Birth Cohort Studies Assessing Norovirus Infection and Immunity in Young Children: A Review

Abstract

Globally, noroviruses are among the foremost causes of acute diarrheal disease, yet there are many unanswered questions on norovirus immunity, particularly following natural infection in young children during the first 2 years of life when the disease burden is highest. We conducted a literature review on birth cohort studies assessing norovirus infections in children from birth to early childhood. Data on infection, immunity, and risk factors are summarized from 10 community-based birth cohort studies conducted in low- and middle-income countries. Up to 90% of children experienced atleast one norovirus infection and up to 70% experienced norovirus-associated diarrhea, most often affecting children 6 months of age and older. Data from these studies help to fill critical knowledge gaps for vaccine development, yet study design and methodological differences limit comparison between studies, particularly for immunity and risk factors for disease. Considerations for conducting future birth cohort studies on norovirus are discussed.

Birth cohort studies show that noroviruses are high-incidence pathogens associated with diarrhea among young children, particularly those aged ≥6 months in low/middle-income countries. Evidence for immunity and genetically determined risk factors are provided and important limitations and considerations are discussed.

Noroviruses are one of the foremost causes of diarrheal illnesses globally, associated with nearly 20% of all diarrheal episodes globally and an estimated 685 million episodes and 212 000 deaths annually [1, 2]. While rotaviruses are responsible for the largest global proportion of morbidity and mortality in children <5 years of age [3], noroviruses are associated with approximately 18% of all cases in this age group worldwide [1]. In some countries with universal rotavirus vaccine coverage, norovirus cases exceed those of rotavirus among children seeking medical care [4–6].

Noroviruses are genetically and antigenically diverse [7]. Infections among humans are primarily caused by genogroup I and II (GI and GII) viruses of which new variants of the GII.4 genotype have emerged every 2–4 years, sometimes causing global epidemics [8]. Clear evidence of efficacy of infection control strategies, as well as vaccine development (a candidate is in phase 2b clinical trials), has been hindered [9] because until recently in vitro culture of human strains was not possible [10, 11]. Despite significant progress furthering our understanding of the immune response and correlates of immune protection following vaccination, many questions remain unanswered including the role of host genetic susceptibility, which is associated with expression of histo-blood group antigens (HBGAs) within the intestinal lumen [12–14].

Birth cohorts provide an important observational design to gain understanding of immunity following natural infection and disease in contrast to studies in older children and adults whose infection history is not known. Infants are naive to norovirus exposure at birth and their prospective history of infection and corresponding immunologic reactions can be ascertained. In addition to testing for pathogen presence in diarrheal stools, in most studies routine stool specimens are collected on a frequent and reoccurring basis to assess asymptomatic infections. Thus, the impact of repeat norovirus infection(s) on protection against subsequent infection and disease can be studied. Estimates of diarrheal disease attribution can also be estimated when presence of infections or coinfections with bacterial, viral, or parasitic pathogens are investigated additionally. If genotyping of strains is performed, responses to viruses from other genotypes (heterotypic immune responses) within a genogroup as well as virus-specific disease determinants (homologous immune responses) can also be studied.

We conducted a literature review of birth cohort studies describing norovirus prevalence, protection from subsequent illness after infection, and determinants associated with the risk of disease. We describe significant findings and discuss limitations of currently published studies. As birth cohort studies can help to fill critical knowledge gaps in vaccine development [15], considerations for conducting future birth cohort studies of norovirus exposure and infection are discussed.

METHODS

A literature review was conducted using the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [16] (details of strategy and exclusion criteria are reported in Supplementary Figure 1). For inclusion, studies must have been (1) birth cohorts, defined as longitudinal studies enrolling mothers of newborn human children followed from within 1 month of birth to a predetermined age in early childhood (1–5 years), and (2) used reverse-transcription polymerase chain reaction for detection of norovirus prevalence in the children’s stools.

RESULTS

Ten birth cohort studies described in 21 manuscripts (Supplementary Figure 1; Supplementary Table 1) meeting the inclusion criteria were selected for qualitative review [17–37]. All were among children living in low- or middle-income countries (LMICs) in South and Central America, Asia, and Africa (Supplementary Table 1). A summary of the study designs and methodological details is included in Supplementary Tables 1–3.

PRINCIPAL RESULTS FOUND FOR THE 10 BIRTH COHORT STUDIES

Exposure to Norovirus Is Common in Young Children in LMICs

While there was heterogeneity in the number of routine stool samples tested (Supplementary Table 2), among birth cohort studies where at least 1000 routine stool samples were tested (Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development [MAL-ED], Peru, Ecuador, and Chile) from a range of 194 to 291 children, incidence estimates ranged from 51 to 157 infections per 100 child-years (Table 1). In these studies, approximately 66%–90% of children experienced at least 1 norovirus infection (symptomatic or asymptomatic) in early childhood (Table 1).

Table 1.

Incidence and Percentage of Children Infected by Norovirus and With Norovirus-associated Diarrhea as Reported by Birth Cohort Studies

Birth Cohorta [Reference]	Incidence of Norovirus Infection (per 100 Child-Years)	Percentage of Children Infected by Norovirus	Incidence of Norovirus Diarrhea (per 100 Child-Years)b	Percentage of Children With Norovirus-associated Diarrhea
MAL-ED [29]	Incidence overall in MAL-ED: 157. Nepal: 40; Tanzania: 105; Brazil: 120; Pakistan: 148; South Africa: 189 India: 194; Bangladesh: 221; Peru: 235	86% had at least 1 infection by age 1; 89% had at least 1 infection by age 2; 54% had at least 1 GI infection by age 2; 84% had at least 1 GII infection by age 2	Incidence overall in MAL-ED not reported. Nepal: 36; Peru: 216; other countries not reported	16% had at least 1 diarrheal episode with GI present by age 2; 36% had at least 1 diarrheal episode with GII present by age 2
Peru [24]	146	80% had at least 1 infection by age 1; 38% had at least 2 infections by age 1	66	38% had at least 1 diarrheal episode with norovirus present by age 1; 71% had at least 1 diarrheal episode with norovirus present by age 2
Ecuador [28]	51 (95% CI, 45–58)	66% had at least 1 infection by age 3; 40% had 2 infections by age 3; 16% had >2 infections by age 3	17 (95% CI, 14–21)	30% had at least 1 diarrheal episode with norovirus present by age 3; 10% had >1 diarrheal episode with norovirus present by age 3
India [20, 21]	Not reported	4% had at least 1 asymptomatic infection by age 3	14	40% had at least 1 diarrheal episode with norovirus present by age 3
Brazil [17]	Not reported	60% had at least 1 asymptomatic infection by age 3	Not reported	55% had at least 1 diarrheal episode with norovirus present by age 3
Chile [23]	112	57% had at least 1 asymptomatic infection by 18 mo	13	Percentages not reported

Abbreviations: CI, confidence interval; GI, genogroup I; GII, genogroup II; MAL-ED, Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development.

^aBirth cohort studies in Mexico [18, 19], Bangladesh [26, 36], and Vietnam [35] did not report incidence or the percentage of children infected with norovirus or with norovirus-associated diarrhea. Birth cohort studies in India, Brazil, and Chile reported the percentage of children asymptomatically infected by norovirus rather than the percentage of total norovirus infections.

^bPercentages were reported directly from the manuscript text. The following numbers of children were followed for the analysis conducted in each study: 199 (MAL-ED), 291 (Peru), 194 (Ecuador), 173 (asymptomatic) [21] or 373 (diarrheal) [20] (India), 20 (Brazil), 198 (Chile). Incidence estimates were reported directly from text (Ecuador) or calculated from the reported number of infections divided by the child-months of follow-up (MAL-ED, Peru, Chile). All incidence estimates were converted to units of 100 child-years for ease of comparison.

Norovirus Is Associated With Diarrhea Among Young Children in LMICs

For the Peru, Ecuador, and India birth cohorts, all diarrheal stools collected were tested (ranging from 438 to 1495 samples from 194 to 452 children; Supplementary Table 2). From these studies, incidence estimates for norovirus-associated diarrhea ranged from 14 to 66 per 100 child-years, and approximately 30%–70% of children experienced norovirus-associated diarrhea in early childhood (Table 1).

Asymptomatic Infections and Coinfections With Other Enteropathogens Complicate Diarrheal Disease Attribution in Birth Cohort Studies

Norovirus detection in routine and diarrheal stools ranged widely among individual birth cohort countries (Table 2). In the Peruvian cohort study, the norovirus-attributable fraction of diarrhea was 8% for children in the first year of life and increased to 23% by 2 years of age [24]. In contrast, there was no difference in the percentage of diarrheal vs routine stools positive for norovirus found in the Ecuadorian cohort [28]. Several birth cohort studies tested for presence of other enteropathogens (Table 3), but norovirus-attributable incidences were reported among a broad range of diarrheal pathogens only for the MAL-ED and Bangladesh cohorts. Despite high (23%) prevalence in diarrheal stools, the adjusted attributable fraction for norovirus was only 5.1% in the overall MAL-ED study due to a high percentage of asymptomatic cases (19%) [29, 30, 32]. Norovirus ranked fourth and sixth among enteropathogens with the highest attributable incidences during the first year of life in the MAL-ED and Bangladesh studies, respectively [32, 36].

Table 2.

Percentage of Routine and Diarrheal Stools Positive for Norovirus by Country

Country [Reference]	Routine Stools, % (No. Positive/No. Tested)	Diarrheal Stools, % (No. Positive/ No. Tested)
Nepala [29]	2 (7/325)	…
Tanzaniaa [29]	…	7 (12/171)
Mexicob [19]	5 (3/66)	8 (9/115)
Indiab [20, 21]	4 (7/173)	11 (207/1856)
Chilec [23]	8 (187/2278)	18 (26/145)
Vietnamb [35]	…	13 (169/1309)
Brazilc [17]	20 (25/124)	11 (12/105)
Bangladeshd [26]	~10 (NR/1385)	~10 (NR/420)
Ecuadorc [28]	18 (181/1016)	18 (79/438)
Bangladeshd [36]	…	~20 (NR/1741)
Peruc [24]	13 (491/3690)	23 (341/1495)
All country-MAL-EDa [29]	19 (438/2307)	24 (1699/7077)
South Africaa [29]	30 (75/250)	…
Perua [29]	…	33 (608/1843)

Abbreviations: MAL-ED, Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development; NR, not reported.

^aPercentages are reported in the manuscript; number of positive samples was calculated from total number of samples and percentage positive values.

^bPercentages were calculated from reported number of confirmed positive samples and total number of samples reported in the manuscript.

^cPercentages, number of positives, and total number of samples are reported in the manuscript.

^dPercentages were estimated from manuscript figures as the actual number or percentage of norovirus-positive stools was not reported.

Table 3.

Summary of Data Reported in Birth Cohort Studies on Coinfections by Other Enteropathogens and Norovirus-attributable Incidence of Pathogen-attributable Diarrhea

Birth Cohort [Reference]	Enteropathogens Investigated	Major Findings Related to Coinfections and Norovirus-attributable Incidence
MAL-ED [29–32]	Rotavirus, adenovirus, astrovirus, Aeromonas, Campylobacter, EAEC, EIEC, atypical EPEC, typical EPEC, ETEC, Plesiomonas, STEC, Salmonella, Shigella, Vibrio, Yersinia, Cryptosporidium, Giardia, Entamoeba histolytica. Expanded study [32] used TaqMan array cards and expanded targets to include 11 additional microorganisms: sapovirus, Helicobacter, Enterocytozoon, Encephalitozoon, Cyclospora, Cystoisospora, Ancylostoma, Ascaris, Necator, Strongyloides, Trichuris	•  After Campylobacter spp. and enterovirulent Escherichia coli, norovirus was the third most common pathogen detected in 7077 diarrheal stools collected among 1457 children in the global MAL-ED collaborative, with detection rates of 41%, 40%, and 23%, respectively •  78% of norovirus-positive diarrheal stools were also positive for copathogens • Campylobacter, EAEC, Giardia, and ETEC were the most common copathogens among 1607 norovirus-positive diarrheal stools (43%, 24%, 19%, and 12%, respectively) • In the first and second years of life, the norovirus-attributable fraction was 5%–6% • Norovirus-attributable incidences were 16.4 (95% CI, 13.7–22.6) and 15.4 (95% CI, 12.5–20.1) attributable episodes per 100 child-years for the first year and first 2 years of life • Genogroup I norovirus infections were not attributable to diarrheal disease
Peru [24, 25]	Rotavirus, sapovirus	• 4% (2/54) and 7% (4/54) of norovirus-positive asymptomatic and diarrheal stools were rotavirus positive • 4 diarrheal stools tested sapovirus-positive
Chile [23]	Rotavirus, sapovirus, Salmonella, Shigella, Campylobacter, Yersinia	• 0.2% (4/2278) of asymptomatic norovirus-positive stools were coinfected with rotavirus • 4% (1/26) of symptomatic norovirus-positive stools were coinfected with Campylobacter • Diagnostic methods may lack sensitivity
India [20]	Rotavirus, Aeromonas, Vibrio, Shigella, Giardia, Cryptosporidium, Ascaris	• 27% (56/207) of norovirus-associated diarrheal stools contained coinfections • The most common coinfections were rotavirus, Giardia, and Cryptosporidium (17% [36/207], 4% [9/207], and 2% [5/207], respectively)
Vietnam [35]	Rotavirus, Shigella, Campylobacter, Salmonella	• Of the 1309 diarrheal stools available, detection rates for rotavirus, norovirus, Campylobacter, Salmonella, and Shigella were 30%, 13%, 12%, 10%, and 9%, respectively. • 15% (192/1309) of stools were positive for >1 pathogen • At least 1 additional pathogen was detected in 50% (88/176) of norovirus-positive stools
Bangladesh [26]	Rotavirus, adenovirus, astrovirus, sapovirus, Campylobacter, EAEC, EPEC, ETEC, STEC, Salmonella, Shigella/EIEC, Vibrio, Yersinia, Enterocytozoon, Encephalitozoon, Cyclospora, Cystoisospora, Cryptosporidium, Giardia, E. histolytica, Ancylostoma, Ascaris, Necator, Strongyloides, Trichuris	• Roughly 10% of stools (for each diarrheal and routine) tested positive for norovirus. • Attributable fractions were highest for Campylobacter, EAEC, EPEC, and rotavirus.
Bangladesh [36]	Rotavirus, adenovirus, astrovirus, sapovirus, Aeromonas, Bacteroides fragilis, Campylobacter (spp. jejuni, coli), Clostridium difficile, EAEC, Shigella/EIEC, atypical EPEC, typical EPEC, ETEC, STEC, Helicobacter pylori, Salmonella, Vibrio cholerae, Cyclospora cayetanensis, Cystoisosporiasis belli, Cryptosporidium, Giardia, Entamoeba histolytica, Enterocytozoon bieneusi, Encephalitozoon intestinalis, Ancylostoma duodenale, Ascaris lumbricoides, Necator americanus, Strongyloides stercoralis, Trichuris trichiura	• Of the 1741 episodes of diarrhea, norovirus was detected in ~20% • Norovirus ranked sixth (~90 attributable episodes) among the 32 enteropathogens attributable to diarrhea among children in the first year of life. • Preceding norovirus in ranking were C. jejuni/coli, rotavirus, adenovirus 40/41, Shigella/EIEC, and STEC associated with 187.4, 181.5, 142.3, 124.7, and 107.5 attributable episodes, respectively, during the first year of life

Abbreviations: CI, confidence interval; EAEC, enteroaggregative Escherichia coli; EIEC, enteroinvasive Escherichia coli; EPEC, enteropathogenic Escherichia coli; ETEC, enterotoxigenic Escherichia coli; MAL-ED, Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development; STEC, Shiga toxin–producing Escherichia coli.

Risk of Norovirus Infection Associated With Child Age, Growth, and Breastfeeding

More children were infected or had norovirus-associated diarrhea after 6 months of age as compared to infants <6 months old in all 6 birth cohorts where child age stratification was reported (Table 4). The lowest incidence rates for infections (symptomatic or asymptomatic) occurred before 6 months of age (5–154 infections per 100 child-years), and older infants (6–11 months of age) experienced the highest incidence rates (76–221 infections per 100 child-years) for infection (Table 4). A decrease in norovirus-associated diarrheal severity was observed with increasing age in the MAL-ED, Chilean, and Indian birth cohorts [20, 23, 29].

Table 4.

Data Presented in Birth Cohort Studies Relating to Child Age and Risk for Norovirus Disease

Birth Cohort [Reference]	0–5 mo	6–11 mo	12–24 mo
Perua [24]	103 infections per 100 child-years; 29 episodes of norovirus-associated diarrhea per 100 child-years; 55% (66/119) of infections were asymptomatic	188 infections per 100 child-years; 79 episodes of norovirus-associated diarrhea per 100 child-years; 36% (73/202) of infections were asymptomatic	84 episodes of norovirus-associated diarrhea per 100 child-years
MAL-EDb [29]	154 GI or GII infections per 100 child-years	221 GI or GII infections per 100 child-years; 71% of infections and 82% of norovirus-associated diarrhea reported among children at least 6 mo of age	216 GI or GII infections per 100 child-years
Ecuadorc [28]	5 infections per 100 child-years	76 infections per 100 child-years	64 infections per 100 child-years
India [20]		26 episodes of norovirus-associated diarrhea per 100 child-years occurred among children 12 mo or younger; median age of children experiencing norovirus-associated diarrhea was 7 mo (IQR, 4.5–12 mo)	9 episodes of norovirus-associated diarrhea per 100 child-years occurred among children in the second year of life
Chiled [23]	68 asymptomatic norovirus infections per 100 child-years	100 asymptomatic norovirus infections per 100 child-years; episodes of norovirus-associated diarrhea were similar for all 3 age categories	48 asymptomatic norovirus infections per 100 child-years
Brazil [17]		Children experienced first episode of norovirus-associated diarrhea by 10 mo of age on average (range, 7–11 mo)

Abbreviations: GI, genogroup I; GII, genogroup II; IQR, interquartile range; MAL-ED, Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development.

^aIncidence estimates reported here were calculated from the incidence per 100 child-months reported in the manuscript.

^bSum of GI and GII incidence estimates reported here were calculated from individual GI and GII incidence per 100 child-months reported in manuscript. To derive the incidence estimates reported for the 12–24-mo age category, the sum of incidence estimates for the age categories of 12–17 and 18–24 mo were used. Since only percentages of infections and diarrhea were reported in manuscript, the fraction of children infected that were at least 6 months of age over the total number of children infected could not be derived.

^cThe table heading 12–24 months includes data reported for children aged 12–23 months.

^dIncidence estimates reported here were calculated from incidence per child for 6 months (semester) of life. Also, data presented for age categories of 0–6, 7–12, and 13–18 months are listed under table headings of 0–5, 6–11, and 12–24 months, respectively.

Relationships between norovirus infection and child growth were studied in 4 birth cohorts. In the Peruvian study, length-for-age z score (LAZ) and weight-for-age z score (WAZ) were lower for infants infected with norovirus, and the effect persisted at 24 months of age [24]. In India, malnutrition indicators were comparable for children with and without norovirus infection, but a there was a higher proportion of reinfections among stunted and wasted children [20]. In the MAL-ED study, an increase in one LAZ correlated with a 17% reduction in the odds of having norovirus-associated diarrhea [29], and diarrheal episodes attributable to norovirus were negatively associated with child growth at 3, 24, and 60 months of age [33]. In Bangladesh, norovirus-attributable diarrhea was associated with improved length attainment [36].

In the MAL-ED study, there was a non–statistically significant decrease in norovirus-associated diarrhea severity with each additional month of breastfeeding [29]. In the Indian study, there was no difference in the severity of norovirus-associated diarrhea pre- or postweaning [20].

Genogroup- and Genotype-specific Findings on Norovirus Infection Immunity

From 4 birth cohort studies examining infections and immunity at the genogroup level (considered heterotypic immunity since the GI and GII are divided into at least 9 and 22 genotypes, respectively) [7], it appears that previous infection with GII can be (but is not necessarily) protective of subsequent GII infections (Table 5). Genotype-specific findings provide further insight on the extent of homotypic protection (conferred by repeat infection by the same genotype or GII.4 variant), but such data were described only in the Peruvian and Indian cohort studies and a small study related to the Bangladesh site of the MAL-ED study, where 97%, 86%, and 100% of repeat infections were caused by either a different genotype or by a different GII.4 variant, respectively (Table 5).

Table 5.

Summary of Genogroup- and Genotype-specific Findings on Norovirus Infection and Immunity Compiled From Birth Cohort Studies

	Prior GI Infection		Prior GII Infection		>1 Prior GI or GII Infection		Percentage (Ratioa) of Repeat Infections due to
Birth Cohort [Reference]	Reduced Likelihood of Subsequent GI Infection	Reduced Likelihood of Subsequent GI Diarrhea	Reduced Likelihood of Subsequent GII Infection	Reduced Likelihood of Subsequent GII Diarrhea	Reduced Likelihood of Subsequent GI Infection	Reduced Likelihood of Subsequent GII Infection	Different Genotype or GII.4 Variant	Same Genotype or GII.4 Variant
Chileb [23]	NR	NR	✘	✓	NR	NR	NR	NR
Peruc [24]	✘	✘	✓	✓	✘	✓	97% (147/151)	3% (4/151)
MAL-EDd [29]	✘	✘	✓	✓	✘	✓	NR	NR
Ecuadore [28]	✘	✘	✘	✘	✘	✘	NR	NR
India [20]	NR	NR	NR	NR	NR	NR	86% (31/36)	14% (5/36)
MAL-ED Bangladesh site [34]	NR	NR	NR	NR	NR	NR	100% (11/11)	0% (0/11)

Abbreviations: ✓, evidence in the birth cohort study is supportive; ✘, evidence in the birth cohort study is not supportive; GI, genogroup I; GII, genogroup II; MAL-ED, Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development; NR, not reported.

^aRatio of repeat infections due to different and the same genotype or GII.4 variant over the total number of repeat infections reported in each study.

^bNone of the children with the 19 confirmed symptomatic GII infections had a previous documented infection with a GII virus, compared with 10 of 31 asymptomatic GII infections.

^cMore than 12 different GII genotypes were detected; previous infection with GII was protective against subsequent infection and diarrhea (hazard ratio [HR], 0.55 for both [95% confidence interval {CI}, .41–.74 and .34–.87], respectively); there was a further reduction in the hazard of subsequent infections and diarrhea with ≥2 previous GII infections (HR, 0.23 [95% CI, .11–.48] and 0.18 [95% CI, .05–.68], respectively). When risk of subsequent GI infections were examined among children with previous GI infections, no decrease was observed (HR, 0.93 [95% CI, .40–2.18]).

^dPrior infection with GII provided a 27% and 26% reduction in subsequent infection and diarrhea (HR, 0.727 [95% CI, .571–.926] and 0.761 [95% CI, .504–1.150], respectively) [26]. There was also a further reduction of a subsequent symptomatic infection with GII when children experienced ≥2 prior GII infections (HR, 0.668 [95% CI, .381–1.72]). When risk of subsequent GI infections was examined among children with previous GI infections, no decrease was observed (HR, 0.973 [95% CI, .682–1.388]).

^ePrior GII infection was not associated with risk of subsequent GII infection (relative risk [RR], 0.81 [95% CI, .39–1.69]). When risk of subsequent GI infections was examined among children with previous GI infections, no decrease was observed (RR, 1.12 [95% CI, .69–1.83]).

Genetic Predisposition Is a Risk Factor for Norovirus Infection

The presence of HBGA on the gut epithelium (a phenotype referred to as “secretor positive”) is essential for susceptibility to infections by most norovirus genotypes (GII.4 in particular), but polymorphism in the genes coding for fucosyltransferases (such as FUT2, which determines the secretor phenotype) required for the synthesis of HBGA may vary among different ethnicities, having implications for vaccine efficacy as genetically protected individuals may not respond to vaccination [14]. Two studies investigated HBGA and blood group profiles to explore genetic protection from norovirus infection [20, 28]. In the Ecuadorian cohort, where 88% of the children were secretor positive, no significant risk for infection was found in association with secretor status or blood group. However, all GII.4 infections and all 4 coinfections with GII.4 viruses were in secretor-positive children. By contrast, the infection rate for non-GII.4 viruses was nearly double among secretor-negative children. In the Indian birth cohort, where 61% of children were secretor positive, risk of norovirus diarrhea associated with secretor status was only observed for GI viruses and no correlation was found between disease risk and blood group. More GII.4 diarrheal episodes (75%) were among secretor-positive children, and fewer secretor-positive children (58%) were infected with non-GII.4 strains. In addition, all GI.1 infections were among secretor-positive children.

DISCUSSION

The potential to further our understanding of natural infection and immunity in community settings is a fundamental strength of the birth cohort study design, which can help to fill critical knowledge gaps in norovirus vaccine development [15]. For example, hosts that are immunologically naive to norovirus infection are studied longitudinally, and when routine (control) stools are collected on a frequent and recurring basis from the same children submitting diarrheal stools, homotypic and heterotypic immune responses following repeat infections and illnesses can be studied. Enhanced discrimination between asymptomatic infection and long-term shedding is possible with the birth cohort design, as well as the study of genetic susceptibility and other determinants of disease. When presence and quantity of other enteropathogens in stools are also investigated, estimates of overall diarrheal disease etiology and burden can be refined. The 10 published birth cohorts of this review provide important information on norovirus infection and immunity, but there is a tremendous opportunity to deepen this understanding with future studies. We thus provide a list of considerations for future birth cohorts on norovirus (Table 6).

Table 6.

Considerations for Future Birth Cohort Studies

Methodological • Use real-time RT-PCR detection assays for norovirus screening  o Report Ct values for relative quantification of viral load and establish Ct cutoff values for real-time amplification • Perform conventional RT-PCR and genotyping and/or next-generation sequencing on real-time positive results  o Sequence regions of the capsid for genotyping and polymerase for dual-typing; amplify longer regions of the capsid and/or genome when possible • When screening for coinfections/other enteropathogens, use sensitive methods targeting a broad panel of viral, bacterial, and parasitic pathogens for better resolution of disease attribution • Implement quality assurance and quality control parameters for detection assays and laboratorians (ie, standard operating procedures, proficiency panels, internal controls) • Defining illness: Use ≥3 stools within 24 hours for diarrheal illness • Define events of vomiting-only gastroenteritis and mixed (diarrhea and vomiting) gastroenteritis • Routine stool collection: as frequent as practical for increased resolution of “asymptomatic” vs “long-term shedding” • To better capture episodes of diarrhea, visit enrolled participant homes as frequently as possible rather than relying on them to contact study or clinic personnel

Reporting for incidence, prevalence, and disease attribution • Incidence of infections • Prevalence in asymptomatic and diarrheal stools • Incidence of norovirus-associated diarrhea • Disease attribution estimates • Include values for each community/country studied if multiple locations are included in the study

Immunity and molecular epidemiology • Examine genogroup-specific protection following prior infection/illness • Examine genotype-specific protection following prior infection/illness • Determine if >1 infection/illness confers a higher degree of protection • Further study reinfections by same genotype or GII.4 variant (sequencing longer genome regions, determine immunostatus of patients) • Include a timeline of infections for each child to better assess community exposures • Further assess community exposures by testing specimens from other household (such as siblings) or community members

Collect data for disease determinants • Child age with each infection/illness • Severity of illness • Virus (genotype)–specific disease determinants • Coinfections with other enteropathogens • Genetic predisposition (HBGA phenotypes, genotypes) • Nutritional and lifestyle factors (height/weight, exclusive breastfeeding, socioeconomic status, mother’s education, access to clean water and basic sanitation, etc) • Risk factors for pregnant and nursing mothers • Microbiome: comparison between infected and uninfected populations, between different communities, and changes in microbiome that occur with age, breastfeeding status, and with infection(s) • Collect convalescent sera (sera collection 2 weeks after diarrheal event or infection) after repeat infections and perform HBGA blockade or in vitro neutralization

Abbreviations: Ct, cycle threshold; GII, genogroup II; HBGA, histo-blood group antigen; RT-PCR, reverse-transcription polymerase chain reaction.

Data indicate that exposure to norovirus is extremely common and there is a high incidence of norovirus-associated infection and diarrhea among very young children living in LMICs. This is apparent in studies where the number of children enrolled and number of stools tested was substantial, and is supported by the less extensive studies. Rates of asymptomatic norovirus infection and prevalence in diarrheal stools were variable in birth cohort sites, complicating estimates of the attributable fraction. Also complicating attribution, there was heterogeneity in the frequency in which routine stool samples were collected (from weekly, to every few months, to not at all; Supplementary Table 1). Reanalysis of MAL-ED data to include a more stringent definition of nondiarrheal stools allowed better discrimination between subclinical infection and long-term shedding, which commonly occurs for days to months after infection [42, 43], and increased estimates of the norovirus-attributable incidence [33]. But even within the MAL-ED study, where the study design and detection methodologies were standardized across all sites and where a robust system of quality assurance to ensure proficiency among laboratorians was developed [31], country-to-country variation was reported. While this suggests underlying differences in risk factors between the populations studied, confidence would be improved if the study design was powered for site-level comparisons [29, 32] (Table 6).

Also complicating disease attribution, coinfection with other enteric pathogens was common in the studies where it was investigated. Reanalysis of MAL-ED data using quantitative molecular methods to calculate etiologic burdens found that 65% of diarrhea episodes were pathogen attributable [32]. Similar estimates were found in the recent Bangladesh study [36]. These studies found norovirus to be ranked among the top 10 etiologic agents of pathogen-attributable diarrhea [32, 36], despite significant differences in study designs (the Bangladesh cohort was a single site, did not include routine stool collection from children in the study, and instead used matched controls from the Bangladesh Global Enterics Multicenter Study site [38]. However, uniquely, both studies included estimates of pathogen load to better distinguish between diarrheal disease and subclinical illness in disease attribution estimates [32, 36, 38]. Considering pathogen load, as well as severity and other risk factors for disease, may help to clarify the issue of attributing diarrheal illness to a specific pathogen [38–47] (Table 6). Furthermore, investigating vomiting-only events, which have been shown to cause significant portion of norovirus infections (at least in high-income countries), may also be helpful, but only one birth cohort (India) considered vomit-associated gastroenteritis, without a clear definition [20, 48] (Table 6).

Human challenge and epidemiologic studies demonstrate compelling evidence of homotypic immunity (genotype- or GII.4 variant- specific) and even some degree of heterotypic immunity (across genotypes within the same genogroup) imparted after infection [13, 49]. Birth cohort studies are generally supportive that some degree of heterotypic protection is conferred after infection by GII norovirus, but few studies provided molecular surveillance, so deciphering whether lack of reinfection is due to immunity or lack of exposure could not be discerned. Evidence of homotypic protection is more robust, but only 2 studies reported such findings. Furthermore, no birth cohorts included serological data to assess cross-reactivity and protection. Sequencing longer regions of the capsid to identify amino acid substitutions contributing to lack of cross-protection among norovirus strains would be beneficial. Reporting a timeline for genotypes and variants in circulation among study children and even among other community or household members, such as siblings, would enhance assessment of exposure. Collecting baseline and convalescent sera after infection and performing immunological assays, such as in vitro neutralization or its surrogate, HBGA blockade, would help to clarify immunological findings [13]. Field studies with a longer follow-up are also needed for investigating the duration of immunity, which is estimated mathematically to extend to nearly 10 years [50] (Table 6).

A consistent finding across birth cohort studies was that children <6 months of age are less frequently infected than older infants and children, which is in agreement with a recent meta-analysis that estimated that <15% of pediatric norovirus gastroenteritis occurred among infants <6 months of age [51]. Similar findings of protection among younger infants have been reported for rotavirus and was the impetus for its current vaccine schedule recommendations [52]. Antibodies passed to infants transplacentally and through breastmilk likely contribute to decreased risk for younger infants [43], and as opportunities for both person-to-person and foodborne exposures increase with age, infection risk is also likely to increase. However, more clarification is needed on the role of each, as well as the contribution of risk factors associated with the mother during pregnancy and the first 6 months postpartum if the mother breastfeeds her infant (Table 6).

HBGA secretion appeared to contribute at least partially to the molecular epidemiology of noroviruses in the 2 communities where it was studied [20, 28], a finding that has implications for infectivity, disease severity, and vaccine efficacy [14, 43]. Several other risk factors (ie, nutrition status, breastfeeding duration, access to clean water and adequate sanitation, mother’s education, and other measures of socioeconomic status [SES]) were examined in one or more birth cohort studies. Discrepant findings on the impact of norovirus infection on child growth and the preventive role of breastfeeding were found, but differences in methodological details and the calculations used to assess risk in the studies prevent direct comparison. Further study of these and other risk factors is warranted, particularly across different populations (with diverse genetic profiles and SES characteristics) (Table 6). However, the high infectivity of norovirus is such that it is also possible that the contribution of individual-level factors to norovirus infection or disease risk may be comparatively less significant [53]. More generally for infectious diseases, risk factors interact with individual- and population-level immunity [54] and there are limits to what case-control studies can achieve [55].

Norovirus infection can also be modulated by commensal microbiota, either directly by binding of certain bacteria to norovirus particles [10], or indirectly by modulation of the innate immune response [56]. Repeated or chronic exposure to enteropathogens may also result in more infections or greater severity if environmental enteropathy develops, which has been associated with lower seroresponse rates and immunity following rotavirus vaccination [57]. The intestinal microbiota almost certainly contributes to infection and immunity, but no birth cohort studies to date have included microbiome studies. Of interest would be to compare the microbiomes of infected vs uninfected populations, and differences between communities (including developed and less developed countries), and to investigate changes to the microbiome that occur with child age, breastfeeding status, and with coinfections by enteropathogens (Table 6).

Taken together, current birth cohort studies show that noroviruses are high-exposure pathogens associated with diarrhea among young children, particularly after 6 months of age in LMICs. However, with variable rates of asymptomatic norovirus infection and presence of diverse coinfections with other enteropathogens among different study regions, it is possible that there will be differences in vaccine efficacy for some low-income settings as have been observed for rotavirus [58]. Birth cohort studies are uniquely able to test the validity of laboratory and human challenge studies and clarify questions on immunity and risk factors for disease, but important limitations to the currently published studies exist. Thus, considerations for future studies are presented and summarized in Table 6, many of which are being addressed by extension studies in Peru and India and new studies initiated in Nicaragua and the United States [59, 60] (personal communication with study investigators). Studying norovirus immunity following natural infection in birth cohort studies aids in vaccine development and other control efforts to reduce the burden of disease.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions. J. L. C. and J. V. designed the study and interpreted the data. J. C. performed the literature search, prepared the tables and figure, and wrote the manuscript. J. V., B. A. L., and D. C. P. critically reviewed the manuscript, revised the final version for content, and read and approved the final version.

Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention (CDC).

Financial support. This study was partially supported by a grant from the National Institute of Food and Agriculture (grant number 100005825) and by the intramural food safety program at the CDC.

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.