Analysis of clinical and environmental strains of nontoxigenic Vibrio cholerae for susceptibility to CTXPhi: molecular basis for origination of new strains with epidemic potential.

Toxigenic Vibrio cholerae strains are lysogens of CTXPhi, a filamentous phage which encodes cholera toxin. The receptor for CTXPhi for invading V. cholerae cells is the toxin-coregulated pilus (TCP), the genes for which reside in a larger genetic element, the TCP pathogenicity island. We analyzed 146 CTX-negative strains of V. cholerae O1 or non-O1 isolated from patients or surface waters in five different countries for the presence of the TCP pathogenicity island, the regulatory gene toxR, and the CTXPhi attachment sequence attRS, as well as for susceptibility of the strains to CTXPhi, to investigate the molecular basis for the emergence of new clones of toxigenic V. cholerae. DNA probe or PCR assays for tcpA, tcpI, acfB, toxR, and attRS revealed that 6.85% of the strains, all of which belonged to the O1 serogroup, carried the TCP pathogenicity island, toxR, and multiple copies of attRS, whereas the remaining 93.15% of the strains were negative for TCP but positive for either one or both or neither of toxR and attRS. An analysis of the strains for susceptibility to CTXPhi, using a genetically marked derivative of the phage CTX-KmPhi, showed that all TCP-positive CTX-negative strains and 1 of 136 TCP-negative strains were infected by the phage either in vitro or in the intestines of infant mice. The phage genome integrated into the chromosome of infected V. cholerae O1 cells forming stable lysogens. Comparative analysis of rRNA gene restriction patterns revealed that the lysogens derived from nontoxigenic progenitors were either closely related to or distinctly different from previously described clones of toxigenic V. cholerae. To our knowledge, this is the first demonstration of lysogenic conversion of naturally occurring nontoxigenic V. cholerae strains by CTXPhi. The results of this study further indicated that strains belonging to the O1 serogroup of V. cholerae are more likely to possess the TCP pathogenicity island and hence to be infected by CTXPhi, leading to the origination of potential new epidemic clones.

Induction of the lysogenic phage encoding cholera toxin in naturally occurring strains of toxigenic Vibrio cholerae O1 and O139.

In toxigenic Vibrio cholerae, the CTX genetic element which carries the genes for cholera toxin (CT) is the genome of a lysogenic bacteriophage (CTXPhi). Clinical and environmental strains of V. cholerae O1 or O139 and stools that were culture positive for cholera were analyzed to study the induction and transmission of CTXPhi. To our knowledge, this is the first report of the examination of CTXPhi in clinical materials and in naturally occurring strains. DNA probe analysis revealed that 4.25% (6 of 141) of the isolated V. cholerae strains spontaneously produced a detectable level of extracellular CTXPhi particles in the culture supernatants whereas another 34.04% (48 of 141) produced CTXPhi particles when induced with mitomycin C. CTXPhi isolated from 10 clinical or environmental strains infected a CT-negative recipient strain, CVD103, both inside the intestines of infant mice and under laboratory conditions. All culture-positive stools analyzed were negative for the presence of CTXPhi both in the DNA probe assay and by in vivo assay for the infection of the recipient strain in infant mice. These results suggested that naturally occurring strains of toxigenic V. cholerae are inducible lysogens of CTXPhi but that cholera pathogenesis in humans is not associated with the excretion of CTXPhi particles in stools, indicating that induction of the phage may not occur efficiently inside the human intestine. However, in view of the efficient transmission of the phage under conditions conducive to the expression of toxin-coregulated pili, it appears that propagation of CTXPhi in the natural habitat may involve both environmental and host factors.

Molecular analysis of toxigenic Vibrio cholerae O139 Bengal strains isolated in Bangladesh between 1993 and 1996: evidence for emergence of a new clone of the Bengal vibrios.

Vibrio cholerae O139 Bengal emerged in 1992 and rapidly spread in an epidemic form, in which it replaced existing strains of V. cholerae O1 in Bangladesh during 1992 and 1993. The subsequent emergence of a new clone of V. cholerae O1 of the El Tor biotype that transiently displaced the O139 vibrios during 1994 to 1995 and the recent reemergence of V. cholerae O139 and its coexistence with the El Tor vibrios demonstrated temporal changes in the epidemiology of cholera in Bangladesh. We studied clonal diversity among V. cholerae O139 strains isolated from cholera patients and environmental surface water since their first appearance until their transient disappearance in 1994 as well as the O139 strains that reemerged during 1995 to 1996 and were isolated in the capital Dhaka and four rural districts of Bangladesh to investigate the origin of the reemerged strains. Analysis of restriction fragment length polymorphisms in genes for conserved rRNA and cholera toxin (CT) (ctxA) or in DNA sequences flanking these genes revealed four different ribotypes and four different ctx genotypes among the 93 strains of V. cholerae O139 studied. Ribotypes I and II and ctx genotypes A through C were shared by strains isolated from the epidemic outbreak during 1992 and 1993 in Bangladesh and India, ribotype III was represented by a single CT-negative O139 strain from Argentina, and 16 of 27 (59.2%) of the reemerged strains isolated during 1995 and 1996 belonged to a new ribotype of O139 vibrios designated ribotype IV. All 16 strains belonging to ribotype IV also belonged to a new ctx genotype (genotype 4). These results provide evidence for the emergence of a new clone of toxigenic V. cholerae O139 in Bangladesh. Further analysis of the rfb gene cluster by PCR revealed the absence of a large region of the O1-specific rfb operon and the presence of an O139-specific genomic region in all O139 strains. The PCR amplicon corresponding to the rfaD gene of a CT-negative O139 strain from Argentina was smaller in length than those of the toxigenic O139 strains but was identical to those of seven non-O1 and non-O139 strains. All O139 strains except the CT-negative strain carried structural and regulatory genes for CT and toxin-coregulated pili (ctxA, tcpA, tcpI, and toxR). These results suggest that the O139 Bengal strains possibly emerged from an El Tor strain but that the CT-negative non-Bengal O139 strain might have emerged from a non-O1, non-O139 strain. Thus, strains belonging to the O139 serogroup may have emerged from similar serotype-specific genetic changes in more than one progenitor strain of V. cholerae.

Emergence of a new clone of toxigenic Vibrio cholerae O1 biotype El Tor displacing V. cholerae O139 Bengal in Bangladesh.

The emergence of Vibrio cholerae O139 Bengal in 1993, its rapid spread in an epidemic form, in which it replaced existing strains of V. cholerae O1 during 1992 and 1993, and the subsequent reemergence of V. cholerae O1 of the El Tor biotype in Bangladesh since 1994 have raised questions regarding the origin of the reemerged El Tor vibrios. We studied 50 El Tor vibrio strains isolated in Bangladesh and four other countries in Asia and Africa before the emergence of V. cholerae O139 and 32 strains isolated in Bangladesh during and after the epidemic caused by V. cholerae O139 and 32 strains isolated in Bangladesh during and after the epidemic caused by V. cholerae O139 to determine whether the reemerged El Tor vibrios were genetically different from the El Tor vibrios which existed before the emergence of V. cholerae O139. Analysis of restriction fragment length polymorphisms in genes for conserved rRNA, cholera toxin (ctxA), and zonula occludens toxin (zot) or in DNA sequences flanking the genes showed that the El Tor strains isolated before the emergence of V. cholerae O139 belonged to four different ribotypes and four different ctx genotypes. Of 32 El Tor strains isolated after the emergence of O139 vibrios, 30 strains (93.7%) including all the clinical isolates belonged to a single new ribotype and a distinctly different ctx genotype. These results provide evidence that the reemerged El Tor strains represent a new clone of El Tor vibrios distinctly different from the earlier clones of El Tor vibrios which were replaced by the O139 vibrios. Further analysis showed that all the strains carried the structural and regulatory genes for toxin-coregulated pilus (tcpA, tcpI, and toxR). All strains of the new clone produced cholera toxin (CT) in vitro, as assayed by the GM1-dependent enzyme-linked immunosorbent assay, and the level of CT production was comparable to that of previous epidemic isolates of El Tor vibrios. Further studies are required to assess the epidemic potential of the newly emerged clone of V. cholerae O1 and to understand the mechanism of emergence of new clones of toxigenic V. cholerae.

Molecular epidemiology of toxigenic Vibrio cholerae in Bangladesh studied by numerical analysis of rRNA gene restriction patterns.

Cholera is endemic in Bangladesh, and a regular seasonal pattern of cholera epidemics occurs. We examined the clonal relationships among 103 clinical and environmental Vibrio cholerae isolates belonging to O1, O139, or non-O1 non-O139 serogroups isolated during epidemic and interepidemic periods in Bangladesh and compared them with those of 51 V. cholerae isolates from four countries in Asia and Africa. These studies were done by a computer-assisted numerical analysis of the restriction endonuclease cleavage patterns of rRNA genes (ribotypes). Unweighed pair-group cluster analysis of BglI- and HindIII-generated band patterns revealed 16 clusters. Ribotypes were defined as clusters of strains possessing > 98% similarity. The results showed that 154 isolates could be differentiated into 15 different ribotypes, and strains belonging to 3 of these ribotypes (ribotypes I, V, and VIIIA and VIIIB) were isolated more frequently during the epidemic periods than during interepidemic periods in Bangladesh. Classical vibrios belonged to six different ribotypes (ribotypes I to VI), with a mean similarity coefficient of 0.84, and the El Tor vibrios belonged to five different ribotypes (ribotypes VIIIA and IX to XII), with a mean similarity coefficient of 0.82. A single clone of El Tor vibrios (ribotype XII) was resident in Tanzania, whereas Nigeria, Syria, and India shared toxigenic El Tor strains with Bangladesh. Cholera toxin (CT)-positive O139 vibrios isolated from Bangladesh and India belonged to a single ribotype (ribotype VIIIB) and were > 98% similar to one of the ribotypes of El Tor vibrios (ribotype VIIIA), but a CT-negative O139 vibrio from Argentina (ribotype XIII) was < 75% similar to the same cluster of El Tor vibrios, thus suggesting more than one possible origin for O139 vibrios. Strains belonging to the same ribotypes (ribotypes VIII to X) were isolated from both patients and surface water in Bangladesh, indicating possible transmission through surface water. A clone of a CT-positive environmental isolate of non-O1 V. cholerae (ribotype VII) was found to be closely related (76.3% similarity) to a clone of classical vibrios (ribotype I) and was only between 27.2 and 56.1% similar to clusters of El Tor, O139, and two other non-O1 nontoxigenic clones.

Molecular analysis of rRNA and cholera toxin genes carried by the new epidemic strain of toxigenic Vibrio cholerae O139 synonym Bengal.

Vibrio cholerae O139 synonym Bengal recently caused large epidemics of cholera-like disease in Bangladesh and India. We compared the restriction fragment length polymorphisms of ctxA and rRNA genes (ribotypes) in 27 isolates of V. cholerae O139 from patients in Bangladesh and India with those of 48 isolates of V. cholerae O1 from patients and 21 V. cholerae isolates from surface waters in Bangladesh, which included 2 O139 and 19 other non-O1 isolates. Ribotyping of the isolates with BglI revealed that all 29 isolates of O139 vibrios belonged to a single ribotype, suggesting a clonal nature of the infection. However, the O139 vibrios comprised two ctxA genotypes and carried three or more copies of the ctxA gene, and the chromosomal locations of these copies were unlike those of the El Tor or classical vibrios. Analysis of the restriction fragment length polymorphisms of the rRNA genes suggested that V. cholerae O139 isolates are more closely related to El Tor strains of V. cholerae O1 than were 19 other non-O1 vibrios and 33 classical V. cholerae O1 isolates that were studied. However, further studies are needed to determine whether V. cholerae O139 originated from mutations and genetic changes in a V. cholerae O1 strain or was due to the acquisition of virulence genes by a previously unknown V. cholerae non-O1 strain.

Clonal relationships among classical Vibrio cholerae O1 strains isolated between 1961 and 1992 in Bangladesh.

In Bangladesh, the replacement of classical Vibrio cholerae by the E1 Tor biotype in 1968 and the reappearance of the classical biotype and its coexistence with the E1 Tor biotype after 1982 were never adequately explained. We have analyzed 23 classical V. cholerae isolates collected between 1961 and 1968, 14 classical isolates collected between 1982 and 1992 from the capital city, Dhaka, and 6 classical V. cholerae isolates collected from two southern districts of Bangladesh and studied restriction endonuclease cleavage patterns of rRNA genes (ribotypes) to investigate the clonal relationships among the isolates. Southern blots of total DNA digested with restriction enzyme BamHI, BglI, EcoRI, HindIII, or PstI were probed, using a cloned Escherichia coli rRNA operon. While restriction enzymes BamHI, EcoRI, and PstI failed to differentiate the isolates on the basis of ribotyping, BglI and HindIII produced digestion patterns that allowed differentiation. Ribotyping the isolates with BglI and HindIII revealed five different clones (ribotypes IA, IB, IIA, IC, and IIC) of classical vibrios in Bangladesh. Strains belonging to ribotypes IA and IB were isolated in Dhaka before 1968, and one ribotype (IA) was again isolated between 1982 and 1992. Ribotype IIA was isolated in 1988 and 1989, when both clones (IA and IIA) of classical vibrios coexisted with the EI Tor vibrios. Isolates belonging to ribotypes IC and IIC were collected in the southern districts of Bangladesh and were clearly different from those collected in Dhaka between 1968 and 1992 by ribotyping analysis with BglI. These results support the previous assumption that classical vibrios were never completely replaced in Bangladesh and also demonstrate the existence of more than one genetically different clone of classical V. cholerae.

Antibiotic resistance pattern of heat-labile enterotoxin (LT) producing Escherichia coli isolated from children with diarrhoea in Bangladesh: clonal relationships among isolates with different resistant phenotypes.

Fifty-six heat-labile, enterotoxin-producing (LT+) Escherichia coli isolated from 33 children less than 5 years of age with diarrhoea were analysed for resistance to antibiotics, plasmid contents, and clonal relationships among isolates by ribosomal RNA (rRNA) fingerprinting (ribotyping). Fifty-five (98.2%) of the LT+ isolates were resistant either to tetracycline alone (48.2%) or to tetracycline and one or more other antibiotic, i.e. ampicillin, streptomycin, chloramphenicol, trimethoprim-sulfamethoxazole, or nalidixic acid. Most of the isolates harboured one or more plasmid but antibiotic resistance patterns did not always correlate with particular plasmid patterns. Ribotyping of the isolates using the restriction endonuclease EcoRI revealed a total of 7 different ribotypes, and ribotypes were shared by E. coli isolates with different antibiotic resistant phenotypes. The results indicate that in Bangladesh at least 7 different clones of LT+ E. coli acquired resistance to one or more different antibiotics in various combinations. However, a similar drug resistance pattern was not mediated by the same set of plasmids in all strains. The mechanism for the emergence and spread of antibiotic resistance among E. coli should be investigated further in Bangladesh, where LT+ E. coli is an important agent of early childhood diarrhoea.

Differentiation of Shigella flexneri strains by rRNA gene restriction patterns.

We studied the restriction endonuclease cleavage patterns of rRNA genes (ribotypes) of 72 clinical isolates of Shigella flexneri representing eight serotypes to determine whether ribotyping could be used to distinguish S. flexneri strains and to compare the discriminating ability of the method with that of serotyping. By using a cloned Escherichia coli rRNA operon as the probe, Southern blot hybridization of restriction endonuclease-digested total DNA was carried out. Ribotyping of the isolates with each of the five restriction endonucleases BamHI, EcoRI, HindIII, PstI, and SalI generated reproducible restriction patterns. However, HindIII produced the optimum digestion pattern of the rRNA genes and was more useful than the other enzymes used in differentiating strains. Analysis of the 72 isolates showed 11 different HindIII cleavage patterns of their rRNA genes. Four of these HindIII-generated ribotypes could be further differentiated into two to four subribotypes by using PstI. The results indicate that ribotyping has an application for differentiation of S. flexneri strains and can complement serotyping. Definition of strains in terms of both serotype and ribotype may be of greater use in epidemiological studies.

Enteropathogens associated with acute and persistent diarrhea in Bangladeshi children less than 5 years of age.

A longitudinal study of diarrhea was carried out from May 1988 to April 1989 by household surveillance of 705 children less than 5 years old in rural Bangladesh. Stool samples were examined for enteric pathogens at the beginning of each diarrheal episode. For persistent episodes, stool examination was repeated on days 15-17 of the illness. For each case of persistent diarrhea, stool samples from age-matched acute diarrheal and healthy controls were examined. Compared with healthy controls, cases of diarrhea were associated with Shigella species (P = .07) and rotavirus (P less than .05). Diffusely adherent Escherichia coli (P less than .05) and cryptosporidia (P = .07) were the only enteropathogens associated with persistent diarrhea in comparison with acute diarrhea. No more than 15% of children had the same class of pathogen identified from stool on both days 1-3 and days 15-17, indicating that persistent infection was uncommon. However, a different enteropathogen was frequently found on days 15-17, suggesting that sequential infection may be a cause of persistent diarrhea.

Fluorescent antibody staining test for rapid diagnosis of Shigella dysenteriae 1 infection.

An indirect fluorescent antibody test for rapid detection of Shigella dysenteriae 1 in diarrheal stools was developed. A diagnosis could be made within 90 min of submission of specimens to the laboratory. On comparison with culture results, the test had a sensitivity of 92%, a specificity of 93%, and positive and negative predictive values of 94% and 92%, respectively.

Evaluation of a DNA probe to identify enteroaggregative Escherichia coli from children with diarrhoea in Bangladesh.

Six hundred and seventy-five Escherichia coli isolates obtained from 225 diarrhoeal children less than five years of age were tested for adherence to HeLa cells and for hybridisation with DNA probes for genes conferring aggregative adherence (AggA), localised adherence (LA) and diffuse adherence (DA) to assess the usefulness of a recently developed DNA probe for AggA of enteroaggregative E. coli (EAggEC). The strains were further analysed with the DNA probes for heat--labile enterotoxin (LT), heat--stable enterotoxin (ST), Shiga--like toxins (SLT I and SLT II) and for enteroinvasiveness and adherent strains were all negative for these properties. The HeLa cell assay and DNA probe assays showed excellent agreement in identifying LA and DA positive isolates. However, significant disparities occurred in the case of AggA positive isolates, and the DNA probe failed to identify 31.9% (15 of 47) of the EAggEC identified by the HeLa cell adherence assay. The failure of the DNA probe to identify all the EAggEC indicated that there may be a high degree of genetic heterogeneity for the expression of AggA, and development of more DNA probes is necessary to detect all the possible genetic variants of EAggEC.

Rotavirus as an important cause of diarrhoea in a hospital for children in Dammam, Saudi Arabia.

A major cause of morbidity and mortality among children in developing countries is acute diarrhoea and rotavirus, a reovirus-like agent, is found to be a leading causative agent. We report here the incidence of rotavirus infection among infants and young children with gastro-enteritis treated at the Children's Hospital in Dammam, Kingdom of Saudi Arabia. One hundred and fifty children, aged between 1 and 60 months, suffering from diarrhoeal illness, were selected as the study cohort; another 150 hospitalized children of the same age group, but not suffering from diarrhoea, served as controls. Sixty-two per cent of the study children complained of watery diarrhoea and 39% of vomiting; all had mild to moderate dehydration. Seventy-two per cent had fever and 30% associated respiratory illness. Worst affected were those 7-12 months old, among whom rotavirus was detected by ELISA techniques in 39.6%, compared with 7.5% of the control children of comparable age group. We could not detect any rotavirus from control children aged over 36 months, whereas approximately 15% of the study children in this age group yielded rotavirus from the stool. The second most common pathogen isolated was Shigella ssp. It was isolated from 18.7% of children under study as against 3% of the controls.

An outbreak of nosocomial cholera in a rural Bangladesh hospital.

At a time of year when Vibrio cholerae infection accounted for over 50% of admissions to a rural Bangladeshi diarrhoea treatment centre, 29% of 48 patients hospitalized with non-cholera diarrhoea developed nosocomial V. cholerae infection. During an investigation of the 8-week outbreak, only the severity of the non-cholera diarrhoea which prompted hospital admission emerged as an important risk factor for nosocomial infection; food, intravenous solutions, oral rehydration fluid, patient attendants and hospital personnel could not be implicated as transmission sources. Patients receiving antibiotics while hospitalized did not develop nosocomial infection. Antecedent non-cholera diarrhoea may represent an important risk factor in some cases of V. cholerae infection occurring in persons who reside in cholera-endemic areas where rates of non-cholera diarrhoea are also high.

Cholera in Indonesia: epidemiologic studies of transmission in Aceh Province.

To determine the modes of transmission of cholera in the regency of Pidie, Indonesia, and to consider strategies for its control, we set up a laboratory to identify Vibrio cholerae 01 from patients with severe diarrhea in all government clinics in the regency and questioned culture-positive cases and neighborhood controls about possible exposures to V. cholerae 01. Between 12 July and 15 August 1982, 63 of 138 suspected cholera cases were confirmed by the laboratory; 53 of these patients were seen and followed up. We were unable to identify a single, indisputable mode of transmission for cholera which was amenable to immediate control. Nonetheless, a number of factors, including exposure to water from the Tiro-Sigli River and consumption of ice, were associated with disease. Other findings bring into question the value of current practices of chlorinating dugwells and disinfecting homes with Lysol during a cholera outbreak. The case-control approach to investigating the mode of transmission of cholera has distinct limitations when applied in endemic setting where there may not be a single predominant vehicle of transmission, or where the vehicle such as river water is used by all and is only periodically contaminated.

Usefulness of plasmid profiles for differentiation of Shigella isolates in Bangladesh.

We studied the plasmid profiles of 136 Shigella isolates in Bangladesh to determine whether plasmid profiles could be used for differentiation of strains for epidemiological studies. Many different plasmid patterns were observed within each species, indicating that many genetically different strains of Shigella are responsible for illness in Bangladesh.

Plasmid-borne multiple drug resistance in Vibrio cholerae serogroup O1, biotype El Tor: evidence for a point-source outbreak in Bangladesh.

In 1979, an outbreak of plasmid-borne, multiply drug-resistant Vibrio cholerae serogroup O1 biotype El Tor (V. cholerae O1) occurred in the Matlab area of Bangladesh. The outbreak could have resulted from the introduction into the area of a single resistant strain or from multiple conjugations of drug-sensitive V. cholerae O1 with C plasmids in other environmental flora. Resistant strains were phage typed to determine their relatedness, and plasmid studies were conducted to determine the frequency of C plasmids in nonvibrio flora of family contacts of cholera patients. Forty-one (85%) of 48 resistant strains of V. cholerae O1 examined belonged to two closely related phage types new to the area, whereas 59 drug-sensitive strains from the same period were primarily of two different phage types. Group C plasmids were in nonvibrio strains from five of 36 family contacts of patients with drug-resistant cholera but none of 82 family contacts of patients with sensitive cholera. This outbreak most likely began from the introduction into the area of a single, multiply drug-resistant strain of V. cholerae O1. C plasmids detected in the nonvibrio flora of family contacts probably came from the resistant strain of V. cholerae O1.

Patterns of Shigella infection in families in rural Bangladesh.

To assess the mode of transmission of Shigella infection in rural Bangladesh, questionnaire and culture surveys were conducted in baris (neighborhoods) where persons with diarrhea associated with Shigella infection and index controls with non-Shigella diarrhea lived. Nineteen percent of persons in Shigella baris and 7% of persons in control baris were infected during the survey periods (P less than 0.001). The prevalence of Shigella infection was highest for children 1-9 years of age and for females than 39 years and was not related to socioeconomic status, family size or household crowding. Use of surface water for drinking was not a risk factor for Shigella infection; in fact, use of river water was more frequent in control baris. Both household and bari contacts of Shigella index cases frequently excreted different serotypes from that excreted by the person with the index case. In Shigella baris, families with infection were significantly more likely than uninfected families to have a history of an overnight stay away from home by a family member during the previous week. These observations suggest there were multiple introductions of Shigella into some families and that the epidemiology of Shigella infection for families in rural Bangladesh differs from that observed for families living in more industrialized countries.

Longitudinal studies of infectious diseases and physical growth of children in rural Bangladesh. II. Incidence of diarrhea and association with known pathogens.

Longitudinal studies were done in two villages rural Bangladesh to learn more about the interactions between infectious diseases and the nutritional status of children. Diarrheal diseases, identified by surveillance of 197 children aged 2-60 months, were studied for bacterial, viral and parasitic enteropathogens in 1978-1979. The annual incidence of diarrhea was highest in children aged 2-11 months, and declined progressively with age from seven to four episodes per child per year. An enteropathogen was identified from rectal cultures taken during diarrhea in 51% of episodes and from 6% of monthly cultures taken when diarrhea was not present. Enterotoxigenic Escherichia coli were the pathogens found most frequently, followed by shigellae and rotaviruses. Diarrheal episodes associated with shigellae had the longest duration, while episodes associated with Vibrio cholerae or with rotavirus were more frequently associated with dehydration. E. coli diarrhea had a peak incidence during the hot months, and shigellosis was more frequent during the cool, dry months.