Field Exchange Dec 2000: Lessons learnt from the 1998 Bangladesh floods

This article is based on research conducted by Dr. Moazzem Hossain who was working as a Programme Manager in Health and Nutrition of Save the Children Fund UK at the time of the study. Elizabeth Stevens, Policy Officer at Save the Children UK, wrote the following article based on the research.

Bangladesh experienced the worst flood of the century between July - October 1998. Some 30 million persons were affected by the crisis. The people of Bangladesh have extraordinary coping mechanisms for dealing with floods, but the scale and duration of the 1998 flood placed these under considerable strain, and there were serious concerns about the possibility of a large-scale famine.

The government and aid agencies responded in a variety of ways. Flood shelters were built; medical care was provided in treatment centres and by mobile teams; damaged homes were repaired; agricultural products were supplied; cash for work programmes were initiated as well as grants and interest free loans. Food assistance was targeted at those in urgent need and the poorest members of the community. In the areas where the study was carried out, poorest households were identified on the basis of income statements provided by each household. In other areas where SCF and NGOs were working the poorest were identified through discussions with the community. Nutrition interventions included supplementary feeding for households with a malnourished child, and the provision of Vitamin A capsules where night blindness was detected. Questions have been asked as to the effectiveness of the targeting and the extent to which interventions influenced the nutritional status. A study of the results of a Save the Children UK survey indicates some answers to these questions.

The nutrition survey

Save the Children UK conducted a survey in six severely flood affected districts in August 1998 to look at the nutrition situation, coping mechanisms and community priorities for suitable rehabilitation. The survey was repeated in the same areas in December 1998. Multiple stage cluster sampling was done for each round of survey following WHO guidelines (WHO 1983). Villages were selected randomly as Primary Sampling Units by probability proportionate to size. Villages were then divided into clusters of 30-40 households and one of these clusters was chosen randomly for sampling. A total of 60 clusters in 6 areas were surveyed during the first round, and 56 clusters were surveyed during the second round. Children aged between 9-59 months were measured to assess their nutritional status; 1597 children were measured during the first round and 1451 in the second. The children were weighed and measured and their weight for age, weight for height and height for age were calculated. In the study, wasting (weight for height) was mostly used as the most appropriate indicator for acute malnutrition, and it avoided difficulties in assessing the age of the child. Data were analysed using statistical package programmes (SPSSPC 9.0 and EPI Info 6.0). Anthropometric measurements were compared with NCHS median and z scores were calculated using EPINUT.

In an attempt to assess the extent to which the interventions during the flood influenced the nutrition status of children, a secondary analysis was carried out on the situation of 180 children who were included in both the first and second surveys. The analysis identified factors which influenced the recovery pattern of the children, and which may assist in the development of policy guidelines promoting more effective interventions.

The analysis found that while moving from crisis (flood period) to rehabilitation (post-flood) phase there was clear evidence of a cross-over phenomenon in the recovery pattern of nutrition status. Of the 180 children, 17% were acutely malnourished during August whereas by December the proportion had reduced to 12%. Although this represents an improvement of 5% in the malnutrition prevalence, the recovery pattern was not straightforward. An internal shift had occurred which reduced the net effect. 90% of the malnourished children in August improved by December, with 82% returning to a normal nutritional status. In contrast, half of the children who were well in August saw their nutritional status worsen so that in December 9% of them had fallen into the malnourished category. In December, 14 out of 23 malnourished (64%) were "new" cases or previously normal children. We can thus observe an important cross-over phenomenon.

Implications for targeting

An interesting finding of this study was that the criteria for targeting assistance were not always appropriate especially during the rehabilitation phase. Results of the study showed that those who were poor and in need during the flood period were helped, and those who had a malnourished child during the crisis were also more likely to receive assistance. These observations fit with the statements of agencies involved in the flood response. In a rehabilitation or recovery phase, however, the situation becomes different. The flood affected all people regardless of their socio-economic condition. A family that coped and survived well during the flood might have exhausted all their assets. They would therefore be more vulnerable in the rehabilitation phase in comparison to a poorer and more vulnerable family that was assisted during the flood by local or external agencies. People who have lost most of their assets are considered to be too great a risk for "normal" and "soft" loans and are forced to find money at very high interest rates. A high proportion of asset loss was significantly associated with deterioration from normal nutritional status to malnutrition.

This study suggested that criteria for targeting assistance should be established for both the crisis and recovery phase right at the start of any relief programme to ensure that the impact of interventions is sustained.

Another interesting observation was made regarding the assistance and membership of NGOs. The study found that although criteria for targeting assistance were used, there was a greater proportion of NGO members among the beneficiaries. This implies that the very poorest did not necessarily receive assistance. An evaluation of the DEC-funded response to the floods also noted that NGO members might have benefited disproportionately: "There is some criticism that NGOs in general targeted their own group members disproportionately. The most disadvantaged members of a community may not always have benefited from some NGOs disaster response." It would appear that agencies need to be aware of this potential problem when targeting aid.

Factors affecting recovery

A number of variables - loan burden, loss of assets, illness, source of food - were analysed for their impact on nutritional status. The loan burden had a strong association with the recovery pattern. The proportion of assets lost, episodes of diarrhoea and source of food during the week preceding the survey in December were also found to have a significant relationship with the cross-over phenomenon.

Access to loans

Almost all people had to take some form of loan during or after the flood. Mohajons are money-lenders who lend money at very high rates (some times 200% a year) at the moment of crisis. Poor people with minimum reserve or collateral are forced to take loans from these mohajons. Inability to pay back the money usually ends in losing the remaining assets or else provide free labour for any amount of days decided by the money-lenders. A loan from a mohajon was negatively correlated with the progress in nutritional status. On the other hand, having access or ability to take loans from other sources (banks, neighbours etc.) with a long-term repayment schedule at a very low or no interest was positively correlated with the improvement of nutrition status. This suggests that flood-affected families need to be assisted before they exhaust their resources and are forced to take loans from a mohajon, or that their accessibility to other loans should be widened.

Access to food assistance

The children of families who had access to a variety of sources of food (e.g. loan from relatives or donation from aid agencies) had a greater chance of improving than declining in nutritional status. The survey in December showed a significant association between children declining to become malnourished and whether they had access to additional sources of food. 78% of those who deteriorated to become malnourished depended only on their own stock for food while only 36% in the group which moved from malnourished to normal did so. Those 78% had a more limited access to food either through a loan from relatives or donation from local or external agencies.

Prevention of diarrhoea

The other factor that seemed to have some influence on the recovery pattern was the repeated attacks of diarrhoea. Although there was association between an attack during flood (August) and deteriorating from normal to malnutrition, it was absent when checked with an episode of diarrhoea during December. But the association was stronger for those who had repeated episode of diarrhoea i.e., during flood (August) and post-flood period (December). This suggests that likelihood of deterioration from normal to malnutrition could be reduced if the repeated episode of diarrhoea could be prevented by taking adequate care of those who become ill during the initial flood period.


This study led to an observation of a cross-over phenomenon in which some children's nutritional status declined from normal to malnourished over the period of four months. This is obviously a cause for concern. Certain factors - type and size of loan, proportion of assets lost, episodes of illness, and sources of food - showed an association with the cross-over of children from normal to malnourished states and the reverse. These findings have implications for the choice of intervention made by aid agencies and government.

The findings also raise the question of the relevance of targeting malnutrition during an emergency of such a large scale in which everyone is affected. Furthermore, it may not be appropriate to use the same criteria for targeting assistance during both the crisis and rehabilitation phases. Appropriate criteria should be established during the design of the project so that the effects of interventions are not lost in the recovery phase.

This research was not of a longitudinal design and the numbers of subjects are small. However, the findings do merit further consideration. It would be useful for well-designed longitudinal studies in flood prone areas to be undertaken, to look at these issues more carefully and develop appropriate policy recommendations.

Note: Dr Hossain would like to acknowledge the contribution of Save the Children Fund UK who funded the initial surveys. He is grateful to Lola Gostelow, Emergency Advisor and Anna Taylor, Nutrition Advisor, SCF-UK, Martine Billanou, Programme Director of SCF-UK Bangladesh, and to Muhammod Shuaib and Abdullah-Al-Harun of SURCH for their support and encouragement. Thanks is also extended to the children and their parents who were affected during the flood but always co-operated during the data collection.

Correspondence: Dr. Moazzem Hossain, Save the Children Fund, House 28, Road 16 (New), Dhanmondi R/A, Dhaka- 1205, Bangladesh. Email: