against RGE hospitalizations and ED visits combined (95% CI: 61%–100%; Table 4). The magnitudes of VE were similar when the analysis was limited to those who received only 2 doses of their respective vaccines (VE: 100%, 95% CI: <0%–99%). In the outpatient setting, VE against RGE was 40%, although the CI included zero. In a supplementary analysis that expanded the outcome definition to include diagnosis codes in any claim position (not just the primary), the VE remained similar for both the combined hospitalization and ED setting (VE: 94%, 95% CI: 61%–100%) and the outpatient setting (VE: 40%, 95% CI: <0%–88%). Adjusted estimates of VE for RGE were not calculated because of the small number of events among the infants receiving RV5.
The VE associated with 2 doses of RV5 against AGE identified in the combined hospitalization and ED setting was 40% (95% CI: 18%–56%; Table 3). The regression result adjusted for age was similar (VE: 37%, 95% CI: 16%–53%), as was the result adjusted for sex and calendar year. In the outpatient setting, VE against AGE was 31% (95% CI: 24%–38%). The sex-adjusted and calendar year–adjusted VE estimate was also similar, as was the adjusted estimate accounting for age, sex and rotavirus season (VE: 26%, 95% CI: 18%–33%), as were the results restricted to infants who received only 2 doses of their respective vaccines.
In this large, nationwide, observational study, an incomplete RV5 regimen consisting of a first or first and second dose was found to be effective in preventing rotavirus gastroenteritis through the 2007 to 2008 rotavirus seasons. Our study yielded estimates of vaccine effectiveness that were similar to those of prelicensure clinical trials and postmarketing effectiveness studies.8–11 In a post hoc analysis of a clinical trial, efficacy between doses 1 and 2 was estimated at 82% whereas efficacy between doses 2 and 3 was estimated at 84% (compared to our findings for RGE of 88% and 94%, respectively).14 A hospital-based case-control study with laboratory testing for rotavirus reported VE against hospital or ED RGE of 69% for 1 dose and 81% for 2 doses of RV5.10 The broad representativeness of our large, underlying database population increases the likelihood that these results are generalizable to real-world scenarios where infants may inadvertently not complete the recommended RV5 vaccination series. Although the approved indication of the vaccine is 3 doses, multiple factors could contribute to an infant not completing a vaccine regimen. Therefore, these results are of particular relevance to healthcare professionals when assessing the impact of a partially completed RV5 series.
The majority of infants in both cohorts received 3 doses of their corresponding cohort vaccine in accordance with the recommended vaccine schedule, with an average length of 6 weeks between the first and the second dose, and an average length of 7 weeks between the second and the third dose. Thus, the estimates of effectiveness most directly reflect short-term protection. In an exploratory analysis restricted to infants who received only 1 dose (average follow-up for RV5 infants: 14.8 weeks) and those who received only 2 doses (average follow-up for RV5 infants: 12.6 weeks), we found similar trends of vaccine effectiveness for rotavirus gastroenteritis as the main analysis. Although the exploratory analysis results may be biased as it is conditioned on the infants not receiving any other doses, the results do provide some reassurance that the short-term protection is sustained beyond the 6-week period.
In this study, we assessed VE during the 5 months of the RV season because such estimates are likely to have the most public health relevance with regards to assessing impact of vaccination on disease burden. As there may be differences in VE during rotavirus seasons of varying severity, it would be of interest to assess potential differences in VE between the 2 rotavirus seasons in our study, as the US 2008 season was milder than previous seasons, as measured by RGE incidence. Unfortunately, we were not able to report separate VE estimates for each of the season because of the small number of cases among the RV5-vaccinated cohort in 2008.
Although there are 2 rotavirus vaccines currently licensed in the United States (RV5 [RotaTeq] and RV1 [Rotarix, GlaxoSmithKline Biologicals, Rixensart, Belgium]), misclassification of rotavirus vaccine dose because of mixed vaccination schedules (infants receiving RV1 instead of RV5) in the study is unlikely. Our study accrued infants who received a first dose from February 2006 through December 2007, with follow-up until May 2008. Since RV1 was approved in April 2008 and not marketed until August 2008,15 RV5 was the sole rotavirus vaccine marketed in the United States throughout the study period.
The observed VE against AGE hospitalizations and ED visits combined was somewhat higher than VE against the same outcomes when they are observed in the outpatient setting. This is consistent with the observed higher efficacy in clinical trials for more severe disease.14,16 In addition, these results may be indicative of greater precision in the use of diagnosis codes for RGE and AGE in the hospital or ED setting compared to the outpatient setting.17
Although relatively few cases of RGE were identified by the ICD-9 diagnosis code for rotavirus, the identified cases are highly likely to be actual RGE cases given that this code is reported to have high positive predictive value, albeit low sensitivity. Hsu et al18 found although 98% of patients with a rotavirus-associated discharge code were confirmed to be rotavirus positive by laboratory testing, only 25% with rotavirus-positive tests were assigned a rotavirus code. Hence, the reported RGE incidence relying solely on this code is likely to be substantially underestimated in this study.
In the absence of laboratory confirmation, epidemiologic studies commonly identify RGE through hospital discharge data by ICD-9 diagnosis codes for AGE that capture diarrhea of determined and undetermined etiology.18,19 It has been reported that 52% of children <5 years of age with a hospital discharge diagnosis of AGE tested positive for rotavirus,18 and that a higher proportion of AGE in children less than age 5 years is related to the rotavirus in the winter months of January through May.19 Thus, by evaluating the incidence of AGE during only the rotavirus seasons, this study increased the concordance between AGE diagnoses and RGE, although AGE incidence as identified in this study is expected to lack some sensitivity as a measure of RV5 VE because some portion of AGE is unrelated to rotavirus.
In this study, the main analysis identified health outcomes through diagnosis codes recorded in the primary position. Although this improved the probability that the study-outcome variables captured only RGE and AGE events, we note that some study outcomes were not counted as the ICD-9 code was recorded in a nonprimary position. This may have led to an underestimated incidence rate, but it is unlikely to have affected our measures of vaccine effectiveness, unless RV5 recipients had a different concordance between the claim code position and the underlying clinical diagnosis from DTaP recipients. As a demonstration of this point, our secondary analyses using codes in any position yielded similar vaccine effectiveness estimates.
Because of these factors, we note that there is likely incomplete ascertainment of both all-cause gastroenteritis and rotavirus gastroenteritis cases within our study, resulting in a lower absolute incidence than actually present in the population. However, the measure of vaccine effectiveness is a relative one, so it will be much less affected by incomplete ascertainment of gastroenteritis with corresponding underestimation of absolute incidence, unless the incomplete ascertainment is substantially different among RV5-exposed infants. For example, if physicians preferentially do not diagnose RGE among RV5 recipients, our measures of VE would be overestimates. For the hospitalized and ED-treated cases, treating physicians are likely unaware of the RV5 vaccination status of the infants. Indeed, the estimates of vaccine effectiveness in this study are similar to those of a prelicensure clinical trial and postmarketing effectiveness studies, external comparisons that suggest a lack of substantial bias in the estimates obtained in this study.10,11,14
There are limitations associated with the use of claims data for healthcare research. Misclassification of exposure is a possibility if all vaccination doses were not captured in the claims database. In this circumstance, estimates of vaccine effectiveness may have been inflated if RV5 doses were missed. However, the similarities of VE estimates in this study compared with other reports provide further assurance that the RGE VE estimates are relatively unbiased.10,14 The diagnosis code on a medical claim may not be definitive because the code may represent a rule-out diagnosis, or may be incorrectly recorded. Diagnosis of RGE requires laboratory confirmation, and laboratory results were not available for this study. Because insurance claims represent financial transactions that translate into reimbursement for providers, a financial incentive exists for providers and insurers to record them correctly, so the billable medical services represented in the database (both vaccine exposure and study outcomes) are likely to be complete.
Finally, measures of association should be interpreted with caution as there may be unmeasured factors associated with both RV5 administration and gastroenteritis, which were not taken into account in the analyses and may bias the observed measures. Because the early differential in RV5 uptake was likely based on provider and governmental characteristics such as initial vaccine availability and state purchasing patterns rather than patient characteristics, we were able to construct a relatively comparable comparison group for the evaluation of effectiveness. As infants were members of the same commercial health plan, they likely had similar socioeconomic backgrounds. There were no substantial demographic differences between the 2 cohorts, and both had similar baseline clinical features, reflecting comparability of healthcare access and utilization. Therefore, the compared groups should have similar exposure to the rotavirus and underlying risk for RGE. Overall, crude and adjusted estimates of VE were similar, suggesting that strong confounding from an unmeasured source is unlikely.
In summary, an incomplete RV5 vaccination regimen was associated with lower incidence of RGE and AGE through the rotavirus seasons. Further research is needed to assess whether the reported magnitude of VE with 1 or 2 RV5 doses is observed in other populations or healthcare settings.
All authors fulfilled the conditions of authorship and have approved the article as submitted.
1. Parashar UD, Hummelman EG, Bresee JS, et al. Global illness and deaths caused by rotavirus
disease in children. Emerging Infect Dis. 2003;9:565–572
2. Tucker AW, Haddix AC, Bresee JS, et al. Cost-effectiveness analysis of a rotavirus immunization
program for the United States. JAMA. 1998;279:1371–1376
3. Widdowson MA, Meltzer MI, Zhang X, et al. Cost-effectiveness and potential impact of rotavirus
vaccination in the United States. Pediatrics. 2007;119:684–697
4. Parashar UD, Alexander JP, Glass RIAdvisory Committee on Immunization
Practices (ACIP), Centers for Disease Control and Prevention (CDC). . Prevention of rotavirus gastroenteritis
among infants and children. Recommendations of the Advisory Committee on Immunization
Practices (ACIP). MMWR Recomm Rep. 2006;55(RR-12):1–13
5. Griffin DD, Kirkwood CD, Parashar UD, et al. Surveillance of rotavirus
strains in the United States: identification of unusual strains. The National Rotavirus
Strain Surveillance System collaborating laboratories. J Clin Microbiol. 2000;38:2784–2787
6. Abdel-Haq NM, Thomas RA, Asmar BI, et al. Increased prevalence of G1P genotype among children with rotavirus
in metropolitan Detroit. J Clin Microbiol. 2003;41:2680–2682
7. RotaTeq® [package insert]. 2007 Whitehouse Station, NJ Merck and Co, Inc
8. Vesikari T, Matson DO, Dennehy P, et al.Rotavirus
Efficacy and Safety Trial (REST) Study Team. Safety and efficacy of a pentavalent human-bovine (WC3) reassortant rotavirus
vaccine. N Engl J Med. 2006;354:23–33
9. Wang FT, Mast TC, Glass RJ, et al. Effectiveness of the pentavalent rotavirus
vaccine in preventing gastroenteritis
in the United States. Pediatrics. 2010;125:e208–e213
10. Boom JA, Tate JE, Sahni LC, et al. Effectiveness of pentavalent rotavirus
vaccine in a large urban population in the United States. Pediatrics. 2010;125:e199–e207
11. Cortes JE, Curns AT, Tate JE, et al. Rotavirus
vaccine and health care utilization for diarrhea in U.S. children. N Engl J Med. 2011;365:1108–1117
12. Centers for Disease Control and Prevention (CDC). . Delayed onset and diminished magnitude of rotavirus
activity–United States, November 2007-May 2008 MMWR Morb Mortal Wkly Rep. 2008;57:697–700
13. Ewell M. Comparing methods for calculating confidence intervals for vaccine efficacy. Stat Med. 1996;15:2379–2392
14. Dennehy PH, Vesikari T, Matson DO, et al. Efficacy of the pentavalent rotavirus
vaccine, RotaTeq® (RV5), between doses of a 3-dose series and with less than 3 doses (incomplete regimen). Hum Vaccin. 2011;7:563–568
15. Rotarix® [prescribing information]. 2011 Rixensart, Belgium GlaxoSmithKline Biologicals
16. 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;370:1757–1763
17. Mast TC, Walter EB, Bulotsky M, et al. Burden of childhood rotavirus
disease on health systems in the United States. Pediatr Infect Dis J. 2010;29:e19–e25
18. Hsu VP, Staat MA, Roberts N, et al. Use of active surveillance to validate international classification of diseases code estimates of rotavirus
hospitalizations in children. Pediatrics. 2005;115:78–82
19. Patel MM, Tate JE, Selvarangan R, et al. Routine laboratory testing data for surveillance of rotavirus
hospitalizations to evaluate the impact of vaccination. Pediatr Infect Dis J. 2007;26:914–919
Keywords:© 2013 Lippincott Williams & Wilkins, Inc.
rotavirus; vaccines; gastroenteritis; immunization