A multi-country study of intussusception in children under 2 years of age in Latin America: analysis of prospective surveillance data

Intussusception (IS) is a form of acute intestinal obstruction that occurs mainly in infants [1]. It is the most frequent cause of acute abdominal emergency in the first two years of life but rarely occurs in adults [1]. Most cases of IS are considered idiopathic although several authors have suggested links with various infectious agents [2–4]. An association between natural rotavirus (RV) infection and IS has not been found [3, 5, 6]. The main argument against this association is the seasonality of RV disease in children between 3–24 months without any similar variability in IS prevalence in the same age group [6–10].

The first RV vaccine licensed in the United States (RotaShield™, Wyeth-Lederle) was withdrawn in 1999 because of concerns about the association of vaccination with IS [11, 12]. In retrospective investigations, administration of this tetravalent rhesus human reassortant RV vaccine (RRV-TV) was associated with a significantly increased risk of IS (case–control analysis, adjusted odds ratio: 21.7; 95% CI: 9.6 to 48.9), especially in the two weeks following the first vaccination dose [5, 13].

Development of a safe and effective RV vaccine has been given high priority by the WHO because of the considerable RV disease burden especially in developing countries [14, 15]. Newly licensed RV vaccines include a human RV vaccine (HRV) (Rotarix™_, GlaxoSmithKline Vaccines) and a human-bovine reassortant vaccine (RotaTeq™, Merck & Co.). Results of large-scale, randomized, double blind, placebo-controlled studies, each involving more than 60,000 infants, show these new RV vaccines to be immunogenic and highly effective for the prevention of RV gastroenteritis and associated hospitalizations during the first year of life. Importantly, both vaccines appear to have a favorable safety profile, with both studies excluding any immediate risk of IS (risk <1 in 20,000) [16, 17].

Universal RV vaccination has been introduced in many Latin American countries; since 2006, 14 countries and one territory have incorporated vaccines in their National Immunization Programs, including 12 countries using Rotarix™ and 2 using RotaTeq™ [18]. Recent post-licensure IS surveillance studies have associated rotavirus vaccination with an increased, short-term risk of IS within 7 days following the first vaccine dose among Mexican infants [19, 20]. However, a recent analysis involving 14 Latin American countries using rotavirus vaccines provided evidence that health benefits of vaccination far outweigh this transient risk and support continuous RV vaccination in Latin America [21].

This study was conducted to provide an understanding of the epidemiology and to estimate the incidence of IS in Latin American infants prior to RV vaccine introduction. Since there is a need for continuous safety surveillance in countries that have adopted RV vaccination into their public sectors, our study will provide useful baseline IS rates regarding further assessment of the potential IS risk versus health benefits in Latin America.

This study estimated the incidence of definite IS in young children in 11 Latin American countries as 1.9 to 62.4 per 100,000 children under 2 years of age and 3.8 to 105.3 per 100,000 children under 1 year of age, with a wide variation in incidence. The estimates for children under 1 year of age are in a similar range to published estimates from other regions, 38 per 100,000 in Switzerland (study period: 2003–2006) [24], 65 per 100,000 in New Zealand (study period: 1998–2003) [10], 66 per 100,000 in the UK [25], 71 per 100,000 in Denmark (study period: 1990–2001) [26] and 78–100 per 100,000 in Hong Kong (study period: 1997–1999). [8] However, the increase in definite IS incidence in the second 6 months of life seen in other studies [24, 26–28], was not demonstrated in this study.

Our estimated IS incidence rates were similar to those previously reported from this region. In a 5-year retrospective study in Panama (study period: 1998–2002), annual rates of 19 to 40 cases of IS per 100,000 children under 1 year of age, and 8 to 15 per 100,000 in children under 3 years of age were reported depending on the year of the study [29]. A study in Chile estimated the annual IS rates in children under 1 year and under 2 years of age to be 47–55 per 100,000 and 32–35 per 100,000, respectively [30]. A study conducted in Venezuela reported an annual IS-related hospitalization rate of 35 per 100,000 in infants under 1 year of age [31]. In Brazil, where this study showed the lowest incidences of definite IS despite case identification efforts that were equally intensive as those in other participating countries, the World Health Organization has reported similarly low annual incidence (3.5 per 100,000 children under 1 year of age) [32]. There seems to be no clear explanations for the variability in observed IS rates between different Latin American countries, although intussusception incidence can vary by region [32]. Country-specific incidence rates do not seem to suggest ethnic or geographic influences (e.g. a North–south gradient). While differences in healthcare access and patient management (i.e., in some areas IS cases could not have been seen by medical personnel) are likely to play a role, other unknown factors that may be linked to IS, such as genetic influence, cultural differences, infectious diseases and gastrointestinal infections, may also be involved [11, 33–35]. It should be pointed out however that, when defining specific catchment areas, it was assumed that most of potential IS cases would reach the sentinel hospitals for proper treatment, where intensive, daily surveillance was carried out. It is however recognized that less control could be exerted over cases that might come from outside the study areas.

We found a relatively high proportion of subjects (27.3%) who experienced IS to have had an upper respiratory tract infection in the 2 weeks before the onset of definite IS. The condition is known to be associated with a number of childhood viral illnesses [33], in particular with adenovirus [34], and a link with respiratory syncytial virus (RSV) infection has recently been reported [35].

The epidemiologic characteristics of definite IS cases in this study were similar to those found in other locations or previously in these countries. The majority of definite IS cases occurred before 1 year of age with a peak around 4–6 months of age, which is similar to that observed in the US [36] and other countries [24, 25, 29, 30]. The excess of male cases has been noted in other studies with male:female ratios of 1.8:1 and 1.4:1 reported from Panama and the US, respectively [29, 36].

The proportion of children undergoing surgery in the present study (84%) was similar to other Latin American studies, which have reported rates of 78% and 88% in Chile and Venezuela, respectively [30, 31, 37]. However, these rates are much higher than those reported from Hong Kong (23%) and the US (53%) [8, 9]. Higher surgery rates in Latin America may be due to delays in seeking medical attention, more severe episodes, or less experience in radiological reduction techniques. Moreover, it seems likely that IS cases that resolved spontaneously were not included, thus possibly leading to an underestimate of the overall rate. The typical clinical presentation (vomiting, pain or irritability, and bloody stools) found in this study is similar to other reports as was the low rate of predisposing conditions, such as Meckel’s diverticulum [30–32, 38]. Of interest, among the 465 subjects for whom the type of vaccine previously administered was known, 27 (5.8%) had received human rotavirus vaccine or placebo because they were participating in an ongoing phase III clinical trial in Latin America. Nevertheless, it seems very unlikely that such a low percentage might have influenced our overall estimates for the occurrence of IS prior to vaccine introduction.

The present study, like several others including some from Latin America, showed no clear seasonality of IS-associated hospitalisations [6, 29, 31]. The distribution of definite IS cases did not correspond with the seasonality of RV in any of these countries. As in other regions, the occurrence of RV is higher in winter months in Latin America, however, the timing and duration varies by geographical location with peaks generally occurring in Mexico between November and April [6, 39], in Argentina between April and August, in Brazil between May and July, and in Chile between June and September [40, 41].

We acknowledge the potential limitations of our study. To estimate the incidence rates, we relied on census data as denominators. The reliability of these census data is variable. Whereas these data may be valid on national or regional levels, they may be less precise for smaller regions such as those that defined the catchment areas for some centers. It is likely that this uncertainty accounts for some of the difference between the estimated incidence rates in different countries. We excluded children who lived outside the defined study areas. However, study area boundaries may not have had the same level of precision for all hospitals, which could affect the incidence rates presented. It is therefore likely that some cases of definite IS may have been referred to sentinel hospitals from neighborhood localities that were surrounding the originally defined study area. Also, in most countries, the study took place in a limited number of regional/national reference hospitals and therefore the incidence rates for each country cannot be regarded as national estimates.

It is possible that some of the cases that occurred in the population counted as denominator may have been treated in another hospital not involved in the study, in particular private institutions. Although a systematic review of private institutions within the study area was also performed during the surveillance and efficacy trial. To ensure that the study procedures conformed across all investigator centers add to limit potential under-reporting, the study protocol, case report form and safety reporting were reviewed with all investigators and other personnel responsible for the conduct of the study prior to study start and at regular interval periods. Adherence to the protocol requirements and verification of data generation accuracy was achieved through monitoring visits to each investigator site. Furthermore, computer checks and blinded review of subject tabulations were performed to ensure consistency of case report form completion. Finally, some parents (n = 22) refused participation. As we did not know the age of these children, we could not adjust our incidence rates to account for these children, which may have resulted in an underestimation.

Our findings are in line with previously reported figures and suggest substantial variation in the rates of IS across Latin America. This study provides a recent estimate of the incidence of IS in Latin American infants, which is helpful in monitoring the safety of new RV vaccines following their introduction in the region. A few of these previously foreseen post-licensure evaluations have recently detected a transient, 4 to 6-fold increased relative risk of IS within 7 days following first rotavirus vaccine dose in Mexico. Nevertheless, recent estimates have shown that hospitalisations and deaths averted as a result of vaccination far outweigh the number of possible vaccine-related IS cases and health authorities still continue to strongly recommend routine rotavirus vaccination of infants in Latin America.