Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Evaluation of the impact of a food-based approach to solving vitamin A deficiency in Bangladesh

Ted Greiner and S.N. Mitra


The impact of the third year (March 1992-March 1993) of a World view International Foundation project to increase the production and consumption of high-carotene foods in Gaibandah district, Bangladesh, was evaluated. The mothers of more than 2,500 children age one to six years, representative of Gaibandah, were interviewed at one-year intervals and compared with a similar sample in a geographically adjacent non-project area of equal size. A 24-hour recall was done regarding green leafy vegetables, yellow fruits and vegetables, oil-rich foods, and non-carotene-rich vegetables.

This project was multidimensional and was based on community participation. It used women volunteers, health assistants, and schools to spread knowledge, skills, and encouragement for growing carotene-rich foods and feeding them to young children. Modern and traditional mass media reinforced the messages. Seeds for mainly local varieties of high-carotene foods were distributed free or in certain cases sold by a network of trained women volunteers.

Knowledge of the problem and its causes increased in response to the communications. Home production of the targeted high-carotene foods increased remarkably in both the project and the non-project areas. Consumption patterns changed markedly but predictably toward more expensive foods during the study year due to an unprecedented drop in the price of rice.

Children's consumption of all but the non-carotene rich vegetables increased significantly in Gaibandah. In the non-project area, consumption of yellow fruits and vegetables increased as much as in Gaibandah, oil-rich foods less than in Gaibandah, and non-carotene-rich vegetables much more than in Gaibandah. However, consumption of green leafy vegetables decreased significantly in the non-project area and increased significantly in Gaibandah. In March 1993,26 % and 52 % of children, respectively, had eaten green leafy vegetables the day before the interview.


The impact of the third year of implementation of the Worldview International Foundation (WIF) Nutritional Blindness Prevention Programme (NBPP) in Gaibandah district, Bangladesh, was evaluated. The 1991 population of Gaibandah was estimated to be l,856,000, divided into seven subdistricts (thanas). The NBPP began by testing various communication approaches to solve vitamin A deficiency in a pilot project in one thana in 1984-1986. This was evaluated by Helen Keller International (HKI) [1]. The recommendations for this evaluation were followed by NBPP and greatly influenced the design of the programme and how it was run during the next five years. Several additions were made in 1991 and 1992, as detailed below. To date, the programme has been implemented throughout four districts in northern Bangladesh with a total population of about 5 million.

In Gaibandah district, NBPP was funded by the Swedish International Development Authority (SIDA), which also funded this evaluation. Although the authors have had a close relationship with NBPP for several years, no NBPP personnel were involved in any stage of this evaluation and analysis. The NBPP staff did not know which areas were to be sampled in either the pretest or the post-test surveys, or where the control areas were.

Description of the NBPP

The NBPP attempted to achieve a cost-effective and sustainable solution to the problem of vitamin A deficiency in Bangladesh. It combined incentives for growing selected vitamin A-rich foods with education and encouragement to increase consumption of

them and other high-carotene foods among young children. Early during the evaluation year, the third year of the programme in Gaibandah, a message was added encouraging the use of increased amounts of fat in young children's diets.

The NBPP worked through modern and traditional mass media and through a network of women volunteers (WVs) recruited from the villages, two in each union, a total of 165 in the district. The WVs, who had a minimum of 10 years of education, were given a three-day training course at the beginning and monthly one-day orientations thereafter. They worked nearly full time and received an honorarium of 600 take (US$1 = 38 take) per month. WVs were expected to visit 25 households daily, five days a week, according to a planned schedule. Late in 1991 a sixth day was added, and their honorarium was increased accordingly. The WVs were supervised by thana officers who had university degrees.

The following description of the programme includes data obtained from NBPP's own records for 1992 as well as from reviews done for SIDA by one of the authors (TG) in 1992 and 1993.

Methods to increase production of vitamin A-rich foods

During the evaluation period in Gaibandah, five activities were pursued to achieve the goal of increasing production of foods that are high in carotene or fat and are therefore likely to improve vitamin A status.

Twenty schools in each of the eight subdistricts were chosen based on their interest in and prospects for gardening (land and water availability). The headmaster and one teacher were trained in relevant skills and provided with funds to purchase the inputs necessary for a school garden focusing on carotene-rich vegetables and fruits. The subsidy was 1,000 take for the first year, 500 for the second year, and none for the third year; however, it was decided to stop the subsidy midway through the project. The schools had not been prepared for the stopping of the subsidy, and 60% of the school gardens disappeared within a few months. Seed was provided free to about 15 highly motivated students per school who wanted to begin gardening at home, for a total of about 2,400 student gardens throughout the district.

Free vegetable seed and other assistance were provided to families of all children in the project areas found to be night-blind. (According to 1992 project statistics, 1,768 new individuals with night-blindness were identified. Of these, 828 were cured during the year, but the disorder reappeared in 208 of them.) Three types of seeds were distributed to all households in the district: bottle gourd, sweet pumpkin, and climbing beans. All of these crops grew on vines that crept up onto rooftops and into trees, thus providing carotene even for landless families.

In late 1991 the WVs became involved not just in nutrition education but also in production. They formed "green banks," consisting of demonstration gardens at their own homes and items for sale, such as seedlings, saplings, manure, and sometimes insecticide. This was intended to encourage other women in the communities to view the WVs as a source of easily available input and expertise for doing their own gardening, and thus to add sustainability to the project after it was phased out.

During 1992 the WVs began forming women's groups, usually consisting of four landholding and five landless women. These groups were given one day training seminars in subjects ranging from horticulture to management skills. A total of 615 women's groups were formed that year, 3.7 per WV. A group garden was then started, with the harvest to be shared equally, although the landless women provided nearly all the labour. Although this was problematic to negotiate and sustain in many cases, it did provide many landless families with an increased supply of carotene-rich foods and some income. An evaluation of this approach in another district, Lal-monirhat, showed that it was successful in targeting the landless: 46% of women's group members were landless, compared with 36% of non-member households in the same area [2].

Nutrition education programme

Nutrition education in Gaibandah had six components geared toward increasing demand for these foods, especially for feeding to young children. A mass media campaign was developed to increase consumption of vitamin A-rich foods. Two different short spot advertisements were broadcast every day from the Rangpur radio station (which reaches Gaibandah). Earlier, NBPP had broadcast short spots on national television a few days a month on average. By 1992 this had become too expensive to continue, but HKI had started doing it by that time. In Bangladesh, radio reaches much of the poorer segment of the population, and television was believed to reach leaders, decision makers, and so on. The NBPP also paid for short vitamin A advertising spots to be shown continuously in local cinemas. Earlier, NBPP funded the production of a professional-quality 33 minute film in traditional melodrama-comedy style that included messages relevant to the programme. Using portable generators, this was shown outdoors at night 1,125 times in the villages in Gaibandah in 1992, attracting large audiences.

The NBPP messages were designed in the mid-1980s by a local group of experienced non-government organizations, scientists, and international agency representatives to reflect the lessons learned during the previous decade or more of efforts to solve vitamin A deficiency problems in the country. These messages were modified over the years in response to four earlier project evaluations; however, they were still somewhat general. The NBPP did not use social marketing methods to tailor messages to the needs of different target groups or to identified resistance points. For example, they did not address popular beliefs such as the belief that consumption of leafy vegetables during breast-feeding causes diarrhoea. However, this was a minor component of the WIF communication effort.

Messages were conveyed through the 160 cooperating schools. In addition, the WVs (according to their own records) made 781,600 home visits. On average, each household would have been reached 2.1 times in 1992 if these visits had been evenly distributed. The WVs also gave talks to groups using a flip chart and handed out 52,560 posters during the year after giving the talks.

The WVs promoted increased production and consumption of the following special foods: 19 highcarotene vegetables consisting of 12 types of dark green leafy vegetables plus carrots, tomato, pumpkin, sweet potato, beans, arum/chara root, and cabbage; 3 types of fruit (papaya, jack fruit, and mango); 2 types of locally common fish; and 2 oil-containing foods (peanuts and mustard seeds). The NBPP had earlier chosen this message in response to a local research project that, apparently using incorrect methods, arrived at the conclusion that these fish were very high in vitamin A, especially the heads. It was later reported that the pigment responsible for this finding did not have vitamin A activity. In any case, our observations suggest that few people in Bangladesh eat either the heads or the innards of fish, meticulously removing the latter from even small fish. The NBPP never dealt with production of fish or any other animal foods.

WVs also promoted the use of colostrum and continuation of breast-feeding for two years, with dietary supplements only from five months of age. The WVs made a total of 185,000 follow-up home visits to children with night-blindness to encourage their continued receipt of these special foods.

A team of four traditional folk singers was recruited and trained for each of the eight subdistricts. They put on outdoor shows in the villages that attracted large audiences. They conveyed messages about vitamin A by singing and chanting in the local dialects, illustrating their act with flip charts. Their own estimates of audience size suggest that about 750,000 people, 40% of the population of the district, witnessed their 3,631 performances during 1992, that is, 208 per performance.

A series of one-day awareness-raising seminars was given three times a year to a total of 200 village leaders. Finally, about twice a month the WVs did home visiting together with local government health assistants (HAs), providing information on dietary sources of vitamin A, the need to boil drinking water for children, and the use of oral rehydration for diarrhoea. Enough of these visits were made in 1992 to have covered about half of the households in the district. The 383 health assistants in the district were also paid 100 take per month if they attended monthly meetings where they received relevant messages.

Even though the range of activities was wide, the entire three-year programme in Gaibandah cost only Swedish kronor (SKr) 4.5 million, approximately SKr 0.8 (under US$0.13, based on an estimated 1992 population of 1.9 million for Gaibandah and an exchange rate of SKr 6.2=US$1) per capita per year.

The evaluation

Evaluations have been conducted by the Institute of Nutrition and Food Science, Dhaka University, of the other three districts in which NBPP has been implemented (although the components have differed somewhat in each). In Lal-monirhat there was no baseline study. In Rangpur and Dinajpur questions about food consumption were asked in a different way than they were for the baseline, and thus changes could not be evaluated. There were no control areas, so it was difficult to know if the other changes found were specific to the areas in which NBPP was working. These latter evaluations found that after three years of programme implementation, about twice as many households grew key vegetables and fruits high in carotene. Night-blindness levels, as reported by mothers to interviewers, were reduced by about half from baseline, when about 5% to 6% of households had a member with the disorder. The prevalence of night-blindness as reported by mothers was found to be a useful and reliable indicator of vitamin A status in Bangladesh [3]. It is the only practical indicator to use in a grass-roots project on the scale at which the NBPP is implemented.

In connection with a consultant visit by one of the authors (TG) to Bangladesh in January 1992, SIDA decided to fund a study that could determine more precisely the impact of NBPP in Gaibandah district. He was responsible for overall study design, interpretive data analysis, and writing this report; the other (SNM), through Mitra and Associates, conducted the field survey interviewing, coded and entered the data in a computer, compiled the data, and was responsible for the statistical analyses.

The two surveys were conducted in March 1992 and March 1993. In both years, all interviews took place during the Islamic month of Ramadan, when most of the population (93%) fasts during daylight hours. In addition to Gaibandah, a non-project comparison area was chosen consisting of seven geographically adjacent subdistricts in the districts of Joypurhat and Bogra. It was assumed that their geographical closeness would ensure that they were similar to Gaibandah, although it was recognized that NBPP activities were likely to influence such nearby areas.

The NBPP started its operations in Gaibandah in early 1990 and actual field operations began in April 1990. The programme continued until mid-1993. Thus this evaluation covered changes that occurred in the third and final year of the programme's field activity. The changes occurring during this one year were compared with those taking place in the non-project area. No baseline study had been done at the beginning and no further evaluation was done at the end. Therefore, this evaluation does not report on the full impact of the NBPP in Gaibandah. Nor can it be generalized to the other three NBPP districts where the set of activities that took place was slightly different and, in the case of Dinajpur and Lal-monirhat, the period of implementation was longer, up to six years.

Survey methods

Interviews were conducted only with households having at least one child age one to six years. Questions were asked of the mother or other guardian most involved in caring for young children. Four attempts were made on other days to obtain interviews with families not at home or unavailable at the first visit. Respondents were assured of confidentiality. Interviewers were instructed to advise parents of children with night-blindness to take them to a health worker.

The sample was chosen to be large enough that a difference of one percentage point in rates of night blindness among children between one and six years of age over time or between the project and the non-project area would be statistically significant at the 95% level. This required a sample of 2,500 children in each area for each survey. The final resulting sample sizes for children and for respondents (mothers of the children in 97% of the interviews) are given in table 1. The interviews were conducted in private and took on average 53 minutes to conduct.

The two surveys were conducted on separate same pies and were not repeat visits to the same households or even to other households in the same clusters. This ensured that any change in attitude or openness toward acquiring relevant new knowledge that might have been occasioned by the first survey had no influence on the results of the second one. (The random selection of clusters for the repeat surveys resulted in 1 of the 70 being chosen again. However, selection of households within that cluster resulted in few, if any, repeat visits.) A description of the sampling method used can be obtained from the authors.

Statistical analyses were performed on only a few key relationships and are reported in all cases in which they were performed. Student's two-tailed tests were performed using standard errors.

Results and discussion

Characterlstics of the sample populations

Results are presented here for both surveys and for project and non-project groups, for a total of four cells. The exact sample sizes were 2,559 and 2,522 for the project area in 1992 and 1993, respectively, and 2,529 and 2,518 for the non-project area. The mean age of about 42 months in all four cells was about the same as the median age. However, the tendency was toward heaping of reported ages in all four cells at five years of age. Ages are difficult to determine in Bangladesh, because they are not recorded and growth monitoring is uncommon. Therefore local calendars were used. Almost 52% of the children in all cells were boys.

The samples of respondents (their guardians) were 1,615 and 1,530 in 1992 and 1993, respectively, for the project area, and 1,733 and 1,599 for the non-project area. Over 99% of the children in all four cells lived at home with one or both parents. There were about 1.5 target children per respondent in all four cells.

The proportion of respondents who were non-Muslim (mainly Hindu) varied from about 6% to about 8% among the cells. The average age of the respondents was also similar, 28 years in the project area and 27 years in the non-project area. Despite their geographical proximity, clear socio-economic differences existed between the two areas. The ecucational level was lower in the project area, where 76% had never attended school, compared with 66% in the non-project area. The project area also was poorer than the non-project area. Average land holdings were somewhat smaller in the project area, with 44% of the respondents landless, compared with 37% in the nonproject area. About 40% of the houses in the project area and 60% of the houses in the non-project area had metal rather than straw roofs. About 4% of the project houses and 10% of the non-project houses had electricity.

TABLE 1. Proportion of households growing different types of foods

Types of foods Project area Non-project area
NBPP action 1992 1993 1992 1993
Green vegetables
bottle gourd leaves 1 94.4 95.2 78.1 87.5
beans (green; on vine) 1 84.0 89.5 79.5 83.3
sweet pumpkin leaves 1 83.7 88.9 69.3 81.8
pui shak (a local vine) 2 60.8 73.6 53.4 64.1
radish leaves 2 45.5 53.8 41.1 46.5
red amaranth 2 37.9 53.6 33.7 40.6
local spinach 2 23.4 37.9 21.5 29.0
cabbage 2 4.8 9.1 8.2 6.5
kang kong 2 4.4 4.8 3.6 5.2
colocasia leaves 3 32.2 46.6 27.0 30.2
colocasia shoots 3 21.1 33.7 14.8 21.3
jute leaves 3 41.5 46.1 35.7 36.5
sajna ("drumstick" tree leaves) 3 12.9 23,9 14.2 22.5
chinese cabbage 3 1.7 5.3 3.4 2.9
mint 3 2.0 3.0 0.6 0.3
amaranth 4 52.3 64.1 43.3 51.4
lafa shak (local shrub) 4 22.8 30.4 1.1 4.2
watercress 4 2.5 8.6 2.9 3.6


8.2 9.1 6.6 9.7
Yellow vegetables and fruits
sweet pumpkin 1 84.4 90.3 71.0 81.6
tomato 2 18.0 53.3 19.5 58.6
papaya 2 30.9 44.5 18.7 27.6
carrot 2 3.5 9.7 0.9 1.6
mango 3 71.8 78.1 61.2 68.3
jack fruit 3 57.6 64.3 53.6 58.5
sweet potato 3 22.3 34.5 24.0 27.5
palmyra palm fruit 4 6.4 10.0 26.2 26.6
Other vegetables and fruits
banana 5 81.6 86.9 60.2 69.6
potato 5 49.0 49.9 64.8 66.1
eggplant 5 46.2 50.5 34.3 43.6
pineapple 5 14.6 20.3 14.9 17.2
dates 5 9.3 15.3 25.1 22.0
cauliflower 5 5.5 9.7 9.2 7.1
wood apple 5 4.6 7.2 14.8 16.8


9.3 10.5 13.8 9.6
Oil-bearing foods
mustard seed 3 31.0 36.0 43.0 44.9
peanuts 3 3.2 7.8 3.3 3.3
sesame 4 10.0 18.7 14.5 16.4
soya beans 4 0.1 0.6 0.1 0.2


4.0 2.7 2.0 1.5
Number of respondents


1,615 1,530 1,733 1,599

1 = distributed by NBPP to all households in Gaibandah district.
2 = distributed by NBPP only to people with night-blindness and student gardeners.
3 = not distributed, but promoted by NBPP.
4 = special food, but not promoted by NBPP.
5 = not a special food (i.e., not high in carotene or fat) and not promoted by NBPP.


Contents - Previous - Next