Prospective surveillance for intussusception in Indian children aged under two years at nineteen tertiary care hospitals

Background India introduced rotavirus vaccines (RVV, monovalent, Rotavac™ and pentavalent, Rotasiil™) in April 2016 with 6, 10 and 14 weeks schedule and expanded countrywide in phases. We describe the epidemiology of intussusception among children aged 2–23 months in India. Methods The prospective surveillance at 19 nationally representative sentinel hospitals from four regions recruited children with intussusception from April 2016 to September 2017. Data on sociodemography, immunization, clinical, treatment and outcome were collected. Along with descriptive analysis, key parameters between four regions were compared using Chi-Square/Fisher’s exact/Mann–Whitney U/Kruskal-Wallis tests. The pre- and post-RVV periods were compared to estimate the risk ratios. Results Six hundred twenty-one children with intussusception from South (n = 262), East (n = 190), North (n = 136) and West (n = 33) regions were recruited. Majority (n = 465, 74.8%) were infants (40.0% aged 4–7 months) with median age 8 months (IQR 5, 13 months), predominantly males (n = 408, 65.7%) and half (n = 311, 50.0%) occurred during March–June months. A shorter interval between weaning and intussusception was observed for ragi based food (median 1 month, IQR 0–4.2 months) compared to rice (median 4 months, IQR 1–9 months) and wheat (median 3 months, IQR 1–7 months) based food (p < 0.01). Abdominal pain or excessive crying (82.8%), vomiting (72.6%), and bloody stool (58.1%) were the leading symptoms. Classical triad (abdominal pain, vomiting and bloody stool) was observed in 34.8% cases (24.4 to 45.8% across regions). 95.3% of the cases were diagnosed by ultrasound. 49.3% (10.5 to 82.4% across regions) cases were managed by reduction, 39.5% (11.5 to 71.1% across regions) cases underwent surgery and 11.1% spontaneously resolved. Eleven (1.8%) cases died. 89.1% cases met Brighton criteria level 1 and 7.6% met Level 2. RVV was received by 12 cases within 1–21 days prior to intussusception. No increase in case load (RR = 0.44; 95% CI 0.22–1.18) or case ratio (RR = 0.5; 95% CI 0.3–1.2) was observed after RVV introduction in select sites. Conclusions Intussusception cases were observed across all sites, although there were variations in cases, presentation and mode of management. The high case load age coincided with age of the RVV third dose. The association with ragi based weaning food in intussusception needs further evaluation.


Background
To reduce the rotavirus diarrhoea related childhood deaths, the World Health Organization (WHO) has recommended introduction of rotavirus vaccine (RVV) in national immunization programmes (NIPs). Some increased risk of intussusception after the first (relative risk, RR: 4.7-13.8) and second (RR: 1. 3-5.3) doses of RVVs was reported from several countries (Mexico, Brazil, Australia, United Kingdom, United States, Spain and Singapore) [1][2][3][4][5][6][7][8]. But no increased risk of intussusception after any dose of RVV was observed in some countries (Ethiopia, Ghana, Kenya, Malawi, Tanzania, Zambia, Zimbabwe and South Africa) [9,10]. The impact of RVV on diarrhoea morbidity and mortality outweigh the risk of intussusception and associated mortality [11]. In view of the concern about intussusception, documentation of baseline and monitoring following RVV introduction have been recommended [12]. India introduced RVV into the NIP in April 2016 and by 2019 expanded countrywide in four phases [13]. Two types of RVV are used in NIP, Rotavac™ (RV1-116E; Bharat Biotech) in 26 states/union territories and Rotasill™ (RV5; Serum. Institute of India) in 11 states/union territories, both follow 6, 10, 14 weeks schedule. Intussusception is an acute severe clinical condition occurring mostly during infancy, which overlaps with the age of primary vaccinations. The nationally representative background epidemiology of intussusception in India is not clear. The reports from India primarily included retrospective data and considerably varied in the epidemiology, presentation and management [14,15]. Information on population level rate in India is limited and the reported incidences of intussusception requiring hospitalization varied from 17.7 (95% CI 5.9-41.4) in Delhi (North India) to 254 (95% CI 102, 524) cases per 100,000 child-years in Vellore (South India) [16,17]. The incidence of intussusception vary widely globally, across the different high and low-middle income countries [18]. The reasons for the variations are unknown. There is no information from India regarding the regional variation in intussusception epidemiology. Thus, documentation of the intussusception epidemiology prior to RVV introduction to establish a reliable baseline for monitoring the trend over time and identify potential risk factors was needed to support the vaccine safety surveillance efforts [19,20]. Under the vaccine safety surveillance effort, we describe the epidemiology, clinical characteristics of intussusception among children aged under-2 years seeking hospital care in India and initial changes with RVV introduction documented through a nationally representative sentinel surveillance network.

Study area and participating hospitals
This prospective active hospital-based sentinel surveillance was conducted over 18 months, at 19 nationally representative tertiary care hospitals (Supplementary Figure 1). From the four regions, 3-6 hospitals per region including medical colleges and private-sector hospitals (North region-5 sites, 3 public and 2 private; South region-5 sites, 2 public and 3 private; East region-6 sites, 5 public and 1 private; West region-3 sites, 2 public and 1 private) were selected through a systematic process. Out of these sites, four sites were located in the states where RVV was introduced in April 2017 under phase 1. At all sites the RVVs were available in private market during the study period, even prior to the introduction in NIP.

Case definition, case selection and data collection
The children aged 2-23 months admitted to these hospitals with diagnosis of intussusception were eligible. All the ageeligible patients admitted were screened to identify the suspected cases (any of the diagnoses: intussusception, acute or subacute intestinal obstruction, acute abdomen, abdominal pain, abdominal distension, and blood in stool with vomiting). These suspected cases were tracked till final diagnosis and all confirmed intussusception cases were recruited after written informed consent from parent or legally authorised representative. A log of screened, suspected and confirmed cases was maintained. The data on socio-demography, feeding and immunization (from immunization card), clinical features, investigation findings, hospital course, treatment and outcome were collected using common case record form (CRF). The symptoms were captured as recorded in the case sheets and reported by the parents. An independent Case Adjudication Committee (CAC) comprised of a paediatrician, paediatric surgeon, and radiologist reviewed the CRFs and investigations to assign the diagnostic certainty levels, according to Brighton Collaboration case definition (BCCD) [21].

Quality assurance
Multilevel quality assurance and data quality-checking processes were put in place to ascertain protocol adherence, rigor and completion of surveillance at all sites. A data team reviewed all the CRFs and data-related query was resolved at the earliest. Each site was visited by external experts (Technical Advisory Group, TAG) to assess the case surveillance and tracking, consent, and data extraction quality and completeness in the CRFs. The TAG members checked CRFs completeness and quality for few randomly identified cases with the case records. Subsequently members from the data team visited the study sites and checked the admissions for the study period from the medical records section using diagnoses and International Classification of Diseases (ICD) codes (ICD-9/10, codes listed in Supplementary Table 1), to identify any missed cases.

Data management and analysis
Double data entry was done for the CRFs using a customised data entry platform. The matched and verified data were stored in the server with authorised access and daily backup. The standard of living index (SLI), representing the socioeconomic status was estimated using the scores for household assets ownership, with reference to the National Family and Health Survey for India [22]. The SLI was categorised into high, medium and low categories. The cases were categorised into levels 1 to 3 based on the BCCD criteria by the CAC [21]. In view of the number of sites, for representation we grouped the sites into four regions, North (5 sites), South (5 sites), East (6 sites) and West (3 sites) regions. The parameters (sociodemographic, feeding, clinical, intervention and outcomes were compared between the regions to detect variations. The intussusception classical triad includes three features; abdominal pain, vomiting and blood in stools. We considered intussusception modified triad with abdominal pain, vomiting and rectal bleeding, detected either as blood in stool or blood on per-rectal examination. The descriptive analysis findings expressed the outputs as proportions, means and standard deviations, or median and interquartile range (IQR), as appropriate. The values between regions and groups were compared for statistical significance using Chi-Square or Fisher's exact tests for the proportions and Mann-Whitney U or Kruskal-Wallis tests for the medians depending on the skewness, sample and number of groups. The missing data were excluded from analysis. The statistical significance was considered if p < 0.05. The statistical analysis was performed using STATA version 15.0 (StataCorp LLC, Texas, USA). In the Indian context, there is no definite population catchment area for hospitals and no definite referral chain, which makes estimation of the intussusception incidence difficult. For comparison of the data across sites and intussusception time trend, in addition to case load (absolute number of cases), we attempted deriving the intussusception case rate per 1000 paediatric hospitalisations at the hospitals. On review, while the admission numbers in the paediatric medicine wards varied widely, the admission numbers in the paediatric surgery wards were relatively stable and the intussusception cases were primarily managed in the paediatric surgery wards.  [23].

Ethical issues
Informed written consent was obtained for all the eligible cases before recruitment and data collection. Confidentiality in data handling was maintained. The study protocol was reviewed and approved by the ethics committees of all participating institutes.

Results
Between   Fig. 1 shows monthly trends of pooled intussusception case load and case rates per 1000 paediatric surgery admission ( Fig. 1a) and the regional case rates per 1000 paediatric surgery admission (Fig. 1b). More cases (n = 311, 50.0%) were seen during March to June months, in the summer season ( Fig. 1). While 42.8% of the patients were resident of the same district where the hospital was based, 41.9% of the patients presented directly to the hospital. Among the children aged > 6 months, 23.2% were exclusively breastfed for 6 months and the median duration of breastfeeding was 4 months. Mixed feeding was initiated before 6 months of age in 53.9% children. The median weaning age was 6 months and 61.0% received rice based food. Ragi (finger millet) was given to 22.8% children, only in the South region. The interval between weaning and intussusception was significantly shorter for ragi (median 1 month, IQR 0-4.2 months) than rice (median 4 months, IQR 1-9 months) and wheat (median 3 months, IQR 1-7 months) based food (    Rice based, median (IQR) Multiple food, median (IQR) 7 (2.5-9) 6 (2.7-12) 6 (3-12) 6 (3-12) 6 (3-12) 0. region only (where ragi based weaning practice was observed), the median interval between weaning and intussusception for ragi based food was significantly shorter (median 1 month, IQR 0-4 months) than rice (median 5 months, IQR 2-9 months) and wheat (median 3 months, IQR 2-8 months) based food (p < 0.01).
The pooled median interval between the onset to hospital admission was 2 days (IQR 1, 3), which varied across the regions; North-2 days (IQR 1, 3), South-1 day (IQR 1, 3), East-2 days (IQR 1, 3) and West-2 days (IQR 1, 2) (p = 0.07). Table 2 summarises the clinicopathological parameters and management for the recruited children. Abdominal pain or excessive crying (82.6%) was the most common symptom followed by vomiting (72.6%) and blood in stool (58.1%), as reported by the parents. More children had abdominal pain or excessive crying in North (84.6%) and East (72.6%) regions and blood in stool in the East region (73.2%). The classical triad (abdominal pain, vomiting and blood in stool) was observed in 34.8% cases, with 24.4 and 45.8% in the South and in East regions, respectively (p < 0.01). When blood on per-rectal examination was included, the modified intussusception triad (abdominal pain, vomiting and blood in stool or blood on per-rectal examination) was Fig. 1 The seasonal distribution of pooled intussusception case load and case rate during the calendar months (a) and regional intussusception case rates during the study period (b). Note: Intussusception case rate: Intussusception cases per 1000 paediatric surgery admissions at the hospitals    Most of the cases (95.3%) were diagnosed by ultrasound. Ileocolic (84.7%) was the most common site of intussusception. A pathological lead point (PLP) was documented in 91 (14.6%) cases and lymph nodes or Payer's patch were the commonest (13.0%) and few had appendix or polyps (Table 2). While 49.3% cases were reduced, 39.5% underwent surgery and 11.1% resolved spontaneously. More children in East region (71.1%) underwent surgery followed by North region (50.0%) (p < 0.01). Among those who underwent surgery, 26.1% children required bowel resection, and 39.5% were from the North region. The onset-admission interval was longer for the cases who underwent surgery (median 2 days; IQR 1, 3) than reduction (median 1 day; IQR 1, 3) (p < 0.01) (Supplementary Table 2). Among the children who presented on the day of onset, 62.3% were managed by reduction (28.9% required surgery), which declined to 35.1% when they presented after 3 days (48.3% required surgery) (p < 0.01) (Supplementary Table 2). Seven hospitals (East-1, North-1, and South-1) were conducting only surgical management. At the hospitals with both facilities (reduction and surgery), 31.0% (127/404) children underwent surgery for indications: failed reduction (32.2%), late presentation (> 3 days since onset, 28.3%) and associated complications (39.3%) (p < 0.01). Most of the cases (97.1%) recovered and were discharged. Eleven cases (North-4 and East-7) died of post-surgery sepsis, shock and multiorgan failure. The median hospitalisation period was 3 days (IQR 2, 6 days) ranging from 2 to 5 days (South: 2 days, IQR 2, 3 days; North: 3 days, IQR 1, 7 days; West: 4 days, IQR 2, 6 days; and East: 5 days, IQR 3, 8 days; p < 0.01). The median hospital stay periods for children who underwent surgery, reduction and spontaneously resolved were 7 days (IQR 5, 9 days), 2 days (IQR 1, 2 days) and 3 days (IQR 2, 5 days), respectively (p < 0.01) (Supplementary Table 2). CAC assigned 89.1% (553/621) and 7.6% cases as Level 1 and 2 respectively according to the BCCD. Diarrhoea and acute respiratory illnesses (ARIs) within 4 weeks prior to intussusception was reported in 12.9 and 23.7% children respectively. More children from South region had history of diarrhoea (21%) and ARIs (39.7%) than other regions (p < 0.05).
Vaccination information was available for 487 (78.4%) children and 391 (80.2%) of those with vaccination information had no RVV exposure. Out of 96 (15.4%) children who received any RVV (RVV-1, n = 96; RVV-2, n = 88; and RVV-3, n = 65), 12 children received RVV in  The location of intussusception at more than one site; such as Ileo-colic + Colo-colic, Ileo-colic + Ileo-colo-colic, Ileo-colic + Ileo-ileo-colic, Ileo-colic + Ileo-ileal f Others include cyst, Meckel's diverticulum, incisional hernia, intraluminal growth and malrotation g Spontaneously resolved: The patient was managed with IV fluids and nil orally, not required any additional intervention h None-No definite treatment could be provided, as the patient(s) was referred or died before any definite treatment i The denominators used for percentage estimation include patients undergone surgery only  Table 3). As shown in Fig. 2, only two cases occurred during 1-7 days after the third dose RVV and one case occurred on the vaccination day. At the four sites from three states where RVV (Rotavac™) was introduced, no increase in either intussusception case load (RR = 0.44; 95% CI 0.22, 1.18) or case rate per 1000 paediatric surgery admission (RR = 0.5; 95% CI 0.3, 1.2) during the post-RVV introduction period were observed (Table 3).

Discussion
We observed regional variances in the intussusception case load and case rate per 1000 paediatric surgery admission, high in the South region and low in the West region. Higher number of cases were observed during March to June months, which was comparable to other reports from India [14,15,17,24,25]. The male predominance (65.7%), median age at 8 months and high case load at 4-7 months of age (40.0%) were similar to other reports from India and globally (proportion of male: 63.0-77.2%; median age: 5-8 months and high case load age: 4-9 months) [14,15,17,18,21,24,26,27]. About 37.0% of cases occurred during 2nd to 6th month of age, the usual age for administration of RVV. The children came from all socioeconomic categories and in all the regions. The proportion of exclusively breastfed children (< 6 months) was lower than national average (52.1%, 2015-16) [22]. The children weaned with ragi food (only in South region) had intussusception earlier (median 1 month) than those weaned with wheat (median 3 months) and rice (median 4 months) based food. A shorter exclusive breastfeeding duration with mixed feeding and some weaning foods may have some role in intussusception occurrence. Exposure to foreign proteins and enteric infections are potential risk factors for intussusception, but we couldn't find any report on association with any specific diet. There were significant variations for several sociodemographic and dietary practice parameters across the regions ( Table 1). The classical triad was documented in 34.8% cases, which was higher than reports  from India (19.0%) and South Korea (7.6%), but lower than Tanzania (42.5%) [15,24,28]. The regional variation in the classical triad appeared to parallel with the interval between illness onset and hospitalization. Ultrasound was the commonest (95.3%) mode of diagnosis, as reported (72.0-100%) from India and other countries [15,18,[28][29][30]. Ileocolic was the most common location, similar to reports (68.0-79.0%) from India and globally [3,7,8,14]. PLPs observed in this study was higher than reported in studies (8.0-9.0%) from India and Tanzania with appendix and lymph nodes as the common [31,32]. The cases who presented early were more frequently managed nonsurgically with shorter stay (2-3 days) compared to those with surgical intervention (7 days). There were significant variations in the clinical features among the children across the regions, which may be due to the interval for presentation or clinical practices followed at the hospitals (Table 2). Death was observed in 1.7% cases, which was comparable to reports from India (1.0%) and other countries from Asia (0.25-6.0%), Latin America (1.0-5.0%), but lower than the African countries (2.0-25.0%) [17,18,24,28,29,[31][32][33]. Level 1 BCCD was met by 89.1% of the cases, similar to other reports from India [5,7,14]. Failed documentation of reduction prevented some cases meeting the BCCD level 1 certainty.
From the limited period of post-RVV (Rotavac™) introduction at four sites (from three states), no increase in the intussusception case load and case rate per 1000 paediatric surgery admission were observed. Intussusception within 1-21 days was observed mostly after the third RVV dose, which also overlaps with the age of natural occurrence. Under the NIP, RVV is given at 6, 10, 14 weeks of age and allowed before 12 months of age [34]. A delay in RVV administration may coincide with the high case load age of natural intussusception (4-7 months age), which may make the interpretation of association with the vaccine difficult. In routine practice, documentation of vaccine exposure should improve and appropriate guidance to parents be given regarding the subsequent RVV vaccination. Monitoring of intussusception risk is recommended for countries as part of the RVV introduction program [35,36]. In the absence of any specific population denominator for tertiary care hospitals, estimation of incidence in Indian context is difficult. The intussusception case rate per 1000 paediatric surgery admissions may be a proxy indicator for monitoring and comparison across the sites and regions with RVV introduction.
Early suspicion, case detection and referral to appropriate hospital are critical for minimizing the surgical interventions and favourable outcomes. Efforts are needed to equip and enable the public health facilities for nonsurgical management in children, though this depends on the timing of the presentation to hospital.
The epidemiology, clinical presentation and management of intussusception in children across the regions may serve as baseline for future studies. The regional representation and mix of private-and public-sector hospitals were the advantages for this surveillance.
This study had some limitations. In absence of definite catchment population and referral pattern, populationbased incidence or case rate estimation was not possible. The etiologies and risk factors for intussusception were not studied.

Conclusion
To conclude, the intussusception cases were seen all across the country and the majority of the cases occurred during the first year of life. The high case load age (4-7 months) for intussusception in children coincided with the age of RVV third dose. Some variations in case loads across the different regions and seasonality were observed. The potential role of diet exposure, weaning practices and food like ragi in intussusception needs further evaluation. No increased occurrence of intussusception cases was observed during the limited RVV post-introduction period. Early case detection, prompt referral and appropriate management are needed to avoid surgical intervention and complications. Immunization exposure must be documented to assess the vaccine associated risk. In absence of populationbased incidence or case rate, the intussusception case rate per 1000 paediatric surgery admission may be used for inter-regional comparison and trend monitoring.