F ZUMRAWI,* J P V A U G H A N , " J C WATERLOWt and B R KIRKWOODtt Zumrawi F [Department of Home Science, University of Khartoum, Democratic Republic of Sudan], Vaughan JP, Waterlow JC and Kirkwood BR. Dried skimmed milk, breast-feeding and illness episodes — a controlled trial in young children in Khartoum Province, Sudan. International Journal of Epidemiology 1981, 10: 303—308. In a controlled trial mothers and children attending urban maternal and child health (MCH) clinics in Khartoum Province were given a fortnightly take home food supplement of 1 Kilogram of dried skimmed milk (DSM) or an equivalent amount of local beans. There were approximately 3 0 0 children aged 6—26 months in each group and each child was followed for 3 to 6 months. A comparison of the two groups showed: a) that the DSM group mothers were more likely to continue breastfeeding; b) there was no evidence to associate DSM with an increased incidence of episodes of diarrhoea, fever or vomiting; c) the utilisation of health institutions was very similar in the two groups; and d) there was no significant difference in the mothers' assessment of the proportion of children with a 'poor' appetite in either group. This trial met with considerable methodological problems and the results should therefore be interpreted cautiously. There is a great need for more and better designed trials to assess the possible adverse effects of DSM. In a recent review of food aid, Maxwell and Singer3 found that about 66% was sold by recipient countries on their domestic markets to extend food supplies and to generate funds, 16% was used on food for work projects, 11 % was used in supplementary feeding programmes and 7 % was used for emergency relief. As the production of DSM in North America and Europe considerably exceeds current requirements, excess stocks are being accumulated. It is likely therefore, that the use of DSM will increase in aid schemes and for emergency food supplies. In a recent authoritative review undertaken for UNICEF on supplementary feeding programmes, many of which used DSM, doubts were raised about the evidence for any benefits of such programmes for young children in developing countries. We undertook this controlled trial on the distribution of DSM in mother and child health (MCH) clinics: (a) to provide information on the possible harmful or beneficial effects of DSM on young children and (b) to gain information on the methodological problems associated with such studies undertaken in developing countries. Another publication examines the effect of DSM on child growth.s This trial was designed as a contribution to elucidating the current world controversy on the interrelationships of breast feeding, powdered milks and child development.1 Strong criticisms have been expressed on the possible harmful effects of distributing powdered milks in Third World Countries, by discouraging breast feeding and promoting diarrhoea! disease, malnutrition and infant deaths. The widespread distribution of dried skimmed milk (DSM) in aid programmes imposes on obligation on the governments, UN agencies and charitable organisations which support these programmes to monitor whether these harmful effects are also associated with DSM. * Department of Home Science, University of Khartoum, Democratic Republic of Sudan. ** Evaluation and Planning Centre, Ross Institute, London School of Hygiene and Tropical Medicine, Gown Street, London, WC1, UK. t Department of Nutrition, London School of Hygiene and Tropical Medicine, London, UK. tt Tropical Epidemiology Unit, Ross Institute, London School of Hygiene and Tropical Medicine, London, UK. Reprints from- J P Vaugban 303 Downloaded from http://ije.oxfordjournals.org/ at London School of Hygiene & Tropical Medicine on January 21, 2016 Dried Skimmed Milk, Breast-Feeding and Illness Episodes- a Controlled Trial in Young Children in Khartoum Province, Sudan 304 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY METHODS feeding practices and on the mother's assessment of her child's appetite and activity. MCH Clinic and borne follow-up visits After the initial visit the mother and the child should have been seen regularly at 2 week intervals, alternately at the MCH clinic and at home. If the mother and child were not seen on 2 consecutive occasions an urgent home visit was made to reduce the non-attendance and drop-out rates. At each fortnnightly follow-up the mother was asked if the child had had any episodes of diarrhoea, fever or vomiting in the previous 2 weeks and if so whether any visits had been made to a health institution. Final follow-up In early July 1979 the mothers were seen at the clinic for the last time and were asked whether they were breast-feeding or not, and for their assessment of their child's appetite and activity. The child's height and weight were also recorded. RESULTS Sample Mothers attending the MCH clinic for the first time with a child aged 6—26 months were accepted into the trial. The children's ages were taken as those given by the mothers. We attempted to exclude children who had clinical signs of malnutrition, a weight for age of less than 60% Harvard standard, or who had been unwell during the previous 2 days. The number of children who entered the trial was 646, of whom 626 were followed for between 3 to 6 months. There were finally 287 children in the DSM group and 339 in the Beans group. First clinic visit At the first clinic visit the nutrition worker weighed the child to the nearest 100 grams with the standard UN1CEF infant beam scales and measured the height in centimetres with the child lying down, using a standard UNICEF infantometer. Other routine clinic procedures were followed and the clinic growth card given to the mother with advice noted on it. Then 1 Kg of DSM or beans was given free to each mother and an arrangement made for a home interview in the next few days. The anthropometric results will be reported separately. Initial borne interview The nutrition worker visited the child's home and administered questionnaires to the mother to obtain information on socio-economic and demographic baseline data, household conditions, breast- The mothers were accepted into the DSM group at a faster rate than into the Beans group due to some initial supervision difficulties, and so there were more child months of follow-up in the DSM group. However, since the Beans mothers attended the follow-ups more regularly, the total number of recorded follow-ups was very similar in both groups. Each child was followed-up for between 3 and 6 months. Less than 5% of mothers missed more than 2 of the monthly clinic follow-ups and only 2.4% did not answer both the first and the final breastfeeding questionnaire. Only 6% of children did not have a complete set of height and weight measurements. The percentage of male children in the DSM and Beans groups was 49.1% and 53.7% respectively and the age distribution of the children in the 2 groups was very similar. A comparison of some of the socio-economic and- demographic variables for the 2 groups is given in Table 1. The DSM group had a lower educational background of parents, a lower average family expenditure on food, and less adequate household water supplies, toilet facilities and general household conditions. This apparent disadvantage of the DSM group makes an interpretation of the results of the trial more difficult. At the start of the trial significantly more of the DSM mothers (78.4%) were breast-feeding compared to the Beans group (64.9%) (p<0.00i). Approximately 70% of both groups of mothers said they Downloaded from http://ije.oxfordjournals.org/ at London School of Hygiene & Tropical Medicine on January 21, 2016 MCH Centres The cities of Khartoum, Khartoum North and Omdurman were selected for this trial, because they had well-organised MCH clinics at which there had been a regular distribution of DSM, provided by a charitable organisation for some time. Eight MCH clinics in urban health centres were selected and then paired for the general similarities of the areas they served. Two pairs were in Omdurman and one in each of Khartoum and Khartoum North. One clinic in each pair distributed DSM and the other acted as a control. It was decided that it would have been unethical to withhold all supplements and so local beans were distributed to the control group. A female nutrition worker was attached to each of the MCH clinics for the duration of the trial and 2 senior nutritional workers separately supervised the 4 DSM and 4 Beans clinics. One of us (FZ) was in overall supervision. The trial was carried out between December 1978 and July 1979. 305 INFANT FEEDING TRIAL IN SUDAN TABLE 1 A companion of some family and household characteristics of the DSM and Beans groups. Beans Mother hiving had primary education or more 44.3% 60.8% xxx Families who had had one or more children who had died 32.0% 26.9% N.S. Families with 4 or more live children 49.6% 45.0% N.S. Average monthly family income (in Sudanese pounds) £71.2 £78.1 N.S. Average monthly expenditure of food £37.5 £44.8 xxx Water tap connected in house 76.0% 85.5% xx Flush toilet in house Household judged .to be dirty by interviewer 1.4% 7.7% xxx 15.2% 6.6% xxx (N.S. = not significant, xx p< 0.01 and xxx p< 0.001). TABLE 2 The observed and expected numbers of mothers stopping breast-feeding during the trial in the DSM and Beans groups. (Both sexes children combined together). Number in group DSM Beans 212 208 Observed number stopping breastfeeding during trial 32 39 Expected number stopping 43.12 27.88 0.74 1.40 Relative stopping rate X 1 = 8 . 5 7 , d f - 1, p< 0.005 gave fresh cows and/or goats milk to their children each day. Risk of mothers stopping breast-feeding The logrank6 test was used to compare the rate of stopping breast-feeding in the 2 groups of mothers, taking into account the ages of each child and their varying length of follow-up. Complete information was available for 420 mothers who were breastfeeding at the start of the trial (DSM = 212 and Beans = 208) and 71 of these stopped breastfeeding during the trial, 32 in the DSM and 39 in the Beans groups (see Table 2). The following summary for all age groups shows the number of mothers who actually stopped breastfeeding and the number that would have been expected to stop if the stopping rate had been the same in the 2 groups. The Beans group mothers were significantly more likely to stop breast-feeding than those given DSM (p< 0.005) or conversely, those given DSM were more likely to continue breast-feeding. A possible reason that might explain this difference is that the Beans group appeared to be at a socio-economic advantage compared to the DSM group at the start of the trial. The analysis was therefore repeated taking into account the influence of the following variables: family expenditure on food of more or less than £S40 per month, whether the mother had had any formal education and whether the household had a piped water supply. The difference in the rate of stopping breastfeeding between the DSM and Beans group was confirmed, which suggests that this may well be a real difference and not one due to confounding variables. Incidence of reported attacks of diarrhoea, fever and vomiting The number of mothers answering the illness episode questionnaire in each 2 week follow-up period was found and the percentage of children having had an episode of diarrhoea, fever or vomiting in that period was calculated. At the start of the trial there were substantial Downloaded from http://ije.oxfordjournals.org/ at London School of Hygiene & Tropical Medicine on January 21, 2016 DSM 306 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY Attendances at health institutions Mothers reporting an episode of diarrhoea were analysed for their subsequent attendances at a dispensary, health centre, hospital, pharmacy and traditional or private practitioner (see Table 3). Although the total number of visits in the DSM group was twice the Beans group, this was found to be the case at the very start of the trial and corresponds with the higher reporting rates for illness episodes in the DSM group. In approximately 60% of diarrhoea episodes the mother and child attended a health centre and only 1—2% went to dispensaries. Approximately 10% did not attend any centre. The remaining 30% was about equally distributed between hospitals, pharmacies, and traditional and private practitioners. The analysis showed that the proportions varied only slightly for diarrhoea and fever. For vomiting, however, the percentage of hospital visits was almost double that for diarrhoea or fever, and pharmacies were consulted proportionately less often. The percentage attendances in the 2 groups were remarkably similar although it should be remembered that the DSM group were apparently using the services about twice as frequently even at the start of the trial. The fact that the percentage of hospital attendcrs in the 2 groups remained so similar during the trial is evidence against any difference in the severity of the illness episodes in the 2 groups. This is particularly pertinent for diarrhoeas because of their association with severe dehydration and subsequent infant mortality. Mother's assessment of their child's appetite and activity The mothers were asked to assess their child's appetite and activity as good, normal or poor at the beginning and end of the trial. The analysis is based only on those mothers who answered both the initial and final questionnaires. The results for a 'poor appetite' are shown in Table 4 for all mothers and separately for those who were breast-feeding throughout the trial. The results for activity were very similar. At the start of the trial more mothers in the DSM group rated their child's appetite (and activity) as poor and there appears to be an improvement at the end of the trial in both groups. The differ- TABLE 3 Reported visits made to bealtb institutions following episodes of diarrhoea (visits expressed as a percentage of all visits shown in brackets). DSM (n = 593) Dispensary Health centre Beans (n = 262) 1.5 0.4 61.9 61.8 Hospital 6.9 6.1 Pharmacy 9.8 8.8 Traditional practitioner 3.2 4.2 Private practitioner 4.2 6.5 No visits 12.5 12.2 TOTAL 100.0 100.0 Downloaded from http://ije.oxfordjournals.org/ at London School of Hygiene & Tropical Medicine on January 21, 2016 differences between the DSM and Beans groups in the frequencies of reported episodes of diarrhoea, fever and vomiting, which may have been due to faulty interview techniques, observer bias and/or to the socio-economic differences between the 2 groups. This reporting difference makes the interpretation of the results more difficult. Both the DSM and Beans groups showed a gradual decline over the period of the trial in the percentage of children having had diarrhoea, which was probably due to a fall off in reporting and/or the effect of seasonal variation. The proportional difference between the 2 groups did not, however, increase or decrease in any significant manner. The analysis was repeated for children breastfed throughout the trial and again no significant differences between the incidence of diarrhoea in the 2 groups was found. It was concluded that despite the overall difference between the two groups in reporting rates there was no evidence for an association of increased reporting of diarrhoea with cither DSM or Beans, even taking into account breastfeeding practices. The same conclusions were reached for episodes of fever and vomiting. 307 INFANT FEEDING TRIAL IN SUDAN TABLE 4 By all mothers %Poor at: By mothers who breast-fed throughout n start finish n start finish ences that existed between the 2 groups at the start of the trial, together with the lower response rate of the Beans group of mothers to the questions in the final questionnaire, makes any conclusions difficult, but there is no good evidence of a real difference between the 2 groups (p >0.05). However, there is a suggestion, particularly for those mothers breast-feeding all through the trial, that appetite may have been rated as poor more often in the DSM group then the Beans group. All those appetites rated as 'poor' at the start of the trial in the Beans group (15) showed an improvement whereas only about three quarters (30 out of 36) of those in the DSM group improved, although this was not a significant difference (p X).05). DISCUSSION Most studies that have attempted to evaluate DSM in 'take home' supplementary feeding programmes have concentrated on the beneficial effects on nutritional status, particularly child growth. Much less attention has been paid to the possible harmful effects and yet such considerations arc perhaps more important for policy decisions on whether DSM should be used or not. In view of the current controversies there is an urgent need for more well controlled studies to assess these possible harmful effects. It must be emphasised that in this study the mothers were given the DSM and beans to 'take home' and we have no means of knowing what actually happened to these food supplements. Also DSM was compared to another food (locally purchased beans) and not with an untreated or 'pure' control group. It was considered to be unethical to withdraw a food supplement from mothers coming to clinics where DSM had been distributed for several years. It was also quite clear that attendance at the clinics was strongly encouraged by such supplements. DSM Beans 283 164 14.8 4.6 175 13.7 4.0 9.1 0.0 91 12.1 0.0 The results of this study must be accepted with caution because the DSM and Beans groups were not strictly comparable on several important background variables, and also because so much of the information had to be collected by interviewers using questionnaires and much of the information was based on memory recall. Also the effects of interviewer bias must be taken into account. The study was explained to the nutrition field workers, who were inevitably already aware of the 'Baby Killer' controversy, and therefore observer and interviewer bias against DSM may explain some of the differences between the groups in the illness episodes data. Also the answers given by the mothers may well have been influenced by being interviewed by nutrition workers. The significant difference in the rate of stopping breast-feeding appears to be a definite finding and is the opposite way round to what might have been expected. Either the DSM supplements were actually encouraging mothers to continue breastfeeding and without them they would have stopped, or the receipt of Beans actually encouraged those mothers about to stop actually to do so. It is difficult to know which of these 2 possibilities is most likely without the use of simple controls. The lack of evidence in this study to associate DSM with an increased incidence of diarrhoea, fever or vomiting is also contrary to what is popularly held, but it is necessary to remember how difficult it is to detect such changes unless they are quite large and also that their incidence was different between the 2 groups at the very start of the trial. The lack of evidence to associate DSM with diarrhoea is also supported by the fact that the utilisation of health institutions, particularly hospitals, was so similar between the DSM and Beans group throughout the trial. Most evaluation studies of supplementary feeding programmes have used 'objective' measurements Downloaded from http://ije.oxfordjournals.org/ at London School of Hygiene & Tropical Medicine on January 21, 2016 % Poor at: Percentage of children's appetites that were rated as poor by their mothers at the start and finish of the trial. 308 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY ACKNOWLEDGEMENTS This study only became possible because of generous financial support from the Overseas Development Administration (ODA) of the United Kingdom Government. We owe many thanks to the Sudanese Ministry of Health, the Assistant Commissioner of Health for Khartoum Province and Dr Mohamed Hassan, all of whom were very helpful in loaning support, staff and equipment. Our special thanks go to Miss Samira Hamo of the Ministry of Health and Miss Sana Arbab of the Ministry of Social Affairs who supervised the field work. Our thanks must also go to Mrs Shirley Harrison of the Tropical Epidemiology Unit for all her special efforts in computing and analysing this study. REFERENCES World Health Organization. Statement and recommendations on infant and young child feeding. Development Dialogue 1980; l i 102-119. Chetley A. The Baby Killer Scandal. London: War on Want, 1979. Maxwell SJ and Singer HW. Food Aid to Developing Countries: A Survey. Discussion paper published by the Institute of Development Studies, University of Sussex, Brighton, UK, 1978. Beaton GH and Ghassemi H. Supplementary Feeding Programmes for Young Children in Developing Countries. Report prepared for UNICEF and the ACC Sub-committee on Nutrition of the United Nations. UNICEF, New York, USA, 1979. Vaughan JP, Zumrawi F, Waterlow JC and Kirkwood BR. An evaluation of dried skimmed milk on children's growth in Khartoum Province, Sudan. Nutrition Research 1981; 1. 2 4 3 - 2 5 2 . Peto R, Pike MC, Armitage P, Breslow NE, Cox DR, Howard SV, Mantel N, McPherson D, Peto J and Smith PG. Design and analysis of randomised clinical trials requiring prolonged observation of each patient. BrJ Cancer 1977; 35i 1-39. (Revised version received 27May 1981) Downloaded from http://ije.oxfordjournals.org/ at London School of Hygiene & Tropical Medicine on January 21, 2016 and very little thought has been given to measuring how mothers (who are after all very astute observers) and families assess their children. Although appetites appeared to improve in both the DSM and Beans group children, there was still a worrying proportion in the DSM group whose appetite did rot appear to improve. The results of this trial lead us to question some of the assumptions now generally held about the effects of DSM on breast-feeding and in producing diarrhoea episodes. Further trials are needed in this highly controversial area where there is so little substantial evidence. However, such studies need to incorporate proper 'untreated' control groups and to pay attention to the many methodological problems involved.