Prevalence, Clinical Features, and Genotypes of Norovirus-Associated Diarrhea in Wuxi, China, 2013–2020

ABSTRACT.

Norovirus (NoV) is a common pathogen that can cause infectious diarrhea. This study aimed to determine the prevalence, clinical features, and genotypes of NoV-associated diarrhea in Wuxi, China. A total of 4,416 stool samples were collected from patients with diarrhea at enteric disease clinics of sentinel hospitals in Wuxi from February 1, 2013 to December 31, 2020. Univariate and Akaike information criterion stepwise logistic regression were used to identify differences as integrated within a clinical setting (NoV positive [+] versus NoV negative [–], NoV+ versus rotavirus [RV]+, NoV+ versus bacteria+, genogroup [G] I and GII genotypes). Norovirus was detected in 9.85% of stool samples, which was greater than other tested pathogens. Excluding coinfection of NoV and other viruses or bacteria, patients infected with NoV had a lower chance of acquiring the virus in summer (P < 0.001; odds ratio [OR], 0.257; 95% CI, 0.189–0.36) when compared with patients without NoV. Patients with diarrhea infected with NoV featured nausea and vomiting (P < 0.001; OR, 2.297, 95% CI, 1.85–2.86) and loose stools (P = 0.006; OR, 2.247; 95% CI, 1.30–4.10), but less abdominal cramping (P = 0.001; OR, 0.676; 95% CI, 0.54–0.84). Patients infected with RV (P < 0.001; OR, 0.413; 95% CI, 0.25–0.68) or bacteria (P < 0.001; OR, 0.422; 95% CI, 0.26–0.67) were more vulnerable to fever than those infected with NoV. A total of 379 GII strains were detected concomitant with 48 GI strains, and there was a seasonal difference between the GI and GII genotypes. Strengthening pathogen detection for infectious diarrhea was helpful for understanding the epidemiological characteristics of infections with NoV and, potentially, for preventing disease outbreaks.

INTRODUCTION

Despite advances in disease prevention initiatives worldwide, infectious diarrhea remains one that has a profound negative impact on public health. It has been reported that diarrhea remains a leading cause of death and disease in children younger than 5 years.^1,2 To our current knowledge, human norovirus (NoV) is the main viral cause of infectious diarrhea. Although people of all ages are at risk for NoV infection, children and the elderly are more likely to develop severe disease.³ A recent updated systematic review and meta-analysis⁴ has found that the pooled global prevalence of NoV among acute gastroenteritis was 16% for the past 20 years. Norovirus infection is characterized by diarrhea, vomiting, nausea, and abdominal cramping.⁵ However, clinical symptom duration, expression, and severity can vary according to age, immune system status, and previous exposure history.^6,7 Norovirus is a positive-sense, single-strand RNA virus that belongs to the genus Norovirus in the Caliciviridae family. Moreover, it has been classified genetically into 10 genogroups (GI–GX), of which GI, GII, GIV, GVIII, and GIX could infect humans.⁸ In addition to NoV, rotavirus (RV) is also considered to be a frequent etiology of gastroenteritis.⁹ It is the leading cause of severe gastroenteritis in children younger than 5 years of age.¹⁰ In China, the incidence of infection and rate of hospitalization resulting from RV in children are 28% to 65% and 30% to 50%, respectively.¹¹ In addition, viral and bacterial pathogens have been identified in 30% to 70% and 6% to 20%, respectively, in patients with infectious diarrhea.¹² A better understanding of the epidemiology, etiology, and seasonality of infectious diarrhea would be valuable for planning and adopting targeted monitoring and preventive measures. If we do not strengthen the monitoring, prevention, and control of the pathogeny of infectious diarrhea, humans will face serious health concerns, large economic losses will ensue, and the medical burden will increase along with public health problems.¹³ Infectious diarrheal viruses, especially NoV, are the leading causes of gastroenteritis. It is important for us to identify the prevalence, clinical features, and genotypes of NoV-associated diarrhea in Wuxi. Therefore, we conducted a study to screen patients with diarrhea for NoV infection and ascertain the prevalence, clinical features, and genotypes of NoV in Wuxi to understand more fully the genetic diversity of NoV and the genetic characteristics of major epidemics in the region, as well as to provide preliminary information for the development of an infectious diarrhea surveillance network.

MATERIALS AND METHODS

Monitoring scheme.

We monitored patients with diarrhea at enteric diseases clinics at sentinel hospitals in Wuxi from February 1, 2013 to December 31, 2020. Fecal specimens were collected from anyone ≥ 1 year who had defecated three or more times a day, from babies who were breastfed within 6 months and produced stools six or more times per day, and from infants fed artificially within 12 months who defecated four or more times a day, accompanied by changes in stool characteristics, such as loose stool and watery stool.

Epidemiological and etiological surveillance of patients with diarrhea was carried out in representative medical institutions at the city/county, district, township, and community levels. There were 59 epidemiological monitoring sites and 16 etiological monitoring sites, which include two municipal hospitals, two county hospitals, seven community hospitals, and five township health centers, covering the urban population (including community hospitals) and rural population (including township health centers) in six districts. The monitoring department includes the digestive clinic, pediatrics clinic, infection intestinal clinic, and emergency department.

The sentinel hospitals were responsible for case finding, sampling, and information collection. The district-level CDCs were in charge of testing bacterial samples and sending virus specimens to the municipal CDC for virological testing. The three levels of the surveillance system share information through a dedicated online system.

Sample collection, detection of pathogens, and NoV genotype.

Stool samples from patients with diarrhea were collected by medical staff trained in surveillance protocols. A 10% suspension was prepared with sterile normal saline that was centrifuged at high speed (10,000 revolutions/minute) for 5 minutes. According to the manufacturer’s instructions, nucleic acid was extracted from 200-μL fecal supernatant specimens using the Roche MagNA Pure LC 2.0 automatic nucleic acid separation and purification system (High Pure Viral RNA Kit of Roche) and was stored at –80°C. Norovirus RNA extraction, detection, and genotyping were conducted as described previously.^14,15

Apart from NoV detection, all samples were also screened for other viruses (astrovirus, sapovirus, RV, and adenovirus) and for bacteria (Shigella, nontyphoidal Salmonella [NTS], Vibrio parahaemolyticus, Campylobacter jejuni, Vibrio cholerae, diarrheagenic Escherichia coli (DEC)]. Primers and probes of viral nucleic acids were synthesized and distributed by Jiangsu CDCs using the Qiagen Probe RT-PCR Kit. Testing of NoV, astrovirus, RV, adenovirus, and sapovirus was performed by polymerase chain reaction (PCR) or reverse-transcription PCR (Supplemental Table S1).^16–18 Bacterial pathogens were tested by performing isolation with or without enrichment procedures at the first step. For DEC and C. jejuni, the isolation was tested subsequently by PCR (Supplemental Table S2); for NTS, V. parahaemolyticus, V. cholerae, and Shigella, the isolation was tested subsequently by biochemical and serological assays.^19,20

Data collection and statistical analysis.

Individual data on demography, clinical manifestations, and laboratory testing results were collected by reviewing medical records. The data were then entered into a standardized database by trained clinicians. All data were uploaded to the online management system structured by the local CDC. Potential demographic and clinical measures for risk prediction included categorical variables such as age, season, occupation, fever, abdominal cramps, nausea or vomiting, and fecal traits.

Distributions of demographic and clinical characteristics were evaluated using the χ² test for categorical variables. The Akaike information criterion, an estimate minimum information theoretical criterion, is designed to measure the goodness of fit and complexity of a model. Stepwise regression, a variable selection method based on the AIC, builds the best regression model by adding or deleting predictor variables gradaully.²¹ Subsequently, an AIC stepwise logistic regression model was used to explore differences as integrated within a clinical setting (NoV positive [+] versus NoV negative [–], NoV+ versus RV+, NoV+ versus bacteria+, genogroup [G] I versus GII genotypes). Effect size is presented with odds ratios (ORs) and the corresponding 95% CI. Data were collected, integrated, and plotted in Excel 2016 (Microsoft Corp., Redmund, WA). All statistical analyses were performed using R (version 3.5.2; R Foundation for Statistical Computing, Vienna, Austria). All statistical tests were two sided and a P-value < 0.05 was set to denote statistical significance.

RESULTS

During the 8-year study period, 683,425 patients with diarrhea, which accounted for 0.6% of the number of outpatient visits, were studied. Children were 0 to 4 years old (47.4%) and adults were 25 to 59 years old (27.7%), which accounted for 75.1% of all cases. A total of 4,416 samples (0.6%) were collected for the detection of pathogens, of which 1,077 (24.4%) were detected for enteric pathogens. There were 435 NoV+ cases (9.9%), of which 48 (11.0%) were GI, 379 (87.1%) were GII, and 8 (1.8%) were coinfections with GI and GII. We also found that there were 43 cases of coinfection (1.0%) with NoV and other viruses or bacteria. Excluding coinfection, 392 cases (8.96%) were confirmed as NoV+ and 3,430 cases (78.4%) were confirmed as NoV– for all tested pathogens, or positive for other enteric pathogens. The positive rates of other enteric pathogens were as follows, excluding coinfections: Shigella, 0.1%; NTS, 2.6%; V. parahaemolyticus, 1.4%; C. jejuni, 0.1%; non-O1/O139 V. cholerae, 0.2%; DEC, 3.6%; astrovirus, 0.8%; RV, 5.4%; sapovirus, 0.3%; and adenovirus, 0.5%.

Comparison of NoV+ and NoV–.

Norovirus was detected throughout the year, and the prevailing season lasted as long as half a year (from October–April) (Figure 1). The seasonal distribution of NoV+ detection was different from NoV– (P < 0.001). Multivariate stepwise logistic regression showed that NoV had a lesser chance of appearing in summer (P < 0.001; OR, 0.257; 95% CI, 0.18–0.36) (Table 1).

Figure 1.

Monthly infection percentage of patients who were norovirus positive (NoV+), NoV negative (NoV–), and rotavirus positive (RV+), and who were infected with bacteria. The percentages detected in this particular month out of the entire study period were calculated.

Table 1.

Multivariate stepwise logistic regression analysis for norovirus positivity versus norovirus negativity, norovirus positivity versus rotavirus positivity, and norovirus positivity versus bacteria positivity

Variable	NoV–, n (frequency) (N = 3,430)	NoV+, n (frequency) (N = 392)	RV+, n (frequency) (N = 226)	Bacteria+, n (frequency) (N = 325)	NoV+ vs. NoV–		NoV+ vs. RV+		NoV+ vs. bacteria+
					OR (95% CI)	P-value	OR (95% CI)	P-value	OR (95% CI)	P-value
Season
Spring (March–May)	844 (24.6)	136 (34.7)	61 (27.0)	70 (21.5)	1.000	–	1.000	–	1.000	–
Summer (June–August)	1,063 (31.0)	44 (11.2)	5 (2.2)	171 (52.6)	0.257 (0.189–0.36)	< 0.001	3.728 (1.50–11.35)	0.009	0.119 (0.07–0.19)	< 0.001
Autumn (September–November)	915 (26.7)	125 (31.9)	31 (13.7)	80 (24.6)	0.830 (0.64–1.08)	0.167	1.870 (1.13–3.15)	0.016	0.763 (0.50–1.15)	0.201
Winter (December–February)	608 (17.7)	87 (22.2)	129 (57.1)	4 (1.2)	0.835 (0.62–1.11)	0.229	0.301 (0.20–0.46)	< 0.001	10.391 (4.07–35.24)	< 0.001
Occupation*
Farmers and migrant workers	1,084 (31.6)	105 (26.8)	57 (25.2)	115 (35.4)	–	–	1.000	–	1.000	–
Preschool and kindergarten children	630 (18.4)	73 (18.6)	76 (33.6)	33 (10.2)	–	–	0.623 (0.37–1.03)	0.067	2.654 (1.51–4.74)	< 0.001
Civil servants and specialists	322 (9.4)	48 (12.2)	11 (4.9)	20 (6.2)	–	–	2.968 (1.38–6.81)	0.007	3.201– (1.64–6.43)	< 0.001
Other	1,394 (40.6)	166 (42.3)	82 (36.3)	157 (48.3)	–	–	1.265 (0.80–2.00)	0.312	1.108 (0.74–1.65)	0.616
Fever
No	2,935 (85.6)	343 (87.5)	176 (77.9)	255 (78.5)	–	–	1.000	–	1.000	–
Yes	495 (14.4)	49 (12.5)	50 (22.1)	70 (21.5)	–	–	0.413 (0.25–0.68)	< 0.001	0.422 (0.26–0.67)	< 0.001
Abdominal cramps
No	1,156 (33.7)	161 (41.1)	101 (44.7)	80 (24.6)	1.000	–	–	–	–	–
Yes	2,274 (66.3)	231 (58.9)	125 (55.3)	245 (75.4)	0.676 (0.54–0.84)	0.001	–	–	–	–
Nausea or vomiting
No	2,379 (69.4)	203 (51.8)	145 (64.2)	204 (62.8)	1.000	–	1.000	–	–	–
Yes	1,051 (30.6)	189 (48.2)	81 (35.8)	121 (37.2)	2.297 (1.85–2.86)	< 0.001	1.439 (0.98–2.12)	0.065	–	–
Fecal traits†
Mucus stools	256 (7.5)	17 (4.3)	7 (3.1)	10 (3.1)	1.000	–	–	–	–	–
Loose stools	374 (10.9)	58 (14.8)	37 (16.4)	32 (9.8)	2.247 (1.29–4.10)	0.006	–	–	–	–
Watery stools	2,691 (78.5)	309 (78.8)	180 (79.6)	277 (85.2)	1.482 (0.91–2.57)	0.133	–	–	–	–
Other	109 (3.2)	8 (2.0)	2 (0.9)	6 (1.8)	0.948 (0.37–2.23)	0.905	–	–	–	–

Bacteria+ = bacteria positivity; NoV– = norovirus negativity; NoV+ = norovirus positivity; OR = odds ratio; Rotavirus+ = rotavirus positivity.

^*Only three of nine kinds of occupations were included and analyzed in a logistic model (as a binary variable); others were interpreted as “others”.

^†Only three of nine kinds of fecal traits were included and analyzed in a logistic model (as a binary variable); others were interpreted as “other.”

Norovirus spanned all ages, from 0 to 87 years old. The proportion of the child population that was NoV+ seemed similar to NoV– (P = 0.819) (Figure 2). However, we found there were statistically different seasonal distributions in NoV+ and NoV– groups for each age group (all P < 0.05) except the 5- to 18-year age group (P = 0.152) (Figure 3). The male-to-female ratio was 1.16:1, but the difference was not statistically significant between the NoV+ and NoV– groups (P = 0.545). Similarly, there was no statistical difference between the NoV+ and NoV– groups in terms of residency and occupation (P > 0.05). In patients who were NoV+, there were no statistically significant differences in sex (P = 0.158) and season (P = 0.675) for different age groups.

Figure 2.

Age distribution of patients who were norovirus positive (NoV+), NoV negative (NoV–), and rotavirus positive (RV+), and who were infected with bacteria.

Figure 3.

Seasonal distribution of patients with (+) without (–) norovirus by age group.

In terms of clinical features, we demonstrated that patients who were NoV+ presented with nausea and vomiting (P < 0.001; OR, 2.297; 95% CI, 1.85–2.86) and loose stools (P = 0.006; OR, 2.247; 95% CI, 1.30–4.10), but with less abdominal cramping (P = 0.001; OR, 0.676; 95% CI, 0.54–0.84) compared with patients who were NoV– (Table 1).

Comparison between NoV+ and RV or/and bacterial infections.

A total of 226 samples were confirmed as RV+ (5.1%) and 325 samples were bacteria+ (7.4%) (coinfections excluded). Our results showed there were significant differences between NoV and RV detection for different seasonal groups (P < 0.001). In addition, RV was detected primarily during the coldest seasons (December–February) (Figure 1). Multivariate logistic regression indicated that civil servants and specialists were more susceptible to NoV (P = 0.007; OR, 2.968; 95% CI, 1.34–6.81). Patients infected with NoV were less likely to manifest fever (P < 0.001; OR, 0.413; 95% CI, 0.25–0.68) than those with RV (Table 1).

Stepwise logistic regression found that the seasonal difference between NoV and bacteria was more obvious. Bacteria were found primarily during warm seasons (June–August) (P < 0.001; OR, 0.119; 95% CI, 0.07–0.19), and NoV occurred mainly during the coldest seasons (December–February) (P < 0.001; OR, 10.391; 95% CI, 4.07–35.24). The proportion of NoV infection was greater than that of bacterial infection among kindergarten/at-home children (P < 0.001; OR, 2.654; 95% CI, 1.51–4.74) and civil servants and specialists (P < 0.001; OR, 3.201; 95% CI, 1.64–6.43). Compared with bacteria, patients with NoV manifested fever less frequently (P < 0.001; OR, 0.422; 95% CI, 0.26–0.67) (Table 1).

Features of GI and GII genotypes.

A total of 379 GII strains were detected concomitant with 48 GI ones. In a stepwise logistic regression model, summer was affected more easily with GI than GII (P < 0.001; OR, 0.128; 95% CI, 0.08–0.20), and winter was less likely (P < 0.001, OR, 11.251; 95% CI, 4.44–37.99). In addition, we found that those older than 45 years had a greater possibility of being affected by GI than GII (Table 2).

Table 2.

Multivariate stepwise logistic regression analysis for genogroup (G) I and GII genotypes

Variable	GI, n (frequency) (N = 48)	GII, n (frequency) (N = 379)	OR (95% CI)	P-value
Season
Spring (March–May)	26 (54.2)	123 (32.5)	1.000	–
Summer (June–August)	8 (16.7)	42 (11.1)	0.128 (0.08–0.20)	< 0.001
Autumn (September–November)	9 (18.8)	123 (32.5)	0.761 (0.51–1.14)	0.190
Winter (December–February)	5 (10.4)	91 (24.0)	11.251 (4.44–37.98)	< 0.001
Age, years
0–4	2 (4.2)	79 (20.8)	1.000	–
5–18	7 (14.6)	35 (9.2)	0.937 (0.44–2.03)	0.867
19–44	27 (56.3)	149 (39.3)	0.601 (0.35–1.02)	0.061
45–59	6 (12.5)	63 (16.6)	0.458 (0.25–0.85)	0.013
≥ 60	6 (12.5)	53 (14.0)	0.445 (0.23–0.84)	0.013

G = genogroup; OR = odds ratio.

Coinfection samples.

Forty-three NoV coinfection samples were discovered (excluding coinfections of NoV GI and GII): 11 with RV, 6 with astrovirus, 1 with sapovirus, 3 with NTS, and 22 with DEC.

DISCUSSION

In 2011, Wuxi began to monitor patients with diarrhea at a pilot site in Jiangsu Province. Since 2012, five common viruses that can cause diarrhea have been monitored throughout the year. Our samples were collected from outpatients of 11 hospitals and 5 township health centers in 6 districts and counties from February 2013 to December 2020, and were acquired from a variety of ages and occupations, which could reflect the morbidity of NoV infections.

The positive rate of NoV was 9.85% (including coinfection with other viruses or bacteria) out of 4,416 stool samples, which is less than previous studies in the Chaoyang District, Beijing (33.9% among sporadic diarrhea samples from adults)²²; Huzhou and Zhejiang (18.1% in patients with acute gastroenteritis patients)²³; Tianjin (26.4% among children hospitalized with acute gastroenteritis)²⁴; and Shanghai (22.9% among outpatients with diarrhea).¹³ Furthermore, we found that, excluding coinfections, NoV had the greatest rate of infection of all pathogens, followed by bacteria; RV was third. Norovirus has been identified as a relevant cause of diarrhea in children younger than 5 years and is the primary agent of acute gastroenteritis outbreaks, which affect individuals of all ages.^25–28 In our study, only 17.62% of 0- to 4-year-old children were included; therefore, we suspect that the low positive rate of NoV might be associated with age composition and geography.

Previous studies^29,30 have shown that NoV peaked mainly in winter or cold weather. In our study, the results show that fewer strains were detected from June to August, and more strains were detected from October to April (when the weather was comparatively cold in Wuxi). Peak seasonality was observed in the winter months, similar to previous systematic reviews, meta-analyses, and research^31,32 about of the seasonality of NoV. Future studies should be conducted to explore the association between NoV infections and temperature.

In terms of clinical features, we found that patients with NoV-associated diarrhea presented with nausea or vomiting and loose stools, which is similar to previous studies.^33–35 In our study, abdominal cramps were seen less commonly in patients who were NoV+ compared with patients who were NoV–. Previous studies^35,36 have shown that acute gastroenteritis caused by NoV may be accompanied by abdominal cramps, but not necessarily. As for genotype, NoV GII was the predominant strain for NoV-associated diarrhea, as described in other studies,^24,37 and our findings (11.03% GI, 87.13% GII, 1.84% mix of both genotypes) were consistent with them. Our research also found that GI was dominant in summer, whereas GII was in winter. This might be a result of variant alternation caused by seasonal changes.

In our study, NoV+ versus RV+ and NoV+ versus bacteria+ also contributed to a better understanding of the pathogen and provided a basis for rough diagnosis. Seasonal effects were consistent with NoV+ and NoV– comparisons. However, RV was also found in cold climates, and bacteria in hot climates. Occupation distribution differences were significant in our study, and preschool and kindergarten child were more vulnerable to NoV than bacteria. Fever was seen less commonly seen in patients who were NoV+ when compared with patients who were RV+ and bacteria+. Similarly, other studies^33,38 also claimed a lower occurrence of fever in patients with NoV than in those with RV and bacteria.

The strength of our study was that it was the first in Wuxi to be concerned with sporadic NoV infections of the entire population. However, there were also several limitations to our study. First, not all data for people with infectious diarrhea were collected and tested for etiology. Second, only one children’s sentinel hospital was enrolled, thus the data for children were very limited. Therefore, the real NoV prevalence might be underestimated. In the future, we will enlarge the range of surveillance in children and experimental detection of etiology. Third, RNA sequencing and phylogenetic analyses of the positive samples were not conducted. Moreover, there was a lack of information on exposure history. Further research regarding specific food risk factors and other routes of transmission could be combined with epidemiological surveys and laboratory tests.

CONCLUSION

Norovirus was the main pathogen of infectious diarrhea in Wuxi, which occurred all year and showed certain seasonal changes. Therefore, strengthening pathogen detection for infectious diarrhea was helpful in understanding the epidemiological characteristics of infections with NoV and, potentially, in preventing disease outbreaks.

Note: Supplemental material appears at www.ajtmh.org.