Causal modelling to explore malnutrition in children in Bangladeshi urban slums

By Sophie Goudet

Sophie is a nutritionist with over nine years experience with international agencies in nutrition and health within developing countries. Her research interest lies in approaches to tackling infant and young child malnutrition in urban slums.

The research in Dhaka, Bangladesh, was supported by Bangladeshi Rural Advancement Committee (BRAC) and Action Contre la Faim. The author thanks BRAC staff working in the slums for their constant support. The author is also grateful to the mothers, their children, and the BRAC community health workers living in the slums for their time, understanding, and willingness to share their knowledge.

Location: Urban slum, Bangladesh

What we know: Undernutrition is highly prevalent in infants and young children in the urban slums of Dhaka. Research reports poor awareness by caregivers of malnutrition and poor health amongst infants and children in urban slums. Understanding IYC caregivers’ perceptions is crucial to design better interventions to improve IYC’s nutritional health.

What this adds: Qualitative research, adapted to the slum challenges, amongst pregnant women and community health workers identified inappropriate care, inappropriate environment, inappropriate food and flooding as the root causes of malnutrition in an urban slum in Dhaka. Policy recommendations emerging are to consider flooding as a major determinant of malnutrition, encourage emergency preventative nutritional interventions and focus on water and sanitation issues during the rainy season.

This paper presents a participant driven technique developed to explore the root causes of malnutrition in infants and young children (IYC) customised to the constraints of conducting research in urban slums. The tool aims to support exploration of the perceptions of slum dwellers regarding root causes of malnutrition. The research took place within urban slums of Dhaka, Bangladesh from November 2008 to May 2009 with the support of BRAC¹ and Action Contre la Faim (ACF). The participants were pregnant women and community health workers. The findings of the research have been published elsewhere².

Perception of root causes of malnutrition

In Bangladesh, research shows that IYC suffer from high levels of malnutrition in the urban slums of Dhaka. IYC are frequently born with a low weight and are exposed to high morbidity and mortality risks. The high food insecurity level in the households combined with poor adult nutritional health and an early engagement of children with income generation activities, mean that IYC chances for catch up growth are extremely low and that the risks of becoming or remaining malnourished are high. Exploring perception of root causes of malnutrition is important in a context where it may not be considered or reported by caregivers while prevalence of malnutrition is high. In Bangladesh, previous research showed that there was a lack of understanding of malnutrition and poor health in IYC by caregivers living in urban slums. Malnutrition was not reported as a cause of death in contrast to diarrhoea, respiratory disease, trauma and stillbirth³. Despite high reported mortality rates, children’s health was predominantly reported by household heads to be good to very good⁴. In this context, it could be hypothesized that a situation where the majority of children are to some extent sick and malnourished is considered ‘normal’ by parents. It could also be that the study was carried out during the dry season during which IYC are in relative better health than in the rainy season. Understanding IYC caregivers’ perceptions is crucial to design better interventions to improve IYC’s nutritional health.

Practical constraints of conducting research in urban slums

Slums in Bangladesh are well known to be violent and stressful places where crime and drug dealing take place. Domestic abuse, fights between household members and neighbours are common. Conducting fieldwork can be extremely challenging and access to communities needs to be constantly negotiated through gatekeepers and local intermediaries. Slum dwellers are often hostile and reluctant to respond to questions/share views with outsiders and researchers are held under constant suspicion until trust can be built which can take significant time. In addition to the time required to build trust, the time constraints of research participants themselves needs to be considered. Women slum dwellers strive for survival by working long days, typically in garment factories, and have limited time available. This tool was designed to take these two constraints into account.

To overcome the barrier of being an outsider, it was decided that the sessions would be hosted by a well known and recognised organisation already working in the slums - BRAC (Bangladeshi Rural Advancement Committee). Since 2007, BRAC ran the ‘Manoshi’ programme⁵ reaching 5.7 million people living in Dhaka slums and focused on the promotion of neonatal and pregnant women’s health. This provided an entry point for this research to access community health workers and pregnant women. BRAC birthing huts located in the slums were used as meeting focal points.

To minimize the time required by participants, the meetings were organised around BRAC training sessions for health workers and antenatal sessions for pregnant women. The choice of focus group discussion (FGD) rather than other ethnographic techniques was to respond to the time constraints, as this technique has the advantage of enabling the identification of emerging themes rapidly.

Causal model

The construction of a causal model is used to better understand the perception of the underlying contributory factors of a phenomenon. Causal models have been used previously in qualitative nutrition research. The determinants of malnutrition can take days to be identified by participants and several sessions are sometimes necessary. Figure 1 shows an extract of a causal model built by health promoters in Peru to explore the poor health and well being of children. In this example, it was interesting to note that some health promoters believed that anaemia could be transmitted by mosquitoes. The causal models were compared to distinguish between the health professionals and the population’s perceptions on the perceived causes of infantile illnesses.

Figure 2: Example of photographs used for the building of conceptual framework with the photograph description/ representation

Picture 1 - A women cooking/food or nutrition
Picture 2 - A view of a slum/environment or health
Picture 3 - A boy standing on an elevated, broken wooden path/environment or care
Picture 4 - A man working/socio-economic status
Picture 5 - A view of a slum next to a garbage pit/ environment or health

For this research, the participants worked together to explore their perception of malnutrition and to define the determinants of malnutrition in IYC. Because of the limited time available, only a short version of the full causal model technique was used. The participants were limited to choosing a maximum of 9 determinants compared to an open construction with an unlimited number of determinants in the full model.

Selecting community health workers and pregnant women

The criteria used (being pregnant or a community health worker) were chosen to compare views between different groups exposed to the issue of malnutrition. It was expected that differences between the participant groups would result as previous research has shown; health professionals tend to consider micro-biological factors while mothers tend to emphasize factors related to child behaviour. In this research, pregnant women and community health workers had different perceptions of the determinants of malnutrition. There were variances in the perception of root causes but also in their level of importance. These differences should be taken into account when designing interventions in the slums as they influence the way nutrition programmes will be received by participants. If interventions are mapped onto knowledge, they are more likely to be successful.

Use of photographs and pile sorting

The photographs used were mainly taken by the researcher during participant observations in the slums. The rest were sourced from the internet. Photographs found online were taken in Bangladesh or India. It was verified that they were no cultural differences that would have made them unfamiliar to participants. The photographs were selected to be representative of the determinants as presented in the UNICEF conceptual framework, e.g. food, environment, socio-economic status, care, health, nutrition, education, water-sanitation and hygiene. Figure 2 gives an example of some of the photographs used. These photographs were selected for their representation of a topic, for example a photograph showing a women cooking could represent the topic food or nutrition. In the first phase, participants were asked to group photographs with the same characteristics into different piles. The purpose was mainly for the participants to become familiar with the photographs and to explore their perceptions associated with the representation in the picture. As the researcher’s own perceptions were different from parcipitants, some picture would be understood in a different way to the researcher’s initial intention. Understanding these differences in perception was also useful for developing a richer comprehension of participants. The results of this initial sorting are not presented here.

Figure 3: Pregnant women building a causal tree using photographs

In the second phase, participants were asked to tell a story beginning with "this child is malnourished because...." The participants were asked to select one photograph representing the most malnourished child and subsequently nine other photographs among 72 explaining potential reasons as to why this child was malnourished. This allowed participants to explore as a group what they saw to be the root causes of malnutrition. Practically, the photographs were organized and ranked in a tree shape (with some guidance from the researcher).The top picture (1st picture selected) representing a malnourished child and nine photographs below symbolizing the roots of malnutrition ranked into immediate and underlying causes. Figure 3 shows participants working on the construction of a causal tree using photographs and Figure 4 shows a completed causal tree.

Data collection and analysis

In total, ten FGDs with an average of 10 participants were organized lasting 60-90 minutes each. Participants’ socio-economic profiles and household food security status was assessed at the beginning of the session through very short questionnaires. Each conceptual framework made in photographs during FGDs was translated into a formal conceptual framework using the theme associated with the picture as defined by the group (example in Figure 4). An overall conceptual framework was constructed, combining all the models and representing all the roots identified.

In this research, the overall conceptual framework showed four main groups of root causes of malnutrition resulting from the analysis of the FGDs. These were: 1) inappropriate care, 2) inappropriate environment, 3) inappropriate food and 4) flooding. A qualitative data analysis and research software was used to identify emerging themes and to compare themes across the different groups.

The tool was successful in meeting the constraints of conducting research in slums and encouraged participation through presenting the tasks as a fun, game like, experience. It led to formative findings that shaped policy recommendations to consider flooding as a major determinant of malnutrition, encourage emergency preventative nutritional interventions and focus on water and sanitation issues experienced during the rainy season.

For more information, contact: Sophie Goudet, email: S.Goudet@lboro.ac.uk

¹BRAC is a development organisation based in Bangladesh dedicated to alleviate poverty by empowering the poor, and helping them to bring about positive changes in their lives by creating opportunities for the poor.

²Goudet SM, Faiz S, Bogin BA, Griffiths PL. "Pregnant Women's and Community Health Workers' Perceptions of Root Causes of Malnutrition Among Infants and Young Children in the Slums of Dhaka, Bangladesh." American Journal of Public Health 101.7 (2011) :1225-33.

³Podymow T, Turnbull J, Islam M, Ahmed M. Health and social conditions in the Dhaka slums. International Society for Urban Health. 2002. Available at: http://www.isuh.org/download/dhaka.pdf. Accessed June 25, 2008.

⁴Hussain A, Ali K, Kvale G. Determinants of mortality among children in the urban slums of Dhaka city, Bangladesh. Trop Med Int Health. 1999;4(11):758.

⁵The Manoshi project was developed by BRAC to establish a community- based health programme targeted at reducing maternal and child mortality in the urban slums of Bangladesh. The programme is funded by the Bill and Melinda Gates Foundation under the Community Health Solutions (CHS) initiative aiming at strengthening and leveraging community organisations and participants to scale up proven interventions in community settings (BRAC 2009, Khan & Ahmed 2006).