Persistent diarrhea is responsible for one third of diarrhea-related childhood deaths in most developing countries (1). Antibiotic treatment directed at enteric pathogens is usually not effective, and dietary management is the mainstay of its treatment (2,3). Currently, the initial dietary treatment of persistent diarrhea in developing countries is based on the use of different lactose-free formulations (4-6). A substantial proportion of hospitalized patients so treated show poor weight gain or persistence of diarrhea beyond a week of hospitalization (7). The role of carbohydrate malabsorption and local gut and systemic infection in these poorly responding patients is not well defined. Apart from controlled studies comparing different dietary regimens or other therapeutic measures, a careful analysis of large series of patients treated according to predetermined standard protocols could help identify factors and possibly provide useful predictors associated with poor outcome on initial diets. In such patients, more complex dietary regimens may be used at the beginning of the treatment itself.
We present an analysis of the results and the prognostic indicators among 75 well-characterized consecutive hospitalized male patients with persistent diarrhea, treated according to a standardized treatment algorithm. The results have implications for development of a standard treatment approach for children with persistent diarrhea, a high research priority of the WHO control program (8).
MATERIALS AND METHODS
Setting and Patient Selection
The study was conducted in the impatient service of the Division of Pediatric Gastroenterology, All India Institute of Medical Sciences, New Delhi. Boys aged between 3 and 48 months with a weight-for-height ≤90% National Center for Health Statistics (NCHS) median and duration of diarrhea of ≥14 days with passage of at least three liquid stools in the previous 24 h were included in the study. Gross blood in stool, any associated severe nondiarrheal illnesses, and declined consent were reasons for exclusion. Girls were excluded because of difficulty in obtaining stools uncontaminated by urine for weighing. Patients who had received prior antibiotic therapy were eligible for the study.
The study protocol was approved by the Ethical Review Committee of the All India Institute of Medical Sciences.
Initial Fluid, Dietary, and Drug Therapy
Fluid. To facilitate duodenal intubation, the patients were kept on a nil oral, intravenous fluids (IVF) regimen in the initial 6 h. Fourteen patients with dehydration were given an initial deficit therapy (0-6 h) with Ringer's lactate solution, whereas those without dehydration received maintenance IVF. After the duodenal fluid was obtained, IVF were discontinued and WHO oral rehydration solution (ORS) was offered on a volume-to-weight basis to replace stool losses.
Diet I. A lactose-free diet composed predominantly of a soy formula (Prosoyal, Fairdeal Pvt. Ltd.; 63 cal/100 ml) was offered at the rate of 100 and 80 kcal/kg body weight (KBW)/day to patients <12 and ≥12 months of age, respectively. Additional calories were provided by a cereal-lentil-oil gruel (85 kcal/100 g) at the rate of 50 and 80 kcal/KBW/day for infants and for those older than 12 months, respectively. Soy formula was offered in four and cereal-based gruel in three divided feeds, a total of seven feeds per day.
Drugs. Systemic antibiotics were provided for associated systemic infection (i.e., pneumonia based on cough plus crepitations or probable septicemia based on lethargy, anorexia, abdominal distention, fever, or hypothermia in the absence of dehydration or electrolyte disturbance). Antidiarrheal drugs were not used.
In addition to the vitamins, trace elements and minerals provided in the supplemented soy formula vitamins A, C, D, B complex, zinc, iron, and folate were supplemented in twice the daily recommended amounts. Calcium was given orally in maintenance doses (9).
Criteria for Change of Diet I to Diet II
Children were considered to be treatment failures on Diet I if they purged at rates exceeding ≥60 g/KBW/24 h on the sixth or the seventh day while on the diet; this was a consensus definition among five senior pediatricians. All poor responders, by this definition, were changed to Diet II if free of systemic infection. When there was a strong suspicion of systemic sepsis, the need for dietary change was reassessed 48 h after antibiotic treatment.
Diet II was composed of 100 g chicken puree, 30 g glucose, and 40 g oil per liter of prepared formula with an energy density of 58 kcal/100 g. It was offered at the rate of 120 kcal/kg/day.
In one patient who developed severe diarrhea on Diet II, with a positive test for stool-reducing sugars, glucose was removed from the diet. He was given chicken and oil, as in Diet II, orally and glucose intravenously.
The intake of water, ORS, and diet was monitored at 3-h intervals and nude weights were obtained daily. Stool output and vomitus were measured by before and after weighing of absorbent diapers.
All patients except for one, after recovery on Diet I or II, were challenged with 7% lactose-containing feeds for a 24-h period, 4 weeks after discharge. If there was no recurrence of symptoms, 100 ml of cow's milk was offered. If there were no adverse reactions, three milk feeds were given daily as part of a mixed diet, and patients were observed at weekly intervals for the next 6 weeks.
At 0, 24, and 72 h after admission, blood urea, serum sodium and potassium levels were determined by standard methods. Stool specimens were collected in duplicate on admission: the first in Cary-Blair transport medium, and the second in a plain vial for microscopy for parasites, pus cells, and red blood cells.
A closed-end method was used for collecting the duodenal fluid. The quantitation of aerobic and anaerobic bacteria was as described in an earlier publication by us (10).
Fecal and Duodenal Fluid Microbiology
Primary identification of bacteria in the fecal specimens was done as per the WHO manual (11). Campylobacter jejuni was isolated using Columbia agar, with incubation under microaerophilic conditions (N2, 85%; CO2, 10%) (12). Giardia lamblia and Entamoeba histolytica cysts and trophozoites were looked for promptly by direct microscopy and by concentration methods.
Characterization of E. Coli
Three lactose-fermenting colonies of E. coli were tested with DNA probes that detect enterotoxigenic E. coli (LT, STp, STh), Whereas three E. coli strains per child were also examined for their pattern of adherence to Hep-2 cells, as described previously (13-15).
E. coli that adhered to Hep-2 cells, did not produce LT, STp, STh, and were negative for invasiveness were defined as enteroadherent E. coli (EAEC); they were classified into three different phenotypes, localized (EAEC-L), diffuse (EAEC-D), or aggregative (EAggEC).
Statistical analysis was done using the SPSS/PC 4.0 software. The binomial outcomes were compared using χ2 analysis and Fisher's exact test when indicated. The two-tailed Student's t test was applied on normally distributed or log-transformed data to compare means between groups. For skewed data, the nonparametric Mann-Whitney U test was used.
To adjust for the simultaneous effect of several potential confounding factors, we performed a logistic regression analysis after creating dummy variables for qualitative factors. In one model, diarrhea treatment failure after 7 days of hospitalization or otherwise was the dependent variable, whereas feeding status (breast feeding, yes/no), age (≤6, 7-12, or >12 mo), antecedent nutritional status (<74.9 or ≥75% of NCHS standards), preadmission diarrheal duration (<30 or ≥30 days), purge rates on day 1 of hospitalization (≤30, 30.1-60, or >60 g/kg of body weight), total calorie intake (≤480, 481-650, or >651), stool or duodenal fluid pathogen (yes/no), intercurrent systemic infection (yes/no), carbohydrate malabsorption while on Diet I (yes/no) were included as independent variables. Interaction between systemic infection and initial purge rate was also examined as an independent variable.
A similar analysis was performed with weight loss or otherwise after 7 days of hospitalization in comparison with the rehydration weight as the dependent factor.
Factors Associated with Diarrhea Treatment Failure After 7 Days on Diet I
Seventy-five patients were initiated on Diet I for the first 7 hospital days; 47 of them recovered from diarrhea by day 7. Among the admission characteristics examined, breast feeding (p = 0.0001) and initial 24-h purge rates (p < 0.001) were related to the treatment failures on Diet I (Table 1). Notably, initial weight for length (p = 0.69) and height for age (p = 0.44) were not related to recovery. Among pathogens, there was a trend toward higher excretion rate of enteroaggregative E. coli (25.9 vs. 11.9%; p = 0.12), Shigella (7.1% vs. 0; p = 0.14), and diffuse enteroadherent E. coli (25.9 vs. 14.3%; p = 0.23) in the diarrhea treatment failures. The admission duodenal fluid aerobic and anaerobic overgrowth (≥105/ml) did not show significant differences between treatment successes and failures (aerobic: 44.4, 35.7%, p = 0.47; anaerobic: 59.3, 42.9%, p = 0.11).
Among intercurrent events (Table 2) that were not apparent at admission, pneumonia (p = 0.003) or probable septicemia (p = 0.03) were significantly more common in the diarrhea treatment failures by day 7.
During the hospital stay, low pH or reducing substances in stools on day 5 and 6 of dietary therapy were significantly (p = 0.03) correlated with delayed recovery; however, a similar trend was seen as early as day 1 (Table 2). The total calorie intake on days 5, 6, and 7 was similar in the two groups (p = 0.7).
Factors Associated with Weight Loss After 7 Days on Diet I
Children were categorized as having weight loss when the weight at day 7 was less than the postrehydration admission weight. These children are compared with those who showed weight gain in Table 3.
The children with weight loss were younger (p = 0.05) and they had significantly lower food intake during the first 24 h after admission (p = 0.05) and during the hospital stay (p < 0.01). The mean stool output during the initial 24 h and on all subsequent days was similar in the two groups. Interestingly, breast feeding did not protect against weight loss; indeed, there was a trend in the other direction. Notably, among all enteropathogens detected, only the isolation of the EAEC was more common in the children who lost weight (32 vs. 8%; p = 0.01). The duodenal fluid aerobic and anaerobic overgrowth (≥105/ml) was marginally more frequent in the patients losing weight (aerobes, 48 vs. 36.6%, p = 0.26; anaerobes, 64 vs. 43%, p = 0.07).
In the logistic regression model, significant risk factors for diarrhea treatment failures included initial purge rates [30-60 g/kg v.s. <30 g/kg; odds ratio (OR), 11.35; 95% confidence interval (CI), 8.79, 13.9], [>60 g/kg vs. <30 g/kg; OR, 37.89; 95% Cl, 35.14, 40.64] carbohydrate malabsorption while on Diet I [OR, 5.09; 95% CI, 3.69, 6.48] and systemic infection [OR, 41.97; 95% CI 39.56, 44.37]. Low food intake [<450 kcal/kg vs. >650 kcal/kg; OR, 17.44; 95% CI, 15.67, 19.21], [450-650 kcal/kg vs. >650 kcal/kg; OR, 5.179; 95% CI, 3.36, 6.99] was significantly associated with an increased risk for weight loss.
A good trade-off between sensitivity and specificity for prediction of diarrhea treatment failures on Diet I at the end of 7 days of hospitalization was not found for any of the admission or intercurrent clinical and laboratory parameters. Stool frequency of eight or more in initial 24 h predicted delayed recovery from diarrhea on Diet I with 82% sensitivity, but the specificity was only 36.2%.
Clinical Response to Diet II
Among the 28 children who became treatment failures on Diet I, 16 recovered with antibiotic treatment for associated systemic infections without dietary change. The remaining 12 were put on Diet II, of whom 11 recovered within the next 7 days. The only patient whose diarrhea did not abate on Diet II had acquired monosaccharide intolerance based on reducing substances >2% while only on ORS by mouth, and he responded gradually to a diet containing chicken and oil supplemented with intravenous glucose, vitamins, minerals, and electrolytes.
Overall, only one child failed treatment with the two-step dietary algorithm of Diets I and II; the child subsequently died a month later. The other children tolerated animal milk as part of a mixed diet 4 to 6 weeks after discharge from the hospital without recurrence of diarrhea or weight loss.
The study demonstrates that children hospitalized with persistent diarrhea and associated malnutrition can be treated successfully with a step-wise enteral feeding regimen supplemented with vitamins and micronutrients, early detection and treatment of systemic infection, and fluid therapy without need for routine total parenteral nutrition. These results do not apply to a small proportion of critically ill infants who require intensive care.
Carbohydrate malabsorption as measured by low stool pH and reducing substances and systemic infection were associated with lack of recovery from diarrhea with a week on Diet I.
The carbohydrates in Diet I were the starch in cereal gruel and maltodextrins in the soy formula. Malabsorption of complex starch and maltodextrins may have been the result of decreased disaccharidases and impaired pancreatic function associated with enteric infection and malnutrition (16-18).
The failure rates on Diet I were rather high, necessitating prolonged hospitalization. Anticipating the problem of more intensive impaired carbohydrate absorption in a proportion of patients, we used glucose in Diet II. Although the majority of failures on Diet I recovered on the chicken-glucose-based formulation, the problem of osmolar overload limits glucose concentration in the diet and thereby the caloric density (58 kcal/100 g of the feed). Addition of modest amounts of starch to Diet II, well within the digestive capacity of the child, would increase the energy density and provide the right balance between osmolarity, digestibility, and caloric density. A diet providing carbohydrate as a mixture of cereal and glucose was used successfully in Bangladesh for feeding infants with severe persistent diarrhea (19). Such diets need to be evaluated as the initial treatment of persistent diarrhea to achieve higher success rates within a week of hospitalization. Chicken or egg white is a potential protein source with a cereal-glucose mixture to which oil is added to increase the energy density (19,20).
Because all patients tolerated milk as part of a mixed diet 4 to 6 weeks after discharge from the hospital, it would seem that milk-protein allergy is not a major causative factor of persistent diarrhea, as defined for this study, in our setting.
A direct association was observed between pneumonia and septicemia and treatment failures. The possible mechanisms may be the lack of nutrient availability for mucosal repair due to anorexia, or perturbations in the intestinal microflora during a respiratory infection. In any event, early identification and prompt treatment of systemic infection is critical to achieve early recovery and weight gain.
Bacterial overgrowth in the upper small intestine has been considered an important event in the pathophysiology of persistent diarrhea (21,22). In an earlier study, also from India, neither aerobic nor anaerobic overgrowth as such was related to poor outcome (10). Recently Penny et al. (23) also found no association between increased duodenal aerobic or anaerobic counts and absorption of nitrogen, fat, or carbohydrate in children with persistent diarrhea and malnutrition.
The association of EAEC with poor weight gain strengthens the etiologic role recently proposed for the organisms in persistent diarrhea (24,25). However, in the same setting, an earlier antibiotic trial showed no significant clinical impact of oral gentamicin therapy on persistent diarrhea, although the stools in the treated children were negative for enteroaggregative and other bacterial stool pathogens at the end of the treatment (2).
Apart from systemic infection, low food intake was a major factor in explaining poor weight gain. It is noteworthy that children with early or delayed recovery from diarrhea had similar food intakes, suggesting that optimal feeding does not worsen diarrhea while simultaneously promoting weight gain.
None of the several clinical or laboratory indicators examined at admission was able to predict failure on soy- or cereal-based diets with the required sensitivity and specificity.
Breast feeding in the initial week of illness was shown not to increase the risk of persistence of acute diarrhea (26). In this study, an apparent association of breast feeding with treatment failures in the univariate analysis during persistent diarrhea seems to have been the result of confounding factors, as it was not identified as an independent risk factor in the logistic regression model.
Some limitations of the study need to be kept in mind. Hospital-acquired infection by itself or the resulting carbohydrate malabsorption could have contributed to the treatment failures. We did not evaluate this factor, as stool analysis at the time of treatment failure was not done. However, the fact that there was a trend (p = 0.06) toward positive correlation between low stool pH or positive stool reducing substances on the first day of dietary treatment suggests that at least the contribution of carbohydrate malabsorption was not the result of hospital-acquired infection. Further, the findings in these sick, hospitalized patients with regard to prevalence of carbohydrate malabsorption and systemic infection and their contribution to treatment failures may not be generalizable to a community setting where the cases are milder.
In conclusion, initial severity of illness, relative maldigestion of complex starch, and systemic infection were the factors associated with poor outcome in children with persistent diarrhea and associated malnutrition. The use of a mixture of sucrose or glucose and modest amounts of starch as the carbohydrate source in the initial diet itself is proposed as a possible approach to hasten recovery and promote rapid catch-up growth. Such diets can be easily prepared in small peripheral hospitals and even at household level in developing countries.
1. Bhan MK, Arora NK, Ghai OP, Ramachandran K, Khoshoo V, Bhandari N. Major factors in diarrhoea related mortality and morbidity among rural children. Indian J Med Res
2. Bhatnagar S, Bhan MK, Sazawal S, et al. Efficacy of massive dose oral gentamicin therapy in non-bloody persistent diarrhea with associated malnutrition. J Pediatr Gastroenterol Nutr
3. Editorial. Chronic diarrhea in children—a nutritional disease. Lancet
4. Maclean WC Jr, Lopez deRomana G, Massa E, Graham GG. Nutritional management of chronic diarrhea and malnutrition: primary reliance on oral feeding. J Pediatr
5. Orenstein SR. Enteral versus parenteral therapy for intractable diarrhea of infancy: a prospective randomized trial. J Pediatr
6. Bhan MK, Arora NK, Khoshoo V, et al. Chronic diarrhea in infants and children. Indian J Pediatr
7. Bhan MK, Bhatnagar S. Persistent diarrhea in infants and young children. Ind Pediatr
8. Persistent diarrhea in developing countries. Memorandum from a WHO Meeting. Bull WHO
9. Subcommittee on the tenth edition of the RDAs Food and Nutrition Board Commission on Life Sciences National Research Council. Recommended dietary allowances
. Washington DC: National Academy Press, 1989:195-46.
10. Bhatnagar S, Bhan MK, George C, Gupta U, Kumar R, Bright D, Saini S. Is small bowel bacterial overgrowth of pathogenic significance in persistent diarrhea? Acta Pediatr
11. Programme for control of Diarrheal Disease. WHO. Manual for laboratory investigations of acute enteric infections. CDD, 1987;83.3 Rev 1.
12. Nayyar S, Bhan MK, Gupta U, et al. Campylobacter jejuni
as a cause of childhood diarrhea in a north Indian community. J Diarrhoeal Dis Res
13. Sommerfelt H, Kalland KH, Raj, P, Moseley SL, Bhan MK, Bjorvatn B. Cloned polynucleotide and synthetic oligonucleotide probes used in colony hybridization are equally efficient in the identification of enterotoxigenic Escherichia coli J. Clin Microbiol
14. Nataro JP, Kaper JB, Robins-Browne R, Prado V, Vial P, Levine MM. Patterns of adherence of diarrheagenic Escherichia coli
to HEp-2 cells. Pediatr Infect Dis J
15. Bhan MK, Raj P, Khoshoo V, et al. Quantitation and properties of fecal and upper small intestinal aerobic microflora in infants and young children with persistent diarrhea. J Pediatr Gastroenterol Nutr
16. Lebenthal E. Prolonged small intestinal mucosal injury as a primary cause of intractable diarrhea of infancy. In: Lebenthal E, ed. Chronic diarrhea in children
. New York: Vevey/Raven Press, 1984:5-29.
17. Lebenthal E, Shwachman H. The pancreas: development, adaptation, and malfunction in infancy and childhood. Clin Gastroenterol
18. Khoshoo V, Bhan MK, Jain R, et al. Intestinal glucoamylase and other disaccharidases in children with protracted diarrhoea. Indian J Med Res
19. Roy SK, Haider R, Akbar MS, Alan AN, Khatun M, Eeckels R. Persistent diarrhea: clinical efficacy and nutrient absorption with a rice based diet. Arch Dis Child
20. Larcher VF, Shepherd R, Francis DEM, Harries JT. Protracted diarrhea in infancy. Arch Dis Child
21. Challacombe DN, Richardson JM, Rowe B, Anderson CM. Bacterial microflora of the upper gastrointestinal tract in infants with protracted diarrhea. Arch Dis Child
22. Gracey M, Stone DE. Small intestinal microflora in Australian aboriginal children with chronic diarrhea. Aust NZ J Med
23. Penny ME, Paredes P, Brown KH, Laughan B, Smith H. Lack of a role of the duodenal flora in pathogenesis of persistent diarrhea and diarrhea related malabsorption in Peruvian children. Pediatr Infect Dis J
24. Bhan MK, Raj P, Levine MM, et al. Enteroaggregative Escherichia coli
associated with persistent diarrhea in a cohort of rural children in India. J Infect Dis
25. Bhatnagar S, Bhan MK, Sommerfelt H, Sazawal S, Kumar R, Saini S. Enteroaggregative Escherichia coli
may be a new pathogen causing acute and persistent diarrhea. Scand J Infect Dis
26. Lanata CF, Black RE, Creed-Kanashiro H, et al. Feeding during acute diarrhea as a risk factor for persistent diarrhea. Acta Pediatr