Acta Pa?diatrSuppl 381: 133-8. 1992
Lactose feeding during persistent diarrhoea Mary E Penny’ and Kenneth H Brown2 Instituto de Investigacion Nutricional’, Apartado Postal 18-0191, Lima- 18, Peru; Program in International Nutrition2. Department of Nutrition, University of California, Davis, C A 95616, USA
Penny ME, Brown KH. Lactose feeding during persistent diarrhoea. Acta Pzdiatr 1992;(suppl 381):133-8. Stockholm. ISSN 0803-5326 Lactose intolerance is a recognized complication of some episodes of diarrhoea. Whereas it is of clinical importance in relatively few children with acute diarrhoea it seems to be a more common problem in children presenting with persistent diarrhoea, especially in malnourished children with severe diarrhoea. We describe a recent study which also demonstrated adverse clinical and nutritional consequences in the majority of children with mild but persistent diarrhoea consuming 6 g lactose/kg body weight/d. It is desirable to identify these children who need to reduce their lactose intake in order to limit unnecessary dietary changes which are expensive to implement and/or nutritionally disadvantageous. Admission clinical characteristics were analyzed for their ability to predict clinical outcome in a group of 33 children with persistent diarrhoea taking a lactose-containing diet. The results of tests of stool reducing substances and a lactose breath hydrogen test (LHBT) were analyzed in the same group whose clinical management had been independent of the test results. Eighty-nine percent of children who deteriorated had one of the following characteristics: age -= 12 months, weight-for-height less than -2 SD below NCHS norms or fever > 37°C. Children who unequivocally improved while continuing to take lactose could not be selected without the risk of including children who deteriorated. Neither tests of stool reducing substances nor the LBHT improved the predictive value of simple clinical parameters. Reduction of lactose in the diet of children with persistent diarrhoea who have any of these parameters is recommended, but mixing of milk with cereal or substituting with fermented milk products is preferable to milk withdrawal or dilution with water. 0 Persistent diarrhoea, lactose intolerance. stool reducing substances, breath hydrogen test M E Penny, Instituto de Investigacion Nutricional, Apartado Postal 18-0191. Lima-18, Peru
The hydrolyzing enzyme lactase is the most vulnerable of the digestive enzymes. Malabsorption of lactose may occur in any circumstances in which the small intestinal mucosa is damaged and infectious diarrhoea is no exception. Increased rates of lactose malabsorption have been demonstrated in acute diarrhoea, particularly in children with rotavirus infection (1-3). If those children who are lactose intolerant continue to take an unmodified milk diet they may develop persistent diarrhoea and this seems to be more likely in children who are severely malnourished and/or have severe diarrhoea (4). In this group carbohydrate malabsorption may lead not only to abdominal distension and severe dehydrating diarrhoea but may also cause metabolic acidosis (5, 6). Nevertheless, clinical studies have now clarified that these children are not typical of the vast majority of children in developing countries with acute diarrhoea1 illness who are able to continue their normal consumption of milk without adverse consequences (7). In children with persistent diarrhoea it appears that lactose intolerance is a more common clinical problem than in acute diarrhoea (8-12). Lactose intolerance may indeed be the cause of the persisting diarrhoea in a sizable number of these patients. Early studies described
patients with post-enteritis persistent diarrhoea whose diarrhoea promptly ceased when lactose was removed from the diet (8). A recent study of Peruvian children with diarrhoea demonstrated lactose intolerance in children recruited from the community whose diarrhoea had lasted for more than 14 days but was relatively mild (13). The clinical and nutritional consequences of continued lactose feeding were studied in these 64 children. They were aged 3-36 months (mean 18 months), only one child was severely acutely malnourished (weight-forheight less than -2 SD with respect to the NCHS norms) and only eight children were moderately or severely dehydrated more than 5%. Children were randomly allocated to receive one of two milk-based diets which differed only in the addition of /l-galactosidase to one diet resulting in the hydrolysis of 95% of the lactose. Only the respective milk diet was given and was offered in six to eight divided feeds providing a maximum of 110 kcal/kg/day, 150 ml/kg/ day and 6 g lactose/kg body weight/day in the lactosecontaining diet. The clinicians and nurses caring for the patients were masked with regard to the identity of the childrens’ dietary assignment. Children in the two groups were similar on admission,
134 M E Penny and K H Brown mean stool output was measured in the boys and was 55 gm/kg/24 h. The diet was discontinued in four children taking the lactose diet (L) and one in the hydrolysed lactose group because of worsening diarrhoea. Stool output was measured daily and urine and stool were collected separately on days 3-5 inclusive in order to measure macronutrient balance. Amongst the boys who continued with the diet, stool output was significantly higher on days 3 , 4 and 5 in the lactose group compared with those receiving the lactose hydrolyzed milk, mean faecal wet weight 74.4 g/kg/24 h compared with 44.8 g/ kg/24 h ( p < 0.01, Student’s t-test). Forty-one percent of the lactose group, but only one (5.5%) of the hydrolysed lactose group purged more than 10% of their body weight in 24 h ( p < 0.02, Fisher’s exact test), a purging rate which could result in severe dehydration if not adequately treated. Diarrhoea also lasted longer in the lactose group than in the hydrolysed lactose group. This difference was due to 1 1 (35.5%) children in the hydrolysed lactose group compared with one in the H L group whose diarrhoea ceased within 30 h of admission. It is presumed that in these children lactose intolerance was the principal cause of their continuing diarrhoea. Continued lactose feeding was also associated with a significantly greater loss of carbohydrate, nitrogen and energy in stools. Faecal excretion of fat, although responsible for the greatest loss of dietary calories was not significantly different between the two groups. This study demonstrated a significantly worse clinical and nutritional outcome when a diet containing 6 gm lactose/kg/day was compared with a similar but low lactose diet in well-nourished children with relatively mild persistent diarrhoea. These results support the need for a reduction in dietary lactose in such children. However, this conclusion presents a dilemma to developing country physicians concerned with the adverse consequences of diarrhoea1 illness on the growth of small children. Whereas in affluent societies and in hospital practice, substitution of lactose milk for a lactose-free formula is feasible, this is not the case in developing countries where cow’s milk may be an important source of energy, protein, and micronutrients and alternative products are too costly or simply not available. In Lima, for example, it was found that cows’ milk provided half the protein requirements and onethird of the energy requirements of one-year-old children (14). Dilution or restriction of milk consumption may result in reduced nutrient intake, thus exacerbating the negative nutritional impact of these episodes of diarrhoea. These considerations make it desirable to distinguish those children who have clinically significant lactose intolerance so that dietary modifications can be limited to this group. Reviews of children with persistent diarrhoea have always demonstrated many children who did not have lactose malabsorption, particularly older children with milder diarrhoea (8, 11, 15); nevertheless, no study has correlated clinical outcome with
ACTA P R D l A T R SUPPL 381 (1992)
clinical characteristics on admission nor with the results of tests of lactose malabsorption in an unselected population of children with persistent diarrhoea. We conducted a retrospective analysis of the result of the Peruvian study in order to determine whether children whose diarrhoea did or did not improve while consuming the lactose-containing diet could be identified on admission. We related the patients’ clinical characteristics, stool reducing substances and pH, and the result of a lactose breath hydrogen test (LBHT) to subsequent clinical course. Thirty-three Peruvian children with persistent diarrhoea who were allocated to receive the lactose-containing diet are included in this analysis. Children were divided retrospectively according to clinical outcome without considering the results of the stool tests and LBHT. The criteria used are given in Table 1. In the case of seven children it was not possible to categorize them because their clinical course fluctuated and these children were classified as a separate group C. The clinical outcomes were compared with the patients’ characteristics on admission and the results of stool reducing substances and pH testing and the LBHT using sensitivity, specificity analysis (1 6 ) . The clinical parameters used were age, weight, height, temperature on admission. Z-scores for weight-for-height, height-for-age and weight-for-age in relation to NCHS median values were calculated. Stool reducing substances were tested using Clinitest tablets as described by Kerry and Anderson (1 7) in all liquid stools of sufficient volume at the end of each 4 h observation period. Results were analysed considering stools with =-0.5% reducing substances as positive. Stool pH was tested with indicator paper (Whatman). Table 1 . Criteria for grouping children according to clinical response to diet.
Group A: Children whose diarrhoea worsened while n= 18 taking the diet Any of the following: Days 1-5 0 Study dietary failures (stool output > 150 g/kg/d or > 100 g/ kg/d plus one of the following: vomiting > 300 ml or refusal of > 33% of diet. (Without new pathogens in stools.) 0 Volume of stools increasing by 150%1 and > 50 g/kg/d 0 Volume of stools increasing and exceeding > 100 g/kg/d 0 Number of liquid stools increasing to or maintained at > 5/day for 5 days n=8 Group B: Children who improved while taking the diet Any of the following: Days 1-5 0 Decrease in stool output/day by 50% 0 Stool output remaining below 30 g/kg/d 0 Passage of formed stools not followed by further liquid stools 0 Permanent reduction in number of liquid stools from > 5/day to < 3/day for > 5 days
Group C: Children whose progress did not fit into either of the above categories
n=l
Lactose,feeding during persistent diarrhoea
ACTA PEDIATR SUPPL 38 I (1 992)
1 35
The comparison of clinical characteristics on admission with the different outcomes is shown in Table 2. Other characteristics, such as breast-feeding and abdominal distension on admission were more common in Group B but were present in only a few children in this study population and their inclusion did not improve the prediction of clinical outcome. Although individual variables were quite specific in identifying children who would get worse, the sensitivities tended to be low and therefore an unacceptably high proportion of children who needed modification of their diet would have been missed. However, 88.9% (16 of 18) children who were destined to do badly on the lactose diet could be identified with a specificity of 53.3% when children with at least one of the following easily measured variables were separated: age < 12 months, weight-for-age < 2 SD below NHCS norms, and presence of fever. Children who would do well despite continuing lactose (Group B versus the rest) are more difficult to predict. Ten children had none of the above-mentioned trio of clinical characteristics, only four of whom were in group B giving a positive predictive value of 40%. However, in this study population the number of children falling into Group B was only eight, thus limiting the power of the analysis. Laboratory tests were studied independently of clinical characteristics for their predictive capacity. Results of the percentage of stools tested which were positive for reducing substances in the A and B groups are given in Table 3. Both false negative and false positive tests were seen in relation to clinical outcome. Three of the eight B group children would have been wrongly classified if stool reducing substances > 0.5% had been used as an indicator of lactose intolerance. Even in those children in Group A who had high diarrhoea1 fluxes and positive reducing substances on at least one occasion, the excretion of reducing substances was intermittent. By
A lactose breath hydrogen test was conducted on day 3. Following a fast of at least 8 h, expired breath was collected into a two litre anaesthetic bag using an infant face mask. A single feed of the lactose-containing milk was given (1 g lactose/kg body weight) and further breath samples were taken at 30 min intervals for 4 h. Duplicate samples of expired breath were analysed using a Quintron hydrogen analyser. Results were expressed as the maximum hydrogen minus the minimum hydrogen reading. This measure was preferred to max-basal because, despite the fast, 15 children had high basal levels of which ten declined during the first hour of the test and were assumed to be due to residual carbohydrate in the colon. The sensitivity, specificity and predictive values were calculated in order to test the ability of the clinical characteristics on admission and the laboratory tests to distinguish two different subgroups of children: (a) those whose clinical condition definitely worsened with lactose feeding, Group A and (b) those whose diarrhoea unequivocally improved despite the lactose, Group B. Recognition of Group A would be useful in circumstances where it is desirable to continue a lactosecontaining diet in as many children as possible while avoiding putting at risk those who would definitely deteriorate. This assumes that there would be further opportunities to assess the children in order to identify additional children who had milder lactose intolerance and would not have been included in Group A but might warrant a trial of lactose reduction if their diarrhoea persisted. Identification of children in Group B would be useful when fear of the possible adverse effect of continued lactose feeding outweighed nutritional considerations; for instance if no further medical supervision of the children was anticipated. In this case it would be prudent to continue lactose only in those who could be confidently expected to improve while taking a lactose-containing diet.
Table 2. Sensitivity and specificity of different clinical variables to predict the clinical response to the lactose diet.
Ability of parameter to predict child who will get worse (Gp A vs the rest) Specificity
Sensitivity
Predictive value
Clinical parameter on admission
No.
?4"
No.
%
Age < 12 months Weight-for-height (W/H)I < - 2 SD Weight-for-age (W/A)I < - 2 SD Fever on admission > 37.5"C Age < 12 months or W/A < - 2 SD or fever > 37.5"C Vomiting in 1st 24 h Vomit > 50 ml
4/18 5/18 8/18 10/18 16/18 6/18
22.2 27.7 44.4 55.6 88.9 33.3
13/15 14/15 11/15 11/15 8/15 13/15
86.6 93.3 73.3 86.7 53.3 86.7
9/18 15/18
50.0 83.3
13/15 11/15
86.7 73.3
9/11 15/19
81.8 78.9
17/18
94.4
9/15
60.0
11/23
13.9
Any of the following: Age < 12 months or W/H < - 2 SD Age < 12 months or W/H < - 2 SD or fever > 37.5"C Age < 12 months or W/H < - 2 SD or fever > 37.5"C or vomiting > 50 ml in 1st 24 h I
Expressed as Z-score in relation to the NHCS mean.
No.
Yo
66.6 83.3 66.6 83.3 69.6 75.0
136 M E Penny and K H Brown
ACTA PEDIATR SUPPL 381 (1992)
Table 3. Stool reducing substances in relation to day of study and clinical outcome. % of children with > O S %
Yo of all tested stools with >0.5% RS
Worse on L diet ( n = 18) Better on L diet ( n = 8) Significance (Fisher exact two-tailed test)
Day 1
Day 2
Day 3
Day 1
Day 2
Day 3
23.0 11.4 > 0.05
22.5 10.0 > 0.05
42.3 12.1 0.05
38.6 12.5 > 0.05
85.7 37.5 =0.026
Table 4. Ability of a test for reducing substances in a single stool sample collected on the first hospital day to predict the clinical response to the lactose diet according to the time of day when the sample was collected.
Ability of positive RS ( > 0.5%) to predict the child who will get worse (group A vs the rest) Sensitivity
Specificity
Predictive value
(Yo)
(%)
(%,)
23.5 23.1 33.3
64.3 100.0
44.4 100.0 45.5
Fasting (06.00-09.00 h) Morning (09.00-14.00 h) Overnight (21.00-06.00 h)
57.1
Table 5. Ability of different values of peak hydrogen excretion (maxmin) to predict clinical response to the lactose diet.
Ability of BHT to predict child who will get worse (group A vs the rest) Sensitivity Number’
%I
Breath hydrogen (Max-min) ppm Max-min 2 2 0 4/17 23.5 Max-min > I 2 10/17 58.8
Specificity Number 13/15 11/15
%I
86.7 73.3
Predictive value Number
YO
4/6
66.7 71.4
10/14
’ One child did not have the BHT performed. the end of the third day, 12 of 14 children in the A group were excreting reducing substances in stool but by this time it was clinically obvious that the children were not responding to treatment. Differences were observed in the proportion of stool samples positive for reducing substances depending on the time of day. Therefore, results were analysed according to the hour at which the stool was tested: 0
0
RS in at least 1 stool
The first stools collected before the first meal of the day after the normal overnight fast of 6 h. Stools analysed between 09.00 and 14.00 after at least one milk feed. This would be the normal circumstances of a child brought to one of the local health posts.
0
Stools sampled overnight 21.00 to 06.00 when the child had consumed most of the day’s diet.
The relationships between stool reducing substances testing on the first hospital day and subsequent clinical course are shown in Table 4. Generally, the stools passed in the morning either before the first meal or between 09.00 and 14.00 were least likely to be positive. All children with positive stools passed between 09.00 and 14.00 on the first day subsequently had worsening diarrhoea, but this result was very insensitive; there was a false negative rate of 76.9%. Negative results were unhelpful in identifying the children who would definitely improve because 82.4% of children who did not improve would also be selected. Stool pH was also tested but only marginally improved the performance of stool testing when combined with reducing substances results. The ability of the LBHT to predict clinical outcome in children taking the lactose-containing diet is shown in Table 5. Low sensitivity limits the usefulness of the test in this context. The test had a specificity of 86.7% when a max-min of 20 ppm was taken as the cut-off, but was very insensitive; there were 13 (50%) false negative test results. Sensitivity improved to 58.8% when the cut-off was lowered to 12 ppm, but is still too low to be clinically useful. The large number of false negative results may be due to preceding diet. In another analysis, LBHT results following one ration of the lactose-containing diet (1 g lactose/kg body weight) were compared between children who had been taking the lactose-containing diet and 51 children taking a diet which differed only in that the lactose was pre-hydrolysed. Both milks had been the only diet consumed in the two days before the LBHT. Only 16.1Yoof those taking lactose had a positive result (hydrogen excretion > 20 ppm) compared with 44.2% of the hydrolysed lactase group ( p 20 ppm) in 41% of children with persistent diarrhoea, 33.3% of children with acute diarrhoea and 41.7% of children of the same age without diarrhoea all taking the same hydrolyzed-lactose diet.
Lactose feeding during persistent diarrhoea
ACTA P E D I A T R SUPPL 381 (19g2)
In this study the test was used in its simplest form; refinements such as repeating the test with lactulose to identify non-hydrogen producers (19) or in vitro testing of stool bacteria for hydrogen production (20) were not undertaken because they limit the practical value of the test as an outpatient tool and increase cost. A relatively small dose of lactose in milk was given 1 g/kg body weight and there would probably have been a higher proportion of children with positive tests if the standard 2 g/kg body weight dose had been administered. However, we were anxious to avoid potentially harmful exacerbation of diarrhoea in these children, an important consideration for a test being evaluated for its applicability in outpatient situations. Even with the small 1 g dose 44% of children who had not been receiving lactose in the days before the test were found to have peak hydrogen excretion > 20 ppm and 18.6% > 50 PPm. The combination of LBHT plus testing stools for reducing substances improved the performance of the tests but required testing of all stools. This would be both impractical and expensive.
137
The recommendation to reduce lactose in the diet should be interpreted with caution. Lactose intolerance is dose related (21) and it is not necessary to exclude lactose completely from the diet. The use of fermented milk products in which lactose is partially hydrolyzed may be a useful manoeuvre (22,23). Lactose intolerance is also modulated by the vehicle in which the lactose is presented. Chocolate milk, for instance, causes less lactose intolerance than the same milk without chocolate (24) and milk cereal mixtures have been successfully used in children with acute diarrhoea (25). Simple fractionation of the daily milk allowance, and/or taking the milk with other foods, might even be sufficient, but further research is required to evaluate the efficacy of these different ways of avoiding lactose intolerance without withdrawing milk or diluting it with water. Acknowledgements.-I am grateful to Patricia Paredes and the staff of the clinical unit of the IIN for their care of the study children, to Enrique Lancho for technical assistance, and to Homero Martinez for his help in getting the study going. Mary Penny was Wellcome Trust Clinical Lecturer in Infectious Diseases in the Department of Paediatrics, University of Oxford. The studies mentioned were funded by the Wellcome Trust, Thrasher Research Fund and the World Health Organization.
Conclusions Three simple clinical characteristics (young age, poor nutritional status, and presence of fever) were able to identify 88.9% of patients who would deteriorate if they continued to consume a diet containing 6 g lactoselkg body weightlday. However, only 33 children were included in this analysis and the usefulness of these clinical parameters should be confirmed with larger numbers of children in different settings before the results can be accepted for inclusion in diarrhoea1 control programmes. Testing for stool reducing substances is a simple test but is expensive; it needs to be performed on fresh stools and is more difficult when children are in nappies. We have not demonstrated any advantage which would justify the use of the test in determining whether dietary modification is indicated in an individual child with persistent diarrhoea. If breath hydrogen testing is available, children who have a max-min hydrogen excretion of > 20 ppm should be given a reduced lactose diet, but this test cannot reliably be used to identify high risk children with confidence. Since the test was less sensitive than the use of clinical criteria, requires a period of fasting and is expensive, it cannot be generally recommended for the routine screening of children with persistent diarrhoea in developing countries. We would recommend that children with persistent diarrhoea who are under 12 months, those who are malnourished (weight-for-age < 2 SD below NHCS norms), and children with fever should receive a reduced-lactose diet. If this information is not available all children with persistent diarrhoea should be given a clinical trial of a reduced-lactose diet.
References I . Brown KH, Black RE, Parry L. The effect of acute diarrhea on lactose malabsorption among Bangladeshi children. Am J Clin Nutr 1980;33:2226-7 2. Kumar V, Chandrasekaran R, Bhaskar R. Carbohydrate intolerance associated with acute gastroenteritis. Clin Pediatr 1977;16:1123-7 3. Hyams JS, Kraus PJ, Gleason PA. Lactose malabsorption following rotavirus infection in young children. J Pediatr 1981;89:916-1 8 4. Lifshitz F, Coello-Ramirez P, Gutierrez-Topete G, CornadoCornet MC. Carbohydrate intolerance in infants with diarrhea. J Pediatr 1971;79:760-7 5. Torres-Pinedo R, Lavastida M, Rivera CL, Rodriguez H, Ortiz A. Studies on infant diarrhea. I . A comparison of the effects of milk feeding and intravenous therapy upon the composition and volume of the stool and urine. J Clin Invest 1966;45:469 6. Lifshitz F. Clinical studies in diarrhea disease and malnutrition associated with carbohydrate intolerance. In: Chavez A, Bourges H, Gasra S, eds. Proceedings of the 9th Congress of Nutrition. Mexico: Karger, Basel, 1975:173-81 7. Brown KH, Lake A. Appropriate use of human and non-human milks for the dietary management of children with diarrhea. J Diarrheal Dis Res 1991;9:168-85 8. Burke V, Kerry KR, Anderson CM. The relationship of dietary lactose to refractory diarrhoea in infancy. Aust Paediatr J 1965;1:147-59 9. Sunshine P, Kretchmer N. Studies of small intestine during development. 111. Infantile diarrhea associated with intolerance to disaccharides. Pediatrics 1964;3438-50 10. Gracey M, Burke V. Sugar-induced diarrhoea in children. Arch Dis Child 1973;48:331-6 1 I . Coello-Ramirez P, Medina-Huerta LA, Diaz-Bensussen S, Zuniga V, Larrosa-Haro A. Etiologia de la diarrea prolongada en ninos. Bol Med Hosp Infant Mex 1984;41:605-10
138
M E Penny and K H Brown
12. Soeparto P, Noerasid H, Satjadibrata K. Carbohydrate intolerance in infants with chronic recurrent diarrhea. Paediatr Indonesiana 1977;17:153-60 13. Penny ME, Paredes P, Brown KH. Clinical and nutritional consequences of lactose feeding during persistent postenteritis diarrhea. Pediatrics 1989;84:8354 14. Brown KH, Black RE, Lopez de Romafia G, Kanashiro HC. Infant feeding practices and their relationship with diarrheal and other diseases in Huascar, an underprivileged community on the periphery of Lima, Peru. Pediatrics 1989;83:3140 15. Larcher VF, Shepherd R, Francis DEM, Harries JT. Protracted diarrhoea in infancy: analysis of 82 cases with particular reference to diagnosis and management. Arch Dis Child 1977;52:597-605 16. Armitage P, Berry G. Statistical methods in medical research. 2nd ed. Oxford: Blackwell 1987:472-8 17. Kerry KR, Anderson CM. A ward test for sugar in faeces. Lancet 1964;i:981-2 18. Vogelsang H, Ferenci P, Frotz S, Meryn S, Gang1 A. Acidic colonic microclimate-possible reason for false negative hydrogen breath tests. Gut 1988;29:21-6 19. Davidson GP, Goodwin D, Robb TA. Incidence and duration of
ACTA PlEDlATR SUPPL 381 (1992)
20. 21.
22.
23. 24. 25.
lactose malabsorption in children hospitalized with acute enteritis: a study in a well nourished urban population. J Pediatr 198%105:587-90 Perman JA, Modler S, Obson AC. Role of pH in production of hydrogen from carbohydrates by colonic bacterial flora. J Clin Invest 1981;67:643-50 Barillas C, Solomons NW. Effective reduction of lactose maldigestion in preschool children by direct addition of B-galactosidase to milk at meal time. Pediatrics 1987;79:766-72 Dewit 0, Boudraa G, Touhami M, Desjeux JF. Breath hydrogen test and stool characteristics after ingestion of milk and yogurt in malnourished children with chronic diarrhoea and lactase deficiency. J Trop Pediatr 1987;33:177-80 Chandrasekaran R, Kumar V, Walia BNS, Moorthy B. Carbohydrate intolerance in infants with acute diarrhoea and its complications. Acta Paediatr Scand 1975;64:483-8 Saavedra JM, Perman JA. Current contents in lactose malabsorption and intolerance. Annu Rev Nutr 1989;9:475-502 Brown KH, Perez F, Gastafiaduy AS. Clinical trial of modified whole milk, or cereal-mixtures during acute childhood diarrhea. J Pediatr Gastroenterol Nutr 1991:12:340-50
Lactose feeding during persistent diarrhoea Mary E Penny’ and Kenneth H Brown2 Instituto de Investigacion Nutricional’, Apartado Postal 18-0191, Lima- 18, Peru; Program in International Nutrition2. Department of Nutrition, University of California, Davis, C A 95616, USA
Penny ME, Brown KH. Lactose feeding during persistent diarrhoea. Acta Pzdiatr 1992;(suppl 381):133-8. Stockholm. ISSN 0803-5326 Lactose intolerance is a recognized complication of some episodes of diarrhoea. Whereas it is of clinical importance in relatively few children with acute diarrhoea it seems to be a more common problem in children presenting with persistent diarrhoea, especially in malnourished children with severe diarrhoea. We describe a recent study which also demonstrated adverse clinical and nutritional consequences in the majority of children with mild but persistent diarrhoea consuming 6 g lactose/kg body weight/d. It is desirable to identify these children who need to reduce their lactose intake in order to limit unnecessary dietary changes which are expensive to implement and/or nutritionally disadvantageous. Admission clinical characteristics were analyzed for their ability to predict clinical outcome in a group of 33 children with persistent diarrhoea taking a lactose-containing diet. The results of tests of stool reducing substances and a lactose breath hydrogen test (LHBT) were analyzed in the same group whose clinical management had been independent of the test results. Eighty-nine percent of children who deteriorated had one of the following characteristics: age -= 12 months, weight-for-height less than -2 SD below NCHS norms or fever > 37°C. Children who unequivocally improved while continuing to take lactose could not be selected without the risk of including children who deteriorated. Neither tests of stool reducing substances nor the LBHT improved the predictive value of simple clinical parameters. Reduction of lactose in the diet of children with persistent diarrhoea who have any of these parameters is recommended, but mixing of milk with cereal or substituting with fermented milk products is preferable to milk withdrawal or dilution with water. 0 Persistent diarrhoea, lactose intolerance. stool reducing substances, breath hydrogen test M E Penny, Instituto de Investigacion Nutricional, Apartado Postal 18-0191. Lima-18, Peru
The hydrolyzing enzyme lactase is the most vulnerable of the digestive enzymes. Malabsorption of lactose may occur in any circumstances in which the small intestinal mucosa is damaged and infectious diarrhoea is no exception. Increased rates of lactose malabsorption have been demonstrated in acute diarrhoea, particularly in children with rotavirus infection (1-3). If those children who are lactose intolerant continue to take an unmodified milk diet they may develop persistent diarrhoea and this seems to be more likely in children who are severely malnourished and/or have severe diarrhoea (4). In this group carbohydrate malabsorption may lead not only to abdominal distension and severe dehydrating diarrhoea but may also cause metabolic acidosis (5, 6). Nevertheless, clinical studies have now clarified that these children are not typical of the vast majority of children in developing countries with acute diarrhoea1 illness who are able to continue their normal consumption of milk without adverse consequences (7). In children with persistent diarrhoea it appears that lactose intolerance is a more common clinical problem than in acute diarrhoea (8-12). Lactose intolerance may indeed be the cause of the persisting diarrhoea in a sizable number of these patients. Early studies described
patients with post-enteritis persistent diarrhoea whose diarrhoea promptly ceased when lactose was removed from the diet (8). A recent study of Peruvian children with diarrhoea demonstrated lactose intolerance in children recruited from the community whose diarrhoea had lasted for more than 14 days but was relatively mild (13). The clinical and nutritional consequences of continued lactose feeding were studied in these 64 children. They were aged 3-36 months (mean 18 months), only one child was severely acutely malnourished (weight-forheight less than -2 SD with respect to the NCHS norms) and only eight children were moderately or severely dehydrated more than 5%. Children were randomly allocated to receive one of two milk-based diets which differed only in the addition of /l-galactosidase to one diet resulting in the hydrolysis of 95% of the lactose. Only the respective milk diet was given and was offered in six to eight divided feeds providing a maximum of 110 kcal/kg/day, 150 ml/kg/ day and 6 g lactose/kg body weight/day in the lactosecontaining diet. The clinicians and nurses caring for the patients were masked with regard to the identity of the childrens’ dietary assignment. Children in the two groups were similar on admission,
134 M E Penny and K H Brown mean stool output was measured in the boys and was 55 gm/kg/24 h. The diet was discontinued in four children taking the lactose diet (L) and one in the hydrolysed lactose group because of worsening diarrhoea. Stool output was measured daily and urine and stool were collected separately on days 3-5 inclusive in order to measure macronutrient balance. Amongst the boys who continued with the diet, stool output was significantly higher on days 3 , 4 and 5 in the lactose group compared with those receiving the lactose hydrolyzed milk, mean faecal wet weight 74.4 g/kg/24 h compared with 44.8 g/ kg/24 h ( p < 0.01, Student’s t-test). Forty-one percent of the lactose group, but only one (5.5%) of the hydrolysed lactose group purged more than 10% of their body weight in 24 h ( p < 0.02, Fisher’s exact test), a purging rate which could result in severe dehydration if not adequately treated. Diarrhoea also lasted longer in the lactose group than in the hydrolysed lactose group. This difference was due to 1 1 (35.5%) children in the hydrolysed lactose group compared with one in the H L group whose diarrhoea ceased within 30 h of admission. It is presumed that in these children lactose intolerance was the principal cause of their continuing diarrhoea. Continued lactose feeding was also associated with a significantly greater loss of carbohydrate, nitrogen and energy in stools. Faecal excretion of fat, although responsible for the greatest loss of dietary calories was not significantly different between the two groups. This study demonstrated a significantly worse clinical and nutritional outcome when a diet containing 6 gm lactose/kg/day was compared with a similar but low lactose diet in well-nourished children with relatively mild persistent diarrhoea. These results support the need for a reduction in dietary lactose in such children. However, this conclusion presents a dilemma to developing country physicians concerned with the adverse consequences of diarrhoea1 illness on the growth of small children. Whereas in affluent societies and in hospital practice, substitution of lactose milk for a lactose-free formula is feasible, this is not the case in developing countries where cow’s milk may be an important source of energy, protein, and micronutrients and alternative products are too costly or simply not available. In Lima, for example, it was found that cows’ milk provided half the protein requirements and onethird of the energy requirements of one-year-old children (14). Dilution or restriction of milk consumption may result in reduced nutrient intake, thus exacerbating the negative nutritional impact of these episodes of diarrhoea. These considerations make it desirable to distinguish those children who have clinically significant lactose intolerance so that dietary modifications can be limited to this group. Reviews of children with persistent diarrhoea have always demonstrated many children who did not have lactose malabsorption, particularly older children with milder diarrhoea (8, 11, 15); nevertheless, no study has correlated clinical outcome with
ACTA P R D l A T R SUPPL 381 (1992)
clinical characteristics on admission nor with the results of tests of lactose malabsorption in an unselected population of children with persistent diarrhoea. We conducted a retrospective analysis of the result of the Peruvian study in order to determine whether children whose diarrhoea did or did not improve while consuming the lactose-containing diet could be identified on admission. We related the patients’ clinical characteristics, stool reducing substances and pH, and the result of a lactose breath hydrogen test (LBHT) to subsequent clinical course. Thirty-three Peruvian children with persistent diarrhoea who were allocated to receive the lactose-containing diet are included in this analysis. Children were divided retrospectively according to clinical outcome without considering the results of the stool tests and LBHT. The criteria used are given in Table 1. In the case of seven children it was not possible to categorize them because their clinical course fluctuated and these children were classified as a separate group C. The clinical outcomes were compared with the patients’ characteristics on admission and the results of stool reducing substances and pH testing and the LBHT using sensitivity, specificity analysis (1 6 ) . The clinical parameters used were age, weight, height, temperature on admission. Z-scores for weight-for-height, height-for-age and weight-for-age in relation to NCHS median values were calculated. Stool reducing substances were tested using Clinitest tablets as described by Kerry and Anderson (1 7) in all liquid stools of sufficient volume at the end of each 4 h observation period. Results were analysed considering stools with =-0.5% reducing substances as positive. Stool pH was tested with indicator paper (Whatman). Table 1 . Criteria for grouping children according to clinical response to diet.
Group A: Children whose diarrhoea worsened while n= 18 taking the diet Any of the following: Days 1-5 0 Study dietary failures (stool output > 150 g/kg/d or > 100 g/ kg/d plus one of the following: vomiting > 300 ml or refusal of > 33% of diet. (Without new pathogens in stools.) 0 Volume of stools increasing by 150%1 and > 50 g/kg/d 0 Volume of stools increasing and exceeding > 100 g/kg/d 0 Number of liquid stools increasing to or maintained at > 5/day for 5 days n=8 Group B: Children who improved while taking the diet Any of the following: Days 1-5 0 Decrease in stool output/day by 50% 0 Stool output remaining below 30 g/kg/d 0 Passage of formed stools not followed by further liquid stools 0 Permanent reduction in number of liquid stools from > 5/day to < 3/day for > 5 days
Group C: Children whose progress did not fit into either of the above categories
n=l
Lactose,feeding during persistent diarrhoea
ACTA PEDIATR SUPPL 38 I (1 992)
1 35
The comparison of clinical characteristics on admission with the different outcomes is shown in Table 2. Other characteristics, such as breast-feeding and abdominal distension on admission were more common in Group B but were present in only a few children in this study population and their inclusion did not improve the prediction of clinical outcome. Although individual variables were quite specific in identifying children who would get worse, the sensitivities tended to be low and therefore an unacceptably high proportion of children who needed modification of their diet would have been missed. However, 88.9% (16 of 18) children who were destined to do badly on the lactose diet could be identified with a specificity of 53.3% when children with at least one of the following easily measured variables were separated: age < 12 months, weight-for-age < 2 SD below NHCS norms, and presence of fever. Children who would do well despite continuing lactose (Group B versus the rest) are more difficult to predict. Ten children had none of the above-mentioned trio of clinical characteristics, only four of whom were in group B giving a positive predictive value of 40%. However, in this study population the number of children falling into Group B was only eight, thus limiting the power of the analysis. Laboratory tests were studied independently of clinical characteristics for their predictive capacity. Results of the percentage of stools tested which were positive for reducing substances in the A and B groups are given in Table 3. Both false negative and false positive tests were seen in relation to clinical outcome. Three of the eight B group children would have been wrongly classified if stool reducing substances > 0.5% had been used as an indicator of lactose intolerance. Even in those children in Group A who had high diarrhoea1 fluxes and positive reducing substances on at least one occasion, the excretion of reducing substances was intermittent. By
A lactose breath hydrogen test was conducted on day 3. Following a fast of at least 8 h, expired breath was collected into a two litre anaesthetic bag using an infant face mask. A single feed of the lactose-containing milk was given (1 g lactose/kg body weight) and further breath samples were taken at 30 min intervals for 4 h. Duplicate samples of expired breath were analysed using a Quintron hydrogen analyser. Results were expressed as the maximum hydrogen minus the minimum hydrogen reading. This measure was preferred to max-basal because, despite the fast, 15 children had high basal levels of which ten declined during the first hour of the test and were assumed to be due to residual carbohydrate in the colon. The sensitivity, specificity and predictive values were calculated in order to test the ability of the clinical characteristics on admission and the laboratory tests to distinguish two different subgroups of children: (a) those whose clinical condition definitely worsened with lactose feeding, Group A and (b) those whose diarrhoea unequivocally improved despite the lactose, Group B. Recognition of Group A would be useful in circumstances where it is desirable to continue a lactosecontaining diet in as many children as possible while avoiding putting at risk those who would definitely deteriorate. This assumes that there would be further opportunities to assess the children in order to identify additional children who had milder lactose intolerance and would not have been included in Group A but might warrant a trial of lactose reduction if their diarrhoea persisted. Identification of children in Group B would be useful when fear of the possible adverse effect of continued lactose feeding outweighed nutritional considerations; for instance if no further medical supervision of the children was anticipated. In this case it would be prudent to continue lactose only in those who could be confidently expected to improve while taking a lactose-containing diet.
Table 2. Sensitivity and specificity of different clinical variables to predict the clinical response to the lactose diet.
Ability of parameter to predict child who will get worse (Gp A vs the rest) Specificity
Sensitivity
Predictive value
Clinical parameter on admission
No.
?4"
No.
%
Age < 12 months Weight-for-height (W/H)I < - 2 SD Weight-for-age (W/A)I < - 2 SD Fever on admission > 37.5"C Age < 12 months or W/A < - 2 SD or fever > 37.5"C Vomiting in 1st 24 h Vomit > 50 ml
4/18 5/18 8/18 10/18 16/18 6/18
22.2 27.7 44.4 55.6 88.9 33.3
13/15 14/15 11/15 11/15 8/15 13/15
86.6 93.3 73.3 86.7 53.3 86.7
9/18 15/18
50.0 83.3
13/15 11/15
86.7 73.3
9/11 15/19
81.8 78.9
17/18
94.4
9/15
60.0
11/23
13.9
Any of the following: Age < 12 months or W/H < - 2 SD Age < 12 months or W/H < - 2 SD or fever > 37.5"C Age < 12 months or W/H < - 2 SD or fever > 37.5"C or vomiting > 50 ml in 1st 24 h I
Expressed as Z-score in relation to the NHCS mean.
No.
Yo
66.6 83.3 66.6 83.3 69.6 75.0
136 M E Penny and K H Brown
ACTA PEDIATR SUPPL 381 (1992)
Table 3. Stool reducing substances in relation to day of study and clinical outcome. % of children with > O S %
Yo of all tested stools with >0.5% RS
Worse on L diet ( n = 18) Better on L diet ( n = 8) Significance (Fisher exact two-tailed test)
Day 1
Day 2
Day 3
Day 1
Day 2
Day 3
23.0 11.4 > 0.05
22.5 10.0 > 0.05
42.3 12.1 0.05
38.6 12.5 > 0.05
85.7 37.5 =0.026
Table 4. Ability of a test for reducing substances in a single stool sample collected on the first hospital day to predict the clinical response to the lactose diet according to the time of day when the sample was collected.
Ability of positive RS ( > 0.5%) to predict the child who will get worse (group A vs the rest) Sensitivity
Specificity
Predictive value
(Yo)
(%)
(%,)
23.5 23.1 33.3
64.3 100.0
44.4 100.0 45.5
Fasting (06.00-09.00 h) Morning (09.00-14.00 h) Overnight (21.00-06.00 h)
57.1
Table 5. Ability of different values of peak hydrogen excretion (maxmin) to predict clinical response to the lactose diet.
Ability of BHT to predict child who will get worse (group A vs the rest) Sensitivity Number’
%I
Breath hydrogen (Max-min) ppm Max-min 2 2 0 4/17 23.5 Max-min > I 2 10/17 58.8
Specificity Number 13/15 11/15
%I
86.7 73.3
Predictive value Number
YO
4/6
66.7 71.4
10/14
’ One child did not have the BHT performed. the end of the third day, 12 of 14 children in the A group were excreting reducing substances in stool but by this time it was clinically obvious that the children were not responding to treatment. Differences were observed in the proportion of stool samples positive for reducing substances depending on the time of day. Therefore, results were analysed according to the hour at which the stool was tested: 0
0
RS in at least 1 stool
The first stools collected before the first meal of the day after the normal overnight fast of 6 h. Stools analysed between 09.00 and 14.00 after at least one milk feed. This would be the normal circumstances of a child brought to one of the local health posts.
0
Stools sampled overnight 21.00 to 06.00 when the child had consumed most of the day’s diet.
The relationships between stool reducing substances testing on the first hospital day and subsequent clinical course are shown in Table 4. Generally, the stools passed in the morning either before the first meal or between 09.00 and 14.00 were least likely to be positive. All children with positive stools passed between 09.00 and 14.00 on the first day subsequently had worsening diarrhoea, but this result was very insensitive; there was a false negative rate of 76.9%. Negative results were unhelpful in identifying the children who would definitely improve because 82.4% of children who did not improve would also be selected. Stool pH was also tested but only marginally improved the performance of stool testing when combined with reducing substances results. The ability of the LBHT to predict clinical outcome in children taking the lactose-containing diet is shown in Table 5. Low sensitivity limits the usefulness of the test in this context. The test had a specificity of 86.7% when a max-min of 20 ppm was taken as the cut-off, but was very insensitive; there were 13 (50%) false negative test results. Sensitivity improved to 58.8% when the cut-off was lowered to 12 ppm, but is still too low to be clinically useful. The large number of false negative results may be due to preceding diet. In another analysis, LBHT results following one ration of the lactose-containing diet (1 g lactose/kg body weight) were compared between children who had been taking the lactose-containing diet and 51 children taking a diet which differed only in that the lactose was pre-hydrolysed. Both milks had been the only diet consumed in the two days before the LBHT. Only 16.1Yoof those taking lactose had a positive result (hydrogen excretion > 20 ppm) compared with 44.2% of the hydrolysed lactase group ( p 20 ppm) in 41% of children with persistent diarrhoea, 33.3% of children with acute diarrhoea and 41.7% of children of the same age without diarrhoea all taking the same hydrolyzed-lactose diet.
Lactose feeding during persistent diarrhoea
ACTA P E D I A T R SUPPL 381 (19g2)
In this study the test was used in its simplest form; refinements such as repeating the test with lactulose to identify non-hydrogen producers (19) or in vitro testing of stool bacteria for hydrogen production (20) were not undertaken because they limit the practical value of the test as an outpatient tool and increase cost. A relatively small dose of lactose in milk was given 1 g/kg body weight and there would probably have been a higher proportion of children with positive tests if the standard 2 g/kg body weight dose had been administered. However, we were anxious to avoid potentially harmful exacerbation of diarrhoea in these children, an important consideration for a test being evaluated for its applicability in outpatient situations. Even with the small 1 g dose 44% of children who had not been receiving lactose in the days before the test were found to have peak hydrogen excretion > 20 ppm and 18.6% > 50 PPm. The combination of LBHT plus testing stools for reducing substances improved the performance of the tests but required testing of all stools. This would be both impractical and expensive.
137
The recommendation to reduce lactose in the diet should be interpreted with caution. Lactose intolerance is dose related (21) and it is not necessary to exclude lactose completely from the diet. The use of fermented milk products in which lactose is partially hydrolyzed may be a useful manoeuvre (22,23). Lactose intolerance is also modulated by the vehicle in which the lactose is presented. Chocolate milk, for instance, causes less lactose intolerance than the same milk without chocolate (24) and milk cereal mixtures have been successfully used in children with acute diarrhoea (25). Simple fractionation of the daily milk allowance, and/or taking the milk with other foods, might even be sufficient, but further research is required to evaluate the efficacy of these different ways of avoiding lactose intolerance without withdrawing milk or diluting it with water. Acknowledgements.-I am grateful to Patricia Paredes and the staff of the clinical unit of the IIN for their care of the study children, to Enrique Lancho for technical assistance, and to Homero Martinez for his help in getting the study going. Mary Penny was Wellcome Trust Clinical Lecturer in Infectious Diseases in the Department of Paediatrics, University of Oxford. The studies mentioned were funded by the Wellcome Trust, Thrasher Research Fund and the World Health Organization.
Conclusions Three simple clinical characteristics (young age, poor nutritional status, and presence of fever) were able to identify 88.9% of patients who would deteriorate if they continued to consume a diet containing 6 g lactoselkg body weightlday. However, only 33 children were included in this analysis and the usefulness of these clinical parameters should be confirmed with larger numbers of children in different settings before the results can be accepted for inclusion in diarrhoea1 control programmes. Testing for stool reducing substances is a simple test but is expensive; it needs to be performed on fresh stools and is more difficult when children are in nappies. We have not demonstrated any advantage which would justify the use of the test in determining whether dietary modification is indicated in an individual child with persistent diarrhoea. If breath hydrogen testing is available, children who have a max-min hydrogen excretion of > 20 ppm should be given a reduced lactose diet, but this test cannot reliably be used to identify high risk children with confidence. Since the test was less sensitive than the use of clinical criteria, requires a period of fasting and is expensive, it cannot be generally recommended for the routine screening of children with persistent diarrhoea in developing countries. We would recommend that children with persistent diarrhoea who are under 12 months, those who are malnourished (weight-for-age < 2 SD below NHCS norms), and children with fever should receive a reduced-lactose diet. If this information is not available all children with persistent diarrhoea should be given a clinical trial of a reduced-lactose diet.
References I . Brown KH, Black RE, Parry L. The effect of acute diarrhea on lactose malabsorption among Bangladeshi children. Am J Clin Nutr 1980;33:2226-7 2. Kumar V, Chandrasekaran R, Bhaskar R. Carbohydrate intolerance associated with acute gastroenteritis. Clin Pediatr 1977;16:1123-7 3. Hyams JS, Kraus PJ, Gleason PA. Lactose malabsorption following rotavirus infection in young children. J Pediatr 1981;89:916-1 8 4. Lifshitz F, Coello-Ramirez P, Gutierrez-Topete G, CornadoCornet MC. Carbohydrate intolerance in infants with diarrhea. J Pediatr 1971;79:760-7 5. Torres-Pinedo R, Lavastida M, Rivera CL, Rodriguez H, Ortiz A. Studies on infant diarrhea. I . A comparison of the effects of milk feeding and intravenous therapy upon the composition and volume of the stool and urine. J Clin Invest 1966;45:469 6. Lifshitz F. Clinical studies in diarrhea disease and malnutrition associated with carbohydrate intolerance. In: Chavez A, Bourges H, Gasra S, eds. Proceedings of the 9th Congress of Nutrition. Mexico: Karger, Basel, 1975:173-81 7. Brown KH, Lake A. Appropriate use of human and non-human milks for the dietary management of children with diarrhea. J Diarrheal Dis Res 1991;9:168-85 8. Burke V, Kerry KR, Anderson CM. The relationship of dietary lactose to refractory diarrhoea in infancy. Aust Paediatr J 1965;1:147-59 9. Sunshine P, Kretchmer N. Studies of small intestine during development. 111. Infantile diarrhea associated with intolerance to disaccharides. Pediatrics 1964;3438-50 10. Gracey M, Burke V. Sugar-induced diarrhoea in children. Arch Dis Child 1973;48:331-6 1 I . Coello-Ramirez P, Medina-Huerta LA, Diaz-Bensussen S, Zuniga V, Larrosa-Haro A. Etiologia de la diarrea prolongada en ninos. Bol Med Hosp Infant Mex 1984;41:605-10
138
M E Penny and K H Brown
12. Soeparto P, Noerasid H, Satjadibrata K. Carbohydrate intolerance in infants with chronic recurrent diarrhea. Paediatr Indonesiana 1977;17:153-60 13. Penny ME, Paredes P, Brown KH. Clinical and nutritional consequences of lactose feeding during persistent postenteritis diarrhea. Pediatrics 1989;84:8354 14. Brown KH, Black RE, Lopez de Romafia G, Kanashiro HC. Infant feeding practices and their relationship with diarrheal and other diseases in Huascar, an underprivileged community on the periphery of Lima, Peru. Pediatrics 1989;83:3140 15. Larcher VF, Shepherd R, Francis DEM, Harries JT. Protracted diarrhoea in infancy: analysis of 82 cases with particular reference to diagnosis and management. Arch Dis Child 1977;52:597-605 16. Armitage P, Berry G. Statistical methods in medical research. 2nd ed. Oxford: Blackwell 1987:472-8 17. Kerry KR, Anderson CM. A ward test for sugar in faeces. Lancet 1964;i:981-2 18. Vogelsang H, Ferenci P, Frotz S, Meryn S, Gang1 A. Acidic colonic microclimate-possible reason for false negative hydrogen breath tests. Gut 1988;29:21-6 19. Davidson GP, Goodwin D, Robb TA. Incidence and duration of
ACTA PlEDlATR SUPPL 381 (1992)
20. 21.
22.
23. 24. 25.
lactose malabsorption in children hospitalized with acute enteritis: a study in a well nourished urban population. J Pediatr 198%105:587-90 Perman JA, Modler S, Obson AC. Role of pH in production of hydrogen from carbohydrates by colonic bacterial flora. J Clin Invest 1981;67:643-50 Barillas C, Solomons NW. Effective reduction of lactose maldigestion in preschool children by direct addition of B-galactosidase to milk at meal time. Pediatrics 1987;79:766-72 Dewit 0, Boudraa G, Touhami M, Desjeux JF. Breath hydrogen test and stool characteristics after ingestion of milk and yogurt in malnourished children with chronic diarrhoea and lactase deficiency. J Trop Pediatr 1987;33:177-80 Chandrasekaran R, Kumar V, Walia BNS, Moorthy B. Carbohydrate intolerance in infants with acute diarrhoea and its complications. Acta Paediatr Scand 1975;64:483-8 Saavedra JM, Perman JA. Current contents in lactose malabsorption and intolerance. Annu Rev Nutr 1989;9:475-502 Brown KH, Perez F, Gastafiaduy AS. Clinical trial of modified whole milk, or cereal-mixtures during acute childhood diarrhea. J Pediatr Gastroenterol Nutr 1991:12:340-50
