Viral

Rotavirus is the most common cause of pediatric diarrhea. In the prevaccine era, greater than 90% of circulating human rotavirus strains globally belonged to the one of five common genotypes: G1P[8], G2P[4], G3P[8], G4P[8], and G9P[8]. In developing countries, there is greater genetic diversity and emergence of new and unusual strains. The two genera of Calicivirus that cause diarrheal disease in humans are norovirus and sapovirus, each of which is further divided into genogroups and genotypes. Genetic drift among Calicivirus caused by point mutations and recombination events is common, resulting in emergence of antigenic variants. Since the 1990s, GII.4 norovirus has caused most infections worldwide. Among the more than 50 serotypes of adenovirus, types 40 and 41 are most often associated with diarrhea. Astroviruses that cause disease in humans belong to the Mamastrovirus genera (types 1–8).

Bacterial

The major bacterial enteropathogens are Shigella , nontyphoidal Salmonella (NTS), Campylobacter , and Yersinia . Four species of Shigella cause human disease: S flexneri (the major cause of shigellosis in low-resource countries), S sonnei (the second most common cause of shigellosis in low-income countries and the major cause in industrialized and transitional countries), and less commonly S boydii and S dysenteriae . Salmonella are classically divided into the human-restricted typhoidal Salmonella ( S typhi and S paratyphi A and B), which cause enteric fever, and NTS, which contains most other serovars causing human diarrheal disease. Salmonella typhimurium and Salmonella enteritidis are the most common human NTS serovars globally. Two species of Campylobacter affect humans: C jejuni (90%–95% of infections) and C coli . Only 2 of the 11 species of Yersinia cause diarrhea in humans ( Y enterocolitica and Y pseudotuberculosis ). Five pathotypes of Escherichia coli infect humans: enterotoxigenic (ETEC), enteropathogenic, Shiga toxin-producing (also known as enterohemorrhagic) (STEC), enteroinvasive, and enteroaggregative. Pathogenic E coli are identified according to genotypic or phenotypic features that indicate virulence factors that they produce. There are greater than 200 serogroups of Vibrio cholerae but only two (O1 or O139) have been associated with epidemic cholera and cause nearly all sporadic cases. The O1 serogroup is further classified by serotype (eg, Ogawa and Inaba) and biotype (El Tor or classical). The seventh pandemic, which began in 1961 and is ongoing, is caused by V cholerae O1 El Tor. El Tor variants have emerged with genetic and phenotypic characteristics of classical biotype and seem more virulent.

Clostridium difficile that produce toxin are pathogenic for humans. Since 2000, a hypervirulent strain, North American Pulsed Field Type 1, polymerase chain reaction (PCR) ribotype 027 (NAP1/B1/027), has produced outbreaks of disease worldwide. Other bacterial pathogens are common causes of foodborne illness because of their ability to produce emetic and/or enterotoxins. These include Bacillus cereus , Clostridium perfringens , and Staphylococcus aureus. The burden of these pathogens has not been well-documented in developing countries and they are not discussed in detail.

Protozoal

Cryptosporidium is an oocyst-forming coccidian protozoa. At least 13 of the more than 60 species have been found to cause human disease, but C hominus , and to a lesser extent anthroponotic strains of C parvum , account for most (∼90%) human infections. Cyclospora cayetanesis is another coccidian protozoa. The genus Entamoeba includes six species that colonize humans; Entamoeba histolytica is thought to be the sole pathogen causing intestinal illness. Giardia intestinalis (formerly Giardia lamblia and Giardia duodenalis ) is a flagellate protozoan that infects the small intestine and biliary tract.

Etiology of Diarrhea in Developing Countries

Two important studies conducted during the previous decade have advanced the understanding of the burden of diarrheal disease among young children living in developing countries. The GEMS (Global Enteric Multicentre Study) was a large, 3-year, population-based case-control study of acute, medically attended moderate-to-severe diarrhea (MSD) among children younger than 5 years living in Sub-Saharan Africa and South Asia. Three age strata were included: 0 to 11 months, 12 to 23 months, and 24 to 59 months. The second was the MAL-ED (Malnutrition and Enteric Disease) Study of the cause, risk factors, and interactions of enteric infections and malnutrition and the consequences for child health. MAL-ED was a longitudinal community-based study with evaluation and sampling of newborn cohorts during health and acute diarrheal illnesses in eight low- and middle-income countries across Africa, Asia, and South America. Children were followed until their second birthday.

Four major differences in the study design between GEMS and MAL-ED are notable. Similarities in study design are also worthy of mention as they allow comparisons of the findings to be made. First, GEMS used a case definition to enroll more severely ill children with blood in stool, evidence of dehydration (sunken eyes or decreased skin turgor), hospitalization, or administration of intravenous fluids. MAL-ED recruited milder cases from the community, with fewer than one-third meeting GEMS enrollment criteria. Together these studies capture the severe and less severe diarrheal diseases, which may have different etiologies. Second, GEMS sites were generally less developed; four of the seven GEMS sites and no MAL-ED sites were located in countries with the 35 highest under-5 mortality rates in 2010, so exposures and host vulnerability might differ. Third, GEMS was a case-control study in which children were visited a second time 2 to 3 months after enrollment to detect adverse outcomes. Although MAL-ED detected fewer adverse outcomes presumably because milder illnesses were captured, the longitudinal design was better suited to assess the sequence of events to determine causality and to measure disease burden over time (incidence). Both studies measured the proportion of diarrheal disease that was attributable to a broad array of pathogens, adjusting for asymptomatic detection of pathogens in controls. The end point, designated attributable fraction (AF), thus represented the proportion of cases significantly associated with diarrhea that could be prevented if an effective intervention were implemented. Finally, GEMS was conducted before introduction of rotavirus vaccine at any site, whereas three MAL-ED sites introduced vaccine before study initiation.

Across sites, most attributable cases in GEMS were caused by four pathogens (rotavirus, Cryptosporidium , Shigella , and ETEC producing heat-stable toxin [ST] alone or with labile-toxin [LT], herein termed ST-ETEC), and to a lesser extent, adenovirus 40/41. Reanalysis of GEMS data using quantitative PCR (qPCR) substantially increased the AF of pathogens compared with estimates based on culture or immunoassay for key agents, such as Shigella and C jejuni/coli (two-fold), ST-ETEC (1.5-fold), and adenovirus 40/41 (five-fold).

The public health importance of rotavirus in developing countries is a resounding message in GEMS. Rotavirus was the leading pathogen at every site during the first year of life, and at seven sites during the second year of life. The incidence of rotavirus among infants was more than two times higher than that seen for any other pathogen. GEMS data can be used to predict the public health impact of rotavirus vaccine introduction in developing countries. For example, a vaccine with 60% efficacy would prevent 4.2 episodes of rotavirus MSD per 100 child-years in the first year of life alone. In a low-resource African setting, such as Mali, with a birth cohort of 758,000 in 2016, this means that approximately 31,500 cases of life threatening rotavirus infection during the first year of life would be averted annually.

An unexpected observation in GEMS was the high prevalence of Cryptosporidum among episodes of MSD. Cryptosporidium ranked second among infants at all sites regardless of human immunodeficiency virus prevalence, and third among children 12 to 23 months of age. Although recognized as a cause of diarrhea and malnutrition in Sub-Saharan Africa, the importance of this pathogen in Asia had not previously been appreciated. Moreover, children with Cryptosporidium in the 12-to-23-month age group had a significantly higher risk of death during the ensuing 2 to 3 months, consistent with observations of excess mortality associated with this pathogen among infants and toddlers from Guinea-Bissau 15 years earlier.

GEMS also illustrated the strong contribution of Shigella to the MSD burden at every study site. In contrast to rotavirus and Cryptosporidium , whose incidence declined with age, the incidence of Shigella increased with age, becoming the second most common pathogen identified among children 12 to 23 months, and the leading pathogen at 24 to 59 months of age. The qPCR analysis demonstrated not only that Shigella was the major pathogen associated with dysentery, which was expected (AF, 63.8%), but also the second most common agent associated with watery diarrhea (12.9%). The qPCR used in this analysis was not able to distinguish Shigella from enteroinvasive; however, further analysis suggested that most of these strains are Shigella (C. Stine and J. Nataro, unpublished data).

Several pathogens were important only in Asia ( Aeromonas ) or in Asia plus Mozambique ( C jejuni/coli and V cholerae O1). Historically, an association between Aeromonas and diarrhea has been observed inconsistently, raising the possibility that only certain species or pathotypes were capable of causing diarrhea, or that Aeromonas was a cotraveler with pathogens acquired by the same route. Although coinfections involving Aeromonas were common in GEMS, particularly with Shigella , the association with diarrhea persisted when the analysis controlled for the presence of other pathogens and when Aeromonas was the only pathogen identified in a diarrheal episode. In accordance with other reports, 26% of episodes were dysenteric. These provocative data deserve further investigation.

Pathogens significantly associated with MSD at only one or two sites included (1) norovirus GII and E histolytica among infants; (2) norovirus GII, NTS, typical enteropathogenic (EPEC), and enteroaggregative E. coli among children aged 12 to 23 months; and (3) norovirus GII, NTS, and sapovirus in the 24-to-59-month age group. The paucity of NTS as a cause of diarrhea in developing countries deserves special mention. During the past several decades, distinct clones of NTS that have arisen in Sub-Saharan Africa are a frequent cause of often-fatal bloodstream infection in young children (particularly those with coincident malaria or malnutrition) and in adults infected with human immunodeficiency virus. In contrast to most NTS elsewhere, these strains do not seem to arise from zoonotic reservoirs and seldom cause diarrheal disease. A common feature involves high levels of genetic degradation (a characteristic of S typhi ), rather than acquisition of new virulence factors.

At least three pathogens are notable for their absence as a cause of MSD. First, STEC, a zoonotic infection that causes sporadic cases and outbreaks of diarrhea and hemorrhagic colitis linked to hemolytic uremic syndrome in high-income countries, was not found in any site. Second, ETEC strains producing LT only were not associated with diarrhea. This may be because LT-only ETEC produce less severe disease. Third, Yersinia , generally found in cool climates, has occasionally been reported to occur in developing countries but was not detected in either GEMS or MAL-ED.

Interestingly, Giardia was not associated with MSD but instead was found significantly more often in control subjects than in cases 12 to 59 months of age at most sites. This interesting finding has been observed by others and suggests the hypothesis that in developing countries Giardia may actually interfere with the pathogenic mechanisms of other enteric pathogens. Along the same lines, one wonders whether Giardia is a factor associated with suboptimal colonization and immune responses to live enteric vaccines that is commonly seen in developing countries.

In MAL-ED no pathogen exhibited a high AF in all individual sites. When an AF for all sites combined was calculated, the most common agents associated with diarrhea, in descending order, were norovirus GII, rotavirus, Campylobacter , astrovirus, and Cryptosporidium during infancy and Campylobacter , norovirus GII, rotavirus, astrovirus, and Shigella during the second year of life. Rotavirus had the highest AF for sites without vaccine introduction and the fifth highest AF fraction for sites with vaccine introduction. During infancy, the AF in sites without vaccine introduction ranged from 3.2% to 9.6% compared with 16.3% to 27.8% in GEMS. These differences may reflect the observed trend for rotavirus to be identified with increasing frequency as the severity of illness increases; the clinical venue where cases are identified (community vs health center or hospital) is thought to be a proxy for disease severity.

At first glance, it seems that norovirus GII played a leading role in MAL-ED, which is at odds with the GEMS results. However, close examination reveals that norovirus GII has the highest AF during infancy in MAL-ED and the second highest during the second year of life for all sites combined. When sites were examined individually, however, norovirus GII was associated with diarrhea at only three sites during infancy and three sites during the second year of life. Two sites where norovirus GII predominated had introduced rotavirus vaccine, thus increasing the relative proportion of norovirus disease. In GEMS, norovirus GII was significantly associated with MSD in The Gambia (all age groups) and India (12–23 months old). The lack of association of norovirus with diarrhea at some sites in GEMS and MAL-ED was caused by high rates of asymptomatic carriage ; considerable geographic diversity in the prevalence of norovirus GII also contributed. Factors that might influence this diversity include geographic differences in the prevalence of genetic factors that mediate virus binding, which are less common in certain African and Latin American populations. Temporal variations in the prevalence and severity of disease are seen when strains mutate in the hypervariable regions of the capsid genes resulting in emergence of new strains. GEMS and MAL-ED may have underestimated the burden of norovirus by excluding children who present with vomiting alone. Finally, GEMS may have captured less norovirus diarrhea because it produces a less severe illness.

As with norovirus, Campylobacter was identified in most children in MAL-ED by 1 year of age. Despite high prevalence and high AFs for diarrhea when all sites were combined, infection was associated with diarrhea at only a few sites. Both norovirus and Campylobacter were associated with subsequent linear growth faltering in MAL-ED.

Outcomes of Moderate-To-Severe Diarrhea

An observation in both GEMS and MAL-ED was the geographic heterogeneity of pathogens. Nonetheless, both studies found rotavirus, Shigella , Cryptosporidium , and ST-ETEC to be associated with diarrhea in multiple sites. The widespread prevalence and high incidence of these four pathogens suggest that they should be prioritized for development and implementation of interventions to reduce the diarrheal disease burden.