Etiology of diarrhea by multiplex polymerase chain reaction among young children in the United Arab Emirates: a case-control study

The study was conducted at two major government-run referral hospitals (Al Ain and Tawam hospitals) in Al Ain, a large inland city in the Eastern Region of the Emirate of Abu Dhabi, UAE with a population estimates of 766,936 people [9]. The UAE is a federal union of seven distinct States-Emirates with the Emirate of Abu Dhabi being the largest and the political capital of the country. Health care is provided for all nationals as mandated by the constitution. National healthcare indicators are equivalent to those in high-income countries with a total expenditure on health as a percentage of gross domestic product being 4.0 in 2010 [10].

Between December 2017 and April 2019, after obtaining parental informed consent, stool specimens were collected from children less than 5 years of age who presented to the two hospitals with diarrhea, defined as at least three loose stools within the previous 24 h, defined as “cases”. During the same period, stool specimens were collected from children attending the same two hospitals without diarrheal disease in the last 30 days, defined as “controls”. Only one stool sample was collected from each participating child. The exclusion criteria for cases included current antibiotic treatment, chronic conditions (e.g., immune deficiency or cystic fibrosis), long-term immunosuppressive therapy and severe or life- threatening co-morbidity. The exclusion criteria for controls included acute infection (e.g., respiratory infection) and having a sibling recruited as a case or control. We also obtained data on demographics, clinical characteristics and exposures (e.g., history of sick contact with diarrhea) through a brief survey of parents at time of enrollment.

Following collection, fecal specimens were transported on a daily basis to the principal investigator’s research laboratory at the UAE University and stored at − 80 °C until further analysis. An automated nucleic acid extraction instrument (Microlab Nimbus IVD system; Hamilton, Reno, Nevada, USA) was used for the extraction of both RNA and DNA from the stool samples. A weighed aliquot of stool sample was suspended in 1-ml of ASL buffer (Qiagen, Valencia, CA, USA) and incubated for 10 min at room temperature. After clarification by high-speed centrifugation, the sample was loaded into the Nimbus equipment for nucleic acid extraction. After nucleic acid extraction, PCR reactions were set up and run on the CFX96 Real Time Detection System (Bio-Rad, Hercules, CA, USA). For that purpose, we used a multiplex one-step real-time PCR platform, the Allplex™ Gastrointestinal Full Panel Assay (Seegene, Seoul, South Korea) as per manufactures’ instructions [11, 12]. This assay has four panels that test for six viruses, 13 bacteria and six parasites. The viral panel detects adenovirus, astrovirus, norovirus GI, norovirus GII, rotavirus and sapovirus. The first bacterial panel detects Enteroaggregative E. coli, EAEC (aggR), Enteropathogenic E. coli, EPEC (eaeA), Escherichia coli O157 (E. coli O157), Enterotoxigenic E. coli, ETEC (lt/st), hypervirulent Clostridium difficile and Enterohemorrhagic E. coli, Shiga toxin-producing E. coli, STEC (stx1/2). The second bacterial panel detects Aeromonas spp., Campylobacter spp., Clostridium difficile toxin B, Salmonella spp., Shigella spp./ Enteroinvasive E. coli (EIEC), Vibrio spp. and Yersinia enterocolitica. The parasitic panel detects Blastocystis hominis, Cryptosporidium spp., Cyclospora cayetanensis, Dientamoeba fragilis, Entamoeba histolytica and Giardia lamblia. The PCR reaction was performed under the following cycling conditions: 20 min at 50 °C for one cycle; 15 min at 95 °C for one cycle; 10 s at 95 °C, 1 min at 60 °C and 30 s at 72 °C for 45 cycles; 10 s at 95 °C, 44 times. Seegene Viewer Software (Seegene, Seoul, South Korea) was used for detection and data analysis. Samples were classified as pathogen positive at a cycle threshold value of < 40 as per manufacturer’s instructions [12].

The descriptive results were expressed as number of participants and percentages. Proportions were compared with the Chi squared or Fisher exact test for small values. As the continuous variables did not follow a normal distribution (as confirmed by the Shapiro-Wilk test), they were expressed as median values and interquartile range (IQR, i.e. 25th and 75th percentile) and were compared with the non-parametric Kruskal Wallis test.

Adjusting for potential confounders in the above model, we also calculated, for each pathogen significantly associated with diarrhea in the logistic model, the adjusted attributable fraction (aAF), which is the proportional reduction in diarrhea that would occur if exposure to the respective pathogen was eliminated. aAF (with 95% ci) was calculated from the aOR, with the Woolf approximation for small samples, using the equation:

A total of 276 stool samples were analyzed involving 203 samples from cases and 73 samples from controls (Fig. 1). Cases (median age of 17 months, IQR: 8–23) were significantly older than controls (P = 0.01) who had a median age of 11 months (IQR: 2–26). History of sick contact was provided in 30.6% of cases in comparison to approximately 5% in controls (Table 1).

A positive PCR result for at least one pathogen was identified more often in 87 samples (42.8%) from cases than in 17 (23.3%) from controls (P < 0.001) (Table S1). Overall, the OR of isolating a pathogen, especially a virus (OR = 15.8), were always significantly higher in cases than in controls, in both the unmatched and age-matched multivariate analysis (Table 2). From all the six viruses tested for, only rotavirus, norovirus GII and adenovirus were found significantly more often in the cases than in controls, in both the unmatched and age-matched multivariate analysis (Table 3). However, none of the 13 bacteria tested for were more commonly found in the cases than in controls, in both the unmatched and age-matched multivariate analysis (Table 3). From all the six parasites tested for, only Cryptosporidium was found significantly more often in cases than in controls in the unmatched analysis, but that significance disappeared in the age-matched multivariate analysis (Table 3). Overall, the top three pathogens that were significantly more prevalent in a multivariable logistic regression model include rotavirus, norovirus GII and adenovirus (in a descending order). Their adjusted attributable fractions (aAFs) with 95% ci were 0.95 (0.64, 1.0) for rotavirus, 0.86 (0.38, 0.97) for norovirus GII and 0.84 (0.29, 0.96) for adenovirus (Table S2). Of note, these prevalent pathogens were detected throughout the year (Figure S1). Further, rotavirus and norovirus GII were detected across all studied age groups. However, adenovirus was more prevalent among children younger than 24 months when compared with older children (Table S3).

Co-infections occurred in 27.9% of the children, with parasites being present with either viruses or bacteria in 23 children (8.3%). Although amongst all participants, there were no interactions among the three groups of pathogens (all P values > 0.05) in the logistic regression analysis (Table 4), viruses and parasites were significantly more likely to occur together only in the cases (Table 4). Cases had a significant higher number of pathogens as compared to controls, with ≥3 pathogens detected in 7.5% of cases, compared with 2.7% of controls (Table S1). The heatmap (Figure S2) shows that EPEC and norovirus GII infection were the most frequent co-infections occurring in 14 children followed by co-infection with EPEC and EAEC in nine children. Of note, EPEC was identified in (17.7%) of cases and (8.2%) of controls (P = 0.07).