Rotavirus is a major cause of severe gastroenteritis in infants and young children worldwide1 and causes approximately 215,000 preventable deaths annually.2 More than 85% of these deaths occur in low-income countries, particularly in Africa and Asia.3–6 To reduce the burden of rotavirus disease, the World Health Organization recommends inclusion of rotavirus vaccines into all national immunization programs.7 Two live, attenuated, orally administered rotavirus vaccines, a pentavalent bovine–human reassortant vaccine (RotaTeq, Merck Vaccines, Whitehouse Station, NJ) and a monovalent vaccine (RV1; Rotarix, GlaxoSmithKline Biologicals, Rixensart, Belgium) based on a human rotavirus strain, have been licensed and implemented in 81 countries worldwide, including 38 low-income countries.8 The efficacy of both vaccines is much higher in middle- and high-income economies (80%–95%)9 than in developing countries (50%–64%).10 Recently, RV1 was found to have effectiveness between 59% and 64% against severe rotavirus diarrhea in South Africa, Malawi and Botswana.11–13
A key question that remains is why rotavirus vaccines appear to provide significantly reduced protection for children in low-income countries.14 While the exact reasons for this observation are unclear, a range of hypotheses have been proposed. These have included possible poor vaccine immune response because of host factors such as “environmental enteropathy” or undernutrition, reduced immunogenicity of the vaccines because of wider diversity of rotavirus strains seen in many of these regions and high levels of breast milk antibodies resulting in interference with rotavirus oral vaccines.15 This phenomenon of reduced effectiveness in lower resource settings has been described for other existing live oral vaccines against enteric infections including typhoid, cholera and polio.15 , 16
To date, there has been very little study of the potential impact that enteric coinfections may have on rotavirus vaccine effectiveness (VE) observed in test-negative case–control studies. In studies using the same commercial multitarget assay (LuminexxTAG Gastrointestinal Pathogen Panel [GPP]), children from high-income regions have generally shown much lower rates of coinfection than children from low-income regions; one French study found a 7% coinfection rate,17 compared with a 77% coinfection rate observed in children living in Ghana.18 Studies using broad molecular-based testing have confirmed that particularly across Africa and Asia, copathogen detection is fairly common.19–22 In this context, it should be highlighted that pathogen detection does not necessarily imply causation; investigators have used quantitative detection and serial sampling in an effort to better identify enteropathogens more likely to “cause” a clinical episode of diarrhea.23 Although rotavirus detection is more strongly associated with diarrhea than most pathogens, a recent large multisite case–control analysis using quantitative polymerase chain reaction (PCR) thresholds estimated that in 20.8% of all rotavirus-positive samples, a diarrheal pathogen other than rotavirus was detected at a quantity that was more strongly associated with diarrhea.22 These other coinfecting pathogens might partially explain the lower observed VE found in studies in these high-burden settings, as the inclusion of participants vaccinated against rotavirus, presenting with acute diarrhea caused by another enteropathogen, who happened to be shedding rotavirus in the stool (eg, because of an asymptomatic rotavirus infection) would bias estimates of VE toward zero.
In July 2012, Botswana implemented the 2-dose RV vaccine (Rotarix) as part of the nationally funded immunization program. Analysis of our postvaccine introduction case–control study found that effectiveness of a full series (2 doses) was 54% (95% confidence interval [CI]: 23–73),13 which is similar to that seen in prior vaccine efficacy trials in Malawi12 and South Africa.11 In this study, we aimed to determine whether coinfections have a role to play in explaining the reduced observed VE found in Botswana; we have previously shown that children admitted to Botswana hospitals with gastroenteritis commonly have multiple enteropathogens detected.24 We calculated rotavirus VE observed in children with rotavirus monoinfection (cases in which rotavirus was the only pathogen detected) and compared with VE observed in children with mixed infections of rotavirus and other key enteric pathogens.
Study Design and Participants
From June 2013 to April 2015, stool and swab samples were prospectively collected from children presenting with severe gastroenteritis at 4 different clinical sites across Botswana: 2 referral hospitals in the south and north of the country (Princess Marina Hospital, Gaborone and Nyangabgwe Referral Hospital, Francistown); a primary care facility (Bobonong Primary Hospital, Bobonong) and a district-level facility (Letsholathebe II Memorial Hospital, Maun). These sites were chosen to include a range of geographic and socioeconomic locales, serving populations with varying health indicators in Botswana.
Case patients and controls were enrolled at participating hospitals during diarrhea surveillance if they had either ≥3 episodes of diarrhea (looser than normal stools) or ≥2 episodes of emesis and any episodes of diarrhea within a 24-hour period, lasting <7 days.13 A case patient was defined as a child ≥4 months of age hospitalized for laboratory-confirmed rotavirus diarrhea by enzyme immunoassay (EIA) who was age eligible to receive rotavirus vaccination (born after May 1, 2012). A control was defined as a child ≥4 months of age hospitalized for rotavirus-negative diarrhea by EIA (ie, who tested negative for rotavirus) and who was age eligible to receive rotavirus vaccination.13 Rotavirus vaccination status was confirmed by review of the child’s medical record carried by the parent. A photocopy or photograph of the vaccination portion of the card was obtained to corroborate with transcribed vaccination data.
A bulk stool specimen (≈5 mL) or flocked rectal swab specimen was obtained from each suspected rotavirus case on the day of presentation to the hospital or within 48 hours of hospital admission to avoid the detection of nosocomial infection. All samples collected were transported frozen to the Botswana National Health Laboratory where they were stored at −80°C until analyzed.
All samples were tested with a commercial rotavirus EIA (Premier Rotaclone, Meridian Bioscience, Cincinnati, OH) according to manufacturer’s instructions. All rotavirus EIA positive samples were further tested on-site with a previously validated in-house multiplex PCR panel, which comprise in total 3 assays that detect 9 enteropathogen targets (enterotoxigenic Escherichia coli, Campylobacter, Shigella, Salmonella, Giardia, Cryptosporidium, rotavirus, norovirus and adenovirus).25 The laboratory technologists were blinded as to vaccine status. Amplification and detection were performed as previously described.24
An aliquot of stool or rectal swab was also shipped on dry ice to the Center for Disease Control and Prevention in Atlanta, Georgia, for extraction using the easyMAG platform and PCR testing with the 15 pathogen target LuminexxTAG Gastrointestinal Pathogen Panel (Luminex Molecular Diagnostics, Toronto, ON) assay on the MAGPIX system according to the manufacturer’s instructions. This assay has previously been evaluated and can simultaneously detect the following: Giardia, Cryptosporidium, Entamoeba histolytica, Yersinia enterocolitica, Salmonella, E. coli heat stable (ST) enterotoxin, E. coli heat-labile (LT) enterotoxin, Shigella, Clostridium difficile toxin A, C. difficile toxin B, Campylobacter, Vibrio cholerae, E. coli O157, Shiga toxin 1, Shiga toxin 2, norovirus GI, norovirus GII, rotavirus A and adenovirus 40/41.26 , 27 Not all stool specimens were tested with GPP assay because of insufficient sample.
VE was computed using unconditional logistic regression models adjusting for age, birth month/year and hospital as previously described.13
The VE was calculated using only data from children with card-confirmed vaccination status, ensuring use of accurately ascertained vaccination information. Analyses assessed effectiveness of 2 doses (full series), 1 dose (partial series) and 1 or more doses (intention to vaccinate) of RV1 versus no vaccination. Rotavirus VE was stratified by nonmixed (rotavirus only cases) and mixed infections defined as detection of rotavirus plus detection of an additional diarrhea-associated pathogen(s).
For each testing method (in-house PCR and GPP), the primary analysis restricted the definition of coinfection to detection of 5 other pathogens found to be most common in diarrhea cases as compared with well controls (ie, had the highest attributable fractions [AF] for diarrhea in children under the age of 2 in the Global Enteric Multisite Study quantitative PCR reanalysis—Shigella spp, Cryptosporidium spp, adenovirus 40/41, Campylobacter spp and enterotoxigenic E. coli).22 A secondary analysis defined coinfection as the detection of any non-rotavirus pathogen target. A subanalysis of this secondary analysis was performed, including coinfections, only if the pathogen nucleic acid was present in higher quantity (ie, the PCR cycle threshold value [Ct] < 30). Logistic regression models were used to calculate the odds ratios and 95% CI of rotavirus vaccination, adjusting for age, birth month/year and hospital. Adjusted VE (aVE) was calculated as (1 – adjusted odds ratio) × 100%. Statistical significance was designated as P < 0.05. Analyses were performed using SAS statistical software (version 9.3).
Informed consent was obtained from parents or guardians of the children. Ethical approval was obtained from the Botswana Ministry of Health, from ethics committees of individual study sites, from the University of Pennsylvania and from McMaster University.
A total of 667 children ≥4 months of age met the inclusion criteria and were of eligible age for vaccination. Of these, 57 were excluded either because there was no verifiable documentation of rotavirus vaccination or because there was insufficient stool sample for testing or both, and the remaining 610 children were studied. Using EIA testing, 242 children (40%) tested positive for rotavirus and 368 children (60%) tested negative. Testing for coinfections was performed on 182 rotavirus EIA-positive subjects using the GPP commercial assay; rotavirus was the only pathogen detected in 122/182 (67%) subjects, whereas 60/182 cases (33%) were coinfected with 1 or more enteropathogens with a high attributable fraction. Testing for coinfections was performed on 213 rotavirus EIA-positive subjects using in-house multiplex PCR panel. Rotavirus was the only pathogen detected in 115/213 (54%) subjects, whereas 98/213 (46%) cases were coinfected with rotavirus plus at least 1 of 5 significant enteropathogens.
No differences in demographic or socioeconomic variables (Table, Supplemental Digital Content 1, http://links.lww.com/INF/C873) were noted between all rotavirus cases (rotavirus-positive cases and rotavirus-negative diarrhea controls) and nonmixed (pure rotavirus) and mixed infections (coinfections) groups, except for the proportion of moderate to severely malnourished children in the rotavirus-only cases versus rotavirus coinfected cases (11/76 versus 17/58, respectively; P = 0·04) and the proportion of children with electricity at home in the rotavirus-only cases versus rotavirus coinfected cases (69/93 versus 41/74, respectively; P = 0·01).
Coinfecting Pathogens per Vaccination Status
Of 368 rotavirus EIA test–negative controls, 29 (8%) were not vaccinated, 51 (14%) and 288 (78%) had received 1 or 2 doses of the monovalent rotavirus vaccine, respectively (Table 1). Of the 242 rotavirus EIA-positive cases, 43 (18%) were not vaccinated and 37 (15%) and 162 (67%) had received 1 or 2 doses of the monovalent rotavirus vaccine, respectively. Using EIA assay, the overall aVE was 48% (95% CI: 1–72), 54% (95% CI: 23–73) and 53% (95% CI: 21–72) for 1 dose, 2 doses and at least 1 dose, respectively (Table 1). In the subset tested with the GPP assay, the overall aVE was 46% (95% CI: −9 to 73), 59% (95% CI: 27–77) and 56% (95% CI: 24–75) for 1 dose, 2 doses and at least 1 dose, respectively. When using the in-house multiplex PCR assay, the overall aVE was 48% (95% CI: 0–73), 56% (95% CI: 25–75) and 55% (95% CI: 23–73) for 1 dose, 2 doses and at least 1 dose, respectively.
In the rotavirus monoinfection subgroup using the GPP assay, aVE was 39% (95% CI: −29 to 71), 63% (95% CI: 30–80) and 58% (95% CI: 23–77) for 1 dose, 2 doses and at least 1 dose, respectively. In the coinfected subgroup which we defined as detection of rotavirus and any of the 5 significant non-rotavirus pathogens, aVE was 63% (95% CI: −15 to 88), 51% (95% CI: −14 to 79) and 53% (95% CI: −8 to 79) for 1 dose, 2 doses and at least 1 dose, respectively.
In the rotavirus monoinfection subgroup using the in-house multiplex PCR, aVE was 49% (95% CI: −13 to 77), 53% (95% CI: 11–74) and 53% (95% CI: 11–75) for 1 dose, 2 doses and at least 1 dose, respectively. In the coinfected subgroup, aVE was 50% (95% CI: −13 to 78), 62% (95% CI: 25–80) and 59% (95% CI: 22–79) for 1 dose, 2 doses and at least 1 dose, respectively. We expanded the analysis of coinfection to include detection of any other enteropathogen by both assays, and the calculated VE rates were comparable (Table 2).
A subanalysis was performed restricting coinfection to detection of pathogen at higher concentrations via the in-house multiplex PCR (Ct value < 30; Table 3). This allowed for detection of pathogens, which were the most likely cause of acute diarrhea, and not just “innocent bystanders.” Ct values for rotavirus were similar between those with copathogens detected and those with only rotavirus detected (20 versus 18, respectively). There again was no significant difference between aVE for rotavirus monoinfections and aVE for rotavirus mixed infections in this analysis.
The observed effectiveness of rotavirus vaccine was similar in children in whom rotavirus was the only pathogen detected as compared with children with coinfections involving rotavirus and other enteropathogens, suggesting that enteric coinfections do not appear to play a significant role in explaining the reduced VE that is observed in Botswana. We also calculated P values using χ2 test. The values were not statistically significant (P > 0.05—as shown in all the tables). This shows that the aVE for rotavirus monoinfection and aVE for rotavirus plus detection of any other pathogen is not significantly different for all the doses of the rotavirus vaccine given. This is further evidence that rotavirus coinfections do not appear to have any effect on the observed rotavirus VE. We are aware of one other study that carried out a similar analysis; the investigators did not find any impact of coinfection on aVE but restricted their laboratory testing to a limited number of viruses (adenovirus, norovirus genogroup I and II, astrovirus and sapovirus).11 Our findings would also suggest that rotavirus is likely to be the “cause” of diarrhea for most children hospitalized with gastroenteritis who test positive by EIA, regardless of whether other enteropathogens are detected using broad-range testing.
A potential limitation of our study is the use of qualitative PCR assays to identify coinfections. These extremely sensitive detection methods may lead to higher detection rates of possible “innocent bystanders” that are not actually causative agents with respect to the acute diarrheal illness presentation. Stated another way, it is possible that—had we possessed a test modality to identify the true microbiologic cause of a given episode of diarrhea—we would have been able to demonstrate that reduced VE of rotavirus vaccine in Botswana is in fact caused by false attribution of illness to rotavirus detected in enteric specimens. There is, however, now convincing data showing that conventional microbiology techniques (eg, bacterial culture and parasite microscopy) will miss in some cases the majority of attributable infections.22 We also specifically performed subanalyses using more strict definitions of coinfection (ie, limited to pathogens with high attributable fractions and also to detections with higher amounts of nucleic acid detected [Ct < 30], with no significant impact on our point estimates of VE). We should also note that both our molecular assays did not distinguish ST-positive from LT-positive Enterotoxigenic Escherichia coli (ETEC), despite the fact that ST-positive ETEC has been more highly associated with diarrheal disease.22 However, given the relatively small number of ETEC detections (2% by GPP, 11% for multiplex PCR and 5% for multiplex PCR with Ct < 30), we feel that this is unlikely to affect our main conclusions. The coinfection rates differed between the 2 assays 54% versus 33% for in-house assays and GPP assays, respectively. In a previous study, we observed that the in-house multiplex PCR assay detected slightly more pathogen targets as compared with the GPP assay, especially bacterial pathogens.25
Our findings suggest that coinfections alone are not likely to explain the observed lower VE reported in low-income countries. We believe this observation is not caused by the testing method because with 2 different assays, we found similar results. There are other factors, such as malnutrition, which have been found to contribute to lower VE with other orally administered vaccines and are very common in resource-limited settings.13 , 28 We were unable to study the potential effect of these factors, in conjunction with coinfections, because of our small sample size. However, despite the lack of explanation for the lower VE, there is still compelling evidence of the public health value of rotavirus vaccination in an African setting and of its potential to curb severe rotavirus infection where the burden of fatal disease is highest.13 , 29
The greatest limitation of this study is that it is underpowered to make a definite conclusion that enteric coinfections do have an effect on the effectiveness of rotavirus vaccine. Therefore, more research is still needed to understand factors that contribute to lower rotavirus VE in high burden settings to hopefully help realize the full potential of this intervention.
1. Matthijnssens J, Zeller M, Heylen E, et al; RotaBel study group. Higher proportion of G2P rotaviruses in vaccinated hospitalized cases compared with unvaccinated hospitalized cases, despite high vaccine effectiveness
against heterotypic G2P rotaviruses. Clin Microbiol Infect. 2014;20:O702–O710.
2. Tate J, Burton A, Boschi-Pinto C, et al. Global, regional, and national estimates of rotavirus mortality in children <5 years of age, 2000–2013. Clin Infect Dis. 2016;62(suppl 2):S96–S105.2.
3. Lin CL, Chen SC, Liu SY, et al. Disease caused by rotavirus infection. Open Virol J. 2014;8:14–19.
4. Bayard V, DeAntonio R, Contreras R, et al. Impact of rotavirus vaccination on childhood gastroenteritis
-related mortality and hospital discharges in Panama. Int J Infect Dis. 2012;16:e94–e98.
5. Ustrup M, Madsen LB, Bygbjerg IC, et al. Outstanding challenges for rotavirus vaccine introduction in low-income countries—a systematic review. Dan Med Bull. 2011;58:A4323.
6. Parashar UD, Hummelman EG, Bresee JS, et al. Global illness and deaths caused by rotavirus disease in children. Emerg Infect Dis. 2003;9:565–572.
7. World Health Organization. Manual of Rotavirus Detection and Characterization Methods. 2009.Geneva: World Health Organization.
9. Clarke E, Desselberger U. Correlates of protection against human rotavirus disease and the factors influencing protection in low-income settings. Mucosal Immunol. 2015;8:1–17.
10. Parashar UD, Johnson H, Steele AD, et al. Health impact of rotavirus vaccination in developing countries: progress and way forward. Clin Infect Dis. 2016;62(suppl 2):S91–S95.
11. Groome MJ, Page N, Cortese MM, et al. Effectiveness of monovalent human rotavirus vaccine against admission to hospital for acute rotavirus diarrhoea in South African children: a case-control study. Lancet Infect Dis. 2014;14:1096–1104.
12. Bar-Zeev N, Kapanda L, Tate JE, et al; VacSurv Consortium. Effectiveness of a monovalent rotavirus vaccine in infants in Malawi after programmatic roll-out: an observational and case-control study. Lancet Infect Dis. 2015;15:422–428.
13. Gastañaduy PA, Steenhoff AP, Mokomane M, et al. Effectiveness of monovalent rotavirus vaccine after programmatic implementation in Botswana: a multisite prospective case-control study. Clin Infect Dis. 2016;62(suppl 2):S161–S167.
14. Afrad M, Hassan Z, Farjana S, et al. Changing profile of rotavirus genotypes in Bangladesh, 2006–2012. BMC Infect Dis. 2013;13(1):320.
15. Lopman BA, Pitzer VE, Sarkar R, et al. Understanding reduced rotavirus vaccine efficacy in low socio-economic settings. PLoS One. 2012;7:e41720.
16. Jiang V, Jiang B, Tate J, et al. Performance of rotavirus vaccines in developed and developing countries. Hum Vaccin. 2010;6:532–542.
17. Mengelle C, Mansuy JM, Prere MF, et al. Simultaneous detection of gastrointestinal pathogens with a multiplex Luminex-based molecular assay in stool samples from diarrhoeic patients. Clin Microbiol Infect. 2013;19:E458–E465.
18. Eibach D, Krumkamp R, Hahn A, et al. Application of a multiplex PCR assay for the detection of gastrointestinal pathogens in a rural African setting. BMC Infect Dis. 2016;16:150.
19. Kotloff KL, Nataro JP, Blackwelder WC, et al. Burden and aetiology of diarrhoeal disease in infants and young children in developing countries (the Global Enteric Multicenter Study, GEMS): a prospective, case-control study. Lancet. 2013;382:209–222.
20. Platts-Mills J, Babji S, Bodhidatta L, et al. Pathogen-specific burdens of community diarrhoea in developing countries: a multisite birth cohort study (MAL-ED). Lancet Glob Health. 2015;3(9):e564–e575.
21. Bhavnani D, Goldstick JE, Cevallos W, et al. Synergistic effects between rotavirus and coinfecting pathogens on diarrheal disease: evidence from a community-based study in northwestern Ecuador. Am J Epidemiol. 2012;176:387–395.
22. Liu J, Platts-Mills JA, Juma J, et al. Use of quantitative molecular diagnostic methods to identify causes of diarrhoea in children: a reanalysis of the GEMS case-control study. Lancet. 2016;388:1291–1301.
23. Taniuchi M, Sobuz SU, Begum S, et al. Etiology of diarrhea in Bangladeshi infants in the first year of life analyzed using molecular methods. J Infect Dis. 2013;208:1794–1802.
24. Pernica JM, Steenhoff AP, Welch H, et al. Correlation of clinical outcomes with multiplex molecular testing of stool from children admitted to hospital with gastroenteritis
in Botswana. J Pediatric Infect Dis Soc. 2016;5:312–318.
25. Goldfarb DM, Steenhoff AP, Pernica JM, et al. Evaluation of anatomically designed flocked rectal swabs for molecular detection of enteric pathogens in children admitted to hospital with severe gastroenteritis
in Botswana. J Clin Microbiol. 2014;52:3922–3927.
26. Navidad JF, Griswold DJ, Gradus MS, et al. Evaluation of Luminex xTAG gastrointestinal pathogen analyte-specific reagents for high-throughput, simultaneous detection of bacteria, viruses, and parasites of clinical and public health importance. J Clin Microbiol. 2013;51:3018–3024.
27. Claas EC, Burnham CA, Mazzulli T, et al. Performance of the xTAG® gastrointestinal pathogen panel, a multiplex molecular assay for simultaneous detection of bacterial, viral, and parasitic causes of infectious gastroenteritis
. J Microbiol Biotechnol. 2013;23:1041–1045.
28. Linhares AC, Carmo KB, Oliveira KK, et al. Nutritional status in relation to the efficacy of the rhesus-human reassortant, tetravalent rotavirus vaccine (RRV-TV) in infants from Belém, pará state, Brazil. Rev Inst Med Trop Sao Paulo. 2002;44:13–16.
29. Enane LA, Gastañaduy PA, Goldfarb DM, et al. Impact of Rotavirus Vaccination on Hospitalizations and Deaths From Childhood Gastroenteritis
in Botswana. Clin Infect Dis. 2016;62(suppl 2):S168–S174.