Burden and Etiology of Moderate and Severe Diarrhea in Chinese Children Less than 5 years of Age: Prospective, Population-based Surveillance

Background. Diarrhea remains the leading cause of childhood illness in China. Better understanding of burden and etiology of diarrheal diseases is important for development of effective prevention measures. Methods. Population-based diarrhea surveillance was conducted in Sanjiang (southern China) year-round and Zhengding (northern China) in autumn/winter. Stool specimens were collected from children <5 years of age experiencing diarrhea. The TaqMan Array Card (TAC), based on multiplex real-time PCR, was applied to detect multiple enteric microbial agents simultaneously. Results using these methods were compared to those derived from conventional PCR assays. moderate-to-severe with of The and sapovirus all the in by during the cool season. Compared to conventional PCR the incremental detection via

Population-based diarrhea surveillance in children under age 5 was conducted in 2 rural counties: Zhengding county of Hebei province (north) and Sanjiang county of Guangxi province (south) in China in 2012 and 2013. The occurrence of rotavirus, noroviruses, sapoviruses, astroviruses, and enteric adenoviruses were analyzed. However, only a low percent (~ 25%) of diarrhea cases contained one of abovementioned 4 viruses during the 1-year surveillance period using PCR methods (5). Even if the focus was on the epidemic season, no viral agents were detected in half of diarrhea episodes. Currently, the cause for most diarrhea episodes in Chinese children remains unknown. To address this knowledge gap, a multipathogen TaqMan Array Card (TAC) (6) was employed to investigate etiologies in children with moderate-to-severe diarrhea (MSD), utilizing specimens collected from the abovementioned populationbased surveillance effort.

Study population
As described elsewhere, in Sanjiang County, 3,613 children were monitored between January 1 and December 31, 2013. In Zhengding County, 6,443 children were followed from October 1, 2011 to March 31, 2012 (5). Children who presented to a participating health care facility (including village clinics, township hospitals and county hospitals) with acute diarrhea and whose parents or guardians provided informed consent were enrolled. Information on demographics, medical history, physical examination, and the management plan during the full course of illness, as well as a stool specimen at presentation were collected. After assessing severity of illness for each patient based on a modi ed Vesikari scoring system (7), fecal specimens from children who presented with MSD were selected for TAC testing. This study was reviewed and approved by the Institutional Review Board (IRB) of the Institutes of Biomedical Sciences, Fudan University. Written informed consent was obtained from the parent/guardian of each child. This research did not require review by the CDC institutional review board because the CDC tested preexisting, anonymous specimens.

Conventional PCR assay
All stool samples were tested for rotavirus, norovirus, sapovirus, astrovirus, and adenovirus using conventional PCR assays in the laboratory of Fudan university (5). To identify the genotype of viruses, all PCR products were puri ed and sequenced using Sanger dideoxy termination sequencing (Biosune Co., Ltd., Shanghai).

Multipathogen TaqMan Array Card assay
Nucleic acid extraction Since surveillance in Sanjiang was year-round, covering the peak(s) of seasonal diarrhea resulting from bacterial and viral infections, TNA of bacteria and viruses were extracted using MagMAX™ Total Nucleic Acid Isolation Kit (Thermo Fisher Scienti c, Waltham, MA, USA) on KingFisher Flex (Thermo Fisher Scienti c, Waltham, MA, USA). As an extrinsic control, bacteriophage MS2 (ATCC, Manassas, VA, USA) was spiked in lysis buffer to monitor the e ciency or inhibition of extraction and ampli cation of each sample. While surveillance in Zhengding only focused on viral diarrhea, TNA of viruses were extracted using the Tianlong virus RNA/DNA extraction kit on the NP968 3S system (Tianlong, Xi'an, China). Procedures were performed according to the manufacturer's instructions unless otherwise stated. All isolated TNA were stored at -80 °C prior to assays.

TAC design and procedure
The panel of TAC (Applied BiosystemsTM, USA) in US CDC used in this study included 6 viral, 13 bacterial, and 5 parasitic agents as previously described (Supplementary Table 1) (6). Any quanti cation cycle (Cq) result of < 45 was considered positive.
The Luminex xTAG gastrointestinal pathogen panel assay (GPP) (Luminex Molecular Diagnostics, Austin, TX, USA) as a qualitative multiplex test has been shown to be highly sensitive and speci c for a variety of pathogens in clinical settings globally (8)(9)(10). In this study, the xTAG GPP platform in US CDC was used to evaluate the results of rotavirus A, norovirus GI/GII, and adenovirus 40/41 from TAC assays.

Statistical analysis
We used SAS (SAS Institute Inc., Cary, NC, USA) for statistical analysis. For binary data, the Chi square test was used, or Fisher's exact test when data were sparse. For comparison of Ct values, the Mann-Whitney U test was used. For validation of the TAC assay, sensitivity, speci city, and kappa coe cient with 95% con dence intervals (95% CI) were calculated using GPP test results as references. The incidence rate of viral diarrhea was calculated as the number of diarrheal illness episodes with stool specimens testing positive for rotaviruses, noroviruses, sapoviruses, adenoviruses, and astroviruses among the surveillance participants per 1,000 child-years. The proportions of cases positive for each agent were calculated. Ct values were presented as the mean of two duplicate wells for specimens yielding positive results. If these two duplicate wells presented discordant results (one was positive via Ct value and one was negative), "indeterminate" comments were assigned to the wells; these wells were not included in subsequent calculations. All P values were two-tailed with < 0.05 considered statistically signi cant.

Results
Summary of the study population and incidence of MSD During the study period, 6,380 children in Zhengding and 3,581 children in Sanjiang under 5 years of age participated, and experienced 1,211 and 1,258 diarrhea episodes respectively. The overall incidence of diarrhea was 215.0 and 398.2 episodes per 1000 child-years in Zhengding and Sanjiang. Of these episodes, 340 (31.2%) and 279 (22.9%) episodes were identi ed as moderate-to-severe diarrhea with incidence of 60.4 and 88.3 per 1000 child-years in Zhengding and Sanjiang, respectively ( Table 1). As enterococcus is a common commensal organism in the human intestine, not considered a pathogen Table 1 a. Incidence (per 1000 child-years) of total MSD, MSD with any agent, and MSD with a speci c agent, with 95% con dence interval, by age stratum in Sanjiang.   (Table 1a).
The prevalence of microbial agents in Sanjiang was examined by season (Fig. 2). For bacterial infections, EAEC and ETEC were more frequently detected during summer (50.0%, and 26.3%), while Campylobacter and C. di cile appeared in spring (16.7%, and 8.3%). By contrast, all viral agents were detected throughout the year, except summer when adenovirus 40/41 peaked sharply and the rotavirus detection rate was low. For parasitic infections, only A. lumbricoides was detected frequently and year-round, though a winter peak was observed.
No signi cant differences among tested agents were observed in the age distribution of children infected (Fig. 3). Comparatively, the bacterial agents, represented by EAEC, ETEC, and Campylobacter tended to infect children aged 12-59 months. For bacterial agents with a detection rate > 10% (EAEC, ETEC, and Campylobacter), the highest incidence was found in children aged 12-23 months (49.9, 22.5, 22.5 per 1,000 child-years), and followed by incidence in children aged 0-11 months (37. 8

Comparison on etiological and epidemiological characteristics of viral infection at peak season between Sanjiang and Zhengding
Since we only collected fecal specimens from children with MSD in Zhengding from October through March when viral agents like rotavirus usually peaked each year, we compared virus-speci c incidence pro les between Zhengding and Sanjiang in the same time frame (Fig. 4). Of these viral agents, Zhengding had higher incidence of rotavirus (36.9 per 1,000 child-years), astrovirus (22.5 per 1,000 childyears), and adenovirus 40/41 (19.0 per 1,000 child-years) than Sanjiang. Conversely, Sanjiang had higher incidence of enterovirus (27.9 per 1,000 child-years) than Zhengding. Both places had high incidence of adenovirus (22.4-27.9 per 1,000 child-years) (Fig. 4a).
After age strati cation, increased burdens due to viral pathogens were measured in children aged 0-11 months and 12-23 months (Fig. 4b and c); incidence decreased signi cantly in children aged 24-59 months (Fig. 4d). It was particularly noteworthy that rotavirus, norovirus, astrovirus, and adenovirus 40/41 caused a contrasting incidence of MSD in children aged 0-11 months in Zhengding in comparison with that in Sanjiang; incidence was 77.9, 34.8, 48.9 and 37.3 per 1,000 child-years, respectively, accounting for 45.2%, 53.8%, 46.5% and 42.1% of all cases in children < 5 years of age.

Comparison between GPP assay and TAC assay
In order to validate the TAC results, diagnostic testing using GPP test results as reference was performed ( Table 2). Of 265 specimens tested, the overall sensitivity and speci city for all three viruses of TAC was 89.5% and 92.7% respectively, with individual virus assay sensitivities ranging from 77.0-95.7% and speci cities from 83.5-96.6%. The TAC assays for norovirus GI/GII exhibited the lowest sensitivity, ranging from 68.1-85.9%, whereas the assays for rotavirus A had the lowest speci city, ranging from 77.0-89.9%. Overall, though, the kappa coe cient for each virus was greater than 0.75, indicating substantial consistency between TAC and GPP results. Comparison between conventional PCR assay and TAC assay Compared to conventional PCR assay, TAC displayed more sensitivity-an average increment of 2-fold. However, the increase in sensitivity varied among different viruses. The highest increase was found with adenovirus detection, which was 18.6-fold, followed by astrovirus (9.4-fold) and adenovirus 40/41 (4.1fold). The lowest increase was found with rotavirus detection; the difference between the two assay methods was only 0.4-fold (Fig. 5a). Ct values derived from TAC assay were grouped by conventional PCR result for each viral agent. Overall, the Ct values of conventional PCR-negative samples were signi cantly higher than those of PCR-positive samples for all ve viruses (p < 0.001). Out of these ve viral agents, the average Ct values between PCR-negative and -positive samples in astrovirus, adenovirus 40/41, and sapovirus were notably more discrete (Fig. 5b).

Discussion
We previously reported that viral diarrhea, especially that caused by rotavirus and norovirus in Chinese children < 5 years, remained an unmet public health concern with a signi cant disease burden (5,12). Even so, our estimates were likely conservative, due to the limits of the laboratory methods applied, since viral agents were detected in 52.5% of samples (5). Recently, TAC platforms detecting multiple pathogens simultaneously were developed for surveillance of respiratory and enteric infectious disease with heightened assay e ciency and sensitivity (6,(13)(14)(15)(16)(17).
When applying the TAC assay to reanalyze enteric pathogens in samples derived from our previous population-based surveillance, a signi cant increase of detection rates was observed compared to the conventional PCR assays. In particular, a ~ 30% increase was observed in detection of adenovirus (40/41 excluded) and astrovirus. Further analysis indicated that the increases resulted from detection in specimens containing lower quantities of microbial nucleic acids. Based on the TAC assay, revised incidence of rotavirus infection in children with MSD was 36.9 and 21.8 cases per 1,000 child-years in Zhengding and Sanjiang, respectively, followed by norovirus (13.8 vs 14.6 cases per 1,000 child-years in Zhengding and Sanjiang, respectively). These estimates were 0.4-1.5 fold higher than those derived from conventional PCR assays. This will undoubtedly further enhance demand for rotavirus and norovirus vaccines for disease control (5). At the same time, a diagnostic test using GPP as reference further validated the accuracy of TAC results-valuable evidence for future applications.
Surprisingly, after adjusting the sensitivity of laboratory assay, whether in Sanjiang (South China) or Zhengding (North China), the detection rate of adenovirus infections was consistently elevated, and incidence even exceeded that of norovirus and rotavirus infections in Sanjiang. In addition, high rates of enterovirus were observed in Sanjiang. Classically, adenovirus 40/41 and type 52 are recognized as the causative agents of gastroenteritis, while other types contributed to an array of clinical disease including conjunctivitis, hepatitis, myocarditis, and pneumonia (18). Enterovirus is generally more likely to lead to respiratory symptoms and neurologic diseases (19). Thus, the clinical and public health signi cance of such infections in Chinese children need to be further studied. More frequent detection of enterovirus in Sanjiang than in Zhengding might be attributed to higher tropical temperatures. Enterovirus appears to occur year-round with higher prevalence in late spring throughout early autumn (20). Besides, the high rate of enterovirus detection might be due to the administration of live oral poliovirus vaccine (OPV) that caused increased enterovirus detected in stool samples (21).
Though it is the most populous country in the world, China is still a developing country, with unbalanced economic conditions across the country. Generally speaking, Sanjiang and Zhengding are less developed with relatively lower per capita gross domestic product (22). China has bene tted from a booming economy over the past two decades, and the spectrum of pathogens causing diarrhea has changed signi cantly. Viral infections typi ed by rotaviruses, noroviruses, and sapoviruses have become major pathogens, rather than bacterial infections (5,12). Although more bacterial targets were included in the TAC assay, in Sanjiang, the top ve agents detected in children with MSD were adenovirus, enterovirus, EAEC, rotavirus and sapovirus-mostly viruses. If all agents with a detection rate of > 10% were considered, norovirus, ETEC, astrovirus, and Campylobacter would also be included. This spectrum differed remarkably from that produced by the Global Enteric Multicenter Study (GEMS), which found rotavirus, Cryptosporidium, ETEC, and Shigella to be most common (23). The most likely explanations for the discrepancy are the study design and differences in health-care access, economic development, and environmental conditions between our study sites and GEMS' study sites.
Comparatively, bacterial pathogens may now play a less important role in children's intestinal infection; more resources should be focused on viral infection in China. However, surveillance for bacterial pathogens must continue, since these bacteria are all foodborne pathogens and can rapidly cause largescale outbreaks in certain circumstances (24). A recent published meta-analysis indicated that abovementioned bacteria are distributed widely in food commodities in China (25).
ETEC is primarily associated with diarrhea in children < 5 years of age, as well as being a frequent cause of diarrhea in tourists visiting developing countries. EAEC was initially considered to be an opportunistic pathogen associated with diarrhea in immunocompromised patients or malnourished children (26). Currently, EAEC is recognized as a cause of chronic diarrheal disease in children and adults in both developing and industrialized countries (27). However, compared to ETEC, EAEC did not exhibit a prolonged course of illness in our study. Neither E. coli nor Campylobacter cause frequently observed infectious disease in China. The prevalence of these two bacteria in Chinese population is scarce, particularly among children. To date, just a few studies have been conducted in children < 5 years of age in Wuhan and Shanghai (28,29). However, the detection rates of EAEC, ETEC, and Campylobacter obtained from each study were inconsistent. The detection rates in Wuhan, a city in central China, were similar to those obtained in our study and were markedly higher than those in Shanghai, though conventional PCR methods, other than TAC assay were applied in Wuhan and Shanghai studies. These differences may be attributable to differences in sanitation and economic development between eastern and central regions. Notably, we observed, in a previous study, a statistical difference in prevalence of EAEC and ETEC between children with diarrhea and healthy controls, but no signi cant difference was observed in prevalence of Campylobacter (28). This suggests that EAEC/ETEC may have greater public health signi cance than Campylobacter among Chinese children with diarrhea, and further research should focus on these pathogens. Shigellosis no longer appears to be a signi cant public health concern in China according to this study. The total number of cases infected with shigella in 2019 was 80,483, an incidence of 6.5 cases per 100,000 person-years (30).
In addition to study limitations noted elsewhere (5), we could not calculate odds ratios to con rm speci c pathogens that caused diarrhea due to the absence of a healthy control group. In particular, subtypes of some microbial agents (such as E. coli, adenovirus, and enterovirus) may not be associated with diarrhea. Nevertheless, for the rst time, our study provides a population-based, possible spectrum of pathogens for diarrhea among young children in China. Further veri cation of public health signi cance is planned to be performed through a birth cohort study being conducted in Zhengding, where a stool specimen is required from each participating child once a week regardless of their health status.

Conclusion
Despite signi cant advances in ghting childhood diarrhea through sanitation improvement, safe water supplies, and vaccine development, diarrheal diseases remain one of the leading causes of childhood illness in China. Clearly, comprehensive understanding of prevalent etiologies is essential for the design and implementation of effective prevention strategies. Findings from our study support continued surveillance and attention for pathogens such as rotavirus and calicivirus that have gained increased awareness among the public and policymakers. More importantly, for neglected pathogens, such as EAEC, ETEC, Campylobacter, and non-enteric adenovirus, strengthened research should be performed to clarify epidemiological and clinical characteristics, and eventually con rm their signi cance for public health.

Declarations
Ethics approval and consent to participate