Community Based Management of Severe Acute Malnutrition: The MSF Experience from an Urban Slum Setting in Bangladesh

Background: Until recently, the experience on implementing community based 
management of acute malnutrition (CMAM) among children has been largely based 
in African settings. While the government in Bangladesh is yet to scale up CMAM 
approach, there is still paucity of knowledge on the experience of CMAM within 
the complex milieu of an urban slum context. In Kamrangirchar slum, Dhaka, 
Bangladesh, this paper describes a CMAM programme performance and outcomes run 
by Medecins Sans Frontieres (MSF)/Doctors without Borders, in light to 
performance indicators set by MSF and the Sphere minimum standards. Methods: This was a descriptive retrospective study using routinely collected programme 
data of children admitted with severe acute malnutrition between May 2010 and 
November 2011. Kamrangirchar is an urban slum of a large migrant population in 
Dhaka, Bangladesh. Results: There was a total of 640 new admissions, of 
whom 333 (52%) were males. The median age was 18 months (Inter-quartile range 
(IQR) 12 - 41). 599 children had a reported nutritional outcome at discharge 
from ambulatory therapeutic feeding centre (ATFC), this included: cure rate of 
69% with an average length of stay of 68.8 (SD ± 46.0) days and average weight 
gain of 3.8 g/kg/day (SD ± 2.7). The lost-to-follow-up rate was 18% and 5% 
reported to the programme that they will leave the slum and go back to their 
villages. These performance indicators did not meet the threshold level 
indicators set by MSF and Sphere standards. Conclusions: Our experience 
highlights the need for developing more adapted and contextualised indicators 
for assessing the performance of CMAM programmes in settings such as urban 
slums. Community engagement in the process of developing relevant standards is 
crucial. Nutrition humanitarian actors have a vital role to collaborate with 
local authorities to contextualize and refine these standards.


Introduction
Children in slums are at high risk of severe acute malnutrition (SAM) and its associated morbidity and negative effects on their growth and cognitive development later in life [1]. In Bangladesh, an alarming rapid urbanization has forced poor rural population to migrate to urban areas, particularly in the capital city Dhaka, settling down in poor overcrowded informal slum settings. It is predicted that half of the population in Bangladesh will be living in urban settings by 2050 [2]. Children in overcrowded urban slums are particularly vulnerable to malnutrition and disease due to poor living condition and lack of access to essential basic services [3]. This is a major public health concern in Bangladesh which requires well adapted and targeted interventions, as the country has one of the highest burdens of children with acute malnutrition in the world [4]. According to the latest Bangladesh Demographic and Health Survey (BDHS) in 2014, 14% of children under 5 were identified to be wasted [5].
The innovation of community-based management of acute malnutrition (CMAM) has been endorsed by the World Health Organization (WHO) to overcome the challenges of inpatient based approach such as low coverage, high mortality, and lost-to-follow-up [6] [7]. In brief, CMAM involves early detection of SAM children in the community: nutritional rehabilitation with ready-to-use therapeutic food (RUTF) (or other nutrient-dense foods) and ambulatory follow-up of children within the community. Children are only referred for inpatient care if they have medical complications. The widespread adoption of CMAM strategy has proved to improve geographic access, considerably reduce the burden on health services, foster community empowerment and enhances task-shifting initiatives [8] [9].
Assessing the performance of CMAM services supports on-going and future programmes through generating evidence on CMAM experiences. This is often performed by comparing to nutrition sector global performance indicators, including the widely used Sphere minimum standards [10]. The latter is a cornerstone reference in humanitarian practice, which provides universal benchmark to guide relief efforts in five sectors that dominate humanitarian settings, including nutrition, food aid, shelters, water and sanitation and health services [11]. Although the Sphere standards are claimed to be universal, they were largely based on experience in relief programmes, particularly in Africa. Therefore, the utility and relevance of the use of these standards in the complex milieu of urban slums

Study design
This was a descriptive retrospective study using routine programme data of children with severe acute malnutrition (SAM) enrolled in MSF nutrition programme between May 2011 and November 2012.

Study setting and population
The CMAM programme was located in Kamrangirchar-an urban slum setting in Dhaka. It is the largest slum cluster in the Dhaka Metropolitan Area with an estimated population of 400,000 inhabitants living in an area of 3.1 km 2 . It is officially not part of Dhaka city and there were no governmental health structures in the slum. Provision of health care services was outsourced to non-governmental organisations. The majority of households in this highly dense slum are tin-sheds with shared latrine and cooking area. During the severe rainfall, water running through the latrines also makes the slum inhabitants more prone to water-borne diseases. The majority of inhabitants are migrant population from rural areas, and engaged in low income revenue activities, mostly waste recycling, ready-made garment (RMG) factories, informal manufacturing industry (e.g. tanneries) and in service jobs such as rickshaw pullers [14]. Most of the RMG factories are located in the vicinity of Dhaka which is a huge driving force for migration from rural to deprived informal urban slums in Dhaka including Kamrangirchar. The majority of the RMG workers are women with long working hours and low wage, so young children are often left alone or with their elder siblings for long hours [15].
MSF has been providing health services in Kamrangirchar through two primary health care (PHC) centres and all services were offered free of charge. The and f) death. The main outcomes were compared to the existing MSF [17] and Sphere minimum standards [10].
Box: Programme's admission categories and nutritional outcomes Admission categories: • New admission: A malnourished child admitted to the nutritional program for the first time • Readmission: A child who is readmitted to the nutritional programme after being lost to follow-up for less than two months or having been transferred out to hospital.
• Relapse: A child who presents back to the nutritional programme with severe acute malnutrition within two months of discharge as cured.
• Lost-to-follow up: Absent for more than two consecutive weeks from the community and the ATFC follow up visits.
• Left Kamrangichar: Caretaker informing the programme that the child will leave the slum and return back to the village.
• Non responder: Child not attaining the target weight after more than 90 days of treatment after exclusion of underlying medical and social conditions.
• Transferred to hospital: Referral needed to inpatient health care due to deterioration.
• Died: Died whilst registered in the programme.

Characteristics of the study population
There was a total of 640 new admissions in the nutrition programme, of  Table 2).
Of the 412 children who were discharged as cured, 16 (4%) presented with relapse during the study period of one year. The nutrition outcome of readmitted children and those presented with relapse are shown in Table 2.

Discussion
To  First, the programme had an overall recovery rate of 69% which is below the 75% minimum set by The Sphere standards and the 80% by MSF. This programme cure rate was in line with a large scale CMAM project in Mumbai slums, India that reported a 56% cure rate [18], and lower than the 91.9% cure rate reported in another CMAM programme in rural Southern Bangladesh [1].
The underlying reasons for the low cure rate in our study lies in the fact that nearly one-in-four new admissions were either "lost to follow-up" or had actually "left Kamrangirchar" to another setting, and thus negatively influenced the achievable cure rate. Loss-to-follow up was high despite the presence of a well-supported and resourced community follow-up and tracing system. Moreover, CHWs were operating within a relatively small geographic area of four km 2 . One might therefore think that such losses could not be avoided considering the predominantly migrant setting. It might thus be realistic to assume that 15% -20% of all new admissions would end up as a combination of "lost to follow-up" and "left Kamrangirchar" as observed in our experience. This then begs the question about what the target cure rate of an ATFC should be in such a slum setting? If the acceptable threshold for death is maintained at <10%, then an achievable operational threshold for the cure rate in such a setting would be 70% in a best case scenario.
An additional limitation of existing thresholds is the lack of any indicator for deterioration rate expressed as transfer out to hospital for inpatient care. Such rates are subject to factors such as the baseline severity of malnutrition, the medical condition of the child and the capacity and quality of clinical management. High transfer rates to inpatient facilities will negatively influence the cure rate and programme performance, these need to be monitored.
Second, beside a considerable proportion of the population in Kamrangirchar are migrants in transitory phase, the high lost to follow-up rate observed in the programme, might be due to household factors. The majority of mothers were form the core of "daily waged workers", mainly in RMG factories. Such work requires working long hours with limited availability of extended family [15], making attendance at the scheduled weekly follow-up appointments difficult.
Adaptation of ATFC open hours and follow-up schedule (e.g. fortnightly visits) might be needed in this kind of setting. We also observed temporary displacements of population for social and culture events. This included migration during the harvest season or travel of pregnant women back to home-villages for delivery. The latter is a distinct group which needs to be distinguished from the classical "lost to follow-up" category as it involves formal contact with the programme staff with a clear justification for discontinuation of the nutritional care for their children. Such an outcome is so far unreported in the literature but merits being included in standard nutritional outcomes for such settings. cohort. This could be strongly linked to the mediocre or poor compliance to the provided peanut based RUTF which proved to have low acceptability among caretakers and CHWs in our programme [19]. Care takers availability and food sharing within the family could be also contributing factors to this long ALS.
These factors need to be explored in order to set an appropriate threshold for the ALS in such setting.
Fourth, we used fixed community outreach clinics to offer ATFC services despite the small coverage area of three km 2 . This was justified on the basis that the small geographic slum area is characterized by winding and congested roads and a dense population making travel difficult for mothers and children. In retrospect, the down-side of fixed clinics is that it requires displacement from home, time loss and associated indirect costs albeit to a lesser extent than hospital based care. A study from rural Bangladesh demonstrated that well-trained and supervised CHWs were able to effectively manage uncomplicated cases of SAM.
This CMAM approach was centred mainly on the CHW identifying SAM children in the community and providing follow-up and treatment at home. The approach was also well accepted by the care takers [20]. A similar approach would seem the right way to go forward in the slums.
Fifth, in a setting where stunting has been prevalent for decades, there is a general perception that malnutrition is not a "disease". As such, attention is given to these children only when an associated morbidity develops. This may also explain why seven of ten children admitted to the nutritional programme came in through the routine screening at the PHC. A qualitative study from a rural context in Bihar, India, showed that undernutrition was seen not as a disease per se but as a result of different disease conditions and was often related to witchcraft [21]. It is crucial to understand these aspects which might have substantial impact on the uptake and attrition in the CMAM programme. Increased community awareness and empowerment is needed to address this issue. New messages and approaches to communicate effectively in such context need to be also discovered.
Sixth, the relapse rate was 4%, despite an important focus in the programme on nutrition education and cooking demonstration. Through anecdotal data of health promotion team during routine discussions with mothers and caretakers, we discovered that low purchasing power of families is an obstacle to adherence to nutrition counselling: even if mothers knew what is good for their child, they often could not afford it. On the other hand, the definition of relapse is still poorly understood and undefined [22]. A performance indicator for deciding acceptable levels of relapse also needs attention.

Conclusion
In conclusion, our experience from an urban slum setting in Bangladesh has highlighted the need to develop more adapted and contextualized indicators for assessing the performance of CMAM programme in settings such as urban slums. Community engagement in the process of developing relevant standards to their context is crucial. Nutrition humanitarian actors have a vital role to collaborate with local authorities to contextualize and refine these standards.

Funding
The study was funded by MSF-Brussels Operational Centre.