Association of Rotavirus Vaccines With Reduction in Rotavirus Gastroenteritis in Children Younger Than 5 Years

Key Points

Question

Is there a strong association of rotavirus vaccines and preventing rotavirus gastroenteritis (RVGE)?

Findings

Meta-analysis revealed that Rotarix and RotaTeq reduced RVGE in children younger than 5 years by 68.4% and 63.6%, respectively, and this was confirmed in case-control studies (65.3% and 72.8%, respectively). Adjusted indirect comparisons indicated no significant differences in the protection of Rotarix and RotaTeq; other rotavirus vaccines, including Rotavac, Rotasiil, and Lanzhou lamb rotavirus vaccine, also showed positive associations with reduced RVGE risk.

Meaning

The findings favor the worldwide introduction of rotavirus vaccines to prevent RVGE, but head-to-head comparisons are needed to compare the benefit and risk of different rotavirus vaccines.

This meta-analysis synthesizes randomized clinical trials and observational studies to evaluate the comparative benefit, risk, and immunogenicity of different rotavirus vaccines.

Abstract

Importance

Rotavirus vaccines have been introduced worldwide, and the clinical association of different rotavirus vaccines with reduction in rotavirus gastroenteritis (RVGE) after introduction are noteworthy.

Objective

To evaluate the comparative benefit, risk, and immunogenicity of different rotavirus vaccines by synthesizing randomized clinical trials (RCTs) and observational studies.

Data Sources

Relevant studies published in 4 databases: Embase, PubMed, the Cochrane Library, and Web of Science were searched until July 1, 2020, using search terms including “rotavirus” and “vaccin*.”

Study Selection

Randomized clinical trials and cohort and case-control studies involving more than 100 children younger than 5 years that reported the effectiveness, safety, or immunogenicity of rotavirus vaccines were included.

Data Extraction and Synthesis

A random-effects model was used to calculate relative risks (RRs), odds ratios (ORs), risk differences, and 95% CIs. Adjusted indirect treatment comparison was performed to assess the differences in the protection of Rotarix and RotaTeq.

Main Outcomes and Measures

The primary outcomes were RVGE, severe RVGE, and RVGE hospitalization. Safety-associated outcomes involved serious adverse events, intussusception, and mortality.

Results

A meta-analysis of 20 RCTs and 38 case-control studies revealed that Rotarix (RV1) significantly reduced RVGE (RR, 0.316 [95% CI, 0.224-0.345]) and RVGE hospitalization risk (OR, 0.347 [95% CI, 0.279-0.432]) among children fully vaccinated; RotaTeq (RV5) had similar outcomes (RVGE: RR, 0.350 [95% CI, 0.275-0.445]; RVGE hospitalization risk: OR, 0.272 [95% CI, 0.197-0.376]). Rotavirus vaccines also demonstrated higher protection against severe RVGE. Additionally, no significant differences in the protection of RV1 and RV5 against rotavirus disease were noted in adjusted indirect comparisons. Moderate associations were found between reduced RVGE risk and Rotavac (RR, 0.664 [95% CI, 0.548-0.804]), Rotasiil (RR, 0.705 [95% CI, 0.605-0.821]), and Lanzhou lamb rotavirus vaccine (RR, 0.407 [95% CI, 0.332-0.499]). All rotavirus vaccines demonstrated no risk of serious adverse events. A positive correlation was also found between immunogenicity and vaccine protection (eg, association of RVGE with RV1: coefficient, −1.599; adjusted R², 99.7%).

Conclusions and Relevance

The high protection and low risk of serious adverse events for rotavirus vaccines in children who were fully vaccinated emphasized the importance of worldwide introduction of rotavirus vaccination. Similar protection provided by Rotarix and RotaTeq relieves the pressure of vaccines selection for health care authorities.

Introduction

Worldwide, diarrhea, accounting for approximately 70.6 deaths per 100 000 population and 1.75 episodes per child in 2016, is the fifth leading cause of death among children younger than 5 years.¹ Rotavirus gastroenteritis, which results in 28.8% of the deaths from diarrhea, is the leading causative mechanism for diarrhea in children younger than 5 years.^1,2 The mortality and morbidity of RVGE varies by location, with the highest in sub-Saharan Africa, Southeast Asia, and South Asia.² Fortunately, the mortality of RVGE decreased by 43.6% from 2005 to 2015,³ which most likely owing to the introduction of rotavirus vaccines.

In 2018, 101 countries have introduced rotavirus vaccine into their national immunization programs, with global coverage at 53%.⁴ At present, 6 oral rotavirus vaccines have been widely used. Two live attenuated oral rotavirus vaccines, Rotarix (RV1), a 2-dose monovalent (G1P[8]) vaccine and RotaTeq [RV5]), a 3-dose pentavalent (G1, G2, G3, G4, and P[8]) vaccine, are globally introduced.⁵ Another 2 novel vaccines, Rotasiil (BRV-PV) and Rotavac (116E), are currently licensed in India only.⁶ Besides, domestically licensed rotavirus vaccines are also available in China (Lanzhou lamb rotavirus [LLR] vaccine) and Vietnam (Rotavin). Although a decline in the morbidity and mortality of RVGE has been reported in many countries following the introduction of rotavirus vaccines, concerns about serious adverse events still exist. Furthermore, little is known about the comparative benefit and risk of different rotavirus vaccines because of the lack of powerful head-to-head comparisons. Therefore, by synthesizing randomized clinical trials (RCTs) and case-control and cohort studies, we undertook a systematic review and meta-analysis to evaluate the association of different rotavirus vaccines with RVGE in aspects of benefit, risk, and immunogenicity and analyzed comparative protection of different vaccines by indirect comparisons.

Methods

Search Strategy and Selection Criteria

We searched for relevant studies published until July 1, 2018, and further updated until July 1, 2020, in 4 databases: Embase, PubMed, the Cochrane Library, and Web of Science, using search terms including “rotavirus” and “vaccin*.” Randomized clinical trials and cohort and case-control studies reporting the efficacy, effectiveness, safety, or immunogenicity of rotavirus vaccine were included. Studies with fewer than 100 enrolled participants were excluded. The selection criteria are further outlined in detail in eTable 1 in the Supplement.

Procedures

The study selection and data collection process were explained in detail in the eMethods in the Supplement. Using EndNote X8 (Clarivate), 2 reviewers (Z.-W.S. and Y.F.) independently screened all obtained articles for relevance, with a third reviewer (H.L.L.) consulted when necessary. We developed a standardized data extraction form, and Z.-W.S. rechecked the extracted data of included trials identified by Y.F. The quality of RCTs and observational studies were accessed in accordance with the Cochrane Reviewers’ Handbook and the Newcastle-Ottawa Scales (eTable 2 in the Supplement). For 10% of included studies, data were doubly extracted by a third author (H.L.L.). The procedure was supervised and arbitrated by a fourth author (H.G.X.).

Statistical Analysis

In preliminary analyses, estimates of relative risks (RRs), odds ratios (ORs), and risk differences using raw data were similar to reported results; thus, we opted to perform more detailed statistical analyses with raw data, using Stata version 12.0 (StataCorp) and Revman version 5.3 (Cochrane Library). The RRs, ORs, and risk differences were calculated in a Mantel-Haenszel random-effects model. We used the per-protocol estimates in RCTs and combined control groups, including both hospital and community controls, in case-control studies. Considering the diminished vaccine efficacy in low-income countries (LICs) compared with middle-income countries and high-income countries, included studies were stratified by the economic development of countries, using the World Bank’s classification (eTable 3 in the Supplement).⁷ For multicenter RCTs, we included each individual country as a separate observation point whenever possible. If not, we used the sample size in each site to calculate a weighted level of economic development and used this estimate to assign the trial to a specific stratum.

Adjusted indirect treatment comparison was performed to assess the differences in vaccine protection between different subgroups, adopting P < .05 as the level of statistical significance. We performed a metaregression model to estimate the association between vaccine protection in 1 to 2 years of follow-up and the rate of seropositivity at 1 to 2 months after the last dose (IgA antibody concentration ≥20 units/mL or ≥3-fold increase from baseline), and the adjusted R² index was used to quantify the proportion of variance explained by the covariates. For the outcomes obtained from fewer than 3 studies, we conducted a systematic review. A sensitivity analysis was performed by excluding each study to identify the stability and consistency of our results. The Q test and I² statistic was applied to determine heterogeneity (P < .10 or I² > 50% indicated significant heterogeneity). Publication bias was assessed using funnel plots (eFigure 6 in the Supplement).

Results

Study Selection and Characteristics

Initial literature retrieval produced 3998 articles, and 241 full-text articles were assessed for eligibility. Seventeen studies were further identified from updated literature retrieval, and 121 studies were finally included. The selection process is summarized in eFigure 1 in the Supplement. Included studies varied by study design (57 RCTs,^{8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64} 50 case-control studies,^{65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114} and 14 cohort studies^{115,116,117,118,119,120,121,122,123,124,125,126,127,128}), rotavirus vaccine type (74 for RV1^{8,9,11,12,15,16,19,20,21,22,23,24,25,26,31,32,34,35,40,41,42,43,44,48,49,50,51,52,58,65,66,72,73,74,75,76,77,78,79,80,81,82,83,84,85,87,88,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,113,114,118,119,120,121,122,124,125,126,127,128}; 45 for RV5^{10,13,14,17,18,27,28,29,33,45,53,54,55,56,57,59,60,61,62,63,64,65,66,67,68,69,70,71,76,78,81,84,86,88,93,94,95,96,111,112,115,116,117,122,123}; 5 for LLR^{30,89,90,91,92}; and 3 each for Rotavac^36,37,38 and Rotasiil^39,46,47), or study population. The characteristics of included studies are reported in eTable 4 in the Supplement.

Benefits and Risks of Rotavirus Vaccines, Stratified by Vaccine Type

During the first year of follow-up, more children in placebo groups developed RVGE compared with children vaccinated with full-dose RV1 (RR, 0.316 [95% CI, 0.224-0.345]) or RV5 (RR, 0.350 [95% CI, 0.275-0.445]) (Figure 1 and Figure 2). In case-control studies, a low risk of RVGE hospitalization was also estimated among children fully vaccinated with RV1 (OR, 0.347 [95% CI, 0.279-0.432]) or RV5 (OR, 0.272 [95% CI, 0.197-0.376]). A systemic review of cohort studies revealed RRs of 0.125 (95% CI, 0.086-0.182) for RV1 and 0.049 (95% CI, 0.028-0.083) for RV5 for the prevention of RVGE hospitalization, regardless of the cohort year (eTable 5 in the Supplement). Rotavirus vaccines demonstrated higher protection against severe RVGE but less against severe all-cause gastroenteritis. A clear gradient in vaccine protection was noted by country income level, with the highest in high-income countries and the lowest in LICs (RVGE hospitalization, P = .002; RVGE, P < .001; eTables 6-12 in the Supplement).

Figure 1. Random-Effects Model of Rotavirus Vaccine Protection Against Rotavirus Gastroenteritis (RVGE) and RVGE Hospitalization, by Country Income Level, in Randomized Clinical Trials.

HIC indicates high-income countries; LIC, low-income countries; LLR, Lanzhou lamb rotavirus; MIC, middle-income countries; RR, relative risk; RV1, monovalent rotavirus vaccine.

Figure 2. Random-Effects Model of Rotavirus Vaccine Protection against Rotavirus Gastroenteritis (RVGE) and RVGE Hospitalization, by Country Income Level, in Case-Control Studies.

HIC indicates high-income countries; LIC, low-income countries; LLR, Lanzhou lamb rotavirus; MIC, middle-income countries; OR, odds ratio; RV1, monovalent rotavirus vaccine; RV5, pentavalent rotavirus vaccine.

Rotavac reduced RVGE and severe RVGE risk in India by 33.6% (95% CI, 19.6%-45.2%) and 56.0% (95% CI, 37.3%-69.2%), respectively. Rotasiil reduced RVGE and severe RVGE risk by 29.5% (95% CI, 17.9%-39.5%) and 52.2% (95% CI, 12.1%-74.0%) in India and Niger, respectively. In China, LLR was associated with a decrease in RVGE (RR, 0.407 [95% CI, 0.332-0.499]; OR, 0.348 [95% CI, 0.121-0.999]), severe RVGE (RR, 0.248 [95% CI, 0.144-0.427]), and RVGE hospitalization (OR, 0.405 [95% CI, 0.309-0.531]).

In indirect treatment comparisons, no significant differences were noted in the protection of RV1 and RV5 against RVGE (RR, 0.865 [95% CI, 0.565-1.325]; P = .51; OR, 1.264 [95% CI, 0.866-1.844]; P = .23) or severe RVGE (RR, 0.768 [95% CI, 0.335-1.758]; P = .53; OR, 0.944 [95% CI, 0.603-1.476]; P = .80) (eTable 13 in the Supplement). When stratified by the World Bank classification, there were also no significant differences in vaccine protection between RV1 and RV5. Furthermore, to alleviate the bias of sociodemographic factors, only studies conducted in the same region were included to perform adjusted indirect treatment comparison, and the results also indicated little difference in vaccine protection between RV1 and RV5 (eTable 14 in the Supplement).

We identified 36 RCTs,^{9,11,12,13,14,16,17,18,19,20,23,24,26,27,29,30,32,33,34,35,37,38,39,40,41,42,43,45,47,48,50,53,58,60,61,63} 4 case-control studies,^{71,81,101,107} and 2 cohort studies^117,119 evaluating the safety of rotavirus vaccines (eFigures 2-4 in the Supplement). The incidence of serious adverse events in the vaccine group was similar to that of the placebo group. The overall estimate of risk differences showed no increased risk of intussusception and death in children vaccinated with RV1, RV5, Rotavac, Rotasiil, or LLR during 1 or 2 years of follow-up.

Stratified Analyses of Rotavirus Vaccine Benefit by Duration and Vaccination Schedule

In stratified analyses of the duration of vaccine protection, we found that the protection of RV1 or RV5 against RVGE was lower in the second year of follow-up (RV1: RR, 0.494 [95% CI, 0.255-0.955]; RV5: RR, 0.622 [95% CI, 0.388-0.996]) in comparisons with the first year of follow-up (RV1: RR, 0.297 [95% CI, 0.207-0.425]; RV5: RR, 0.344 [95% CI, 0.271-0.436]), while similar in 2 years of follow-up (RV1: RR, 0.348 [95% CI, 0.196-0.618]; RV5: RR, 0.500 [95% CI, 0.288-0.869]; Figure 3). Also, the reductions in vaccine protection during the second year of follow-up were small in the high-income countries (RV1: RR, 0.281 [95% CI, 0.207-0.381]; RV5: RR, 0.497 [95% CI, 0.353-0.699]) but pronounced in the LICs (RV1: RR, 1.288 [95% CI, 0.738-2.248]; RV5: RR, 0.815 [95% CI, 0.659-1.007]) (eTable 9 in the Supplement). Here, estimates should be interpreted with caution, because there was only 1 study for an LIC.⁴⁸ Similar results were observed for Rotavac and Rotasiil in India. In case-control studies, RV1 and RV5 provided similar protection among children aged younger than 12 months, 12-24 months, or ≥12 months. By contrast, the OR of RV1 vs control in the LIC and lower- and middle-income countries was significantly lower among children aged 12 to 24 months (OR, 0.528 [95% CI, 0.249-1.120]; P < .001) or those 12 months or older (OR, 0.526 [95% CI, 0.370-0.750]; P = .008), compared with children younger than 12 months (OR, 0.356 [95% CI, 0.266-0.476]).

Figure 3. Subgroup Analysis for Rotavirus Vaccine Protection, Stratified by Follow-up Duration and Vaccination Schedule.

In randomized clinical trials (RCTs), the first column shows the number of cases in the vaccine group and the total population in the vaccine group. In case-control studies, these numbers denote the number of vaccinated children in the case group and the sum of children with no vaccination plus those receiving full doses in the case group. In RCTs, the second column shows the number of cases in the placebo group and the total population in placebo group. In case-control studies, this column shows the number of vaccinated children in the control group and the sum of children with no vaccination plus those receiving full doses in the control group. Odds ratios (ORs) were used for case-control studies; relative risks (RRs), for RCTs. RV1 indicates monovalent rotavirus vaccine; RV5, pentavalent rotavirus vaccine.

In the second comparison, we divided 35 case-control studies^{65,66,67,68,69,70,71,72,73,74,79,80,85,86,87,88,93,94,95,96,98,99,100,102,103,104,105,106,108,109,110,111,112,113,114} depending on whether the enrolled children received complete vaccination (Figure 3). Studies reported a nonsignificantly lower risk of RVGE hospitalization among children vaccinated with 3-dose RV5 compared with 1 dose, but a similar risk between 2 doses and 3 doses. When stratified by the World Bank classification, no significant differences in vaccine protection between 3 doses, 2 doses, and 1 dose of RV5 were observed. Two-dose RV1 showed stronger association with reduced risk of RVGE hospitalization than only 1 dose (OR, 0.347 [95% CIs, 0.279-0.432] vs 0.561 [95% CIs, 0.493-0.639]; P < .001), especially in the middle-income countries (OR, 0.396 [95% CIs, 0.338-0.465] vs 0.559 [95% CIs, 0.489-0.640]; P = .001).

Strain-Specific Protection of Rotavirus Vaccine

Pooled data from 13 RCTs^{11,16,20,24,27,40,41,42,48,49,54,59,61} suggested that RV1 conferred protection against severe RVGE caused by G1, G2, G3, G4, G9, and P[8] strains, respectively, whereas the protection of RV5 was low (and nonsignificant) against G1, G2, and G3 strain (eTable 15 in the Supplement). The Table showed the strain-specific protection of RV1,¹²⁹ and no significant differences were noted in vaccine protection against partly heterotypic or fully heterotypic strains compared with homotypic strains in middle-income countries. There were also no significant differences in vaccine protection against single-antigen vaccine type and single-antigen nonvaccine type strains for RV1 and RV5. However, higher protection of RV1 against homotypic strains (OR, 0.116 [95% CIs, 0.065-0.217]) than heterotypic strains (OR, 0.457, [95% CIs, 0.264-0.579]) was estimated in case-control studies in middle-income countries (P = .005).

Table. Strain-Stratified Vaccine Protection During the 2-Year Efficacy Period.

Characteristic	RV1			RV5
	Observations, No.	RR or OR (95% CI)a	P valueb	Observations, No.	RR or OR (95% CI)a	P valueb
RCTsc
High-income countries
Single-antigen vaccine type strain	4	0.054 (0.022-0.130)	NA	4	0.056 (0.036-0.088)	NA
Single-antigen nonvaccine type strain	9	0.123 (0.076-0.198)	.11	2	0.098 (0.010-0.945)	.63
Middle-income countries
Homotypic strain	3	0.248 (0.139-0.444)	NA	NA	NA	NA
Partly heterotypic strain	7	0.200 (0.131-0.306)	.56	NA	NA	NA
Fully heterotypic strain	3	0.333 (0.191-0.581)	.47	NA	NA	NA
Single-antigen vaccine type strain	4	0.293 (0.196-0.438)	NA	13	0.533 (0.423-0.672)	NA
Single-antigen nonvaccine type strain	10	0.204 (0.151-0.276)	.16	8	0.443 (0.285-0.690)	.47
Low-income countries
Single-antigen vaccine type strain	2	0.558 (0.315-0.991)	NA	NA	NA	NA
Single-antigen nonvaccine type strain	6	0.485 (0.336-0.701)	.69	NA	NA	NA
Case-control studiesd
High-income countries
Homotypic strain	3	0.096 (0.030-0.313)	NA	2	0.156 (0.091-0.268)	NA
Partly heterotypic strain	3	0.188 (0.063-0.555)	.41	3	0.135 (0.041-0.445)	.83
Fully heterotypic strain	3	0.178 (0.107-0.295)	.35	NA	NA	NA
Single-antigen vaccine type strain	NA	NA	NA	3	0.165 (0.101-0.271)	NA
Single-antigen nonvaccine type strain	NA	NA	NA	2	0.215 (0.105-0.441)	.56
Middle-income countries
Homotypic strain	2	0.116 (0.065-0.217)	NA	NA	NA	NA
Partly heterotypic strain	4	0.457 (0.264-0.579)	.001	NA	NA	NA
Fully heterotypic strain	6	0.335 (0.197-0.569)	.005	NA	NA	NA
Single-antigen vaccine type strain	2	0.458 (0.147-1.427)	NA	5	0.354 (0.249-0.503)	NA
Single-antigen nonvaccine type strain	4	0.592 (0.310-1.129)	.70	3	0.176 (0.039-0.793)	.38
Low-income countries
Single-antigen vaccine type strain	1	0.201 (0.065-0.617)	NA	NA	NA	NA
Single-antigen nonvaccine type strain	2	0.516 (0.232-1.147)	.21	NA	NA	NA
Cohort studiese
Middle-income countries
Homotypic strain	2	0.276 (0.096-0.794)	NA	NA	NA	NA
Partly heterotypic strain	3	0.436 (0.183-1.039)	.25	NA	NA	NA
Fully heterotypic strain	4	0.429 (0.185-0.996)	.26	NA	NA	NA

Abbreviations: NA, not applicable; OR, odds ratio; RCTs, randomized clinical trials; RR, relative risk; RV1, monovalent rotavirus vaccine; RV5, pentavalent rotavirus vaccine; RVGE, rotavirus gastroenteritis.

^aRelative risks are used in the RCT portion of this Table, and ORs are used in the case-control studies portion.

^bP values were the differences of vaccine protection against partly heterotypic or fully heterotypic strains compared with homotypic strains; the differences of vaccine protection against single-antigen nonvaccine type strains compared with single-antigen vaccine type strains.

Metaregression Between Immunogenicity and Protection of Rotavirus Vaccine

Circulating antirotavirus IgA antibodies have been used as the early proxy for vaccine uptake,^130,131 which is a convenient method to monitor vaccine effectiveness at the population level. Pooled data from 24 RCTs^{8,9,12,14,15,16,18,19,22,24,28,29,31,32,33,34,35,42,43,48,53,57,60,61} showed that the percentage of seropositivity among children who were fully vaccinated was 69.3% (95% CIs, 60.0%-78.6%) for RV1 and 89.5% (95% CIs, 84.5%-94.5%) for RV5, much higher than in placebo group (11.9% [95% CIs, 8.5%-15.2%]). Moreover, the percentage of vaccinated children with seropositivity exhibited a positive association with vaccine protection (R² >0; (eg, association of RVGE with Rotarix: coefficient, −1.599; adjusted R², 99.7%; Figure 4). In metaregression analyses, with the difference of the rate of seropositivity between vaccine groups and placebo groups as the abscissa, the adjusted R² values for the associations between immunogenicity and vaccine protection were 86.5% for RV1 and 15.6% for RV5 against severe RVGE and 99.7% for RV1 against RVGE. We did not evaluate the association between immunogenicity and vaccine protection of RV5 against RVGE because only 3 studies^15,60,61 were included.

Figure 4. Metaregression Between Immunogenicity and Vaccine Protection.

The coefficient is the regression correlation coefficient, and the adjusted R² is the proportion of between-study variance explained. Metaregression between immunogenicity and logRR against rotavirus gastroenteritis (RVGE) for RV1 and RV5 (A), RVGE for RV1 (B), severe RVGE for RV1 and RV5 (C), severe RVGE for RV1 (D), and severe RVGE for RV5 (E). RV1 indicates monovalent rotavirus vaccine; RV5, pentavalent rotavirus vaccine.

Sensitivity and Heterogeneity Analyses

Our systematic review found considerable heterogeneity between included studies. To investigate the potential sources of heterogeneity, a subgroup analysis was performed using economic development as a variable, and the heterogeneity was subsequently shown to be dealt with in varying degrees when I² was greater than 50%. Sensitivity analyses for all outcomes did not identify any substantial effects resulting from differences in study quality (eFigure 5 in the Supplement).

Discussion

Our findings from RCTs, case-control studies, and cohort studies corroborated that RV1 and RV5 have substantial and sustained protection against rotavirus disease, especially against severe RVGE, which is in line with previously reported data.^{132,133,134,135} Considering that vaccine administrations are not always followed by recommendations, comorbidities may be present, and sociodemographic factors vary in real world, the consistency of results from observational studies and RCTs reconfirmed the high protection of rotavirus vaccination. Moreover, pooled data showed no increased risk of serious adverse events including intussusception among children who were vaccinated.^136,137 However, a study about intussusception conducted in Australia reported a smaller increased risk of intussusception after RV1 and RV5 vaccination.¹³⁸ Therefore, continuous surveillance of the benefits and adverse effects of rotavirus vaccines is required after vaccination.

The protection against rotavirus diseases varied by time interval after vaccination, and rotavirus vaccines, particularly RV5, provided lower protection against RVGE in the second efficacy period. Although our results indicated that rotavirus vaccines can provide substantial protection against RVGE during the first 2 years of life,¹³⁹ more studies following up the vaccine efficacy for more than 2 years are required. The reduced vaccine protection might be caused by declining vaccine-induced antibodies, acquisition of protection against RVGE through indirect effects of vaccine, or exposure to natural asymptomatic and mild infections among control populations who are unvaccinated.¹⁴⁰ The wane of vaccine protection over time highlights the importance of monitoring the morbidity of rotavirus diarrhea after vaccination; more children may become infected at older ages, and evaluation of alternative vaccination schedules is useful.

During the subgroup analyses, vaccination schedule may affect vaccine performance. Data from case-control studies identified that a partial vaccination provided considerable protection, but not to the same level as a full series.^133,141 Several phase 3 RCTs also showed that RV1 and RV5 conferred early protection against RVGE before completion of a 2-dose or 3-dose schedule.^11,142 This finding is encouraging so that numerous children who are partly vaccinated in LICs and children vaccinated during the periods of intensive rotavirus circulation can receive protection. Nonetheless, the protection of partial vaccination was lower than full vaccination, and the duration of protection from partial vaccination was not clear. Therefore, more efforts should be made to ensure full vaccination as recommended to achieve optimal protection.

The wide variety of rotavirus strain is a challenge for improving vaccine effectiveness. It is encouraging that RV1 and RV5 work well against heterotypic strains. The heterotypic protective immunity is important for low-income and lower middle–income countries, where greater strain diversity and concurrent circulation of several strains is a common phenomenon.¹⁴³ However, prevalent rotavirus strain varied by time and region, and the dominance of 1 strain was often followed by the replacement with other strains.¹⁴⁴ The changes of serotype distribution was also reported in some countries after vaccine introduction.^145,146,147 Therefore, the characterization of rotavirus strains after vaccination should be monitored to avoid population-based selection of so-called escape strains, especially fully heterotypic strains and new strains, because of the long-term pressure of vaccine immunity.¹⁴⁸

A clear gradient in rotavirus vaccine protection was noted by country income level, with the highest in high-income countries.^{11,33,48,53,149} Possible reasons for weaker vaccine protection in LICs include host characteristics, such as malnutrition¹⁵⁰; environmental enteropathy¹⁵¹; concomitant enteric infections¹⁵²; poor maternal health¹⁵¹; high titers of RV-specific maternal antibodies in breast milk^153,154; and interference by coadministration of oral poliovirus vaccine.¹⁵⁵ Besides, most children in LICs were not vaccinated as per the recommended schedule and subsequently received lower protection from partial vaccination. The lower-than-expected rotavirus protection in LICs can also be explained by high natural rotavirus infections before vaccination, which confer protection against subsequent RVGE and may cause a biased outcome.^43,156,157 Furthermore, the scarcity of clean water can increase the risk of rotavirus spread by fecal-oral transmission in LICs. However, since the greater burden of severe RVGE in middle-income countries and LICs, the cases of severe RVGE prevented by rotavirus vaccines seem to be more in these settings. In addition, rotavirus vaccination was found to be cost-effective in LICs, suggesting a potential benefit of vaccination.^158,159

Currently there are 6 rotavirus vaccines licensed in the market, but little was known about the interchangeability of these vaccines. Therefore, we performed adjusted indirect comparisons, which showed similar protection of RV1 and RV5, Rotavac, and Rotasiil, particularly at the same economic level or in the same country. This relieves the pressure of vaccine selection and suggests that health care authorities should weigh not only vaccine effectiveness but also economic factors associated with vaccine procurement and introduction, such as unit price, cold-chain volume, the cost of storage, and wastage. Cost-effectiveness models in Kenya and Palestine have indicated that 2-dose RV1 vaccinations seems to be more cost-effective and create less strain on a cold chain than 3-dose RV5 vaccinations.^160,161 Additionally, the duration of high vaccine protection and the reduced vaccine protection resulted from partial vaccination should be taken into account. Furthermore, herd effects induced by rotavirus vaccination should be estimated to further compare the social benefits of different vaccines for children who are unvaccinated. It has been reported in Europe and the US that a herd effect of the rotavirus vaccine may enhance its clinical performance when implemented at a large scale under routine conditions.^162,163 Considering the inherent limitations of indirect comparisons, a well-designed head-to-head study should be conducted to further compare the efficacy, cost-effectiveness, and strain-specific protection of different vaccines.

Limitations

There are several limitations of our meta-analysis. First, despite a systematic search of published studies, the final estimates were identified in only 45 settings, and an exhaustive review of gray literature was not included. Especially in stratified analyses, sparse data in some subgroups limit generalizability; for example, there was only 1 available study conducted in an LIC. But the number of children enrolled in our meta-analysis was more than 100 000, suggesting the value and reliability of our results. Second, considering that the introduction and protection of rotavirus vaccines vary by regions, it may prevent a fair comparison of RV1 and RV5 at a global level. So, we also performed indirect comparisons in the same region. The most accurate method, head-to-head comparisons, to evaluate the comparative efficacy of different vaccines is required in further studies. Third, the missing data and low quality in some included studies may influence our results, although we have excluded studies with small enrolled populations (<100 children). Well-designed observational trials and RCTs are still required to evaluate the clinical performance of rotavirus vaccines.

Conclusions

In conclusion, based on a large worldwide data set, we identified reasonable evidence of sustained high protection and low risk of adverse effects for rotavirus vaccines in children aged 2 years or younger, which is important to combat vaccine hesitancy. Also, the differences in vaccine performance between 4 licensed rotavirus vaccines were not surprising. Although the global introduction of rotavirus vaccines faces many scientific, programmatic, and financial challenges, these licensed vaccines hold promise to have immediate and measurable effectiveness to improve child health and survival from rotavirus disease. Our findings and prelicensing evidence reinforce the importance of optimizing uptake rates of rotavirus vaccines worldwide. Continued surveillance after vaccine introduction is also required to monitor the long-term changes in rotavirus incidence and the potential emergence of heterotypic strains.

Supplement.

eMethods. Methods and Statistical Analysis

eTable 1. Included Criteria During Selection Process

eTable 2. The Criteria for grading methodological quality

eTable 3. The basis for grouping in sub-analysis

eTable 4. Characteristics of the Included RCTs, Case-control and Cohort studies

eTable 5. Estimated Pooled Vaccine Protection in Cohort Studies

eTable 6. Estimated Pooled Vaccine Protection against Severe RVGE Stratified by Income Development at Different Follow-up Times

eTable 7. Estimated Pooled Vaccine Protection against Severe RVGE Stratified by Income Development

eTable 8. Estimated Pooled Vaccine Protection against Severe AGE Stratified by Income Development at Different Follow-up Times

eTable 9. Estimated Pooled Vaccine Protection against RVGE Stratified by Income Development at Different Follow-up Times

eTable 10. Estimated Pooled Vaccine Protection against RVGE Hospitalization Stratified by Income Development at Different Follow-up Times

eTable 11. Estimated Pooled Vaccine Protection against RVGE Stratified by Income Development

eTable 12. Estimated Pooled Vaccine Protection against RVGE Hospitalization stratified by income development

eTable 13. Indirect Comparisons between RV1 and RV5 Stratified by Income Development

eTable 14. Indirect Comparisons between Different Vaccines in the Same Region

eTable 15. Random-effect Model of Strain-specific Rotavirus Vaccine Protection

eFigure 1. Flow Diagram

eFigure 2. The Risk of Serious Adverse Events Between Vaccine and Placebo Groups at Different Follow-up Times

eFigure 3. The Risk of Intussusception Between Vaccine and Placebo Groups at Different Follow-up Times

eFigure 4. All-Cause Mortality in Vaccine and Placebo Groups During the Follow-up Period

eFigure 5. Sensitivity Analysis

eFigure 6. Funnel Plots

eReferences.