From April 17, 2017-July 16, 2018, PREVAIL enrolled mothers who resided in the Cincinnati, Ohio region during their third trimester of pregnancy. Maternal demographic information and saliva samples were collected at enrollment. Children were followed until their second birthday with weekly symptom surveys, weekly stool sample collection, and additional data collection during episodes of AGE. Information on childcare was collected at 6 weeks and 4, 6, 12, 18, and 24 months of age, along with duration of breastfeeding. Additional methodology details have been previously described [14]. The study protocol was reviewed and approved by the Centers for Disease Control and Prevention, Cincinnati Children’s Hospital Medical Center and the participating hospitals’ IRBs.

All stool samples underwent reverse transcriptase-realtime quantitative polymerase chain reaction (RT-qPCR) testing to detect GI and GII noroviruses. RT-qPCR positive samples were genotyped after sequencing of conventional reverse transcription-polymerase chain reaction products (Supplemental Material). Cycle-threshold (Ct) values from RT-qPCR were used as a proxy for viral load (lower Ct values represent higher viral load). To identify potential co-infections, 16 additional enteropathogens were tested by RT-qPCR and/or the Luminex xTAG^® Gastrointestinal Pathogens Panel (Supplemental Material).

Norovirus infections were defined based on timing and genotype of norovirus detected in stool. The initial positive sample was considered the start of an infection. Subsequent detections of the same genotype were classified as the same infection if within 30 days of the prior detection, regardless of any intervening negative samples. Similar to other birth cohort studies [13,15], any detection of a new genotype or detection >30 days from the prior detection was considered a new infection. The duration of viral shedding was defined as the number of days from the first to last detection of norovirus within an infection.

Symptom data from weekly surveys, illness surveys, and medical records were combined to identify episodes of AGE, defined as ≥3 looser than normal stools and/or ≥1 episode of vomiting in 24-hours. An infection was considered symptomatic if it started ≤4 days prior to or ≤14 days after the start of AGE symptoms; infections that started outside this range were considered asymptomatic. Severity of symptoms were classified using a 20-point modified Vesikari score (MVS) [16] (Supplemental Material).

Incidence rate was calculated by dividing the total number of norovirus infections by the total weeks at risk in the cohort. Weeks at risk was the sum of weeks in which a stool sample was submitted by participants. Norovirus-positive samples were not subtracted from weeks at risk since infections were based on genotype detected and children remained at risk of infection from other genotypes throughout the follow-up period.

Fucosyltransferase 2 (FUT2) genotype (secretor status), a genetic mutation which impacts binding of norovirus to its target cells, of mothers and children was determined from saliva samples to evaluate potential genetic susceptibility to infection [17] (Supplemental Material).

Characteristics of mothers and children and a description of norovirus infections are initially provided for all participants and infections. Subsequent analyses were restricted to participants who submitted ≥70% of weekly samples and who were followed for ≥18 months (adherent participants). Co-infections were excluded from symptom analyses. The chi-square, Mann-Whitney U and Kruskal-Wallis tests were used for comparing data in descriptive analyses.

Generalized estimating equation (GEE) models utilizing a binomial family function and exchangeable correlation structure were fit to estimate predictors of infection symptom status among adherent children. Four GEE models were fit including an unadjusted model using all infections and three models adjusted for predictors selected a priori using 1) all infections, 2) infections during the first year of life and 3) infections in the second year of life. An adjusted generalized linear model (GLM) was similarly fit among first infection only. Predictors included health insurance type, childcare at the time of infection, breastfeeding status, mother and child secretor status, norovirus genogroup/genotype, whether the child had previous norovirus infections, child’s age, and minimum Ct value detected during the infection (Supplemental Material). Results are presented as odds ratios (ORs) with corresponding 95% confidence intervals (95% CIs). All analyses were conducted in R version 4.4.0 [18].

In the entire cohort, 72 GI and 330 GII norovirus infections were identified among 156 children (Table 1). One in five (21.6%, n=53) children experienced ≥1 norovirus infection by 6 months of age, increasing to over half (51.0%, n=125) within the first year of life and nearly two-thirds (63.7%, n=156) by age 2. The median age at infection among all children was 11.9 months [IQR: 7.6-17.3] (Table 1). Eighty-nine children (36.3%) did not have any detected norovirus infection during follow-up; 34.8% (n=23/66) of these children were FUT2 non-secretors (status unknown=23). Overall, 13.6% (n=53/391) infections were among FUT2 non-secretors (status unknown=11). The incidence of norovirus infection was 1.7 (95% CI: 1.5-1.9) per child-year, peaking during winter months (Figure 1).

Among GI infections, GI.3 was the most frequently detected genotype (67.6%, n=46/68) followed by GI.6 (25.0%, n=17/68). GII genotypes were more widely distributed with GII.4 Sydney (27.9%, n=86/308) and GII.6 (23.7%, n=73/308) detected in approximately one in five of all genotyped infections (Supplemental Figure 1).

Few co-infections were identified (n=34/402, 8.5%) [19]. Excluding coinfections, 65.2% (n=214/328) of norovirus infections with known symptom status were asymptomatic.

We observed 253 (45 GI and 208 GII) norovirus infections among 101 adherent children (Table 2). Genotypic distributions were similar as those described for the full cohort (Supplemental Figure 1). Eighteen co-infections were identified. Among norovirus-only infections with known symptom status, two-thirds (65.3%, n=143/219) were asymptomatic, a proportion nearly identical to the full cohort (Table 2).

The median age at first norovirus infection for adherent children was 9.4 months [IQR: 6.1-12.3]. By 2 months of age, 4.0% of children had experienced a norovirus infection, and that proportion increased rapidly in early life to 61.4% (n=62) and 84.2% (n=85) at 1 and 2 years of age (Table 1, Figure 2). 71.1% (n=64/90) of infections that occurred during the first year of life were asymptomatic compared to 61.2% (n=79/129) in the second year. No children experienced a symptomatic infection before 4 months of age (Figure 2) and only 3.0% had by by 6 months of age. The proportion increased to 20.8% by 1 year of age and further rose to 51.5% by age 2 (Figure 2). Asymptomatic infections generally occurred at a higher level than symptomatic infections (Figure 3). This relationship was most pronounced with GI and GII.other genotypes, but less consistent for GII.4 Sydney (Figure 3).

The median duration of viral shedding was 13.0 days [IQR: 1.0-23.0] with longer shedding seen for GII infections (GII.4 Sydney: 12.0 [IQR: 1.0-23.0]; other GII: 15.0 [IQR: 1.0-25.5]) compared to GI infections (8.0 [IQR: 1.0-21.0]) (Table 2). When all genogroups were combined, children with symptomatic norovirus infections shed virus longer than those with asymptomatic infections (median=18 days [IQR: 7.5-28.0] vs. 10 days [IQR: 1.0-22.0], p-value=0.004).

Minimum Ct values were comparable by genogroup/genotype with median values ranging from 24.5-26.0 (p=0.358). The median Ct value was significantly lower in symptomatic infections (23.0 [IQR: 21.0-27.0]) compared to asymptomatic infections (27.0 [IQR: 23.0-31.0], p-value<0.001). Yet, over half of infections with Ct values less than 25, representing infections with the highest viral loads, were asymptomatic (Ct <20: 52.6%; Ct 20-24: 51.3%) (Figure 4). Among the infections with Ct values <25, the proportion of children considered secretors was comparable for both asymptomatic (84.3% secretors) and symptomatic (87.8% secretors) infections.

The median duration of symptoms was 3 days [IQR: 2-5] and similar across genogroups/genotypes (p-value=0.200) (Table 2). Vomiting was the most common symptom reported, present in 84.2% (n=64/76) of symptomatic infections, followed by diarrhea (77.6%, n=59/76) and fever (31.5%, n=23/73). The median duration of vomiting and diarrhea were 1.0 day [IQR: 1.0-2.0] and 3.0 days [IQR: 2.0-4.3], respectively. Children experienced vomiting without diarrhea in 22.4% (n=17/76) of symptomatic infections. Children typically experienced a maximum of 3.0 [IQR: 2.0-5.0] vomiting and 4.0 [IQR: 3.0-5.0] diarrhea episodes in a 24-hour period (median values). Vomiting was more frequently reported in GII.4 Sydney (85.7%, n=18/21) and other GII (87.5%, n=42/48) infections compared to GI (57.1%, n=4/7) infections but not significantly (p=0.117). Nearly all children with GII.4 Sydney infections experienced diarrhea (95.2%, n=20/21); diarrhea was also common in both GI (71.4%, n=5/7) and other GII (70.8%, n=34/48) infections. Fever was reported frequently in GI infections (50.0%, n=3/6), but this was not significantly higher than in GII.4 Sydney (23.8%, n=5/21) or other GII (32.6%, n=15/46) infections (p=0.460) (Table 2). Symptom characteristics were similar when comparing infections occurring in the first and second years of life (Supplemental Table 1).

Most symptomatic infections (88.2%, n=67/76) were considered mild according to their MVS scores (Table 2). Severity by genogroup/genotype did not differ significantly, though the highest proportion of moderate and severe infections occurred among GII.4 Sydney infections (moderate=14.3%, n=3/21; severe=9.5%, n=2/21). Severity did not differ by child’s year of life (Supplemental Table 1).

Overall, medical care was sought for one-third (32.9%, n=25/76) of symptomatic infections, resulting in 15 outpatient visits and 10 emergency department or hospital admissions among 23 children (Table 2). Differences in care seeking were observed by genogroup/genotype, though not significantly (p=0.167). The majority (85.7%, n=6/7) of symptomatic GI infections were treated at home. One in five (21.7%, n=15/69) children with a symptomatic GII infection (GII.4 Sydney or other GII) sought care in the outpatient setting while an additional 13.0% (n=9/69) presented to the ED or were hospitalized. Out of 51 GII.4 Sydney infections with known symptom status among adherent children, 41.2% (n=21) were symptomatic and 11.8% (n=6) resulted in an ED visit or hospital admission (Table 2).

Unadjusted models using all infections among adherent children suggest the strongest predictor of experiencing symptoms was infection with GII.4 Sydney (OR=2.95, 95% CI: 1.11-7.82) (Supplemental Table 2). When adjusting for all predictors (Figure 5), infection with GII.4 Sydney (OR=3.45, 95% CI: 1.18-10.12) remained a significant predictor of experiencing symptoms. Increased age at infection (OR=1.10, 95% CI: 1.01-1.19) had a significant but modest impact on symptomatic status. Lower viral load (higher Ct value) was associated with reduced odds of symptomatic infection (OR=0.89 for a one unit increase in Ct value, 95% CI: 0.83-0.95). The high OR (3.68, 95% CI: 0.90-15.07) observed for children considered FUT2 secretors suggests secretors may have increased odds of symptomatic infection, however, the CI was wide and crossed the null. Childcare setting, breastfeeding, mother’s secretor status, and whether the child had prior infections were not predictive of symptom status.

Model convergence was not achieved when including all predictors in models stratified by child’s year of life (Supplemental Table 2). In the first year of life, children who attended large childcare settings were at greatest risk of symptomatic infection (OR=4.97, 95% CI: 1.42-17.40) compared with those who were cared for solely at home or in another’s persons home; an inverse relationship was found in the second year of life (OR=0.31, 95% CI: 0.14-0.73). Breastfeeding increased the odds of symptomatic infection (OR=5.92, 95% CI: 1.64-21.33) during the second year of life only. Higher Ct values were associated with lower odds of symptomatic infection during the second year of life (OR=0.85, 95% CI: 0.77-0.95), as well as for first infections (OR=0.87, 95% CI: 0.74-0.99) (Supplemental Table 2).