Association Between Rotavirus Vaccination and Risk of Intussusception Among Neonates and Infants

This systematic review and meta-analysis assesses the association of rotavirus vaccination among neonates and infants with risk of intussusception for up to 2 years after vaccination overall and by type of vaccine.


Introduction
Vaccines play an important role in the prevention of infectious diseases. Rotavirus (RV) vaccination has significantly reduced the occurrence and severity of RV-related gastroenteritis and mortality among infants and young children. 1 Although data from some clinical trials show that the efficacy of RV vaccines for prevention of RV gastroenteritis reaches 36% to 96% within a year of follow-up, 2,3 the Global Advisory Committee on Vaccine Safety has noted that the use of RV vaccines may be associated with an increased risk of intussusception. 4 Even though the efficacy of the vaccine might outweigh the small potential risk of intussusception, the Global Advisory Committee on Vaccine Safety has also suggested performance of active surveillance to ensure that the long-term benefit and safety of RV vaccines are entirely assessed. 4 Data from randomized clinical trials (RCTs) regarding the efficacy and safety of RV vaccines show conflicting evidence on the incidence of intussusception. Whether or not there was an association between vaccination and an increased risk of intussusception, the answer varied across studies. [5][6][7][8] The varying data from the RCTs about the difference in the incidence of intussusception could be attributable to multiple variables such as age, sex, geographic and population distribution of the participants, and the different types of RV vaccines used in these studies. However, the reasons behind the differing risk of intussusception are still not clear. Therefore, we conducted this systematic review and meta-analysis of published RCTs to further assess the risk of development of intussusception after RV vaccination.

Methods
This systematic review and meta-analysis followed the Cochrane Collaboration Group 9

and Preferred
Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) 10

Literature Search and Study Selection
The PubMed, Web of Science, Cochrane library, and Embase databases were searched from January This meta-analysis focused on the association between RV vaccination and the risk of intussusception. Only RCTs that compared the risk of intussusception between the vaccine and placebo groups for neonates or infants were included in the analysis. Other inclusion criteria included the following: (1) use of any type of vaccination, such as monovalent (RV1) (Rotarix; GlaxoSmithKline), pentavalent (RV5) (RotaTeq; Merck & Co, Inc), monovalent human-bovine (116E) (Rotavac; Bharat Biotech), oral bovine pentavalent (BRV-PV) (Rotasiil; Serum Institute of India), and human neonatal (RV3-BB); (2) a sample size of at least 100 participants; and (3) data on the incidence of intussusception. Exclusion criteria were (1) no data on intussusception; (2) no placebo group; (3) use of human reassortant rotavirus tetravalent vaccine (RRV-TV) (Rotashield; Wyeth Laboratories Inc) because the US Advisory Committee on Immunization no longer recommends the use of this vaccine owing to high risk of intussusception; and (4) duplicate publications. Some studies included the participants from the same general population, in which case, the most comprehensive and up-to-date study was selected for inclusion in this meta-analysis. Data from unpublished trials or conference abstracts were also excluded from the final analysis (Figure 1).

Data Collection and Extraction
Data were extracted using a standardized data extraction form. Information collected included the study title, authors, publication year, study period, phase and the registration number of the RCT, city, country, continent, sex, weight, vaccine type, time of vaccination, sample size in the vaccine or placebo group, number of intussusception cases, number of days between the administration of the vaccine and diagnosis of intussusception, the risk estimates or data used to calculate the risk estimates, and 95% CIs or data used to calculate 95% CIs. If a trial had more than 2 groups or differed in the vaccine component or concentration, the extracted information and data on the vaccine used were those of the similar composition and with the closest concentration to the approved vaccine.

Statistical Analysis
The association between RV vaccination and intussusception, the pooled results of relative risk (RR), the risk difference (RD), and 95% CIs in the 3 different follow-up periods were calculated using the Mantel-Haenszel method. Because of the very low incidence of intussusception, the number of intussusceptions in the vaccine and placebo groups during the observation period was 0 at the same time in some trials. To accurately and objectively reflect the facts, we evaluated RR and RD.
Undefined RRs (ie, RR of 0 or infinity) were not included in the statistical calculations but were included in the calculation of RD.
Separate analyses were performed for RV1 and RV5. Because the 116E, BRV-PV, and RV3-BB vaccines were used only in local areas and there were fewer related trials, these 3 vaccines were combined into 1 group for analysis. If only 1 trial evaluated the vaccine, a systematic review was conducted, and if more than 1 trial evaluated the vaccine, a meta-analysis was conducted. The results  were separately calculated as subtotal results. The pooled and subtotal results were presented in forest plots as RRs, RDs, and 95% CIs of intussusception after RV vaccination. Both the pooled and subtotal results of RR and RD of intussusception were estimated at 31 days, 1 year, and 2 years after the vaccination period. If the number of trials was small during a follow-up period (31 days, 1 year, and 2 years), analysis was not conducted on publication bias.
The Q test and I 2 statistic index were used to assess the degree of heterogeneity between different studies: low (<25%), moderate (26%-50%), and high (51%-75%). Results were calculated using a fixed-or random-effect model depending on the statistical results of the heterogeneity test.
Sensitivity analysis was performed for identifying the heterogeneity among the studies. All statistical analyses were performed using Stata, version 14.2 (StataCorp). A 2-sided P < .05 was considered to indicate statistical significance.

Study Characteristics
As shown in Table 1

Study Quality
Quality assessment of the trials was performed according to the Cochrane collaboration′s tool for assessing the risk of bias. 9 Of the 25 RCTs, 19 were high quality and 6 were moderate quality.

Risk of Intussusception Within 31 Days After Rotavirus Vaccination
As shown in Table 2 and Figure 2, 20 cases of intussusception were diagnosed within 31 days after any RV vaccination, with 11 cases (55%) in the vaccine group and 9 cases (45%) in the placebo group.
The RR of intussusception ranged from 0.85 to 3.15 among the 4 studies that reported intussusception risk within 31 days after vaccination. Heterogeneity among these studies was low (Q = 0.78; P = .85; I 2 = 0%). The RD of intussusception ranged from −0.32 per 10 000 infants to 15.11 per 10 000 infants (Table 2). Heterogeneity among those studies was also very low (Q = 1.01; P > .99; Fourteen cases of intussusception were diagnosed within 31 days after RV1 vaccination (7 cases [50%] in the vaccine group and 7 cases [50%] in the placebo group). The subtotal estimate of RR of intussusception within 31 days after each dose of RV1 for the fixed-effect model was 0.91 (95% CI, 0.33-2.55; P = .86). The subtotal estimate of RD of intussusception within 31 days after each dose for the fixed-effect model was −0.08 per 10 000 infants (95% CI, −2.22 to 2.06 per 10 000 infants; P = .94).     Relative risk and 95% CIs were calculated using the Mantel-Haenszel method, with a fixed-effects model used to pool data. Randomized clinical trials with 0 cases of intussusception among the vaccine and placebo groups were not included in the relative risk statistics but were included in the statistics of the risk difference. Other vaccines included monovalent human-bovine (116E) (Rotavac), human neonatal (RV3-BB), and oral bovine pentavalent (BRV-PV). Boxes represent means, with the size of the box corresponding with the weight; horizontal lines represent 95% CIs; and diamonds indicate pooled means with the horizontal points of the diamonds representing 95% CIs. RV1 indicates monovalent rotavirus vaccine (Rotarix); RV5, pentavalent rotavirus vaccine (Rotateq).

Risk of Intussusception Within 1 Year of Vaccination
As shown in Table 2 and Figure 2, a total of 74 cases of definite intussusception were diagnosed within 1 year after any RV vaccination (37 cases [50%] in each group). The RR of intussusception ranged from 0.33 to 5.01 among 10 studies that reported intussusception outcome at 1 year.
The pooled effects were calculated using the fixed-effect model. The overall estimate of RR of intussusception within 1 year of RV vaccination for the fixed-effect model was 0.84 (95% CI, 0.53-1.32; P = .45). The overall estimate of RD of intussusception within 1 year after each RV dose for the fixed-effect model was -0.65 per 10 000 infants (95% CI, −2.68 to 1.39 per 10 000 infants; P = .53).
Thirty-five cases of definite intussusception were diagnosed within 1 year after RV1 vaccination

Risk of Intussusception Within 2 Years of Vaccination
As shown in Table 2

Discussion
Intussusception is a potentially life-threatening condition in children, and recent evidence has indicated an association between the RV vaccination and intussusception. 5 and RV5 vaccines showed an increased risk of intussusception within 7 days, especially after the first dose. However, only SCCS and self-controlled risk interval studies were included in the analysis.
Another meta-analysis of 10 SCCSs showed that RR for intussusception was 5.71 (95% CI, 4.50-7.25) from 1 to 7 days after the first dose, 1.69 (95% CI, 1.33-2.14) after the second dose, and 1.14 (95% CI, 0.75-1.74) after the third dose. 43 The SCCS evaluation is increasingly being used during the active vaccine safety surveillance, whereas an SCCS has its own limitations of measuring only the incidence of reported cases with a descriptive design rather than an analytic study. Thus, SCCSs could include potential referral bias because variation of the treatment application has no control. 44 When comparing different study designs to determine the best design for surveillance of vaccine safety, the limitations of those studies are evident, especially based on their heterogeneity. Thus, these positive results need to be carefully considered and further investigated.
Another possible reason for no association could be the exclusion of RRV-TV vaccination in this meta-analysis. Many previous studies [45][46][47] investigating RRV-TV demonstrated that RRV-TV was associated with a strong increased risk of intussusception; it was suspended in 1999 because of the safety issues.

Limitations
This study has limitations. The power was low for analysis of RCTs assessing the risk of intussusception with 116E, BRV-PV, and RV3-BB vaccines because of the limited number of trials.
Another limitation was the inability to assess whether there was a difference in the risk of intussusception among infants from various geographic regions because of unavailability of sufficiently large trials in the same region.

Conclusions
In this systematic review and meta-analysis of RCTs of the RV1, RV5, 116E, BRV-PV, and RV3-BB vaccines, we found no association of vaccination with increased risk of intussusception compared with placebo among infants for up to 2 years after vaccination. Our results contradict the postmarketing monitoring suggestion about the risk of intussusception after the RV vaccination. We suggest that the benefit of the vaccination exceeds the potential risk of intussusception.