Intussusception following rotavirus vaccination: an updated review of the available evidence

Abstract

In 1999, the first rotavirus vaccine licensed in the USA was withdrawn 9 months after introduction due to an association with intussusception that was detected in post-licensure surveillance. This association prompted large clinical trials designed to ensure the safety of two current live oral rotavirus vaccines, RotaTeq and Rotarix, which have since been recommended for use worldwide. Following their introduction, post-licensure studies have focused not only on the effectiveness and impact of these vaccines, but also on continued surveillance for intussusception. Most recent evidence from several countries shows a small increased risk of intussusception following vaccination with Rotarix and RotaTeq within the context of their demonstrated benefits. This review summarizes the available data on the safety of rotavirus vaccines with regards to intussusception.

Intussusception, a condition in which a segment of intestine invaginates within a more distal segment, is the most common cause of intestinal obstruction among infants and children [1]. As the process progresses, vascular supply to the involved segments can be compromised, potentially leading to bowel ischemia, perforation and possible death if left untreated. Intussusception is a rare event but occurs worldwide, with the majority of cases occurring in children <1 year of age, and peak incidence among those 4–7 months of age [2,3]. The average global incidence of intussusception among children <1 year of age is 74 per 100,000 infant-years, but incidence varies by region, with reported rates as high as 328 cases per 100,000 infant-years reported in some countries in Asia [3]. In developed countries, death due to intussusception is rare, with case fatality rates of <1%. Data from lower income countries are sparse, but evidence of higher case fatality in Africa has been observed (9.4% [range = 2–25%]) [3].

In the pre-vaccine era, rotavirus was the leading cause of severe gastroenteritis among children <5 years of age worldwide and is responsible for approximately 450,000 deaths each year, with >95% occurring in developing countries with higher child mortality [4]. Due to the high burden of disease, the WHO recommends routine rotavirus vaccination in infants worldwide. Two live oral vaccines are licensed globally: a pentavalent human-bovine reassortant rotavirus vaccine given as a three-dose series (RotaTeq [RV5]; Merck and Co.) and a monovalent human attenuated rotavirus vaccine given as a two-dose series (Rotarix [RV1]; GSK Biologicals). Although postulated, an association between naturally occurring rotavirus infection and intussusception has not been affirmed [2,5-7]. However, the association of intussusception with a previous tetravalent rotavirus vaccine (RotaShield [RRV-TV]; Wyeth Lederle Vaccines) led to its withdrawal and prompted specific monitoring for this rare event in clinical trials and post-marketing studies of the two currently licensed vaccines [8-10]. Data on rotavirus vaccines and intussusception were reviewed previously in 2009 [11]; this review considers data through 2013.

Experience with RRV-TV

RRV-TV was licensed in the USA on 31 August 1998, with distribution beginning in October 1998. It was recommended for use as a three-dose series in infants based on its efficacy and safety as demonstrated in clinical trials [12]. Pre-licensure studies determined that this live oral vaccine was generally well tolerated, with adverse effects that included fever and abdominal cramping [13]. These studies also identified five cases of intussusception that all occurred after the second or third dose of RRV-TV at varying dosage levels among children 4–7 months of age. Although these five cases were observed among 10,054 infant vaccine recipients (0.05%) in these studies, the rate was not significantly increased over that among placebo recipients, where one case occurred among 4633 infants (0.02%, p = 0.45) [6]. Nevertheless, intussusception was listed as a potential adverse reaction of the vaccine, and the US Advisory Community on Immunization Practices recommended post-licensure surveillance for intussusception [12].

Within 1 year of its introduction in the US, concern about an association between RRV-TV and intussusception emerged. By 27 May 1999, the Vaccine Adverse Event Reporting System (VAERS), a US database for passive surveillance of adverse events following vaccination, had received nine spontaneous reports of intussusception among infants who had received RRV-TV, prompting suspension of RRV-TV in July 1999 while a potential association was further investigated [14,15].

A nationwide case–control study identified an elevated risk in the 3–14 days following vaccination with any dose (adjusted odds ratio [aOR] = 10.6 [95% CI: 5.7–19.6]) [15]. The increased risk of intussusception was higher during the 3- to 14-day risk window following the first dose (aOR: 21.7 [95% CI: 9.6–48.9]) than the second (aOR: 3.3 [95% CI: 1.1–9.8]). The highest increased risk was found in the 3–7 days following the first dose of RRV–TV, with an aOR of 37.2 (95% CI: 12.6–110.1). A self-controlled case-series analysis confirmed these findings. These results translated to an estimated attributable risk of one additional case of intussusception for every 4670–9474 infants vaccinated, which corresponded to 10.6–21.4 excess cases per 100,000 vaccinated infants (Table 1) [15]. A retrospective cohort study conducted around the same time (December 1998–July 1999) using data from 10 managed care organizations yielded similar findings, estimating an attributable risk of one additional case of intussusception per 11,073 children vaccinated with RRV-TV (Table 1) [16].

Table 1.

Attributable risk estimates of intussusception associated with rotavirus vaccines among infants by study.

Vaccine	Setting	Study period	Age inclusion criteria	Calculation methods used for AR estimate		AR (per 100,000 vaccinated infants)		Ref.
				AR based on data from this analysis type	Risk for dose: interval	Point estimate	95% CI
RRV-TV	USA: hospitals in 19 states	November 1998–June 1999	1–11 months	Case–control	All doses: days 0–21	10.6	NR	[15]
				Self-controlled case series	All doses: days 0–21	21.4	NR
	USA: 10 managed care organizations	December 1998–July 1999	1–11 months	Vaccinated cohort versus unexposed person-time	All doses: days 0–21	9.0	NR	[16]
RV5	Australia: hospitals in six jurisdictions	July 2007–June 2010	1–11 months	Self-controlled case series	Dose 1: days 1–21; Dose 2: days 1–7	7.0	1.5–33.1	[27]
	USA: National passive surveillance of adverse events (VAERS)	February 2006–April 2012	NA	Self-controlled risk interval	Dose 1: days 3–6	0.7	0.2–1.7	[36]
	USA: National network of three health plans (PRISM)	January 2004–September 2011	5–36.9 weeks	Self-controlled risk interval	Dose 1: days 1–7	1.1	0.3–2.7	[37]
					Dose 1: days 1–21	1.5	0.2–3.2
				Vaccinated cohort versus unexposed person-time	Dose 1: days 1–21	1.2	0.2–3.2
RV1	Mexico: 16 hospitals in 10 states	August 2008–August 2010	6–35 weeks	Self-controlled case series	Dose 1: days 1–7	2.0	NR	[24]
	Brazil: 53 hospitals in seven states	August 2008–August 2010	6–35 weeks	Self-controlled case series	Dose 2: days 1–7	1.5	NR	[24]
	Mexico: 221 hospitals	January 2008–October 2010	<1 year	Self-controlled case series	Dose 1: days 0–30	3.7	1.2–3.7	[25]
	Australia: hospitals in six jurisdictions	July 2007–June 2010	1–11 months	Self-controlled case series	Dose 1: days 1–21; Dose 2: days 1–7	4.3	0.8–23.3	[27]
	USA: six health care organizations in five states (VSD)	April 2008–March 2013	4–34 weeks	Vaccinated cohort versus expected rates	Dose 1: days 1–7 Dose 2: days 1–7	5.3	NR	[38]
	USA: National network of three health plans (PRISM)	January 2004–September 2011	5–36.9 weeks	Vaccinated cohort versus unexposed person-time	Dose 2: days 1–21	7.3	0.8–22.5	[37]

AR: Attributable risk; CI: Confidence interval; NA: Not applicable; NR: Not reported; PRISM: Post-licensure rapid immunization safety monitoring program; RRV-TV: RotaShield tetravalent rotavirus vaccine; RV1: Rotarix monovalent human attenuated rotavirus vaccine; RV5: RotaTeq pentavalent human-bovine reassortant rotavirus vaccine.

This evidence prompted the withdrawal of RRV-TV from the US in 1999 less than 1 year after its introduction [8]. A subsequent workshop in the US in 2001 reviewed the available data and came to the consensus that the attributable risk of intussusception due to RRV-TV was approximately 1 per 10,000 vaccinated children (range of 1 per 5000 to 1 per 12,000) [17]. Discussion among the international medical community in February 2000 acknowledged the need for rotavirus vaccines, particularly in settings of higher rotavirus disease burden. However, the manufacturer stopped the production of the vaccine and further efforts to study RRV-TV in other countries, while deemed ethical, did not occur at that time [18].

These studies also prompted a debate about whether the risk of intussusception increased with age (i.e., effect modification) [11]. A re-analysis of the RRV-TV case–control data showed that catch-up immunization during the first year after vaccine introduction, where vaccination was permitted for all children until 6 months of age, contributed disproportionately to the occurrence of intussusception after vaccination. While only 38% of the first vaccine doses were administered to infants ≥90 days of age, approximately 80% of the intussusception cases occurred in this age group. Because baseline incidence of intussusception is higher among children ≥90 days of age during the first 6 months of life, it would be expected that the absolute number of cases would also be higher in this age group [19]. Another re-analysis by Rothman et al. also suggested possible increase in the risk of intussusception with age, relative to background, with the lowest risk among infants 60–90 days of age [20]. However, the authors of the original case–control study did not find significant difference among odds ratio estimates across age categories at which RRV-TV was administered. Because only a few cases were among children <60 days of age, confidence limits were too wide to conclusively determine risk in this age group [21]. Although RRV-TV was removed from the market, the possibility of an age-related risk of intussusception had to be considered with any new vaccines.

Safety & efficacy trials of RV5 & RV1

In the years following the withdrawal of RRV-TV, development of the two oral rotavirus vaccines currently recommended, RV5 and RV1, involved large clinical trials that were designed to permit evaluation of safety with respect to intussusception; the results of these trials were published in 2006 [9,10]. Both vaccines were shown to be highly efficacious in clinical trials conducted in high- and high-middle income countries, mainly in Europe and the Americas [9,10,22]. Based on the results of these trials, the WHO initially recommended the use of RV1 and RV5 among infants in those regions in 2007, and later expanded their recommendations to include all national immunization programs worldwide in 2009 following the availability of efficacy data from low-income settings in Africa and Asia [23]. The WHO recommendations included age restrictions on vaccine administration, such that the first dose of either RV1 or RV5 was to be given to infants aged 6–15 weeks, and the last dose of either vaccine could be administered to infants up to 32 weeks of age. The age restrictions were driven in part by concerns of a higher baseline intussusception risk in older infants and based on age criteria used for subjects enrolled in the clinical trials.

The pivotal clinical trial for RV5 was performed in 11 countries including from Europe and the US and included 68,038 healthy infants. This study did not identify a significant increased risk of intussusception within 42 days of vaccination of any of the three vaccine doses [9]. This primary safety hypothesis was satisfied in the trial, with six vaccine recipients and five placebo recipients who developed confirmed intussusception within 42 days after any dose of RV5 (relative risk [RR]: 1.6 [95% CI: 0.4–6.4]). Extending the surveillance period to 1 year of the first dose also confirmed longer term safety with intussusception occurring in 12 vaccine recipients and 15 placebo recipients (RR: 0.8 [95% CI: 0.3–1.8]) [9].

The pivotal clinical trial for RV1 was conducted among 63,225 infants in 11 Latin American countries and Finland to assess the risk of intussusception within 31 days of any of the two doses of vaccine [10]. Six vaccine recipients and seven placebo recipients had intussusception within 31 days of either dose, with a RR of 0.85 (95% CI: 0.30–2.42). No temporal clustering was observed after either dose. During the entire safety monitoring period (median duration of 100 days after dose one), nine confirmed cases of intussusception occurred among vaccine recipients (incidence rate, 2.84 per 10,000 infants) and 16 among placebo recipients (incidence rate, 5.07 per 10,000 infants). This translated to a relative risk of 0.56 (95% CI: 0.25 to 1.24) when comparing vaccine to placebo recipients [10].

Post-marketing studies of RV5 & RV1

Post-licensure monitoring for intussusception was conducted in countries where vaccines were implemented and data demonstrating an increased risk of intussusception after vaccination in those settings were published beginning in 2011.

Mexico & Brazil (RV1)

A study of RV1 in Mexico and Brazil from August 2008 through August 2010 following national vaccine introduction that used case-series and case–control methods detected a small increase in the risk of intussusception among infants within 1–7 days after receiving the first dose in Mexico and after the second dose in Brazil [24]. In Mexico, an increased risk of intussusception was detected 1–7 days after the first dose of RV1 in both the case-series (incidence ratio, 5.3; 95% CI: 3.0–9.3) and case–control (OR 5.8; 95% CI: 2.6–13.0) analyses. Although no significant increased risk was observed after the first dose in Brazil, a smaller increased risk was observed 1–7 days following the second dose in the case-series (incidence ratio, 2.6; 95% CI: 1.3–5.2) and case–control analyses (OR 1.9; 95% CI: 1.1–3.4). These risks translated to approximately 1 excess case per 51,000 vaccinated infants in Mexico and 1 per 68,000 vaccinated infants in Brazil attributable to RV1, with an estimated 96 cases of intussusception and five deaths caused annually in both countries combined (Table 1). However, these risks were outweighed by the estimated benefits of over 1300 deaths and 80,000 hospitalizations due to rotavirus disease averted each year among children <5 years of age (Table 2) [24].

Table 2.

Risk-benefit estimates of rotavirus disease and intussusception outcomes by country^†.

Country	Outcome	Rotavirus outcomes averted	Intussusception outcomes caused	Rotavirus outcome averted: intussusception outcome caused	Ref.
Mexico	Hospitalizations	11,551	41	282:1	[24]
	Deaths	663	2	331:1
Brazil	Hospitalizations	69,572	55	1265:1	[24]
	Deaths	640	3	213:1
Australia	Hospitalizations	6528	14	466:1	[27]
	Deaths	NR	NR	NR
USA	Hospitalizations	53,444	35–166	322–1530:1	[52]
	Deaths	14	0.1–0.5	28–134:1

^†Estimates based on one vaccinated birth cohort to age 5 years.

NR: Not reported.

Similar results were found in a separate post-marketing study conducted in Mexico by the vaccine manufacturer that used prospective active surveillance for intussusception among infants <1 year of age through the review of medical records at 221 hospitals from January 2008 through October 2010 [25]. For the 701 cases of intussusception that were identified in 698 vaccinated infants, a self-controlled case-series analysis adjusted for age was performed for each RV1 dose separately. The age-adjusted incidence of intussusception occurring 0–30 days after vaccination was compared with those occurring 31 days post-vaccination through 1 year of age and found to be significantly higher following dose 1 but not dose 2. The relative incidence of intussusception following dose 1 was 1.75 (95.5% CI: 1.24–2.48) for the 0- to 30-day period, 3.24 (95.5% CI: 2.15–4.87) for the 0- to 15-day period and 6.49 (95.5% CI: 4.17–10.09) for the 0- to 6-day period. The attributable risk of intussusception with RV1 vaccination was estimated to be 3.7 (95.5% CI: 1.2–7.3) excess cases per 100,000 infants vaccinated (Table 1) [25].

Australia (RV1 & RV5)

The earliest post-marketing data from Australia found some evidence of an increased risk of intussusception following the first dose of both RV1 and RV5 in the first 18 months following national introduction in 2007 [26]. Active surveillance consisting of hospital-based monitoring and pediatrician reporting identified 92 cases of intussusception among vaccinated infants 1 to <9 months of age. These rates were compared to expected incidence rates based on previous hospitalization data from 2000 through 2006. There was no elevated risk of intussusception following any dose (all doses combined) of RV1 or RV5 among infants 1 to <9 months of age in the 1- to 7-day and 1- to 21-day post-vaccination periods. However, restricting analyses to infants 1 to <3 months of age, a possible association between the first dose of rotavirus vaccine with an increased risk of intussusception was detected that invited further study [26].

Continued surveillance in a subsequent larger study in Australia and using different analytic methods demonstrated increased risk of intussusception associated with both RV1 and RV5 [27]. Hospital discharge databases covering >95% of the country’s population were used as the primary source to identify cases of intussusception among infants aged 1 to <12 months of age occurring between July 2007 and June 2010. RV1 and RV5 vaccination histories for all subjects were ascertained from a national immunization register. Self-controlled case-series analyses using age-adjusted regression models compared the risk of intussusception in the 1- to 7-day and 8- to 21-day periods after vaccination dose with the risk observed outside of the 1- to 21-day post-vaccination period among 306 confirmed cases. Elevated risks were observed with both vaccines after the first dose, with a relative incidence of intussusception of 6.8 (95% CI: 2.4–19.0) for RV1 and 9.9 (95% CI: 3.7–26.4) for RV5 in the 1- to 7-day post-vaccine period, and 3.5 (95% CI: 1.3–8.9) for RV1 and 6.3 (95% CI: 2.8–14.4) for RV5 in the 8- to 21-day post-vaccine period. Smaller increased risks were seen in the 1- to 7-day period after the second dose of each vaccine, with a relative incidence of 2.8 (95% CI: 1.1–7.3) for RV1 and 2.8 (95% CI: 1.3–6.8) for RV5. These translated to attributable risks of 4.3 (95% CI: 0.8–23.3) cases per 100,000 vaccinated infants for RV1 and 7.0 (95% CI: 1.5–33.1) cases per 100,000 vaccinated infants for RV5 (Table 1).

Matched case–control analyses were also performed for each vaccine type separately, in which controls selected from the national immunization register were assigned the same ‘event date’ as their matched intussusception case, and the frequency with which the events occurred in the post-vaccination period was compared. These analyses also found an increased risk after the first dose of each vaccine, with odds ratios of 15.6 (95% CI: 3.4–72.6) for RV1 and 11.7 (95% CI: 3.2–43.4) for RV5 in the 1- to 7-day period, and 6.5 (95% CI: 1.7–24.2) for RV1 and 4.7 (95% CI: 1.8–12.0) for RV5 in the 8- to 21-day post-vaccination period. No statistically significant risk was observed after the second dose of either vaccine, with odds ratios of 2.4 (95% CI: 0.8–7.5) for RV1 and 2.5 (95% CI: 0.9–7.2) for RV5 in the 1- to 7-day period. A subsequent risk–benefit analysis considered the attributable risks of both vaccines to be the same, given the overlapping confidence intervals from the self-controlled case-series analyses and used a mid-range estimate of 5.6 additional cases of intussusception per 100,000 vaccinated infants, which translated to 14 excess annual cases nationally. This risk was outweighed by the benefits of vaccination, which was conservatively estimated to prevent 6500 cases of acute gastroenteritis hospitalizations among children <5 years of age each year [27].

USA (RV1 & RV5)

The earliest post-licensure safety monitoring data in the US focused primarily on RV5, which was introduced in the US in 2006, 2 years prior to RV1 introduction in 2008. This included analyses of reports to the VAERS database, which identified 160 reports of confirmed intussusception passively reported from February 2006 through September 2007. Of these, 47 (29%) occurred in infants 1–21 days after receiving RV5, including 27 (17%) that occurred within 1–7 days of RV5 vaccination. Comparing estimated reporting rates to baseline age-adjusted intussusception rates revealed no increase in risk either in the 1- to 21-day (RR: 0.53 [95% CI: 0.37–0.75] or 1- to 7-day (RR: 0.91 [95% CI: 0.59–1.37]) periods after RV5 vaccination [28]. However, the apparent clustering of intussusception cases in the week after vaccination was noted and further monitoring was recommended.

Early analysis of electronic data from the Vaccine Safety Datalink (VSD), a surveillance system managed collaboratively by CDC and several integrated health care organizations that links vaccination data to medical outcomes based on International Classification of Diseases-9 diagnosis codes, found no evidence of increased risk of intussusception following RV5 [28,29]. Initial analysis of early VSD data (May 2006–September 2007) found no increase in the risk of intussusception through 30 days after RV5 vaccination among children 4–48 weeks of age when compared with children of the same age who received any vaccine other than RV5 (RR: 0.84 [95% CI: 0.14–3.92]) [28]. A subsequent cohort study of the VSD population (May 2006–May 2008) compared intussusception rates within 30 days following any dose of RV5 to historical rates (1991–2004) using sequential analyses of weekly data. By the end of the study period, 207,621 doses had been administered and no statistically significant increased risk was found after any dose of RV5 [29].

A later VSD cohort study of infants 4–34 weeks of age from eight managed care facilities assessed intussusception risk with RV5 from May 2006 through February 2010, by which point 786,725 doses of RV5 had been administered [30]. VSD data and medical records were used to identify three cases of intussusception that occurred in the 1- to 7-day period after receiving RV5 and two cases in the same period after receiving a comparison non-rotavirus vaccine. For the 1- to 30-day period after vaccination, 14 cases were identified following RV5 and eight following the comparison vaccine. Rates of intussusception for infants who received RV5 were again not significantly higher than infants who received other recommended vaccines but not RV5, either in the 1–7 days following vaccination (RR: 0.90 [95% CI: 0.10–11.08]) or in the 1–30 days following vaccination (RR: 0.90 [95% CI: 0.10–11.08]). A historical comparison against unexposed rates from 2001 through 2005 from hospital and ED visit data likewise did not show any statistically significant increase in either risk window: standardized incidence ratios (SIR) of 1.01 (95% CI: 0.62–1.54) was observed in the 1- to 30-day window after vaccination following all RV5 doses and 1.23 (95% CI: 0.5–2.54) following the first dose. SIRs of 0.92 (95% CI: 0.25–2.36) and 1.21 (95% CI: 0.03–6.75) were observed for the 1- to 7-day window after all RV5 doses and first RV5 dose, respectively. By using the highest 95% CI upper limit for SIR of 6.75 in the historical comparison, the study estimated a highest theoretical attributable risk of one intussusception case per 65,287 children vaccinated with first dose of RV5.

Ecological studies corroborated these early surveillance data as they gave no evidence of increased rates of intussusception among infants overall in the first years after rotavirus vaccines were reintroduced at the population level [31,32]. These cross-sectional studies relied on baseline estimates of intussusception that were consistent with pre-vaccine data [33,34]. Discharge diagnosis data from over 4000 pediatric hospitals in 44 states (Kids’ Inpatient Database) compared rates of intussusception hospitalizations among children <1 year of age in 2009 with that from 4 years prior to vaccine reintroduction (1997, 2000, 2003 and 2006) and found no significant difference [31]. Between 1997 and 2006, a small decline in intussusception hospitalization rates was observed from 41.6 (95% CI: 36.7–46.5) to 36.5 (95% CI: 31.7–41.2) per 100,000 infants. Based on this historical trend, a predicted rate of intussusception for 2009 was 36.0 (95% CI: 30.2–41.8) per 100,000 infants; the actual observed rate in 2009 showed no increase at 33.3 (95% CI: 29.0–37.6) intussusception discharges per 100,000 infants.

Hospital discharge diagnosis data from a national database (State Inpatient Databases) and corresponding postcensal population estimates were also used to assess trends in intussusception hospitalization rates among infants from 26 states before (2000–2005) and after (2007–2009) rotavirus vaccine introduction [32]. For infants <12 months of age, compared with the 2000–2005 pre-vaccine rate of intussusception (35.3 per 100,000 children), a statistically significant increase was detected in 2007, with a rate of 39.0 per 100,000 (RR: 1.10 [95% CI: 1.04–1.18), but not in 2008 or 2009, where rates were 33.4 per 100,000 (RR: 0.95 [95% CI: 0.89–1.01]) and 32.9 per 100,000 (RR: 0.93 [95% CI: 0.87–0.99]), respectively. Among infants aged 8–11 weeks, to whom most first doses of rotavirus vaccine were given, compared with the 2000–2005 rate of intussusception (6.9 per 100,000), statistically significant increases were observed in each post-vaccine year of 2007 (11.4 per 100,000; RR: 1.64 [95% CI: 1.08–2.50]), 2008 (12.2 per 100,000; RR: 1.76 [95% CI: 1.17–2.65]) and 2009 (11.0 per 100,000; RR: 1.59 [95% CI: 1.04–2.44]). However, no progressive increase of intussusception rates was observed with each subsequent post-vaccine year, as would be expected with increasing vaccine coverage in the population if vaccine were associated with intussusception.

The increased risk detected in other countries prompted continued surveillance and increased caution in the US, including the addition of a history of intussusception as a contraindication to rotavirus vaccination in 2011 by the US FDA and CDC [35]. With the accumulation of additional data, the most recent US post-marketing studies have also identified a small increased risk of intussusception associated with RV5 and RV1 [36-38]. Five hundred and eighty-four confirmed intussusception events were reported to VAERS between 2006 and 2012 after >47 million doses of RV5 administered were analyzed [36]. A self-controlled risk interval analysis using Poisson regression was used to determine that the average daily reports of intussusception during the 3- to 6-day period after the first RV5 dose was 3.75 (95% CI: 1.90–7.39) times higher than that during the 0- to 2-day period after first dose of RV5. This translated to an excess risk of 0.74 cases of intussusception (95% CI: 0.24–1.71) per 100,000 vaccinated infants (Table 1). No significant increases in reporting ratios between the two time periods were found after dose 2 (1.43; 95% CI: 0.85–2.40) or dose 3 (0.75; 95% CI: 0.40–1.42). The total excess risk of intussusception after all three RV5 doses combined was 0.79 (95% CI: −0.04 to1.62) per 100,000 vaccinated infants, which would translate to 33 additional intussusception cases per year (95% CI: 0–66) in the US in a fully mature rotavirus vaccination program with coverage near that of DTaP. The 52 confirmed intussusception reports after RV1 vaccination from April 2008 through April 2012 also appeared to cluster in the 3- to 6-day period after each dose, but were too few, given the low number of doses administered in the US up to that time, to allow for more detailed analysis.

The FDA Post-Licensure Rapid Immunization Safety Monitoring program (PRISM), a surveillance platform for medical outcomes following vaccinations, evaluated data from children who were members of three US health insurance plans to study approximately 613,000 infant-years, 1.3 million total doses of RV5 and 100,000 total doses of RV1 given between January 2004 and September 2011 [37]. Medical insurance claims data for this population of children 5–36.9 weeks of age were used to identify potential cases of intussusception and RV1 and RV5 vaccination status, which were then confirmed by medical record review. Primary analysis employed a self-controlled risk-interval design (SCRI), where counts of intussusception cases in the 1- to 7- and 1- to 21-day risk intervals after each dose of vaccine separately, and all doses combined, were compared with those cases occurring in the 22- to 42-day control interval. A secondary cohort analysis using Poisson regression compared intussusception events occurring in the 1–21 days after any dose against those occurring outside the 21-day post-dose period as well as in infants 5–36 weeks of age who had not received rotavirus vaccine. Age was adjusted for in both analyses.

This study identified 124 confirmed cases of intussusception [37]. In the primary SCRI analysis for the first dose of RV5, a relative risk of 9.1 (95% CI: 2.2–38.6) was observed in the 1- to 7-day post-vaccination period, translating to an attributable risk of 1.1 (95% CI: 0.3–2.7) cases per 100,000 first-dose vaccinated children; a relative risk of 4.2 (95% CI: 1.1–16.0) was found in the 1- to 21-day post-vaccination period, which translated to an attributable risk of 1.5 (95% CI: 0.2–3.2) cases per 100,000 first-dose vaccinated children (Table 1). The secondary cohort analysis also identified a statistically significant elevated risk of intussusception in the 1- to 21-day period after the first dose of RV5 (RR: 2.6 [95% CI: 1.2–5.8]), which translated to an attributable risk of 1.2 (95% CI: 0.2–3.2) cases per 100,000 first-dose vaccinated children. No statistically significant increases in risk were observed for doses two or three of RV5 in either the SCRI or cohort analyses.

The power of analyses for RV1 risk estimates was low, given the fewer number of doses administered in the population and intussusception cases observed than with RV5, and SCRI analysis for RV1 did not reveal significant attributable risk for either dose 1 or 2. Secondary cohort analysis did, however, identify a relative risk of 5.1 (95% CI: 1.6–16.4) for dose 2 of RV1, which translated to an attributable risk of 7.3 (95% CI: 0.8–22.5) cases per 100,000 second-dose vaccinated children; statistically significant results were not obtained for the first dose of RV1. A comparison between RV5 and RV1 was not conducted in this study [37].

Accumulated data from the VSD project confirmed a small but increased risk of intussusception with RV1 [38]. From April 2008 through March 2013, weekly surveillance of discharge data for intussusception coded visits was conducted among infants 4–34 weeks of age who received at least one dose of any childhood vaccine from six health care organizations. Medical records and electronic data concerning vaccination status and International Classification of Diseases-9-CM codes from emergency department and hospital visits were reviewed for intussusception that occurred within 7 days of receiving either dose of RV1. These rates were observed through sequential monitoring and compared to background historical rates of intussusception (2001–2005). In addition, analyses for intussusception risk with RV5 among a separate cohort of infants were updated through March 2013, and analyses comparing RV1 and RV5 were performed for data from April 2008 through March 2013 when both vaccines were in use.

During the study period, approximately 208,000 doses of RV1 were administered, including approximately 116,000 first doses and 92,000 second doses [38]. Six cases of intussusception occurring within 7 days of either RV1 dose were identified, which was significantly higher than the 0.72 cases expected based on background historical rates, for a relative risk of 8.38; the relative risk within 7 days after dose 2 (8.17) was statistically significant; the statistical significance of the dose 1 relative risk (8.82) could not be calculated because of the low number of cases. The attributable risk of intussusception following two RV1 doses was 5.3 per 100,000 vaccinated infants (Table 1). Cluster analyses of all intussusception cases within 30 days of vaccination revealed significant clustering of cases (p = 0.02) in the 3- to 6-day interval after RV1 vaccination.

Updated analyses of the risk of intussusception within 7 days of vaccination with RV5 using data from May 2006 through March 2013 including approximately 1.3 million total doses, with 494,000 first doses, revealed no increased risk in the number of cases of intussusception when compared to the expected number of cases based on historical rates after all doses (RR: 1.13 [95% CI: 0.49–2.22]) and first doses of RV5 (RR: 2.63 [95% CI: 0.72–6.74]). No significant clustering was detected within 30 days after RV5 vaccination. When comparing rates of intussusception following the first two doses of vaccine, the adjusted relative risk of intussusception after RV1 compared to RV5 was 9.4 (95% CI: 1.4–103.8). The risk difference between infants receiving two doses of RV1 compared to those receiving the first two doses of RV5 was 4.4 cases per 100,000 infants [38].

Considerations for developing countries

In December 2013, the Global Advisory Committee on Vaccine Safety found these most recent data to be reassuring in that the risk of intussusception following rotavirus vaccines is small compared to the benefits of vaccination, but also recommended further study in developing countries [39]. Indeed, because estimates of the small increased risk of intussusception following rotavirus vaccination to date are from high- and middle-income countries, additional considerations regarding the risk in lower-income settings are warranted. First, data on the epidemiology of intussusception in developing countries are sparse, including in several where routine immunization has recently started or is about to commence, and further study is needed to characterize baseline rates of intussusception. To that end, intussusception surveillance efforts in some lower-income countries have already begun [40,41].

Second, it is unclear if rotavirus vaccine has a risk of intussusception in populations from developing countries as has been found in higher-income countries because immunogenicity and efficacy is lower in developing countries (39–61% vs 85–98% efficacy against severe rotavirus disease) [9,10,22,42-44]. The reasons for these differences in immunogenicity and efficacy are not clear and may involve several factors, but could have important implications for vaccine safety with regard to intussusception [45]. For example, administration of trivalent oral polio vaccine, which is routinely administered concomitantly with rotavirus vaccine, has been shown to decrease immunogenicity of rotavirus vaccine, particularly after the first dose [46]. It has been hypothesized that the routine use of trivalent oral polio vaccine in Brazil at the time of the study may help account for the lower risk of intussusception following RV1 compared to Mexico, where inactivated poliovirus vaccine was used [24,46]. In this scenario, the lower immunogenicity of the first dose of RV1 may explain the lower risk of intussusception following the first dose in Brazil, whereas the risk is observed in Mexico. It has yet to be determined if and how differences in immunogenicity or efficacy seen in lower-income countries affect intussusception risk in those settings, but the documented lower immune responses in low-income settings support the contention that intussusception risk after rotavirus vaccination may also be lower in those settings.

Third, developing countries stand to benefit the most from the recent relaxation of age restrictions on rotavirus vaccination given the higher child mortality due to severe rotavirus disease [4]. Initial WHO recommendations specified age restrictions of initiation of the series by 15 weeks of age and completion by 32 weeks of age, which was driven by considerations of intussusception risk based on clinical trials [23]. However, it has been recognized that a substantial number of children could be excluded from the benefits of rotavirus vaccine under these age restrictions, especially in developing settings where delays in vaccinations are common. Modeling estimates based on country mortality, vaccine efficacy and intussusception risk data showed that the benefits of removing these age restrictions in low- and low-middle income countries far outweighed the risks, resulting in 154 rotavirus deaths averted for each additional intussusception death due to vaccine [47]. In light of these findings, although vaccination is still recommended with initiation of the vaccine series as soon as possible after 6 weeks of age, and not recommended for children >24 months of age, further age restrictions are no longer specified by WHO [48].

Studying intussusception and any associated risk with rotavirus vaccine in developing countries will likely be challenging due to delays in presentation and diagnosis in some infants given a relative lack of resources and in determining the vaccination status of children with intussusception. However, the case-series method employed in previous studies offers an opportunity to assess risk in these settings. Although the benefits of vaccination are likely to outweigh the risk in these settings where rotavirus disease mortality is highest, further study is nevertheless needed to demonstrate benefits and risks.

Expert commentary

Despite its demonstrated efficacy, RRV-TV was withdrawn due to a significant association with intussusception of approximately one additional case per 10,000 vaccinated infants. The early detection of this rare but serious adverse event informed the development process of RV5 and RV1, and large clinical trials indicated these new vaccines did not have the degree of risk observed with RRV-TV. Due to their efficacy and safety profile, as well as the great burden of rotavirus disease worldwide, both vaccines were recommended for use in infants in both developed and developing countries. As summarized in this review, post-marketing surveillance data have since identified a smaller increased risk of intussusception associated with both vaccines; studies were performed in high- and middle-income countries in different regions of the world. Despite minor differences across studies, considering all the available evidence together, the risk of intussusception for RV1 and RV5 approximates one to six excess cases per 100,000 infants.

In contrast to RRV-TV, where early detection of the associated risk led to its withdrawal within a year of release, by the time an increased risk of intussusception following RV1 and RV5 was suggested, sufficient data about the effectiveness and impact of both currently licensed rotavirus vaccines had accumulated that demonstrated both the direct and indirect benefits in several countries where they had been introduced [49-51]. Available scientific evidence on the benefits and risks of both vaccines informed decisions on their continued use in several countries (Table 2), and to date, consideration of these data has led health organizations around the world to continue to support the routine use of RV1 and RV5 in national infant immunization programs [48,52].

Five-year view

Because background intussusception rates and rotavirus vaccine efficacy can vary by country, surveillance of intussusception and rotavirus disease is needed in multiple settings to best assess the risks and benefits of vaccination. For countries where the risk of intussusception with vaccine has already been documented, continued post-marketing surveillance is beneficial to give the most accurate and up-to-date characterization of risks and benefits as their immunization programs mature. In developing countries, further characterization of the baseline epidemiology of intussusception and any associated risk following rotavirus vaccine is also needed.

Other questions still remain. More study is needed to elucidate the mechanism underlying the association between rotavirus vaccine and intussusception. Further consideration of vaccine properties, such as specific vaccine strains, dosing and timing, as well as host factors such as age, genetic factors, nutritional status and concurrent enteric infections may help in this endeavor. Continued study of the relationship between rotavirus vaccine immunogenicity and the risk of intussusception should also further our understanding, especially as it relates to the effects of concomitant administration of other oral vaccines such as the available oral poliovirus vaccines. Finally, it is not yet known if the recently developed monovalent human-bovine strain (116E) vaccine has any associated risk of intussusception [53].

With regard to RV5 and RV1, additional ecological studies will give further insight into the real-world impact of the small observed attributable risks, including documenting if there are any changes in the overall rates of intussusception among the vaccine age-eligible population. Up-to-date formal risk–benefit assessments and cost–effectiveness analyses in different settings should also help guide policy in the future. A better understanding of the risks and benefits of vaccination, along with continued communication among the scientific community, policy makers and the public, including parents of infants, will promote informed decisions regarding vaccination and ensure that the immense benefits that have already been documented can continue while further minimizing the risks.

Key issues.

• Rotavirus disease is the most common cause of severe gastroenteritis in children <5 years of age worldwide.

• The first US-licensed rotavirus vaccine, Rotashield, was withdrawn in 1999 1 year after introduction due to an association with intussusception of approximately 1 excess case per 10,000 vaccinated children.

• In 2006, large clinical trials for the two currently recommended live oral vaccines, RotaTeq (RV5) and Rotarix (RV1), demonstrated good efficacy and safety with no evidence of association with intussusception.

• Since their introduction, RV5 and RV1 have been shown to be effective with substantial impact in several countries.

• The most recent post-marketing data estimate the risk of intussusception with RV1 and RV5 to be approximately 1–6 excess cases per 100,000 vaccinated infants, which is lower than that of Rotashield.

• The benefits of RV5 and RV1 have been shown to outweigh the small excess risk of intussusception in multiple settings.

• In the light of these data, WHO continues to recommend RV5 and RV1 in national immunization programs for all countries.

• Additional study for risks and benefits of vaccination is needed and ongoing in developing countries, where mortality due to rotavirus disease is highest.