Collins, S., Sadler, K., (2002), Outpatient care for severely malnourished children in emergency relief programmes: a retrospective cohort study, The Lancet (360).
Full published article here
Summary
Background
In emergency nutritional relief programmes, therapeutic feeding centres are the accepted intervention for the treatment of severely malnourished people. These centres often cannot treat all the people requiring care. Consequently, coverage of therapeutic feeding centre programmes can be low, reducing their effectiveness. We aimed to assess the effectiveness of outpatient treatment for severe malnutrition in an emergency relief programme.
Methods
We did a retrospective cohort study in an outpatient therapeutic feeding programme in Ethiopia from September, 2000, to January, 2001. We assessed clinical records for 170 children aged 6–120 months. The children had either marasmus, kwashiorkor, or marasmic kwashiorkor. Outcomes were mortality, default from programme, discharge from programme, rate of weight gain, and length of stay in programme.
Findings
144 (85%) patients recovered, seven (4%) died, 11 (6%) were transferred, and eight (5%) defaulted. Median time to discharge was 42 days (IQR 28–56), days to death 14 (7–26), and days to default 14 (7–28). Median rate of weight gain was 3·16 g kg−1 day−1 (1·86–5·60). In patients who recovered, median rates of weight gain were 4·80 g kg−1 day−1 (2·95–8·07) for marasmic patients, 4·03 g kg−1 day−1 (2·68–4·29) for marasmic kwashiorkor patients, and 2·70 g kg−1 day−1 (0·00–4·76) for kwashiorkor patients.
Interpretation
Outpatient treatment exceeded internationally accepted minimum standards for recovery, default, and mortality rates. Time spent in the programme and rates of weight gain did not meet these standards. Outpatient care could provide a complementary treatment strategy to therapeutic feeding centres. Further research should compare the effectiveness of outpatient and centre-based treatment of severe malnutrition in emergency nutritional interventions.
Jump to SectionIntroductionMethodsResultsDiscussionReferences
Introduction
In 2000, widespread food insecurity and famine spread across Ethiopia. As in many years, humanitarian agencies scaled up operations to implement relief programmes. Nutritional interventions included distribution of general rations to all members of affected populations, blanket supplementary feeding to all members of an identified risk group, targeted dry supplementary feeding programmes for moderately malnourished people, and therapeutic feeding centres for severely malnourished people.
In many areas of Ethiopia, therapeutic feeding centres have been slow and difficult to set up and did not achieve adequate coverage to treat all severely malnourished people. Although there are no data for the effect of specific therapeutic feeding centre programmes, this combination of delays and low coverage probably reduced the effect of therapeutic feeding centres on mortality.
The recommendations of the Sphere project, developed by more than 150 leading humanitarian organisations, are the sole set of internationally accepted standards, and cover all sectors involved in humanitarian interventions. Sphere standards are centre oriented, and do not include indicators for programme coverage or effects of programmes on health infrastructures and communities. Consequently, limitations in areas of care provided by therapeutic feeding centres are rarely shown in programme assessments.
Guidelines for emergency interventions state that provision of inpatient treatment in therapeutic feeding centres is the only way to treat severely malnourished individuals.1, 2, 3, 4 Therapeutic feeding centres often provide high quality care for patients. Inpatients receive Formula 75 and Formula 100 (Nutriset; Malaunay, France), which are appropriate therapeutic foods, in quantities individually tailored to metabolic needs. Systematic medical and supportive care complements this treatment. Such intensive care is essential for initial treatment of patients with complicated malnutrition associated with anorexia, septicaemia, hypothermia, hypoglycaemia, and severe dehydration.5
However, therapeutic feeding centres have huge requirements for resources, skilled staff, and, often, imported therapeutic products,1 which makes them expensive and often too small to attain high population coverage.6 This approach to care could undermine local health infrastructure, disempower communities, and promote congregation of people. Congregation of severely malnourished patients promotes centre-acquired infections, a major difficulty in many feeding centres. Additionally, congregation of communities around therapeutic feeding centres undermines public health-care systems, resulting in mortality and morbidity during famines.7 Admission of a patient to a therapeutic feeding centre usually involves the carer, most often the mother, leaving the family for around 30 days, which reduces families' productive capacity and provision of maternal care for siblings. These negative side effects of therapeutic feeding centres tend to promote poverty and malnutrition in siblings.
Although untried in emergency relief programmes, researchers in stable situations have shown that home-based treatment of severe acute malnutrition is effective and cheap. In Bangladesh, the recovery rate of severely malnourished children treated with 1-week's inpatient care followed by home management was similar to that of children in a specialised nutrition unit.8 Home-based treatment was more than four times cheaper than centre-based care, and was preferred by mothers.9 At 1 year's follow-up, children treated at home had less morbidity than those treated in centres.10
The need for a therapeutic food that can be used safely and stored in the homes of severely malnourished people has complicated home-based treatment. Researchers have developed a ready-to-use therapeutic food (RUTF: plumpynut; Nutriset, France), designed to be nutritionally equivalent to Formula 100,5, 11 that can be used easily and stored safely for several months in a simple, opaque, airtight container (A Briend, personal communication). RUTF is sold as a paste that patients can eat directly from the packet, and does not require mixing or cooking. Preliminary trial results suggest that RUTF is popular with malnourished children and resistant to bacterial contamination.12
Bedawacho Woreda is a district 350 km south of Addis Ababa in the region of Ethiopia governed by the Southern Nations Nationalities and Peoples Regional Government (SNNP). It is a rural area, on the western escarpment of the Rift Valley, ranging from 1650–2150 m above sea level. The population is around 231 000 (200–550 people/km2), most of whom practise subsistence rain-fed agriculture mainly of maize, enset (false banana), sweet potato, and beans, combined with some animal husbandry.
By July, 2000, Bedawacho had had 3 consecutive years of drought and failed harvests, which had led to widespread food insecurity. In August, 2000, Concern Worldwide, a development non-governmental organisation (NGO), implemented a nutritional survey using methods recommended by WHO:4 17% of children aged 6–60 months were malnourished (less than −2 z scores below the reference weight-for-height ratio from the National Centre for Health Statistics13) and 5% were severely malnourished (less than −3 z scores, pitting oedema, or both) (Concern Worldwide, personal communication). In response to these findings, the NGO started a decentralised dry supplementary feeding programme from ten distribution points, all of which were at government-run primary health-care clinics or health centres. 3·5 kg of Famix (a blend of maize, soya, sugar, minerals, and vitamins; Fafa, Addis Ababa, Ethiopia) and 750 g of soap was distributed fortnightly to every child with a weight-for-height ratio less than 80% of the reference. The food ration provided every child with 3·843 MJ/day.
During the first weeks of the feeding programme, 160 children were identified as severely malnourished. The children were dispersed across the Woreda district, much of which was inaccessible. Ideally, we would have started treatment with several days in a phase-1 unit. However, inpatient care facilities were not available in the district, and the regional government did not want agencies to set up new therapeutic feeding centres. Consequently, patients were treated as outpatients with only once-per-week attendance at clinics or distribution sites.
We aimed to assess the effectiveness of outpatient treatment using RUTF for severe malnutrition in emergency relief programmes.
Jump to SectionIntroductionMethodsResultsDiscussionReferences
Methods
Participants
We included 170 patients aged 6–120 months. Between Oct 16, 2000, and Jan 31, 2001, patients were enrolled at any one of the ten supplementary feeding programme distribution points in Bedawacho, all of which were located in government-run primary health-care clinics or health centres. Inclusion criteria were a weight-for-height ratio of 70% or less than the reference ratio, or bilateral pitting oedema. There were no exclusion criteria.
Procedures
The outpatient therapeutic feeding programme was staffed by a nurse supervisor and 12 outreach and distribution workers, recruited and trained during the programme. Five nurses employed by the government worked with the NGO teams after receiving 2 days' training on treatment of severe malnutrition from NGO staff.
One outpatient therapeutic feeding programme site was set up at each of the ten supplementary feeding programme sites. Each participant in the outpatient programme returned to their closest site weekly to receive food and a medical assessment. Every fortnight during the supplementary feeding programme distribution, the outpatient participants were examined and given a 2-week ration of Famix and a weekly supply of RUTF. On the weeks between the supplementary feeding programme, children were examined by local clinic staff and given a 1-week ration of RUTF. NGO outreach workers followed up patients at home once or twice weekly. They checked children's progress and referred ill children back to a clinic.
Diagnosis of severe malnutrition was by physical measurement and brief examination for bilateral pitting oedema in the feet. At admission, the clinic nurse assessed degree of pitting oedema, hydration, dysentery, diarrhoea, anaemia, and signs of chest infection. At each follow-up visit, a nurse recorded weight, extent of pitting oedema, presence of disease, and drugs prescribed, which were recorded on patients' treatment cards. Outcome (death, discharge alive, default, or transfer) was recorded on individual patient's cards and in the programme's registers.
Staff used Salter spring scales and locally constructed height boards to weigh and measure patients. Weights were recorded to the nearest 100 g and heights to the nearest 5 mm. Scales were calibrated before and after the programme using a 1 kg weight, and were regularly adjusted to zero.
At admission, patients were given an oral dose of vitamin A (100 000 IU for children <12 months, 200 000 IU for older children); an oral dose of mebendazole (250 mg for children aged 12–24 months and 500 mg for all other age groups); and an oral dose of folic acid (5 mg). Patients received a 5-day course of co-trimoxazole (25 mg of sulfamethoxazole and 5 mg of trimethoprim/kg orally every 12 h). Dehydration was treated with ReSoMal (Nutriset; Malaunay, France), an oral rehydration solution designed for the treatment of severe malnutrition. On alternate weeks, patients received 5 mg folic acid orally.
At admission, an educational message sheet was discussed with groups of patients' mothers. The message was reinforced by the community outreach team during home visits. Education focused on the feeding regimen and on reducing the risk of complications arising from severe malnutrition.
At admission, each child received 1·28 kg of RUTF (1 week's supply) and a 2-week ration (7·5 kg) of Famix. At weekly clinic visits, patients were given 1·28 kg of RUTF in 14·92 g packets. The RUTF was made from vegetable fat, peanut butter, skimmed milk powder, lactoserum, maltodextrin, sugar, and mineral and vitamin complex (table 1) and provided 4·2 MJ of energy and 24·3 g of protein/day (protein-to-energy ratio 10%). Patients also received 7·5 kg Famix, which provided 8·26 MJ/day and 79 g of protein/day (16%).
Patients were discharged from the outpatient therapeutic feeding programme to the supplementary feeding programme if their weight-for-height ratio was more than 75% of the reference ratio for 2 consecutive weeks and they did not have infectious disease. After discharge into the supplementary feeding programme, outreach workers followed up patients until their weight-for-height ratio was more than 85% of the reference ratio for 2 consecutive weeks. Patients were then discharged from the programme.
Data analysis
Weight and height data, length of stay in the programme, presence of oedema, number of siblings, and outcome were transferred from inpatients' record cards to Epi-info (version 6·04) and Microsoft Excel 97 by use of Epidata data entry programme. Pitting oedema was coded as present or absent. Outcome data were coded as “recovered”, “died”, “transferred at the end of the programme”, or “defaulted”. “Recovery” was defined as discharged from the outpatient therapeutic feeding programme into a supplementary feeding programme. Patients who did not attend the clinic or the supplementary feeding programme on one occasion were followed up at home by the outreach workers. If they were absent on two consecutive occasions and home visits indicated that they were still alive at 1 week after their previous attendance, they were recorded as “defaulters”. If they had died within 1 week of attending the programme they were recorded as “deaths”. Patients who were in the programme when it closed were transferred to the nearest health clinic and coded as “transfers”.
Individual weight gains in marasmic patients were calculated with:
For children admitted with oedema, rates of weight gain after oedema had disappeared were calculated with:
Missing data values were coded as missing. Proportions were compared with X2 with Yates continuity correction.
Role of the funding source
The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.
Jump to SectionIntroductionMethodsResultsDiscussionReferences
Results
Admission weight-for-height data was collected for 167 participants and follow-up weight measurements for 169. Median age of participants was 36 months (IQR 24–48) for girls and 36 months (12–48) for boys. 99 (58%) participants were girls, 64 (38%) had oedema caused by malnutrition (50 [29%] had kwashiorkor malnutrition and 14 [8%] marasmic kwashiorkor), and 106 had marasmus. Median weight at admission was 7·25 kg (5·5–9·6) overall: median weight of marasmic patients was 6·0 kg (5·0–7·5), marasmic kwashiorkor patients 7·0 kg (6·1–8·0), and patients with oedema 9·9 kg (8·3–12·7). Overall, mean weight-for-height ratio at admission, in z scores, was -3·05 (SD 0·77). Mean scores at admission were -3·34 (0·47) for marasmic, -3·77 (0·41) for marasmic kwashiorkor, and -2·20 (0·68) for kwashiorkor admissions. Mean scores at admission for all survivors were -3·03 (0·77) and -3·65 (0·63) for patients who died (table 2).
Table 2
Characteristics of population at admission
*Data missing for three patients.
†Data missing for one patient.
‡Data missing for two patients.
Table 3 shows outcome data. 144 patients (85%) recovered, with a median time to recovery of 42 days (28–56). Seven patients (4%) died, median time to death was 14 days (7–26). Eight patients (5%) defaulted, their median stay in the programme was 14 (7–28) days.
Table 3
Outcome data
*Data missing for one patient.
†Data missing for two patients.
Table 4 shows recovery, mortality, and default rates compared with the Sphere standard indicators14 for therapeutic feeding. Recovery and default rates were significantly better than Sphere indicators (table 4). Mor–ality rates were lower than the Sphere standard, but were not significant (p=0·057).
Table 4
Sphere key indicators14 for therapeutic feeding programmes and study results
*p=0·057.
†p=0·031.
‡p=0·003.
Fatality rates were 5 (5%) of 106 for marasmic patients, 2 (14%) of 14 for marasmic kwashiorkor patients, and 0 for patients with kwashiorkor. Median time to clinical resolution of oedema was 28·0 days (21·0–35·0): 28·0 (24·5–38·5) for marasmic kwashiorkor and 27·0 (21·0–35·0) for kwashiorkor patients. Overall, median rate of weight gain was 3·2 g kg−1 day−1 (1·9–5·6). In patients who recovered, median rates of weight gain were 4·8 g kg−1 day−1 (3·0–8·1) in marasmic, 4·0 g kg−1 day−1(2·7–4·3) in marasmic kwashiorkor, and 2·7 g kg−1 day−1 (0–4·8) in kwashiorkor patients.
At the end of the programme, 11 patients were transferred to government clinics after a median stay in the programme of 70 days (42–105). Mean weight-for-height z scores on discharge to the supplementary feeding programme was -1·88 (0·67).
Jump to SectionIntroductionMethodsResultsDiscussionReferences
Discussion
Our results show that under certain circumstances, outpatient treatment of severe malnutrition is feasible in humanitarian relief programmes and can meet international minimum standards for recovery, mortality, and default rates.
Mortality and recovery rates seemed better than internationally accepted minimum standards for therapeutic feeding centres (table 4). However, cure and mortality rates vary with the epidemiological profile and dynamics of emergency situations, and we did not obtain these data. Seasonal factors such as food availability and incidence of malaria probably affected the outcome.
Rates of weight gain seemed lower than international minimum standards (table 3 and table 4). Caution is required in interpretation of these data. To account for the loss of oedema fluid and weight in oedematous patients in the first days of treatment, international standards recommend daily weighing of patients in phase-one treatment and calculation of weight gain by use of the lowest weight recorded. Slow loss of oedema and weekly rather than daily weight assessment in our study meant that the initial drop in weight followed by rapid weight gain, usually seen in oedematous patients, was not observed. Therefore, we calculated weight gains in oedematous patients using the weight assessed on the first day that oedema was recorded as having disappeared. Thus, weight gain was assessed at a phase of recovery when rates would be expected to have fallen below the initial peak.
Lengths of stay in the programme were longer than international recommendations (table 4). Rapid recovery and high rates of weight gain are important features of successful treatment programmes, since severely malnourished patients are immuno-compromised and vulnerable to infection. In outpatient programmes, severely malnourished patients are not removed from home environments and congregated, thus, exposure to foreign pathogens and the dangers of acquired infection are lower than in inpatient treatment, and rate of recovery is a less important determinant of success. However, we did not investigate factors affecting rates of response; since most children remained well during recovery, slow response was probably caused by sharing food rations. More than 25% of marasmic patients gained weight faster than 8 g kg−1 day−1, presumably because the patient received most of the RUTF and supplementary food as well as positive feeding and caring practices. We are researching use of community therapeutic care groups, in which carers of children with high rates of weight gain are teamed up with mothers of children who are doing less well to try to improve feeding practices.6
The default rate was low, less than a third of the minimum standard (table 4), indicating that outpatient treatment was acceptable to participants. We did not obtain data for programme coverage (proportion of the severely malnourished population admitted), which was a serious omission. Direct estimates of programme coverage, made during the programme, should be done in future research.
Not all groups of patients did equally well, and two of the 14 patients with marasmic kwashiorkor died. The dietary regimen could have been inappropriate for these very vulnerable patients. Conventional treatment guidelines recommend an intake of 335–420 kJ kg−1 day−1 during the initial phase of treatment and of more than 840 kJ kg−1 day−1 during the recovery phase. During the early phases of treatment for all severe cases of malnutrition, dietary composition is important, and guidelines stipulate a protein-to-energy ratio of around 5%, less than 8 mg/kg bodyweight of sodium, and no iron. The only internationally available RUTF at the time of this study had a protein-to-energy ratio of 10%, 18 900 mg/kg of sodium, and 11 500 mg/kg of iron (table 1); the ration provided on average 43·9 mg/kg bodyweight of sodium and 2·7 mg/kg of iron. The association between iron, sodium, and oedematous malnutrition is well recognised15 and in similar settings, high-protein diets (protein-to-energy ratio 16%) have been associated with delayed recovery, persistence of oedema, and raised mortality.16 In our study, oedematous patients lost oedema more slowly than patients given Formula 75, a milk-based food, in therapeutic feeding centres—could poor dietary composition have contributed to their delayed recovery? Formal, controlled tests for the effectiveness of RUTF in oedematous patients are needed to answer this question. We used one regimen of around 1600 kJ kg−1 day−1 since we thought it too complicated to teach mothers to use two regimens in the short duration and difficult circumstances of the study. The large quantity of food provided ensured that even if rations were shared, the patient would still have had sufficient food to recover. If, during the first days of treatment, a severely malnourished child had consumed the entire ration, serious metabolic imbalance could have resulted. However, we thought that overfeeding would have been unlikely, since take-home rations are usually shared with other family members.17 Furthermore, only two of the seven deaths were in the first 2 weeks of treatment, the period usually associated with deaths from overfeeding.18
During the study, the people of Bedewacho began harvesting sweet potato and maize, thereby improving the availability of food, which probably lessened the need to share or sell RUTF. Conversely, the peak incidence of malaria in Bedewacho is from September to October, midway through the study, which might have slowed recovery rates and increased mortality.
We did not obtain information for the epidemiological profile of the population, availability of resources in the community, or additional referrals to clinics by outreach workers. Although outreach workers completed monitoring forms that included a section on RUTF consumption and observed use of RUTF we do not know the exact proportion of the RUTF taken by each patient. Carers often reported that the whole RUTF ration had been eaten by the patient, even when it was obvious to the outreach workers that the RUTF had been shared. Furthermore, variation in weight gain (IQR 1·38–5·71 g kg−1 day−1) suggests that patients consumed varying quantities of the RUTF ration. Rates of weight gain at the top of the range are higher than rates achieved in standard supplementary feeding programmes, indicating that children ate a large proportion of the ration. Rates of weight gain at the bottom of the range are closely similar to those in a normal supplementary feeding programme, indicating that the RUTF was shared.
There were further limitations to our study: follow-up weight and height measurements were done by different workers, thereby increasing the chance of inaccuracies; patients were weighed only once each week and not necessarily at the same time of day; admission heights were not recorded for three patients, and discharge weight or height was not recorded for two patients; and our criterion for discharge from the outpatient to the supplementary programme was lower than the usual 80% of reference weight-for-height ratio. All these factors complicate data interpretation and limit the conclusions that can be drawn.
Large-scale studies in which epidemiological and contextual data are obtained are needed to compare the effectiveness of our approach with that of traditional therapeutic feeding centres. Use of the Prudhon index, which adjusts mortality rates in accordance with the severity of malnutrition,19 would also improve the comparability of results.
Marasmic kwashiorkor patients are very vulnerable and have a high risk of death and complications.14 These patients require the intensive care, monitoring, and cautious Formula 75-based feeding regimens of phase-one therapeutic care, which are not possible in outpatient treatment. In the absence of an RUTF designed for phase-one treatment, outpatient feeding programmes should be run in conjunction with suitable inpatient facilities. The reasonable response to outpatient treatment of marasmic and kwashiorkor patients indicates that only severe cases (marasmic kwashiorkor or those with complications) require inpatient care, which need last only for a few days. Phase-one centres could therefore be smaller and more basic than conventional therapeutic feeding centres, easier to construct, and less resource intensive. Our study took place towards the end of a period of severe food shortage. During more severe famine, with high prevalence of severe malnutrition and infectious disease, the proportion of severe and complicated cases would be higher than in our study region, and the need for phase-one inpatient care correspondingly greater.
An advantage of combining outpatient feeding programmes with phase-one centres is that the outpatient element can be started immediately and begin rehabilitating people while the phase-one centres are being built. Phased implementation is not possible with traditional therapeutic feeding centres, which must be adequately constructed, staffed with trained personnel, and properly organised before any patients are admitted.
The combined approach, with small, decentralised, phase-one centres preferably attached to local health structures, has other advantages over large centralised therapeutic feeding centres: small centres where patients stay for only a few days have lower risks of nosocomial infections; reduce the frequency and duration of inpatient treatment, thus decreasing disruption to carers; and allow emergency funding to strengthen local health infrastructures, rather than setting up competing parallel structures. The feasibility of implementing small phase-one centres will vary. In some cases, the necessary infrastructure will already exist, and such centres would require little more than basic refresher training for staff and the provision of essential drugs and supplies. In other cases, setting up such centres will be much more difficult, requiring additional structures and staff as well as training and supplies.
The slow rates of response to outpatient treatment, even with provision of a large excess of food, indicates that additional inputs to improve patient compliance and decrease sharing of rations are needed. Further studies should assess the role of community groups and mother-to-mother (carer-to-carer) educational techniques to improve compliance under such conditions.
Contributors
S Collins conceived and designed the study and did data analysis. The operational and research outpatient therapeutic feeding programme was established and managed by K Sadler. Both authors wrote and revised the article.
Conflict of interest statement
None declared.
Acknowledgments
Concern Worldwide provided staff, management, and logistical support to the operational feeding programme and supported K Sadler during the research project. ECHO Framework Partnership (CCP No 99/0038) financially supported the operational feeding programme in Ethiopia. Torchbox provided support for the research aspects of the project.