- Email: [email protected]
Prospective study of enteropathogens in children with diarrhea in Houston and Mexico
September 1978
ThelournalofPEDIATRICS
383
Prospective study of enteropathogens in children with diarrhea in Houston and Mexico During a 22-month period, 595 children with diarrhea and 210 age-matched controls attending clinics in Houston (367 children) and Mexico (438) were prospectively evaluated for enteric pathogens.
Enteropathogens associated with disease were Shigella (18%), rotavirus (14%), Salmonella (9%), toxigenic Escherichia coli (6%), and others (12%), Including 14 Proteus isolates that caused rounding of adrenal cells. Enteropathogens were isolated from a greater (P < 0.001) number of children with diarrhea (59%) than from asymptomatic controls (6%). Paired sera tested for antibody to heat-labile toxin of~. coli rarely demonstrated a fourfold rise during episodes of diarrhea. This study demonstrates: (1) more striking Illness in children from Mexico; (2) more common occurrence of Shigella in HOI/stan, and of rotavirus and Salmonella in Mexico; (3) lack of seasonal occurrence of rotavirus isolation in either population and a summertime occurrence of Shigella in Houston; (4) lack of toxigenic E. coli isolation in endemic diarrhea of either population; and (5) a significant (P < 0.001) age-related acquisition of E. coli L T antibodies.
Larry K. Pickering, M.D.,· Doyle J. Evans, Jr., Ph.D., Onofre Mufioz, M.D., Herbert L. DuPont, M.D., Pedro Coello-Ramirez, M.D., John J. Vollet, Ph.D., Richard H. Conklin, M.D., Jorge Olarte, D.Se., and Steve Kohl, M.D., Houston, Texas, and Mexico City, Mexico
TEN YEARS AGO enteropathogens could be isolated from less than 20% of children with diarrhea." 2 Application of the electron microscope has shown that rotavirus accounts for a major portion of diarrheal illness in early childhoodv'" and has been associated with diarrhea in adults.v 11 Knowledge of the mechanisms of disease production by toxigenic Escherichia coli have enabled From the Program ill Infectious Diseases and Clinical Microbiology and the Department of Pediatrics, The University of Texas Medical School at H cusron, and Hospital de Pediatrta del Institute Mexicano del Segura SOCial. Hospital Infantil de Mexico. Supported by grant Al 12699from the National Institutes of Health. Presented in part at the Interscience Conference on Antimicrobial Agents and Chemotherapy, New York, N. Y., October, 1977. "Reprint address: Program ill Infectious Diseases and Clin., Micro.. The University of Texas Medical School, 6431 Fannin, 228 Freeman Bldg.• Houston, TX 77030.
0022-3476/78/0393-0383$00.60/0
e
1978 The C. V. Mosby Co.
enterotoxin-producing E. coli to be recognized as important causes of diarrheal illness in children .2-18 and adults.t"" We conducted a two-year prospective evaluation of enteropathogens in children with diarrhea and compared the findings with those in concomitantly studied age-matched children without diarrhea in Houston and Mexico. Abbreviations used heat-labile toxin LT: ST: heat-stable toxin PIH:
passive immune hemolysis
MATERIALS AND METHODS Populations studied. The study group consisted of 805 children, of whom 595 had diarrhea and 210 were asymptomatic controls matched by age and season. The children attended Hermann Hospital and San Jose outpatient pediatric clinics in Houston, both under the direction of the Department of Pediatrics of The University of
Vol. 93, No.3, pp. 383-388
384
Pickering et al.
The Journal of Pediatrics September 1978
Table I. Age, sex, and race of a pediatric population in Houston and Mexico evaluated for diarrhea, 1976-1977 Houston Sick
Well
Mexico
Sick
Well
No·1 % No·1 % No·1 % No·1 %
Age (yr) <1
43 69 62 274 71 28 36 32 45 50 20 7 6 19 23 9 2
III
1-2 3-5 >5
8l 51 13 20 6 21 6
52 20 21 6
255 100 112 100 340 100 98 100 Sex
Female Male Race or nationality White Black Latin American
44 57 51 146 56 55 49 194
43 37 57 61
51 20 9 8 0 74 29 42 38 1 130 51 61 54 339
0 1 97
Il2
143
38 62
Table n. Clinical illness of children with diarrhea in Houston and Mexico, 1976-1977 Houston
Mexico
No.
%
No.
%
125 74 56
49 29 22
199 104 37
59 31 10
255
100
340
100
No. stools in preceding 24 hr 2-4 88 5-8 107 9-12 44 12+ 16
35 42
11
169 103 57
3 50 30 17
340
100
Length of illness (days) 1-2 3-4 5-7
255
17
-6 100
-
Texas Medical School at Houston (367 children), or the outpatient pediatric clinic at the Social Security Hospital, Mexico City (438children) between December, 1975, and September, 1977. Children selected for study had liquid stools at the time of evaluation and reportedly passed twice the number of stools usually passed within the preceding 24 hours. Further requirements were: (1) diarrhea present for at least 24 hours, but not greater than seven days prior to evaluation and (2) no antibiotic therapy within the preceding week. Each child had a complete physical examination and had stool specimens and a serum sample obtained. A follow-up evaluation
occurred 10 to 21 days later, at which time a repeat physical examination was performed and two serum samples obtained. Informed consent was obtained from parents of children entered into the study. Processing of specimens. All unformed stools were examined for leukocytes, employing the methylene blue wet mount preparation." For the remaining laboratory studies, stools from controls were processed in the same manner as those from children with diarrhea, including electron microscopy and toxin detection assays. For bacteriologic examination, fresh stool specimens were plated onto Tergitol 7, MacConkey, Salmonella-Shigella, and brilliant green agar media. Shigella and Salmonella species were identified according to standard criteria, and at least five isolates of E. coli and any other gram-negative bacilli present in moderate-to-heavy numbers were obtained from each specimen, irrespective of the presence of known bacterial pathogens. Stock cultures of bacteria were maintained on peptone agar slants and transported to the central laboratory in Houston for complete processing within cine month of isolation. Stool examination for parasites was performed on fresh specimens by the zinc sulfate flotation method and also by the formalin ether extraction method. For Entamoeba histolytica to be classified as a pathogen, trophozoites had. to be seen, whereas Giardia lamblia cysts or trophozoites were accepted as possible causative agents. To assay for heat-stable or heat-labile enterotoxins, the gram-negative bacilli (usually E. coli) were grown in 8 ml of casamino acid yeast extract in a 50 ml flask and incubated for I g hours in an incubator shaker. Sterile culture supernatants of all isolates from each specimen were tested for LT, using the y-l adrenal cell assay. 23 Filtrates causing rounding of adrenal cells were verified as LT positive by neutralization with anti-L'I' serum.'·'o" Strains also were tested for ST production in the suckling mouse model." Strains were considered ST positive when individual filtrates produced a gut to remaining body weight ratio of 0.09 or higher." E. coli isolales also were assayed for invasiveness by the Sereny guinea pig eye modelv in the first 375 patients enrolled in the study. Humoral anti-LT antibody was determined by the passive immune hemolysis assay." Fresh sheep red blood cells were sensitized by incubation with a standard amount of purified LT. These cells were hemolyzed by anti-L T antibody in a complement-mediated reaction. Antibody titers were determined by the micro titer technique, employing a standard rabbit anti-LT serum as control. A fourfold rise in titer was considered significant. The presence of virus particles was determined by direct electron microscopy examination of the stool spec-
Enteropathogens in children with diarrhea
Volume 93 Number 3
imens, using the pseudo-replica technique." Negativestained grids with a Formvar coat were examined for the presence of adenovirus and human rota virus, which appeared as a 70 nrn diameter particle with morphologic characteristics described by Bishop and associates.' Statistical analysis. All clinical and laboratory data from each patient enrolled in the study were placed on data sheets, which were prepared for computer an alysis. Data were processed at The University of Texas School of Public Health on a CDC Cyber 73 system computer. The mean, standard deviation, standard error of the mean, and range for each value tested were computed. Significant differences between the study groups for each of the values tested were assessed by analysis of variance and then by the appropriate t test applied to log transformed data. Chi square analysis with Yates correction and linear regression analysis also were utilized.
RESULTS Table I presents the age, sex, and race of the study group in Houston and Mexico. They represent the typical pediatric patient seen at the study clinics during the time interval covered by the study. Seventy-one percent of children with diarrhea from Houston and 94% of the ill children from Mexico were less than two years of age. The majority of children with diarrhea from Houston and Mexico were male. All children evaluated from Houston had not traveled to Mexico. Table II presents the duration of illness prior to evaluation and the number of stools in the 24 hours preceding evaluation. Seventy-eight percent of the children from Houston and 90% of the children from Mexico had been ill for four days or less. Although all patients entered into the study had passed liquid stools more frequently than normal, the number of stools in the preceding 24 hours in the Mexican children was significantly (P < 0.001) greater than in those from Houston. Detection of enteropathogens. Table III identifies the potential enteropathogens which were detected in the stools of 353 (59%) of the 595 pat ients with diarrhea . The remaining 242 children (41) passed unformed stools which were negative for known enteric pathogens. Shigella was the majo r enteropathoge n associated with disease in children from Houston, and was isolated from a significantly (P < 0.005) greater number of children from Houston than from Mexico. Shigella serogroups isolated were S. ftexneri-5o strains (45%), S. sonnei-54 strains (49%), and S. boydii-6 strains (6%). Rotavirus was the major pathogen associated with diarrhea in children from Mexico, and was isolated from a significantly (P < 0.025) greater number of Mexican than Houston children. Salmonella was isolated from a significantly (P < 0.005)
385
Table m. Enteropathogens isolated from children with diarrhea in Houston and Mexico , 1976-1977
Shigella
H5ib5'"
Rotaviris
10
Salmonella Toxigenic E. coli
No.
%
No.
%
64 26
25
14
II
4
46 58 41 24
10
LT ST LT/ST G.lamblia
Adrenal cell positive Proteus
4
5
8 12
2 3 17
7
2
I 3 3
Ascaris
8
Adenovirus Invasive E. coli' E. histolytica
8 I 0
Adrenal cell positive Aero-
I
[7
12 7
4 6 12 5 3
1
4 I
1 I
2 7 0
2
monas
Etiology unknown
107
42
135
40
Total
255
100
340
100
'Done on initial 375 patients.
greater number of Mexican than Houston children. The serogroups of Salmonella isolated were B = 13 strains, C. = 9 strains, C. = 10strains, E = 2 strains, S. washington = 2 strains, S. poona = 14strains, S. derby = 1 strain, and S. heidelberg := I strain. Sterile culture supernatants from each of the non E. coli isolates (Proteus, Aeromonas) produced a positive adrenal cell reaction which was not neutralized by heating at 100°C for 30 minutes, or by cholera or E. coli LT antisera. Multiple enteropathogens were found simultaneously in stool of seven children with diarrhea in Houston and 16 from Mexico. The combinations reflected the prevalent enteropathogens in each location. Rotavirus and Salmonella were each isolated from stool of 11 of the Mexican children with multiple enteropathogens. In children from Houston, isolation of Shigella was significantly (P < 0.(01) greater in the summer months. Isolation of rotavirus in children from Houston and Mexico occurred without significant seasonal distribution. Rota virus was found more freq uently during the first year in Houston; there was no difference in rotavirus isolation between the first and second year of the study in children from Mexico . Sixteen patients (13 ill, 3 control) from whom toxigenic E. coli was isolated from stool had a second serum sample obtained from ten days to four weeks after the initial sample. Only one infant had a fourfold rise in E. coli anti·LT antibody. Controls. During the period of this study, 210 asympto-
386
Pickering et at.
The Journal of Pediatrics September 1978
Table IV. Clinical characterization of children with diarrhea from Houston and Mexico by causative agents, 1916-1977 Rotavirus
Shigella
H
No. Age (mo) Symptoms Vomiting (%) Fever (%) Prior duration ofdi·
I
I
No. patients with 110 enteropathogens isolated
Salmonella
I
I
M
H
II 21.7 ± 4.6
41 8.3 ± I.7
12.0 ± 1.1
48/58 (82)
5111 (45)
19/41 (46)
50/107 (47)
103/135 (76)
10126 (38) 2.6 ± 0.6
30158 (52)
2.6 ± 0.2
3111 (29) 3.0 ± 0.9
21141 (50) 2.3 ± 0.3
43/107 (40) 3.2 ± 0.3
781135 (58) 2.5 ± 0.1
4.8 ± 0.5
9.3 ± 0.7
5.9 ± 1.1
10.1 ± 0.9
6.4 ± 0.5
9.2 ± 0.4
M
H
26 18.7 ± 7
58 10.1 ± 2.2
30/46 (65)
16126 (62)
2.5 ± 0.3
30/46 (65) 2.4 ± 0.2
7.2 ± 0.5
10 ± 0.8
M
H
64 33 ± 3.6
46 19.2 ± 3.2
30/64 (47) 41/64 (46)
107
M 135 7.3 ± 0.8
arrhea (days) No. stools preceding 24 hr
H - Houston; M = Mexico; Mean ± SEM.
matic control children (112 from Houston, 98 from Mexico) were matched by age, season, and geographic location. Fourteen (6%) asymptomatic children had the following enteropathogens isolated: rotavirus-6 (Mexico 3, Houston 3), iT-E. coli-5 (Mexico 3, Houston 2), ST-E. coli-l (Mexico), and Giardia-l (Houston). Enteropathogens were isolated from a significantly (P < 0.001) fewer number of children without diarrhea compared to those with illness. No control patient had multiple enteropathogens isolated. Clinical characteristics of children with diarrhea by presence of enteropathogens. Table IV gives the clinical characteristics of children with diarrhea from Houston and Mexico in relation to the major pathogens detected. Children from Mexico with diarrhea were significantly (P < 0.001) younger and had more intense diarrhea, as determined by the number of stools and greater frequency of vomiting and fever, than those from Houston, regardless of the causative agent. Rotavirus was isolated from significantly younger children in Mexico (P < 0.025) and Houston (P < 0.05) than children from the same location from whom Shigella was isolated. Vomiting was more common in children from whom rotavirus was isolated and less frequent in those from whom Salmonella was identified. Fever occurred more commonly in children from whom Shigella was isolated. There was no significant difference in the number of days ill between the Mexican and Houston children in any group when divided by causative agent. E. coli heat-labile toxin antibody titers. The Figure depicts linear regression analysis of E. coli LT antibody titers in children from Houston and Mexico by age. Sera were obtained for baseline E. coli LT antibody titers at the
time the children were enrolled in the study. There was no significant difference in antibody titers in children from Houston and Mexico in any age group. An age-related acquisition of L T antibodies is demonstrated in Mexico (r = 0.536, P < 0.001) and Houston (r = 0.594, P < 0.001).
DISCUSSION The finding of multiple enteropathogens has been reported previously in adults' and children" from Mexico, but is an uncommon occurrence in the United States. The previously reported finding that the isolation rate of enteropathogens is highest in patients with more intense diarrhea was not substantiated by this study.': 18 The children from Mexicoin general had more striking illness, yet the isolationrate of enteropathogens in this group was no greater than that found in children from Houston. The occurrence of diarrhea in younger children in Mexico may be due to an earlier exposure to numerous enteropathogens. Shigella was isolated from older infants in both locations and generally caused fever. Rotavirus occurred in younger infants and vomiting was a more frequent manifestation of illnessin this group, as has been reported." 7 Salmonella strains and rotavirus were more common in children' with diarrhea in Mexico; Shigella was a more important problem in Houston. These differences may reflect widespread and indiscriminate use of antimicrobials among the general population in Mexico, although the children studied had not received antibiotic therapy within the week prior to evaluation. Antimicrobials may serve as a selection factor for Salmonella at the expense of the more drug sensitive Shigella strains. Rotavirus infection characteristically has occurred
Volume 93 Number 3
during winter months." •. • There are two possible explanations for the lack of seasonal occurrence in our study. We evaluated populations in tropical and semi-tropical areas and most of the children were outpatients. Nearly all previously published studies have been conducted in temperate climates, and hospitalized children with diarrhea were enrolled. A study in the semi-tropical climate of Taiwan also revealed that rotavirus was a common summertime agent, occurring in 56% of children with diarrhea. ,0 Toxigenic E, coli has been shown to be of major importance to U.S. adults traveling to Mexico, and to the animal industry as a cause of scours among the newborns of swine and cattle. Our study agrees with a recent publication in Canada" suggesting that toxigenic E. coli is not commonly associated with endemic diarrhea of children in North America. The small number of toxigenic E. coli isolated from stools of Mexican children in our study is in contrast to two previous studies in which enterotoxigenic strains were detected in 16 and 47% of Mexican children with diarrhea.": IS Several differences in these studies may account for this discrepancy. We evaluated outpatient children over 21 months; other studies evaluated hospitalized children over a shorter time interval. In addition, the population evaluated at the Social Security Hospital is primarily a middle socio-economic class, compared to an indigent population previously studied.": IS Patients with diarrhea due to toxigenic E. coli often develop a rise in serum antibodies to the heat-labile enterotoxin:" 31 Failure to demonstrate a fourfold rise in LT antibodies during episodes of diarrhea among children in the present study supports our stool culture data showing the relative unimportance of toxigenic E. coli as a cause of diarrhea, We did, however, demonstrate that the titers of specific antitoxin antibody in children are related to age; the linear increase in LT-E. coli titers indicates a continual exposure to this or a related antigen with subclinical or mild infection. Other studies have shown that serum antibodies to rotavirus increase steadily from six months to two years of age," 9. " The prevalence of LT-E. coli and rotavirus antibodies at two years of age is comparable to the serum antibody level in adults." 9." Klebsiella and Aeromonas have been incriminated as enterotoxin-producing organisms in children with diarrhea." 13 We detected one Aeromonas and 14 Proteus strains which produced rounding of cells in the Y-l adrenal cell assay. The significance of these strains and the nature of the substance produced are unknown. Gurwith and Williams" evaluated 49 Proteus isolates from children with diarrhea; none caused morphologic changes in the mouse adrenal tumor cell assay.
Enteropathogens in children with diarrhea
387
2.2
1.8
•••• Mexico
-Houston 0.6
The cause of diarrhea eluded us in 40% of children with diarrhea. These cases may be due to virulence characteristics other than enterotoxin production, such as colonization factor'S (which allows adherence of bacteria to the intestinal wall), enterotoxins which are not detected by presently employed methods, or enteric viruses which were not identified. We must now characterize the epidemiology (transmission dynamics and reservoirs) of the known important enteropathogens and intensify the search for other causative agents in patients with diarrhea. We are indebted to Bart Hsi, Ph.D., for biostatistica1 support; P. Byers, P. Gearhart, C. Geib, S. Hamilton, D. Johnson, r. Rutherford, H. West, and D. Satterwhite for technical help; and 1. Addison for assistance in the preparation of the.manuscript, REFERENCES
1. Yow MD, Melnick JL, Blattner RJ, et al: The association of viruses and bacteria with infantile diarrhea, Am J Epide2.
3.
4.
5.
6.
mioI92:33, 1970. Cramblett RO, and Siewers CMF: The etiology of gastroenteritis in infan Is and children, with emphasis on the occurrence ofsimultaneous mixed viral-bacterial infections, Pediatrics 35:885, 1965. Bishop RF, Davidson OP, Holmes JH, et al: Virus particles in epithelial cells of duodenal mucosa from children with acute nonbacterial gastroenteritis, Lancet 2:1281, 1973. Ryder RW, Sack DA, Kapikian AZ, et al: Enterotoxigenic Escherichia coli and reovirus-like agent in rural Bangladesh, Lancet 1:659, 1976. Kapikian AZ, Kim HW, Wyatt RO, et al: Human reoviruslike agent as the major pathogen associated with "winter" gastroenteritis in hospitalized infants and young children, N Engl J Med 294:965, 1976. Konno T, Suzuki H, Imai A, et al: Reovirus-like agent in acute epidemic gastroenteritis in Japanese infants: fecal shedding and serologic response, J Infect' Dis 135:259, 1977.
388
7.
8.
9.
10.
11.
12.
n. 14.
IS.
16.
17.
18.
19.
20.
The Journ al of Pediatrics September 1978
Pickering et of.
Rodriguez WJ, Kim HW, Arro bio JO, Brandt CD, Chanock RM, Kapikian AZ, Wyatt RO , and Parro tt RH: Clinical features of acute gastroenteritis associated with hum an reoviru s-lik e agent in infants and young children, J PEDlATR 91: 188, 1977. Gurwith MI , and Williams TW: Ga stroenteritis in children : a two-year review in Manitoba. I. Etiology, J Infect Dis 136:239, 1977. Gomez-Barreto J, Palmer EL, Nahmias AJ, et al: Acute enteritis ass ociated with reovirus-like agents, JAMA 235:1857,1976. Echeverria P, Ho MT, and B1acklow NR: Relative importance of virus and bacteria in the etiology of pediatric diarrhea in Taiwan, J Infect Dis 136:383, 1977. Bolivar R, Conklin RH, Vollet J. et al: Rotavirus in travelers' diarrhea: Study of an adult student population in Mexico, J Infect Dis 137:324, 1978. Gorbach SL, and Kurana CM : Toxigenic Escherichia coli: A cause of infanlile diarrhea in Chicago, N Engl J Med 287:791, 1972. Guerrant RL, Mo ore RA, Kirschenfeld PM, et al: Role of toxigenic and invasive bacteria in ac ute diarrhea of childhood, N Engl J Med 293:567, 1975. Sack RB, H irsch horn N, Brownlee et al: Enterotoxigenic Escherichia coli-associated d iarrheal disease in Apache children, N Engl J Med 292:1041, 1975. Nalin DR, Raharnan M, McL augh lin JC, et al: Enterotoxigenic Escherichia col i and idiopathic diarrhea in Bangladesh , Lancet 2:1116 , 1975. Ryder RW, Wachsmuth IK, Buxton AE, et al: Infantile diarrhea produced by he at-stable enterotoxigenic Escherichia coli, N Engl J Med 295:849, 1976. Donta ST, Wallace RD, Whipp SC. et al: Enterotoxigenic Escherichia coli and diarrheal disease in Mexican child ren , J Infect Dis 135:482 , 1977. Evans DO, Olarte J, DuPont HL, et al: Enteropathogens associated with pediatric diarrhea in Mexico City, J PEDIATR 91:65, 1977. Gorbach SL, Kean BR, Evans DG, et al: Travelers' diarrhea and toxigenic Escherichia coli, N Engl J Med 292:933, 1975. Merson MH, Morris GK, Sack DA, et al: Travelers' diarrhea in Mexico: A pr ospective study, N Engl J Med 294:1299, 1976.
21.
22. 23.
24.
25.
26.
27.
I:
28.
29.
30.
31.
32.
33.
DuPont HL, Olarte J, Evans DG, et al : Comparative susceptibility of Latin American and United States students to enteric pathogens, N Engl J Med 295:1520, 1976. Pickering LK, DuPont HL , Olarte J , et a1: Fecal leukocytes in enteric infections, Am J e lin Pathol 68:562, 1977. Donta ST, Moon HW , and Whipp SC: Detection of heat-labile Escherichia coli enterotoxin with the use of adrenal cells in tissue culture, Science 183:334, 1974. Donta ST, and Smith OM: Stimulation of steroidogenesis in tissue culture by enterotoxigenic Escherichia coli a nd its neutralization by specific an tiserum, Infect Immun 9:500. 1974. Evans OJ Jr, Evans DO, and Gorbach SL: Production of vascular permeability factor by enterotoxigenic Escherichia coli isolated from man, Infect Immun 8:725, 1973. Dean AG, Ching Y, Williams RG, et al: Tests for Escherichia coli enterotoxin using infant mice: application in a study of diarrhea in children in Honolulu, J Infect Dis 125:407, 1972. Byers PA, DuPont HL, and Evans OJ Jr: A pooling method for screening clinical isolates of Escherichia coli for the production of heat-stable enterotoxin, Annual Meeting of the American Society for Microbiology, New Orleans, May 8-13 , 1977, P 225. Sereny, B.: Experimental shigella keratoconjunctivitis: A preliminary report, Acta Microbiol Acad Sci Hung 2:293, 1955. Evans OJ , Ruiz-Palacios G, Evans DO, et al: Humoral immune response to the heat-labile enterotoxin of Escherichia coli in naturally acquired diarrhea and antitoxin determination by p assive immune hemolysis, Infect Immun 16:781,1977. Portnoy BL, Conklin RH , Menn M , et al: Reliable identification of reovirus-like agent in diarrheal stools, J Lab Clin Med 89:560, 1977. Sack RB, Jacobs B, and Mitra R : Antitoxin responses to infections with enterotoxigenic Escherichia coli, J Infect Dis 129:330, 1974. Blacklow NR, Echeverria P, and Smith DH: Serologic studies with reovirus-like enteritis agent, Infect Immun 13:1563, 1976, Evans DG, Evans DJ Jr, and DuPont HL: Virulence factors of enterotoxigenic Escherichia coli, J Infect Dis 136:S 118, 1977.
ThelournalofPEDIATRICS
383
Prospective study of enteropathogens in children with diarrhea in Houston and Mexico During a 22-month period, 595 children with diarrhea and 210 age-matched controls attending clinics in Houston (367 children) and Mexico (438) were prospectively evaluated for enteric pathogens.
Enteropathogens associated with disease were Shigella (18%), rotavirus (14%), Salmonella (9%), toxigenic Escherichia coli (6%), and others (12%), Including 14 Proteus isolates that caused rounding of adrenal cells. Enteropathogens were isolated from a greater (P < 0.001) number of children with diarrhea (59%) than from asymptomatic controls (6%). Paired sera tested for antibody to heat-labile toxin of~. coli rarely demonstrated a fourfold rise during episodes of diarrhea. This study demonstrates: (1) more striking Illness in children from Mexico; (2) more common occurrence of Shigella in HOI/stan, and of rotavirus and Salmonella in Mexico; (3) lack of seasonal occurrence of rotavirus isolation in either population and a summertime occurrence of Shigella in Houston; (4) lack of toxigenic E. coli isolation in endemic diarrhea of either population; and (5) a significant (P < 0.001) age-related acquisition of E. coli L T antibodies.
Larry K. Pickering, M.D.,· Doyle J. Evans, Jr., Ph.D., Onofre Mufioz, M.D., Herbert L. DuPont, M.D., Pedro Coello-Ramirez, M.D., John J. Vollet, Ph.D., Richard H. Conklin, M.D., Jorge Olarte, D.Se., and Steve Kohl, M.D., Houston, Texas, and Mexico City, Mexico
TEN YEARS AGO enteropathogens could be isolated from less than 20% of children with diarrhea." 2 Application of the electron microscope has shown that rotavirus accounts for a major portion of diarrheal illness in early childhoodv'" and has been associated with diarrhea in adults.v 11 Knowledge of the mechanisms of disease production by toxigenic Escherichia coli have enabled From the Program ill Infectious Diseases and Clinical Microbiology and the Department of Pediatrics, The University of Texas Medical School at H cusron, and Hospital de Pediatrta del Institute Mexicano del Segura SOCial. Hospital Infantil de Mexico. Supported by grant Al 12699from the National Institutes of Health. Presented in part at the Interscience Conference on Antimicrobial Agents and Chemotherapy, New York, N. Y., October, 1977. "Reprint address: Program ill Infectious Diseases and Clin., Micro.. The University of Texas Medical School, 6431 Fannin, 228 Freeman Bldg.• Houston, TX 77030.
0022-3476/78/0393-0383$00.60/0
e
1978 The C. V. Mosby Co.
enterotoxin-producing E. coli to be recognized as important causes of diarrheal illness in children .2-18 and adults.t"" We conducted a two-year prospective evaluation of enteropathogens in children with diarrhea and compared the findings with those in concomitantly studied age-matched children without diarrhea in Houston and Mexico. Abbreviations used heat-labile toxin LT: ST: heat-stable toxin PIH:
passive immune hemolysis
MATERIALS AND METHODS Populations studied. The study group consisted of 805 children, of whom 595 had diarrhea and 210 were asymptomatic controls matched by age and season. The children attended Hermann Hospital and San Jose outpatient pediatric clinics in Houston, both under the direction of the Department of Pediatrics of The University of
Vol. 93, No.3, pp. 383-388
384
Pickering et al.
The Journal of Pediatrics September 1978
Table I. Age, sex, and race of a pediatric population in Houston and Mexico evaluated for diarrhea, 1976-1977 Houston Sick
Well
Mexico
Sick
Well
No·1 % No·1 % No·1 % No·1 %
Age (yr) <1
43 69 62 274 71 28 36 32 45 50 20 7 6 19 23 9 2
III
1-2 3-5 >5
8l 51 13 20 6 21 6
52 20 21 6
255 100 112 100 340 100 98 100 Sex
Female Male Race or nationality White Black Latin American
44 57 51 146 56 55 49 194
43 37 57 61
51 20 9 8 0 74 29 42 38 1 130 51 61 54 339
0 1 97
Il2
143
38 62
Table n. Clinical illness of children with diarrhea in Houston and Mexico, 1976-1977 Houston
Mexico
No.
%
No.
%
125 74 56
49 29 22
199 104 37
59 31 10
255
100
340
100
No. stools in preceding 24 hr 2-4 88 5-8 107 9-12 44 12+ 16
35 42
11
169 103 57
3 50 30 17
340
100
Length of illness (days) 1-2 3-4 5-7
255
17
-6 100
-
Texas Medical School at Houston (367 children), or the outpatient pediatric clinic at the Social Security Hospital, Mexico City (438children) between December, 1975, and September, 1977. Children selected for study had liquid stools at the time of evaluation and reportedly passed twice the number of stools usually passed within the preceding 24 hours. Further requirements were: (1) diarrhea present for at least 24 hours, but not greater than seven days prior to evaluation and (2) no antibiotic therapy within the preceding week. Each child had a complete physical examination and had stool specimens and a serum sample obtained. A follow-up evaluation
occurred 10 to 21 days later, at which time a repeat physical examination was performed and two serum samples obtained. Informed consent was obtained from parents of children entered into the study. Processing of specimens. All unformed stools were examined for leukocytes, employing the methylene blue wet mount preparation." For the remaining laboratory studies, stools from controls were processed in the same manner as those from children with diarrhea, including electron microscopy and toxin detection assays. For bacteriologic examination, fresh stool specimens were plated onto Tergitol 7, MacConkey, Salmonella-Shigella, and brilliant green agar media. Shigella and Salmonella species were identified according to standard criteria, and at least five isolates of E. coli and any other gram-negative bacilli present in moderate-to-heavy numbers were obtained from each specimen, irrespective of the presence of known bacterial pathogens. Stock cultures of bacteria were maintained on peptone agar slants and transported to the central laboratory in Houston for complete processing within cine month of isolation. Stool examination for parasites was performed on fresh specimens by the zinc sulfate flotation method and also by the formalin ether extraction method. For Entamoeba histolytica to be classified as a pathogen, trophozoites had. to be seen, whereas Giardia lamblia cysts or trophozoites were accepted as possible causative agents. To assay for heat-stable or heat-labile enterotoxins, the gram-negative bacilli (usually E. coli) were grown in 8 ml of casamino acid yeast extract in a 50 ml flask and incubated for I g hours in an incubator shaker. Sterile culture supernatants of all isolates from each specimen were tested for LT, using the y-l adrenal cell assay. 23 Filtrates causing rounding of adrenal cells were verified as LT positive by neutralization with anti-L'I' serum.'·'o" Strains also were tested for ST production in the suckling mouse model." Strains were considered ST positive when individual filtrates produced a gut to remaining body weight ratio of 0.09 or higher." E. coli isolales also were assayed for invasiveness by the Sereny guinea pig eye modelv in the first 375 patients enrolled in the study. Humoral anti-LT antibody was determined by the passive immune hemolysis assay." Fresh sheep red blood cells were sensitized by incubation with a standard amount of purified LT. These cells were hemolyzed by anti-L T antibody in a complement-mediated reaction. Antibody titers were determined by the micro titer technique, employing a standard rabbit anti-LT serum as control. A fourfold rise in titer was considered significant. The presence of virus particles was determined by direct electron microscopy examination of the stool spec-
Enteropathogens in children with diarrhea
Volume 93 Number 3
imens, using the pseudo-replica technique." Negativestained grids with a Formvar coat were examined for the presence of adenovirus and human rota virus, which appeared as a 70 nrn diameter particle with morphologic characteristics described by Bishop and associates.' Statistical analysis. All clinical and laboratory data from each patient enrolled in the study were placed on data sheets, which were prepared for computer an alysis. Data were processed at The University of Texas School of Public Health on a CDC Cyber 73 system computer. The mean, standard deviation, standard error of the mean, and range for each value tested were computed. Significant differences between the study groups for each of the values tested were assessed by analysis of variance and then by the appropriate t test applied to log transformed data. Chi square analysis with Yates correction and linear regression analysis also were utilized.
RESULTS Table I presents the age, sex, and race of the study group in Houston and Mexico. They represent the typical pediatric patient seen at the study clinics during the time interval covered by the study. Seventy-one percent of children with diarrhea from Houston and 94% of the ill children from Mexico were less than two years of age. The majority of children with diarrhea from Houston and Mexico were male. All children evaluated from Houston had not traveled to Mexico. Table II presents the duration of illness prior to evaluation and the number of stools in the 24 hours preceding evaluation. Seventy-eight percent of the children from Houston and 90% of the children from Mexico had been ill for four days or less. Although all patients entered into the study had passed liquid stools more frequently than normal, the number of stools in the preceding 24 hours in the Mexican children was significantly (P < 0.001) greater than in those from Houston. Detection of enteropathogens. Table III identifies the potential enteropathogens which were detected in the stools of 353 (59%) of the 595 pat ients with diarrhea . The remaining 242 children (41) passed unformed stools which were negative for known enteric pathogens. Shigella was the majo r enteropathoge n associated with disease in children from Houston, and was isolated from a significantly (P < 0.005) greater number of children from Houston than from Mexico. Shigella serogroups isolated were S. ftexneri-5o strains (45%), S. sonnei-54 strains (49%), and S. boydii-6 strains (6%). Rotavirus was the major pathogen associated with diarrhea in children from Mexico, and was isolated from a significantly (P < 0.025) greater number of Mexican than Houston children. Salmonella was isolated from a significantly (P < 0.005)
385
Table m. Enteropathogens isolated from children with diarrhea in Houston and Mexico , 1976-1977
Shigella
H5ib5'"
Rotaviris
10
Salmonella Toxigenic E. coli
No.
%
No.
%
64 26
25
14
II
4
46 58 41 24
10
LT ST LT/ST G.lamblia
Adrenal cell positive Proteus
4
5
8 12
2 3 17
7
2
I 3 3
Ascaris
8
Adenovirus Invasive E. coli' E. histolytica
8 I 0
Adrenal cell positive Aero-
I
[7
12 7
4 6 12 5 3
1
4 I
1 I
2 7 0
2
monas
Etiology unknown
107
42
135
40
Total
255
100
340
100
'Done on initial 375 patients.
greater number of Mexican than Houston children. The serogroups of Salmonella isolated were B = 13 strains, C. = 9 strains, C. = 10strains, E = 2 strains, S. washington = 2 strains, S. poona = 14strains, S. derby = 1 strain, and S. heidelberg := I strain. Sterile culture supernatants from each of the non E. coli isolates (Proteus, Aeromonas) produced a positive adrenal cell reaction which was not neutralized by heating at 100°C for 30 minutes, or by cholera or E. coli LT antisera. Multiple enteropathogens were found simultaneously in stool of seven children with diarrhea in Houston and 16 from Mexico. The combinations reflected the prevalent enteropathogens in each location. Rotavirus and Salmonella were each isolated from stool of 11 of the Mexican children with multiple enteropathogens. In children from Houston, isolation of Shigella was significantly (P < 0.(01) greater in the summer months. Isolation of rotavirus in children from Houston and Mexico occurred without significant seasonal distribution. Rota virus was found more freq uently during the first year in Houston; there was no difference in rotavirus isolation between the first and second year of the study in children from Mexico . Sixteen patients (13 ill, 3 control) from whom toxigenic E. coli was isolated from stool had a second serum sample obtained from ten days to four weeks after the initial sample. Only one infant had a fourfold rise in E. coli anti·LT antibody. Controls. During the period of this study, 210 asympto-
386
Pickering et at.
The Journal of Pediatrics September 1978
Table IV. Clinical characterization of children with diarrhea from Houston and Mexico by causative agents, 1916-1977 Rotavirus
Shigella
H
No. Age (mo) Symptoms Vomiting (%) Fever (%) Prior duration ofdi·
I
I
No. patients with 110 enteropathogens isolated
Salmonella
I
I
M
H
II 21.7 ± 4.6
41 8.3 ± I.7
12.0 ± 1.1
48/58 (82)
5111 (45)
19/41 (46)
50/107 (47)
103/135 (76)
10126 (38) 2.6 ± 0.6
30158 (52)
2.6 ± 0.2
3111 (29) 3.0 ± 0.9
21141 (50) 2.3 ± 0.3
43/107 (40) 3.2 ± 0.3
781135 (58) 2.5 ± 0.1
4.8 ± 0.5
9.3 ± 0.7
5.9 ± 1.1
10.1 ± 0.9
6.4 ± 0.5
9.2 ± 0.4
M
H
26 18.7 ± 7
58 10.1 ± 2.2
30/46 (65)
16126 (62)
2.5 ± 0.3
30/46 (65) 2.4 ± 0.2
7.2 ± 0.5
10 ± 0.8
M
H
64 33 ± 3.6
46 19.2 ± 3.2
30/64 (47) 41/64 (46)
107
M 135 7.3 ± 0.8
arrhea (days) No. stools preceding 24 hr
H - Houston; M = Mexico; Mean ± SEM.
matic control children (112 from Houston, 98 from Mexico) were matched by age, season, and geographic location. Fourteen (6%) asymptomatic children had the following enteropathogens isolated: rotavirus-6 (Mexico 3, Houston 3), iT-E. coli-5 (Mexico 3, Houston 2), ST-E. coli-l (Mexico), and Giardia-l (Houston). Enteropathogens were isolated from a significantly (P < 0.001) fewer number of children without diarrhea compared to those with illness. No control patient had multiple enteropathogens isolated. Clinical characteristics of children with diarrhea by presence of enteropathogens. Table IV gives the clinical characteristics of children with diarrhea from Houston and Mexico in relation to the major pathogens detected. Children from Mexico with diarrhea were significantly (P < 0.001) younger and had more intense diarrhea, as determined by the number of stools and greater frequency of vomiting and fever, than those from Houston, regardless of the causative agent. Rotavirus was isolated from significantly younger children in Mexico (P < 0.025) and Houston (P < 0.05) than children from the same location from whom Shigella was isolated. Vomiting was more common in children from whom rotavirus was isolated and less frequent in those from whom Salmonella was identified. Fever occurred more commonly in children from whom Shigella was isolated. There was no significant difference in the number of days ill between the Mexican and Houston children in any group when divided by causative agent. E. coli heat-labile toxin antibody titers. The Figure depicts linear regression analysis of E. coli LT antibody titers in children from Houston and Mexico by age. Sera were obtained for baseline E. coli LT antibody titers at the
time the children were enrolled in the study. There was no significant difference in antibody titers in children from Houston and Mexico in any age group. An age-related acquisition of L T antibodies is demonstrated in Mexico (r = 0.536, P < 0.001) and Houston (r = 0.594, P < 0.001).
DISCUSSION The finding of multiple enteropathogens has been reported previously in adults' and children" from Mexico, but is an uncommon occurrence in the United States. The previously reported finding that the isolation rate of enteropathogens is highest in patients with more intense diarrhea was not substantiated by this study.': 18 The children from Mexicoin general had more striking illness, yet the isolationrate of enteropathogens in this group was no greater than that found in children from Houston. The occurrence of diarrhea in younger children in Mexico may be due to an earlier exposure to numerous enteropathogens. Shigella was isolated from older infants in both locations and generally caused fever. Rotavirus occurred in younger infants and vomiting was a more frequent manifestation of illnessin this group, as has been reported." 7 Salmonella strains and rotavirus were more common in children' with diarrhea in Mexico; Shigella was a more important problem in Houston. These differences may reflect widespread and indiscriminate use of antimicrobials among the general population in Mexico, although the children studied had not received antibiotic therapy within the week prior to evaluation. Antimicrobials may serve as a selection factor for Salmonella at the expense of the more drug sensitive Shigella strains. Rotavirus infection characteristically has occurred
Volume 93 Number 3
during winter months." •. • There are two possible explanations for the lack of seasonal occurrence in our study. We evaluated populations in tropical and semi-tropical areas and most of the children were outpatients. Nearly all previously published studies have been conducted in temperate climates, and hospitalized children with diarrhea were enrolled. A study in the semi-tropical climate of Taiwan also revealed that rotavirus was a common summertime agent, occurring in 56% of children with diarrhea. ,0 Toxigenic E, coli has been shown to be of major importance to U.S. adults traveling to Mexico, and to the animal industry as a cause of scours among the newborns of swine and cattle. Our study agrees with a recent publication in Canada" suggesting that toxigenic E. coli is not commonly associated with endemic diarrhea of children in North America. The small number of toxigenic E. coli isolated from stools of Mexican children in our study is in contrast to two previous studies in which enterotoxigenic strains were detected in 16 and 47% of Mexican children with diarrhea.": IS Several differences in these studies may account for this discrepancy. We evaluated outpatient children over 21 months; other studies evaluated hospitalized children over a shorter time interval. In addition, the population evaluated at the Social Security Hospital is primarily a middle socio-economic class, compared to an indigent population previously studied.": IS Patients with diarrhea due to toxigenic E. coli often develop a rise in serum antibodies to the heat-labile enterotoxin:" 31 Failure to demonstrate a fourfold rise in LT antibodies during episodes of diarrhea among children in the present study supports our stool culture data showing the relative unimportance of toxigenic E. coli as a cause of diarrhea, We did, however, demonstrate that the titers of specific antitoxin antibody in children are related to age; the linear increase in LT-E. coli titers indicates a continual exposure to this or a related antigen with subclinical or mild infection. Other studies have shown that serum antibodies to rotavirus increase steadily from six months to two years of age," 9. " The prevalence of LT-E. coli and rotavirus antibodies at two years of age is comparable to the serum antibody level in adults." 9." Klebsiella and Aeromonas have been incriminated as enterotoxin-producing organisms in children with diarrhea." 13 We detected one Aeromonas and 14 Proteus strains which produced rounding of cells in the Y-l adrenal cell assay. The significance of these strains and the nature of the substance produced are unknown. Gurwith and Williams" evaluated 49 Proteus isolates from children with diarrhea; none caused morphologic changes in the mouse adrenal tumor cell assay.
Enteropathogens in children with diarrhea
387
2.2
1.8
•••• Mexico
-Houston 0.6
The cause of diarrhea eluded us in 40% of children with diarrhea. These cases may be due to virulence characteristics other than enterotoxin production, such as colonization factor'S (which allows adherence of bacteria to the intestinal wall), enterotoxins which are not detected by presently employed methods, or enteric viruses which were not identified. We must now characterize the epidemiology (transmission dynamics and reservoirs) of the known important enteropathogens and intensify the search for other causative agents in patients with diarrhea. We are indebted to Bart Hsi, Ph.D., for biostatistica1 support; P. Byers, P. Gearhart, C. Geib, S. Hamilton, D. Johnson, r. Rutherford, H. West, and D. Satterwhite for technical help; and 1. Addison for assistance in the preparation of the.manuscript, REFERENCES
1. Yow MD, Melnick JL, Blattner RJ, et al: The association of viruses and bacteria with infantile diarrhea, Am J Epide2.
3.
4.
5.
6.
mioI92:33, 1970. Cramblett RO, and Siewers CMF: The etiology of gastroenteritis in infan Is and children, with emphasis on the occurrence ofsimultaneous mixed viral-bacterial infections, Pediatrics 35:885, 1965. Bishop RF, Davidson OP, Holmes JH, et al: Virus particles in epithelial cells of duodenal mucosa from children with acute nonbacterial gastroenteritis, Lancet 2:1281, 1973. Ryder RW, Sack DA, Kapikian AZ, et al: Enterotoxigenic Escherichia coli and reovirus-like agent in rural Bangladesh, Lancet 1:659, 1976. Kapikian AZ, Kim HW, Wyatt RO, et al: Human reoviruslike agent as the major pathogen associated with "winter" gastroenteritis in hospitalized infants and young children, N Engl J Med 294:965, 1976. Konno T, Suzuki H, Imai A, et al: Reovirus-like agent in acute epidemic gastroenteritis in Japanese infants: fecal shedding and serologic response, J Infect' Dis 135:259, 1977.
388
7.
8.
9.
10.
11.
12.
n. 14.
IS.
16.
17.
18.
19.
20.
The Journ al of Pediatrics September 1978
Pickering et of.
Rodriguez WJ, Kim HW, Arro bio JO, Brandt CD, Chanock RM, Kapikian AZ, Wyatt RO , and Parro tt RH: Clinical features of acute gastroenteritis associated with hum an reoviru s-lik e agent in infants and young children, J PEDlATR 91: 188, 1977. Gurwith MI , and Williams TW: Ga stroenteritis in children : a two-year review in Manitoba. I. Etiology, J Infect Dis 136:239, 1977. Gomez-Barreto J, Palmer EL, Nahmias AJ, et al: Acute enteritis ass ociated with reovirus-like agents, JAMA 235:1857,1976. Echeverria P, Ho MT, and B1acklow NR: Relative importance of virus and bacteria in the etiology of pediatric diarrhea in Taiwan, J Infect Dis 136:383, 1977. Bolivar R, Conklin RH, Vollet J. et al: Rotavirus in travelers' diarrhea: Study of an adult student population in Mexico, J Infect Dis 137:324, 1978. Gorbach SL, and Kurana CM : Toxigenic Escherichia coli: A cause of infanlile diarrhea in Chicago, N Engl J Med 287:791, 1972. Guerrant RL, Mo ore RA, Kirschenfeld PM, et al: Role of toxigenic and invasive bacteria in ac ute diarrhea of childhood, N Engl J Med 293:567, 1975. Sack RB, H irsch horn N, Brownlee et al: Enterotoxigenic Escherichia coli-associated d iarrheal disease in Apache children, N Engl J Med 292:1041, 1975. Nalin DR, Raharnan M, McL augh lin JC, et al: Enterotoxigenic Escherichia col i and idiopathic diarrhea in Bangladesh , Lancet 2:1116 , 1975. Ryder RW, Wachsmuth IK, Buxton AE, et al: Infantile diarrhea produced by he at-stable enterotoxigenic Escherichia coli, N Engl J Med 295:849, 1976. Donta ST, Wallace RD, Whipp SC. et al: Enterotoxigenic Escherichia coli and diarrheal disease in Mexican child ren , J Infect Dis 135:482 , 1977. Evans DO, Olarte J, DuPont HL, et al: Enteropathogens associated with pediatric diarrhea in Mexico City, J PEDIATR 91:65, 1977. Gorbach SL, Kean BR, Evans DG, et al: Travelers' diarrhea and toxigenic Escherichia coli, N Engl J Med 292:933, 1975. Merson MH, Morris GK, Sack DA, et al: Travelers' diarrhea in Mexico: A pr ospective study, N Engl J Med 294:1299, 1976.
21.
22. 23.
24.
25.
26.
27.
I:
28.
29.
30.
31.
32.
33.
DuPont HL, Olarte J, Evans DG, et al : Comparative susceptibility of Latin American and United States students to enteric pathogens, N Engl J Med 295:1520, 1976. Pickering LK, DuPont HL , Olarte J , et a1: Fecal leukocytes in enteric infections, Am J e lin Pathol 68:562, 1977. Donta ST, Moon HW , and Whipp SC: Detection of heat-labile Escherichia coli enterotoxin with the use of adrenal cells in tissue culture, Science 183:334, 1974. Donta ST, and Smith OM: Stimulation of steroidogenesis in tissue culture by enterotoxigenic Escherichia coli a nd its neutralization by specific an tiserum, Infect Immun 9:500. 1974. Evans OJ Jr, Evans DO, and Gorbach SL: Production of vascular permeability factor by enterotoxigenic Escherichia coli isolated from man, Infect Immun 8:725, 1973. Dean AG, Ching Y, Williams RG, et al: Tests for Escherichia coli enterotoxin using infant mice: application in a study of diarrhea in children in Honolulu, J Infect Dis 125:407, 1972. Byers PA, DuPont HL, and Evans OJ Jr: A pooling method for screening clinical isolates of Escherichia coli for the production of heat-stable enterotoxin, Annual Meeting of the American Society for Microbiology, New Orleans, May 8-13 , 1977, P 225. Sereny, B.: Experimental shigella keratoconjunctivitis: A preliminary report, Acta Microbiol Acad Sci Hung 2:293, 1955. Evans OJ , Ruiz-Palacios G, Evans DO, et al: Humoral immune response to the heat-labile enterotoxin of Escherichia coli in naturally acquired diarrhea and antitoxin determination by p assive immune hemolysis, Infect Immun 16:781,1977. Portnoy BL, Conklin RH , Menn M , et al: Reliable identification of reovirus-like agent in diarrheal stools, J Lab Clin Med 89:560, 1977. Sack RB, Jacobs B, and Mitra R : Antitoxin responses to infections with enterotoxigenic Escherichia coli, J Infect Dis 129:330, 1974. Blacklow NR, Echeverria P, and Smith DH: Serologic studies with reovirus-like enteritis agent, Infect Immun 13:1563, 1976, Evans DG, Evans DJ Jr, and DuPont HL: Virulence factors of enterotoxigenic Escherichia coli, J Infect Dis 136:S 118, 1977.