An integrated community-based outpatient therapeutic feeding programme for severe acute malnutrition in rural Southern Ethiopia: Recovery, fatality, and nutritional status after discharge.

A scaled up and integrated outpatient therapeutic feeding programme (OTP) brings the treatment of severely malnourished children closer to the community. This study assessed recovery from severe acute malnutrition (SAM), fatality, and acute malnutrition up to 14 weeks after admission to a programme integrated in the primary health care system. In this cohort study, 1,048 children admitted to 94 OTPs in Southern Ethiopia were followed for 14 weeks. Independent anthropometric measurements and information on treatment outcome were collected at four home visits. Only 32.7% (248/759) of children with SAM on admission fulfilled the programme recovery criteria at the time of discharge (i.e., gained 15% in weight, or oedema, if present at admission, was resolved at discharge). Of all children admitted to the programme for whom nutritional assessment was done 14 weeks later, 34.6% (321/928) were severely malnourished, and 37.5% (348/928) were moderately malnourished; thus, 72.1% were acutely malnourished. Of the children, 27/982 (2.7%) had died by 14 weeks, of whom all but one had SAM on admission. Children with severe oedema on admission had the highest fatality rate (12.0%, 9/75). The median length of admission to the programme was 6.6 weeks (interquartile range: 5.3, 8.4 weeks). Despite children participating for the recommended duration of the programme, many children with SAM were discharged still acutely malnourished and without reaching programme criteria for recovery. For better outcome of OTP, constraints in service provision by the health system as well as challenges of service utilization by the beneficiaries should be identified and addressed.

children with SAM and can be safely fed to children with SAM at their homes (Kapil, 2009;Nutriset, 2010). Moreover, simplified guidelines for identifying and managing SAM were developed to support the community health workers responsible for these activities in the community (Chamois, 2009;FMOH, 2007).
Although there are reports of good recovery rates from such OTPs (Ashworth, 2006;Chaiken et al., 2006;Chamois, 2009;Collins et al., 2006;Collins & Sadler, 2002), programmes integrated to government health system at the beginning of scaling and integration of OTP had recovery lower than the small-scale programmes (Hedwig, Swindale, Grant, & Navarro-Colorado, 2008). This reflects the knowledge that the context into which nutrition interventions are integrated is modifying their effect (Victora et al., 2005). Reports on the performance of programme after the scaling up and integration into government health systems are scarce. Further, the focus of studies has commonly been the children's health and nutritional status at discharge, usually 6-8 weeks after admission, and little is known to what extent the benefits of the programme are sustained beyond discharge. A concern is also that most of these reports are based on data excerpted from programme records and not by independent assessments (Chaiken et al., 2006;Collins, 2001;Collins & Sadler, 2002). Set in Southern Ethiopia in an area of government-implemented OTP, the aim of our study was to use independently collected data to assess recovery, fatality, and acute malnutrition on and beyond discharge from the programme.

| Study design and setting
This cohort study was conducted from July to December 2011 and included children admitted to OTPs in 94 health posts in four adjacent districts of Wolaita Zone in Southern Ethiopia. With a mostly rural population (88%), the area is characterized by a high population density (342/km 2 ), a fragmentation of farmland ownership and limited incomegenerating opportunities (Hailu & Regassa, 2007;Teklu, 2003). Moreover, recurrent drought and subsequent crop failure in the area have resulted in cyclical nutritional emergencies and chronic food insecurity (Hailu & Regassa, 2007;The Socioeconomic Profile of Southern Nations, Nationalities and Peoples Region, 2000). Food insecurity gets worse during preharvest months (May-July) and starts to improve during harvest (August-November) and postharvest (December-February) time.
Thus, our study period covers a range of food insecurity exposures.
From 2008, the Ethiopian government has scaled up and integrated the OTPs into the existing public health system, including the health posts (Chamois, 2009). Health posts are the lowest tier in the Ethiopian health system hierarchy, serving approximately 5,000 people and being staffed by two female health extension workers, who provide basic curative and preventive health services (Wakabi, 2008). Although the World Health Organization (WHO) has recommended a cut-off at mid-upper arm circumference (MUAC) < 115 mm (WHO, 2013) to define SAM, the Ethiopian national SAM management protocol uses a MUAC cutoff <110 mm or oedema to admit children into the OTP (FMOH, 2007). After admission, the children's caregivers are to be provided with RUTF, medication, and counselling, after which they return home to manage the child with SAM on their own (FMOH, 2007).Visits to the health post for check-ups and refilling of RUTF are usually scheduled weekly.

| Enrolment and follow-up procedures
A weekly visit to all the health posts in the selected districts was undertaken to identify children participating in the OTP. Children aged from 6 to 59 months, from whom the research team collected anthropometric data within 7 days of admission, were eligible to participate in the study.
Data were collected at four households visits, the first within 1 week of admission and then after 4, 8, and 14 weeks. The follow-up at 4 and 8 weeks was selected to capture anthropometric change while in OTP and at around expected time of discharge (termination of treatment) as programme allowed a maximum participation of 8 weeks. The follow-up at 14 weeks was selected arbitrarily to examine acute malnutrition after allowing a reasonable time after expected time of discharge for some change in nutritional status to have occurred.

| Data collection and instruments
Children's MUAC was measured using WHO-recommended MUAC tape and techniques, and the presence of bilateral pitting oedema was assessed using the recommended techniques (WHO & UNICEF, 2009).
Weight (to the nearest 0.1 kg) was measured using UNICEF-recommended digital electronic scale. A structured questionnaire that included questions about socio-demographic information on household, caregiver, and children was used to collect data. Twenty-three female nurses collected data and were trained in anthropometric measurement techniques including repeated standardization sessions to ensure accuracy and precision of measurements according to established guidelines (FMOH, 2007).

| Outcome measures and statistical analysis
The presence of acute malnutrition was defined by use of MUAC or presence of oedema according to WHO and national recommendations (FMOH, 2007;WHO et al., 2007). Children with oedema on admission were further categorized into two groups based on severity:

Key messages
• Children with severe oedema at admission who should have been referred to inpatient care had a high case fatality rate.
• The vast majority of children with severe acute malnutrition on admission failed to reach programme recovery criteria despite programme stay of recommended duration.  FMOH, 2007;Sphere Project, 2011). According to Sphere, the definition of recovery from SAM is based on SAM management guidelines stipulating gain of 15% of weight at admission for nonoedematous children and resolution of oedema for children with oedema in order to discharge children (FMOH, 2007;WHO & UNICEF, 2009). Average weight gain is a programme indicator that according to the national guideline is calculated as weight gained (g) per body weight at admission (kg) per day for children who recovered (FMOH, 2007). We also calculated average weight gain among all children admitted with SAM regardless whether they had recovered or not. The average weight gain indicator was not calculated for children with oedema at Children admitted to the programme n=1659 Assessed for study eligibility n=1640 Not assessed for eligibility Late registration on programme record (n=19) Assessed within 7 days of admission n=1125 Lost to follow-up n=57 Refusal to participate n=5 Caregiver not at home n=9 Child not found at home n=25 Unknown n=18 Eligible for study n=1480 Excluded: Not assessed within 7 days from admission (n=355) Health post closed (n=182) Data collectors failed to visit (n=135) Child not found (n=38) Analysed for nutritional status on admission n=1048 Excluded: Information on key variables missing n= 77 Ever discharged from OTP in 14 weeks n= 898 Analysed for nutritional status at 14 weeks after admission n=928 Oedema/MUAC data at 14 weeks missing n=36 Dead n=27 Analysed at14 weeks after admission n=991 Remained admitted to OTP in 14 weeks n= 66 Data were entered into Epi-Info (Version 6.0), cleaned, and exported to SPSS for Windows (Version 20. 0) for analysis. Children with missing key variables (MUAC or oedema status on admission, age, and sex) were excluded from analysis. Data were explored for levels of missing values, and almost all factors had missing value less than 5%. Length of programme stay and weight gain were not normally distributed when normality test (Shapiro-Wilk) was applied; thus, log transformed and normality were obtained for both variables. Nutritional status on admission, discharge, and 14 weeks after initial admission were presented as proportions of children with 95% confidence intervals (CI) for categorical outcomes.

| Ethical considerations
The research protocol was reviewed and approved by the regional health bureau of South Ethiopia, district health bureaus, the Ethical

Review Board at Addis Continental Institute of Public Health, Addis
Ababa, and Uppsala Regional Ethical Review Board. The Helsinki Declaration was applied in the research project, including gaining of informed consent.

| RESULTS
As shown in Figure 1, children with acute malnutrition were followed from admission in a community-based therapeutic programme up to 14 weeks. Of 1,480 eligible children (6-59 months of age), 355 were excluded from the study because anthropometric assessments were not done within 7 days of admission, and the provided nutritional therapy might have changed the baseline nutritional status. The main reasons for these losses were that health posts were closed so data collectors could not get access to OTP records to look for new admissions (n = 182) and that the data collectors failed to visit mainly due to time constraints in the field (n = 135). Further, of 1,125 children included in the study, nutritional status on admission was analysed for 1,048 children due to missing of key variables (n = 77). During the 14 weeks of the study, 57 children were lost to follow-up mainly as they were not found (n = 25), and 27 children had died. The final analysis was performed on 928 children for whom the nutritional status was measured at follow-up 14 weeks after the initial admission.
The relative large number of children excluded from analyses was compared with children included, and no important differences in terms of baseline characteristics as child age, sex, or caregivers' and household characteristics were found.

| Demographic and socio-economic characteristics of households and caregivers
Most of the participants lived in houses with thatched roof and wood and mud or grass walls (74.7%), had an open-pit latrine (75.5%), and collected water from a public tap or a protected well or spring (77.0%). At the time of the study, 42.4% of the households had more than one child less than 5 years of age. The caregivers of children admitted to the OTP were mostly biological mothers, married, and on average, they were 30.6 years old. The majority of children were below 24 months of age (62.3%), and the female-to-male ratio was 1.3 (Table 1).

| Nutritional status of children on admission to OTP
When using either the oedema or MUAC criteria to define SAM, 78.8% (826/1,048) of the admitted children had SAM on admission, and thus, When all children admitted with SAM were considered, the mean length of stay in the programme was 6.8 weeks (95% CI [6.5, 7.0]).
The median length of admission to the programme was 6.   (Table 3).

| DISCUSSION
Despite adequate stay in the programme, more than one-third of discharged children (35.8%) were severely malnourished, and 37.7% were moderately malnourished on discharge. Only 32.7% of SAM children had gained the required 15% of admission weight or had oedema resolution at discharge. Fourteen weeks after admission to the programme, 4 out of 10 children with SAM on admission remained severely malnourished and almost the same amount progressed to moderate acute malnutrition; thus, three-quarters were acutely malnourished. Moreover, children with severe oedema on admission, that is, complicated SAM, were managed at the OTP and had a high fatality rate.
We use the same definition of recovery as programme and previous studies, that is, achieving the targeted 15% weight gain or oedema resolution at discharge (FMOH, 2007;WHO, 2013). When compared to the reported recovery rates of smaller-scale OTPs that were not integrated into the government health system (Chaiken et al., 2006;Chamois, 2009;Collins et al., 2006;Manary, Ndkeha, Ashorn, Maleta, & Briend, 2004;Manary & Sandige, 2008), our finding of only 32.7% recovery is unacceptably low. Recovery from SAM can be delayed or hindered if the children's intake of RUTF is inadequate that could be due to limited supply by the providers (Tadesse, Ekström, & Berhane, 2016). The OTPs we studied had no external technical and financial support and were part of the government routine health services.
Programmes that depend on the existing health system for provision of resources are known to suffer from shortage and stock-out of essential programme inputs including RUTF (Puett & Guerrero, 2015). During our 6-month stay in the field, we have noticed persistent shortage of RUTF because of lack of transportation of RUTFs from zonal/district health bureau to health posts.
Furthermore, because the actual rehabilitation of SAM takes place at the home of caregivers where chronic food insecurity often prevail and livelihood is precarious, the actual amount of RUTFs consumed by SAM children can be less than the amount provided due to sharing with other children in the household or use of RUTF as a commodity for meeting the household economic and food needs (Collins & Sadler,  2002; Tadesse, Berhane, Hjern, Olsson, & Ekstrom, 2015). Thus, an improvement in the nutritional status of children admitted to the OTP greatly depends on the household contexts and caregivers' perceptions and abilities of usage of RUTFs (Tadesse et al., 2015).
The use of percentage gain as the discharge criterion is suggested to "disfavour" the recovery of children with the lowest MUAC (Dale, Myatt, Prudhon, & Briend, 2013;Forsen, Tadesse, Berhane, & Ekstrom, 2013) and may contribute to the high proportion of children still suf- Most studies assessing SAM management in OTP are based on data excerpted from programme records and performed by organizations involved in the programme implementation. To maximize the quality of collected data, our study based the results on an independent research team that collected data and assessed the children's nutritional status. The research team was not involved in any aspect of programme service provision or other health and nutrition interventions in the study area. All data collectors went through careful training in anthropometric measurements including standardization sessions prior to and throughout the data collection period. A potential limitation of this study was the inclusion of all discharged children in the analysis of programme outcome regardless of longer duration between discharge date and our nutritional assessment after discharge of children from OTP. However, because there was no significant difference in proportion of children with SAM according to duration since discharge, our finding of low recovery (achieving programme criteria for recovery) is likely to be a valid estimate of achievement of programme recovery criteria.

| CONCLUSION
Despite children participating in the OTPs for the recommended duration of the programme, many were discharged without reaching the required weight gain and still acutely malnourished. Children with severe oedema on admission had a higher fatality rate than those with nonoedematous SAM. Further research is needed to understand the contextual factors that constrain the management of SAM at service points and at home.

ACKNOWLEDGMENTS
We like to sincerely thank the caregivers of the SAM children, the CHWs, and the district health offices in the study areas. We are grate-

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest.

CONTRIBUTIONS
EE and YB conceived and designed the study. ET and AW performed the field work, and EE and YB provided supervision throughout the research and writing process. ET carried the analysed data and drafted the paper in close collaboration with EE. All authors revised and approved the final version of the paper.