A positive effect on the nutrition of Philippine children of an oral glucose-electrolyte solution given at home for the treatment of diarrhoea. Report of a field trial by an international study group.

A positive effect on the nutrition of Philippine children of an oral glucose-electrolyte solution given at home for the treatment of diarrhoea Report of a field trial by an international study group * An oral glucose-electrolyte solution administered at home to Philippine children with diarrhoea was associated with a greater average weight gain both during an attack of diarrhoea and over a 7-month period compared with a control group. The longer-term effect on weight, relative to a standard, was more pronounced in children who had more than one attack ofdiarrhoea in the period ofobservation than in those who had only one attack. The size of the longer-term weight gain was 3-5 percentage points towards the standard weight. It is suggested that the vigorous compensation of salt and fluid losses improved children's appetites following diarrhoea attacks. Diarrhoea is a major cause of undernutrition in children, especially when attacks are repeated (1), and is associated with catabolism, malabsorption, anorexia, and imposed fasting (2). It has been sug- gested that a rapid and complete repair of water and electrolyte losses during acute diarrhoea will provide some protection from undernutrition, at least by improving the appetite. Such an effect seems to occur in hospitalized children (2). Our study, carried out in the Philippines, examined this hypothesis in children treated in an outpatient clinic and at home with an oral glucose-electrolyte solution. SUBJECTS AND METHODS Many variables affect body weight in children. To demonstrate with certainty the effect of a single * The international study group comprised: J. C. Azurin, J. J. Dizon, E. Sullesta, J. S. Sumpaico, and C. V. Uylangco of the Philippine Cholera Committee; A. M. Angara, R. Azucena, and L. Zamora of the Philippine Department of Health; D. C. Arisola, C. De La Rama, I. Henares, L. S. Maestral, and R. Suplido of the Regional Health Office and City of Bacolod Health Department, Philippines; H. H. L. Burgess, H. Dhillon, M. Kacic-Dimitri, T. Okuno, and A. C. Reyes of the WHO Regional Office for the Western Pacific, Manila, Philippines; N. Hirschhorn, P. J. Rousselle, and T. N. Warner of Management Sciences for Health, Cam- bridge, MA, USA; and an interdivisional group on diar- rhoeal diseases, World Health Organization, Geneva, Switzerland. Requests for reprints should be addressed to: N. Hirsch- horn, Management Sciences for Health, One Broadway, Cambridge, MA 02142, USA. intervention requires the random assignment of sub- jects to study and control groups, preferably preclas- sified in respect of important variables. If a placebo can be suitably prepared, a " double-blind" study should be conducted. This classic design neutralizes biased assignment and observation, balances known and unknown prognostic variables, and guarantees the validity of statistical tests of significance (3). Unfortunately, our study could not be so designed for several reasons: a placebo for glucose-electrolyte mixtures or solutions cannot be prepared; in our setting, it was not possible to make a random assignment of children to two groups since many people knew each other well and shared information and even medicines; and the random assignment of a number of whole, separate communities would have required resources beyond our means. The study was therefore carried out in two separate communities designated " study " and " control ". Pre-existing variables likely to affect body weight were examined for comparability and the study was designed so as to reduce bias. Communities B (study) and M (control) lie on the shore of Guimaras Strait at opposite ends of Baco- lod City, the capital of the Philippine province of Negros Occidental. Both are crowded with thatched houses, many on stilts over the water, and sanitation is poor. The heads of these households are fisher- men, labourers, or unemployed squatters. El Tor 3583 -87 - BULL. WORLD HEALTH ORGAN., Vol. 55, 1977 88 INTERNATIONAL STUDY GROUP cholera and infant diarrhoeas are endemic and are more prevalent in the rainy season. A health clinic is situated in each community and is readily accessible. Both communities have participated in previous epidemiological studies of cholera. Between 1 July 1975 and 31 January 1976, all children up to the age of 5 years reporting to the clinics with diarrhoea were included in the study. In both communities, a physician or attending nurse examined the children and followed the common routine of prescribing a nonspecific antidiarrhoeal, together with a specific antimicrobial for parenteral infections or dysentery. Stool cultures were not done routinely; prior experience in this area has shown that diarrhoea in most children is not associated with cholera or with Shigella or Salmonella infection. Epidemiological assistants weighed the children, told the mothers the reason for and benefits of continued feeding during diarrhoea, stressed the necessity of giving fluids by mouth, and discussed the sanitary preparation of food and fluids. Foods used included breast milk, bread, fish, bananas, and rice; the fluids used were tea, soft drinks, water, diluted cow's milk, rice water, and fish broth to which some salt had been added. All children were examined daily at home while ill and the educational messages were repeated and assistance offered. Clinical data were systematically recorded and guidelines for referral back to the physician followed. Body weight was recorded the day after diarrhoea ceased. There was almost complete follow-up. Only in community B was a glucose-electrolyte mixture used. This mixture, which we called Oresol, was produced for the study by the Philippines Department of Health in foil packets, each contain- ing sufficient to make 0.5 litre of solution. The composition of the fluid, in mmol/litre, was: Na+ 90 K+ 20 HCO3- 30 Cl- 80 glucose 111 which is that recommended by the World Health Organization (4).a The instructions for the use of Oresol stated that it should be given ad libitum in addition to (or mixed with) other fluids and food, with a target of about 1-2 litres of Oresol per day while diarrhoea continued. The epidemiological as- a The composition in g/litre is: sodium chloride, 3.5; sodium bicarbonate, 2.5; potassium chloride, 1.5; and glucose, 20. sistants were rotated between the two communities monthly; it was made clear to them that this was a study of nutrition and that children with diarrhoea needed fluid and food. Pre- and post-study weights were measured in June and February by the mass weighing (using a portable beam balance) of nearly all children under 5 years of age. Results were expressed as a per- centage of the Philippine median developed by the National Nutrition Survey, " Operation Tim- bang"; b this value is referred to as " relative weight ". The personnel who weighed the children after the study were unaware of the initial weight and in general did not know whether a child had attended the clinic. All weights were read twice. Of those who had been weighed before the study and who subsequently used the clinic, 830% were also weighed after the study. Interviews were conducted before and after the study on a one-third random sample of all families. The enquiries were especially concerned with the care and feeding of children with diarrhoea. Data analysis Standard tests for significance of difference of means were applied. For the comparison of the change in pre- and post-study weights, analysis of covariance was used since the second weight is highly dependent on the first. This is due both to a statistical " regression upon the mean " and to the large relative weight loss incurred by disadvantaged Philippine children in their first year of life and the slight relative weight gain in older children.c The net effect is for those above the median to tend to lose relative weight and for those below the median to tend to gain relative weight in time. For the purposes of classifying age groups (under 1 year and 1-5 years of age), age on 15 September was used. Comparability of study and control communities Table 1 compares the two communities as regards some social, sanitary, and health characteristics. The level of education was high. Community B had a higher prevalence of breast feeding than commu- b " Timbang " means weight; this was a national project to weigh all preschool-age children in each community in order to increase the communities' awareness of the nutrition problem and to identify children needing immediate assist- ance. C CAEDO, M. M. ET AL. Progress report, Integrated Nutrition Family Planning Program. Manila, US Agency for International Development/Philippine Commission on Popu- lation, 1972. EFFECT ON NUTRITION OF AN ORAL GLUCOSE-ELECTROLYTE SOLUTION Table 1. Comparison of the two communities item a Community B CommunityM(study) (control) 1. No. of households/No. of families 1398/1730 712/801 2. Population 9479 4490 3. Persons/house 6.8 6.3 4. Families poor or destitute 71 % 77 % 5. Families with outside well 98 % 99 % 6. Families with toilets 36 % 30 % 7. Mothers with 6th grade educa- tion or higher 81 % 77 % 8. Infants breast fed at time of survey 50 % 34 % P< 0.01 9. Families enrolling children in official nutrition programmes 34% 75 % P< 0.001 10. Mean pre-season weight (% of median) of children < 5 years old 77 % 80% P= 0.06 11. Children who weigh less than 75 % of median 35 % 31 % 1 2. Source of first care for diarrhoea (% responding): Family/friends 55 % 46 % Allopathic 32 % 39 % Traditional 10 % 12 % None 4 % 2 % a Items 4-9 and 12 are based on a random sample survey of one third of all families. nity M at the time of the pre-study survey, yet participation in other official nutrition programmes was significantly greater in M and the pre-study weight in M was marginally higher than in B. Undernutrition was common in both communities: a third of the children weighed less than 75 %Y of the Philippine median. About half of those who were breast-fed were weaned before the age of 6 months. The manner of seeking care for diarrhoea was the same in the two communities: half of the population first consulted friends or members of the family and shared proprietary medicines. Table 2. Diarrhoea attacks in the two communities, July 1975-January 1976 Item Community B CommunityM(study) (control) 1. 2. 3. 4. Attacks Children Children with 2 or more attacks Deaths 5. Boy: girl ratio 6. Children aged < 1 year 7. Mean duration of attack 8. Mean duration from initial visit 9. Children with watery/frequent/ dysenteric stools 10. Children with other (nonenteric) illness 11. Children referred back to physician 12. Children with visible dehydration 13. Children hospitalized 14. Children with consistently good appetite 15. Days during attack when food taken 16. Duration of vomiting 17. Duration of antidiarrhoeal medicines 18. Volume of oral fluids a 1 9. Weight change during attack (a) with other (nonenteric) illness (b) with enteritis only (c) children < 1 year of age (d) children > 1 year of age 464 341 85 246 178 52 0 1 Pooled data: differences of means not significant 1:1 24% 4.4 days 2.2 days 76 %/62 %/1 3 % 19% I 10% 22% 1.8% 56% 3.0 Differences of means significant, P< 0.05 0.4 days 0.2 days 2.3 days 3.5 litres + 129g + 60g (NS) b + 140 g + 61 g (NS) + 149 g 2.6 days 2.5 litres + 74 g + 50 g + 80 g + 62 g + 77 g a Includes Oresol in community B; breast milk not included. b NS = not significant. RESULTS Diarrhoea among children attending clinic During 7 months of observation, 519 children had 710 separate bouts of diarrhoea (Table 2). The approximate ratio of 2: 1 between communities B and M reflects the proportions of the population at risk. The average attack was clinically similar in both communities. In general, illnesses were acute and mild to moderate in severity, with dehydration 89 90 INTERNATIONAL STUDY GROUP (usually mild) in 22% of cases. Diagnoses of addi- tional, nonenteric illness (chiefly infectious) were made during one-fifth of the diarrhoea attacks. Hospital treatment was required for 13 of the attacks (7 from B, 6 from M), nearly all for dehydration or persistent diarrhoea. Of the 7 hospitalized patients from community B, 6 had taken less than 500 ml per day of Oresol; 1 may have had glucose intoler- ance (large intake, voluminous diarrhoea, and con- tinued volume depletion). The duration of vomiting was slightly greater in community B. The appetites of the patients and the number of days on which food was taken were the same in the two communi- ties. The average weight gain in community B (129 g) was nearly twice that in M (74 g). However, when children under 1 year of age and those with other, nonenteric illnesses (two independent populations) were compared, no differences were found between B and M even though the Oresol intakes on a weight basis were, in these two subgroups, higher than the mean (Table 3). Table 3. Average consumption of Oresol (ml/kg body weight) in community B Amount consumed: per attack 151 (range 0-1630) per No. of attack a 1st 129 2nd 193 3rd 195 4th 342 5th-6th 401 by age (years) b < 1 204 1 230 2 198 3 146 4 117 5 80 with other (nonenteric) illness 172 with enteritis only 146 a Consumption during the first attack differed significantly from the average consumption during attacks 2-6 (P< 0.001). b Average consumption at ages < 1-2 years differed significantly from the average consumption at ages 3-5 years (P< 0.02). Characteristics of Oresol consumption Table 3 shows the intake of Oresol during an average diarrhoea attack. An intake of 151 ml per kg of body weight averaged 1.3 litres per attack. The range of intake was wide, 0-11 litres. Intake in- creased with successive attacks, both on a weight basis and in absolute terms, so that during the third to the sixth attacks the intake was 2-3 litres per attack. Intake on a weight basis varied inversely with age. Oresol did not replace other fluids; on the contrary, the correlation between Oresol intake and the consumption of other fluids in community B was positive (r = 0.43; P<0.01). Weight change from the beginning to the end of the study Table 4 presents the prestudy and final relative weights and the difference in children from both the study and the control communities. The data are presented by age (children under 1 year and those 1-5 years of age) and by number of diarrhoeal attacks (children with one attack and those with two or more). The children under 1 year of age showed consider- able relative weight loss, a pattern reported previ- ously for disadvantaged Philippine children.a Analysis of covariance indicated that Oresol (study community B) accounted for a gain of 3 percentage points towards the median; this was statistically significant, however, only in those 1 year of age and over.b The effect on final relative weight of attending clinic B for two or more attacks was over twice as great as for a single episode, amounting to a gain of 5 percentage points towards the median. In those children with paired weights, there was no correlation between relative weight change during the season and weight change during individual attacks of diarrhoea. a CAEDO, M. M. ET AL. Progress report, Integrated Nutrition Family Planning Program. Manila, US Agency for Inter- national Development/Philippine Commission on Popu- lation, 1972. b One would have preferred to include as an additional covariate the pre- and post-study relative weights of children not using the clinic to allow for additional, unknown variables that might differentially affect weight in the two communities. To include such data as a single covariate, however, requires the clinic non-users to be identical in other respects to the clinic users within each community. Unfor- tunately, this was not the case: only 44°% of the non-users had paired weights, the children initially being bigger, older, and better cared-for as a group than the users. Diarrhoeal disease was usually treated at home and was likely to have been milder, and some children in community B were likely to have received Oresol from neighbours. EFFECT ON NUTRITION OF AN ORAL GLUCOSE-ELECTROLYTE SOLUTION 91 Table 4. Pre- and post-study body weights as a percentage of the Philippine median Community B CommunityM (study) (control) 1. By age < 1 year (N =38) (N = 25) pre-study (p) a 87.7 93.9 final (F) a 81.0 80.7 difference a -6.7 -13.2 AC:bF - 41.5+0.4P+2.9 COMVAR C r = 0.69(NS)'d 1-5 years (N = 173) (N = 100) pre-study (P) 75.1 76.3 final (F) 80.1 77.7 difference +5.0 + 1.4 AC: F-32.5+0.6P+3.1 COMVAR r = 0.68 (P< 0.001) 2. By attack, children with: one attack (N = 162) (N = 92) pre-study (P) 77.2 79.9 final (F) 80.3 79.1 difference + 3.1 -0.8 AC: F= 44.4 + 0.4 P + 2.3 COMVAR r-0.65 (P< 0.025) two or more attacks (N = 49) (N = 33) pre-study (P) 78.1 79.6 final (F) 80.4 76.0 difference + 2.3 -3.6 AC: =33.1 +0.5P+5.2 COMVAR -0.62(P< 0.01) For community B: F = 45.2 + 0.5 P -0.3 EPVAR e r = 0.62 (NS) For community M: = 43.2 + 0.5 P -3.0 EPVAR r = 0.70 (P< 0.05) a Pre-study weight (P) = weight in June-July 1975; final weight(F) = weight in February 1976; difference = F-P. b AC analysis of covariance. CCOMVAR = Community Dummy Variable = I if study, 0 if control. d NS = not significant. e EPVAR = Episode (attack) Dummy Variable = 1 if two or more attacks, 0 if one attack. Effect of diarrhoea on weight 1-2 months later If diarrhoea affects growth, the effect should be seen soon after an attack. We were able to analyse 80 attacks that occurred in the 2 months prior to the post-study mass weighing. Table 5 shows that 1-2 months after a bout of diarrhoea, children in the control community M had lost relative weight whereas those in the study community B had gained relative weight; the difference, 4 percentage points towards the median, is marginally significant. Table 5. Effect of an attack of diarrhoea on body weight (percentage of the Philippine median) in the following 1-2 months Community B Community M(study) (control) (N = 45) (N = 35) Weight at start of attack (Decem- ber 1975 or January 1976), WT, 79.8 77.9 Weight 1-2 months later (Feb- ruary 1976), WT2 82.7 76.8 Difference + 2.9 -1.1 P< 0.05 Analysis of covariance: V2 = 1.48 + 0.97 WTi + 3.9 COMVAR a (P< 0.06) a COMVAR = Community Dummy Variable = 1 if study, 0 if control. Feeding practices during diarrhoea Feeding practices were assessed by interview. In both communities taken together, the proportion of those who replied " never " to the question: " Do you stop feeding during diarrhoea?" increased sig- nificantly from one fifth to about half of the respon- dents. However, a significantly larger proportion of respondents in community M (54%) gave this reply at the post-study survey than in the study commu- nity (40%, P<0.05). A slightly lower percentage of mothers of children who did not use the clinic gave this reply than mothers of clinic users (P<0.05). In a sample of 140 families in the control community M, only 2% knew of Oresol and children from 42% of these families, with a history of diarrhoea, attended the health centre. In a sample of 292 families in study community B, 50% knew of Oresol but chil- dren from only 37% of these families, with a history of diarrhoea, attended the health centre. DISCUSSION This study was designed to test the hypothesis that an oral glucose-electrolyte solution (Oresol), used INTERNATIONAL STUDY GROUP freely for outpatient children with diarrhoea, would improve nutrition. Both the study and the control communities were given nutrition education and stress was placed on feeding and the provision of fluids during attacks of diarrhoea. This strategy could be expected to obscure any additional effect Oresol might have; nevertheless, the data support the following conclusions: 1. Oresol was associated with a relative weight gain over a 7-month period in relation to the Philippine median. The increase over the control group due to Oresol averaged 3 %, which was statis- tically significant in the 1-5-year-old group but not in those under 1 year of age. An increase in relative weight of 5% was seen in the study children with two or more bouts of diarrhoea. The effect on relative weight was apparent 1-2 months after an attack of diarrhoea; again, children in the control community lost relative weight while those in the study commu- nity gained. These findings are consistent with Martorell's observation in Guatemala that children who were ill more than 5 Y. of the time with diarrhoea lost weight; incremental growth was unaffected by less frequent disease, by respiratory illness, or by fe- ver (1). Binns (5) also showed, in Papua New Gui- nea, that diarrhoeal episodes resulted in either a loss of weight or lack of growth in 44% and 390% of children, respectively, in the month of illness. Mar- torell compared the weights of children who had diarrhoea frequently with those who did not. When the differences between each 6-month age-cohort (up to the age of 84 months) were summed they amounted to 11 00, which is about the same cumula- tive value that we calculated for our 1-year age- cohort groups from ages 1 to 5. Other studies (6, 7) have shown that nutritional rehabilitation centres dealing with seriously malnourished children pro- duce an average gain of about 4-60% towards the median over 3-6 months. However, these values were uncorrected for regression effects and may therefore be inflated. 2. Children given Oresol gained nearly twice as much weight during an attack of diarrhoea as those not given Oresol. That this was not simply due to salt and water retention is suggested by the failure of Oresol to produce a greater weight gain in children under 1 year of age or in those with additional nonenteric infections, in spite of a larger than aver- age ingestion of Oresol in both subgroups. These children are said to be especially susceptible to salt and water retention (8, 9). The increased weight gain with Oresol may reflect better hydration, lower catabolism, or both. The relationship between weight gain during an attack and long-term relative weight gain is unclear. The possibility that the effects on weight gain- especially long-term gain-were due to a " Haw- thorne effect" (i.e., the effect of intervention alone) must be considered. That possibility is less likely than some effect of Oresol itself for the following reasons: the actual contact time was brief and limited to diarrhoea attacks; feeding and fluid administration during diarrhoea were emphasized to both groups; no differences between the groups as to appetite during attacks were detected; and the clini- cal events of the average attack were nearly identical in both groups except for Oresol intake and weight gain. Oresol is likely to have improved the children's appetites and food intake following the attack by early treatment of volume, sodium, and potassium deficits, each of which is a cause of anorexia or catabolism (10-12). The failure of Oresol to affect relative weight significantly in those aged less than 1 year may have been due to the fact that diarrhoea is a less important cause of relative weight loss than early weaning and insufficient feeding. 3. This study presents the results of considerable field experience with an oral glucose-electrolyte mix- ture used freely for mild or moderate diarrhoea in infants and children. Except for one child, who may have had glucose intolerance, no adverse effects were found. Evidence was found for the dissemination of knowledge of Oresol but within the study commu- nity only. 4. Hospitalization was not significantly reduced by Oresol administration, largely due to the design of the study. However, six out of the seven children from community B who were hospitalized actually failed to take adequate amounts of Oresol. Where the ingredients for an oral glucose-electrolyte solu- tion are not available or prescribed, the introduction of packaged powders like Oresol could reduce the incidence of serious dehydration. ACKNOWLEDGEMENTS The many helpful reviews of our colleagues are appreciated, as well as the dedication of the field staff. 92 EFFECT ON NUTRITION OF AN ORAL GLUCOSE-ELECTROLYTE SOLUTION 93 R1SUMt EFFET POSITIF SUR LA NUTRITION D'ENFANTS PHILIPPINS D'UNE SOLUTION E'LECTROLYTIQUE GLUCOSE'E ORALE ADMINISTREE A DOMICILE POUR LE TRAITEMENT DE LA DIARRHEE La diarrhee est chez les enfants une importante cause de denutrition, les mecanismes qui interviennent etant le catabolisme, la malabsorption, l'anorexie et le jeune impose. Des observations faites anterieurement sur des enfants hospitalises pour diarrhee avaient montre qu'en comblant rapidement le deficit hydro-electrolytique, on ameliorait I'appetit et on prevenait les pertes de poids. I1 ressort de la presente etude que 1'etat nutritionnel d'enfants en traitement ambulatoire progresse apres une rehydrata- tion maintenue grace a une solution electrolytique glucos6e administree par voie orale. Les populations etudi6es sont celles de deux communautes peri-urbaines desheritees habitant au bord du detroit de Guimaras (Negros occi- dentale) aux Philippines. La communaute M a servi de groupe temoin, la communaute B de sujet d'etude. Quoique de dimensions differentes, ces deux communau- tes sont comparables du point de vue de la situation socio- economique et sanitaire ainsi que de l'etat nutritionnel; toutefois, dans la comminaut6 M, I'allaitement au sein est relativement moins pratique mais le taux de frequenta- tion des centres officiels de nutrition est beaucoup plus eleve. Pendant une periode de sept mois, tous les enfants de moins de cinq ans souffrant de diarrhee qui ont et amenes au dispensaire ont requ des medicaments anti- diarrheiques ordinaires, cependant qu'on indiquait aux meres la necessite de ne pas interrompre l'alimentation et de faire prendre des liquides, des visites journalieres de controle 'a domicile etant organis&es. Dans la commu- naute B, on a en outre administre de l'Oresol, solution electrolytique glucosee orale, tant que durait la diarrhee. Presque tous les enfants exposes ont e peses au debut et a la fin de 1'etude; les malades l'ont ete le jour de la pre- miere visite au dispensaire et le jour suivant la fin de la diarrhee. Les donnees cliniques ne font apparaitre aucune difference significative entre les deux communautes, y compris pour le regain d'appetit et l'hospitalisation. Les enfants de la communaute B ont pris environ deux fois plus de poids que ceux de la communaute M au cours d'un episode moyen (129 grammes contre 74), pheno- mene associe a l'absorption d'une quantite de liquide nettement plus elevee. La dose d'Oresol reque a et en correlation positive avec les doses d'autres liquides et avec le nombre d'episodes par enfant mais en correlation inverse avec l'age. Les enfants de un a cinq ans de la com- munaute B ont aussi pris notablement plus de poids en sept mois que ceux de la communaute M (environ 3 %Y de plus vers la mediane). Chez les enfants ayant subi deux episodes ou davantage, la difference d'evolution du poids attribuable a l'Oresol a e d'a peu pres 5%. Un a deux mois apres une diarrhee, les enfants de B avaient progresse en moyenne de 3 % vers la mediane, tandis que ceux deM avaient perdu environ 1 %. Les effets a long terme sur le poids sont semblables ou legerement superieurs a ceux indiques par les centres de recuperation nutri- tionnelle. On estime qu'une action energique de reequilibration hydro-electrolytique chez les enfants en traitement ambu- latoire pour une diarrhee, meme benigne, a un effet immediat et a long terme positif sur le poids, probable- ment en commencant par retablir l'equilibre hydro- electrolytique puis en influant favorablement sur l'appetit a l'issue de 1'episode. REFERENCES 1. MARTORELL, R. ET AL. Acute morbidity and physical growth in rural Guatemalan children. American journal ofdiseases of children, 129: 1296-1301 (1975). 2. HIRSCHHORN, N. & DENNY, K. M. Oral glucose-elec- trolyte therapy for diarrhea: a means to maintain or improve nutrition? American journal of clinical nutri- tion, 28: 189-192 (1975). 3. BYAR, D. P. ET AL. Randomised clinical trials: perspectives on some recent ideas. New England journal of medicine, 295: 74-80 (1976). 4. Treatment and pievention of dehydration in diarrheal diseases. Geneva, World Health Organization, 1976. 5. BINNS, C. W. Food, sickness and death in children of the Highlands of Papua, New Guinea. Journal of tropical pediatrics and environmental child health, 22: 9-11 (1976). 6. BEAUDRY-DARISME, M. & LATHAM, M. C. Nutri- tional rehabilitation centers-an evaluation of their performance. Journal of tropical pediatrics and environmental child health, 19: 299-332 (1973). 7. BEGHIN, I. D. & ViTEiu, F. E. Nutritional rehabilita- tion centres: an evaluation of their performance. Journal of tropical pediatrics and environmental child health, 19: 403-416 (1973). 94 INTERNATIONAL STUDY GROUP 8. TALBOT, N. B. ET AL. Homeostatic limits to safe parenteral fluid therapy. New England journal of medicine, 248: 1100-1108 (1953). 9. BART, K. J. & FINBERG, L. Single solution fro oral therapy of diarrhoea. Lancet, 2: 633-634 (1976). 10. MCCANCE, R. A. Experimental sodium chloride deficiency in man. Proceedings of the Royal Society of Medicine, 119: 245-268 (1936). 11. LEAF, A. & COUTER, W. T. Evidence that renal sodium excretion by normal human subjects is regulated by adreno-cortical activity. Journal ofclini- cal investigation, 28: 1067-1081 (1949). 12. SCHWARTZ, W. B. & RELMAN, A. S. Metabolic and renal studies in chronic potassium depletion resulting from overuse of laxatives. Journal of clinical investi- gation, 32: 258-271 (1953).