Contents - Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
Nutritional evaluation of an agricultural development project in southern Sri Lanka
Gerd Holmboe-Ottesen, Margareta Wandel, and Arne Oshaug
The effect that rural development programmes may have on the nutritional conditions of the target population has become a central issue in the international nutrition debate. The stated goals of such programmes do not usually include nutritional improvement but generally focus on raising the productivity of small-holder agriculturalists and on generating employment for and improving the living standards of the rural poor.
Several reports, however, have indicated likely negative effects on nutrition of various agricultural and rural development programmes [1-5]. There has also been a growing awareness that such programmes have a much greater potential for affecting nutritional conditions (for better or worse) than direct nutrition interventions, because they may have a more thorough and sustained impact on the society than nutrition interventions, which are mostly aimed at relieving symptoms.
This awareness has found expression in various forms in international forums. The Food and Agriculture Organization recommended to its member states that they should evaluate the impact their agricultural programmes had or could be expected to have on nutrition. At the nineteenth FAO General Conference, in 1977, the member states returned this recommendation, asking the FAO to prepare methodologies for evaluating the impact of its programmes and projects on nutrition [6l. Significant in this respect also was the endorsement by the World Conference on Agrarian Reform and Rural Development of including nutrition as an objective of rural development as well as an indicator for monitoring progress . This has led to a number of desk studies and field work over the following years in the FAO and other concerned UN bodies, some bilateral agencies, and the research community [8-17].
This paper reports on a field study to assess the impact of an agricultural programme on an area's food and nutritional situation. The study was part of a larger research programme designed to develop methodology for evaluating the nutritional effects of rural development and agricultural programmes . It was used as a test case to try out methods and indicators.
The study was carried out in the Kirama Oya (river) basin, in the Hambantota District in south-eastern Sri Lanka, where the Norwegian Ministry for Development Co-operation is supporting an integrated rural development programme (HIRDEP). Two HIRDEP projects were in operation in the area, focused on raising the productivity of paddy cultivation: The Kirama Oya Irrigation Scheme dealt with rehabilitating and reconstructing an old irrigation system along the Kirama Oya. The Paddy Cultivation Project was concerned with introducing modern cultivation techniques, a credit system for agricultural inputs, proper water-management practices, and a scheme for crop insurance.
The research project was started in 1981, before the construction work was finished, and ended in 1984. At the end of the research, the two development projects had been in full operation for one to two years, depending on the location of the different project sites in relation to the irrigation system.
The evaluation entailed an annual monitoring of the food and nutrition conditions from 1981 to 1984. To minimize the effect of seasonal variations, data were collected during the same period each year, in the course of a month during August/September, which is the pre-harvest period and the time of the lowest food availability.
The monitoring consisted of two parts: a community-level survey and a household survey. In the former, data were collected through interviews with key informants, discussions with groups of farmers, and visits to project sites. The data included information on price and wage levels, project performance, conditions for agricultural production, and the activities of formal and informal institutions.
The household survey consisted of a questionnaire interview administered to the individual households and anthropometric measurements and clinical examination of selected pre-school children in the households. This survey was done in 1981, 1982, and 1984 only.
Eleven villages, representing every other village on either side of the Kirama Oya, were selected for the initial study in 1981. A subsample of 32 out of 55 sections (godas) within these villages was chosen for subsequent monitoring in 1982-1984. In addition to the original villages, one more village, comprising three Sodas, was included in the last two monitorings. The selection of Sodas was made strategically so as to contain population groups involved in different economic activities and of different degrees of involvement in the Kirama Oya projects. This enable us to study nutritional changes in different groups that were bath directly and indirectly affected by these projects.
A random sampling of the pre-school children, covering one-third of the pre-school population, was carried out each survey year in each goda monitored by taking every other pre-school child in the households visited for anthropological measurement. This method of sampling resulted in the inclusion of more than one child per household in a few cases. For this reason, the number of children included in the total samples for each survey year exceeded the number of households by about 4%.
The sample sizes for the three surveys were: 611 children from 586 households in 1981, 468 children from 450 households in 1982, and 466 children from 449 households in 1984. For the purpose of comparing nutritional status in 1981, 1982, and 1984, a subsample was drawn for each year comprising only children from the godas that were included in all three survey years. These subsamples contained 399, 426, and 424 children respectively (table 1).
For the assessment of nutritional status, the weight and height of children 6 to 60 months old were measured using Salter spring balances and measuring boards. Their ages were determined from their birth certificates. These data were combined for the three indicators of nutritional status: weight for age, height for age, and weight for height. The cut-off points used to classify nutritional status on the basis of these indicators were those recommended by the WHO  and used in the Sri Lankan National Nutrition Surveys: general malnutrition was defined as weight below 75% of the WHO reference standard weight for age, chronic malnutrition as height below 90% of the standard height for age, and acute malnutrition as weight below 80% of the standard weight for height.
The analysis of changes in nutritional status was based on the determination of the prevalence of malnutrition according to the respective cut-off points for the three indicators and the distribution of the sampled population according to these indicators.
A Gaussian normal distribution within the samples of all three years of monitoring was demonstrated for all three indicators. The presentation of data on nutritional status therefore includes mean values and standard deviations expressed in percentages of the WHO standards for the indicators, as well as the proportion of cases falling below the cut-off points (the prevalence rate).
The analysis of differences of change in nutritional status between various socio-economic subgroups revealed that changes had generally occurred in the prevalence of both acute and chronic malnutrition. However, the low prevalence of both types of malnutrition and the small size of the samples in the subgroups gave a very low reliability for the height-forage and weight-for-height indicators. For this reason and for simplicity, the indicator of general malnutrition, weight for age, which may reflect a combination of chronic and acute malnutrition, is presented when the sample is divided into subgroups.
A socio-economic score was calculated for each household according to a method adapted from Smith  and Hesselberg . Data were collected on the possession of certain household items known to be related to socio-economic status in the Kirama area. Ten such items were identified and listed in the questionnaire, and the respondents were asked, for each item, whether they possessed it and in what quantity. A standard profile was worked out for the project area in terms of which and how many of the items an average household had. The score assigned to each household was based on its deviation from the standard profile and a weighted index given to each item according to its necessity and cost.
The socio-economic score was used to classify the households into subgroups of "low," "medium," and "high" socio-economic status. The cut-off points between the three subgroups were chosen on the basis of in-depth data collected from a village selected as a case in the research programme  which was considered to be typical for the area in its socio-economic profile. The cut-off points were chosen so as to distribute the households of this village evenly between the three groups.
Income data were collected from each productive member in the household. Various sources of income, both in cash and in kind, were calculated for the preceding year and added for an estimation of the total household income.
To assess the intake of energy-rich foods, the respondents were asked to recall the previous day's consumption of three staples - rice (uncooked), coconut, and sugar - for the total household. If rice was not eaten at a meal, the quantity of a substitute (e.g. bread, jack-fruit, noodles) was given. These foods were selected as indicators because they represent a large proportion - 70%-90% [6l - of the daily energy intake. In addition, it was easy for the respondents to estimate the quantities of these items eaten by the total household.
Information on how many persons the previous day's meal was cooked for was also obtained and used to estimate the intake per consumption unit. The conversion factors for estimating the number of consumption units per household were based on the FAO/ WHO recommendations for energy intake according to age and sex . A male 16-19 years old was set equal to one consumption unit, and other household members were counted as fractions of a unit in proportion to the recommendations.
The significance of differences in the nutritional-status indicators and other variables was calculated by Student's t-test. Probabilities of differences at the level of .05 were regarded as statistically significant.
Changes in nutritional status
As shown in table 1, there was virtually no change in the prevalence of general, chronic, and acute malnutrition among pre-school children in the total sample from 1981 to 1982. In contrast, the general nutrition status improved significantly from 1982 to 1984, with general malnutrition declining from 31.4% to 24.8%. A slight decline could also be detected in the prevalence of chronic malnutrition, but this change was not significant.
These findings are in agreement with the information obtained in the community survey on the food and agricultural situation in the area. This was very much the same during the monitoring in 1981 and 1982. The area was afflicted by drought in both years, which made cultivation difficult. Furthermore, the projects under HIRDEP were still in their initial phases, and construction work was taking place under the Kirama Oya Irrigation Scheme. Therefore, positive effects could not be expected at that stage. In the period 1982-1984 climatic conditions improved, and the construction work on the irrigation scheme was completed. Also, the farmers were beginning to reap benefits from the Paddy Cultivation Project.
TABLE 1. Changes in children's nutritional status from 1981 to 1984 in total sample
|No. of subjects||% of standard||Malnourished (%)a|
Only sections of villages participating in all surveys are included.
a. Children below 75% of standard for weight for age
(general malnutrition); below 90% of standard for height for age
(chronic malnutrition); below 80% of standard for weight for
height (acute malnutrition).
* Difference between 1982 and 1984 is significant (p = .04).
The following comparison of nutritional data therefore focuses on differences between the 1982 and 1984 surveys. The comparison of the 1981 and 1982 data can be viewed merely as a check on the reliability of the estimations of nutritional status.
It was found that nutritional status improved in all age groups (table 2). It could also be noted that the degree of general malnutrition seemed to increase with age. In the further analysis when the total sample was divided into socio-economic groups, the numbers within each age range were too small to allow a statistically significant comparison between the two years.
The general improvement of nutritional status in the Kirama area occurred upstream, in the northern part, while there was no improvement downstream, in the south (table 3). This was true for changes in all three indicators of nutritional status when expressed as the mean value of the percentage of the WHO reference standards. When expressed as prevalence of malnutrition, the difference between the two areas was seen only for general and chronic malnutrition.
Table 2. Changes in children's nutritional status as indicated by weight for age (general malnutrition) from 1982 to 1984, according to age groups (total sample)
|Age (months) and year||No. of subjects||% of standard||Malnourished
|13 - 24|
a. Children below 75% of standard weight for age
TABLE 3. Changes in children's nutritional status as indicated by weight for age from 1982 to 1984 in various groups in the Kirama Oya basin area
No. of subjects
% of standard
|No paddy land|
a. One township is excluded from the sample.
b. See text for explanation of groups.
* p < .05.
** p < .01.
(*) p = .07.
Table 4. Difference in characteristic of subgroups in 1982 and 1984 samples
|No. of Subjects||Mean age (months)||Paddy land (acres per households)|
|No paddy land|
There was no difference in the mean area of paddy land held by the two groups in the two years that could explain why a positive change was seen only in the children of the Kirama farmers. There was, however, a difference in age distribution between the two years, the mean age of the children in the 1984 sample being higher than that in 1982. However, the effect of this difference would be expected to be an increase in malnutrition rather than a decrease, since malnutrition seems to increase with age in this area (see table 2). The improvement found in the nutritional status of the children of the Kirama farmers' group seems therefore not to be due to differences in the sample characteristics for 1982 and 1984.
TABLE 5 Changes in daily intake of selected staple foods from 1982 to 1984
No of households
Intake (g /CU)
Summed intake (kcal/CU)
|No paddy land|
CU = consumption unit.
a. One township is excluded from the sample.
Changes in intake of staple foods
In the Kirama area as a whole, there was an overall increase of 12% in the energy intake per consumption unit from the three recorded staple foods (table 5). The difference in the total intake of energy from these three foods between 1982 and 1984 was significant. The intake of rice in 1984 was 9.6% above that in 1982; however, this difference was not significant (p = .06). The intake of sugar rose significantly by 38%. A slight, but not significant, decline of 3.4% in the intake of coconut was recorded.
In 1982 the intake of coconut and sugar and of total calories from the three foods was somewhat larger in the southern part of the Kirama Oya basin than in the north. The increase in intake from 1982 to 1984 was larger in the north, however, in both absolute and relative terms.
TABLE 6. Changes in disease patterns from 1982 to 1984 (percentages of households with sick children)
No. of households
|No paddy land|
The greatest positive changes in the intake of the three foods took place in the lowest socio-economic group. This finding is in accordance with the data presented on nutritional status, which show a similar pattern. Apart from a decrease in the intake of coconut, there were no significant changes in intake of the other staples or in total energy intake in the highest socio-economic group.
There was almost no difference in energy intake between 1982 and 1984 in the households of the Kirama farmers. The intake of rice remained almost the same. A slight, but not significant, positive change occurred in the energy intake of the non-Kirama farmers' group. However, the intake of rice remained the same even in this group. In the households without paddy land a significant increase of 18% in the energy intake between 1982 and 1984 was recorded, and this group also showed a significant increase in the consumption of rice. Sugar consumption rose in all three groups. The increase was less in the two paddy farmers' groups, however, than in the group without paddy land, which in turn had the lowest consumption in both years.
Thus, the positive change in the nutritional status of the Kirama farmers' children was not a reflection of increased intake of these selected staple foods in the households. It should be noted, however, that the Kirama farmers had a higher intake of rice and coconut in both years than the other groups.
Changes in food prices
The above intake data must be seen in relation to the increase in food prices that took place during the two year period. The price of rice went up by 40%, that of coconut about 200%, and that of sugar 10%. Right before the monitoring in 1982, the price of sugar had jumped about 41)%. This implies that sugar consump tion was probably below normal in 1982. The high price increase of coconut explains why a general decline in coconut consumption was recorded. A food-basket index especially designed for the Kirama area showed that the overall inflation rate on food was 55% .
Changes in hygienic conditions and disease patterns
General hygienic conditions and access to hygienic toilet facilities did in fact improve somewhat over the two years for the three groups. However, this improvement was of the same magnitude for all the groups.
There were more colds and fever symptoms as well as more diarrhoea in all the groups in 1984 than in 1982 (table 6). The increase in these diseases and symptoms was particularly pronounced among the Kirama farmers. Concomitantly there was a decrease in intestinal-worm infections in this group that was not seen in the other groups. All in all, however, the pattern of change in the prevalence of disease and symptoms could not explain the improvement of nutritional status among the Kirama farmers' children. Since the prevalence rate for most diseases and symptoms, particularly diarrhoea, increased from 1982 to 1984, one would have expected a negative impact on nutritional status rather than the positive change that was recorded.
Changes in production and socio-economic indicators
Because of the drought conditions in 1982, many of the farmers did not cultivate their paddy land, and the failure rate was high among those who did. Therefore, the analysis of paddy production data is focused on a comparison between the harvests of the Kirama farmers' and the non-Kirama farmers' groups in 1984 (table 7). The yields were generally better in the north than in the south. Also, the Kirama farmers had higher yields than the other farmers. It was found that these higher yields were due to the successful operation of the Kirama Oya projects .
TABLE 7. 1984 paddy production data
No. of households
Total harvest (bushels/plot)
Share of harvest (bushels/household)
Plot size (acres)
|northern||56||57.8 (30)||44.0 (24)||35.7 (33)||1.5 (1.0)|
|southern||103||37.5 (27)||27.6 (18)||23.5 (18)||1.9 (1 5)|
|Kirama||87||57.3 (42)||43.0 (27)||30.1 (26)||2.1 (1.5)|
|Non-Kirama||81||30.0 (18)||20.8 (12)||24.3 (20)||1.5 (1.0)|
Principal data are means. Figures in parentheses are medians.
TABLE 8. Changes in household income and socio-economic status from 1982 to 1984
|No. of Households||Total income (rupees/household)||Increase in income (%)||Socio-economic score|
|No paddy land|
a. One township is excluded from the sample.
* p < .02.
** p < .01.
*** p< .001.
The income level in the total sample rose on the average by 55% from 1982 to 1984 (table 8). This increase was at the same level as the rate of inflation for foods in the area. While income was higher in the south than in the north, the increase in income between the two survey years was higher in the north in both relative and absolute terms; it was 88%, far above the inflation rate for foods. The increase in income was significant for all the groups investigated. The non-Kirama farmers had a somewhat larger income than the Kirama farmers in both years; however, the increase was approximately the same in both groups in absolute as well as in relative terms.
The socio-economic score was much lower in the northern part of the area than in the south (table 8). There was no change in this indicator in the south between 1982 and 1984 and a positive, but not significant, change in the north. There were small positive changes in the socio-economic scores of the lower socio-economic groups. The score changed positively, but not significantly, in the Kirama farmers' group; there was virtually no change in the non-Kirama farmers' group and a slight negative change for the households without paddy land.
Evaluation will always be fraught with difficulties with regard to separating the effects of a project from other changes that may have occurred in the socio-ecological environment during the project period. In the present research programme, an attempt was made to overcome this problem by combining quantitative and qualitative data that could give an insight into factors and processes, both internal and external to the two agricultural projects, influencing the nutritional conditions. For this purpose, nutrition monitoring was combined with a case study on the food and nutrition conditions in a village within the project area, with a special focus on women's opportunities to cater for their children's nutritional needs . Furthermore, the monitoring at the household level was combined with community-level data on ecological and institutional factors, as well as the monitoring of prices on food, agricultural inputs, and agricultural wages .
The main purpose of this study was to assess the nutritional impact of an agricultural project on the household, and not on the individual. The nutritional status of children was thus used as an indicator of the food and nutrition situation of the household. This approach has obvious pitfalls. It is built on the assumption that the main causes of undernutrition among the children are related to shortage of food in the household rather than problems related specifically to adequate child feeding. Thus, the relevance of this approach would vary in different societies, depending on the nature of the nutrition problems.
In the case study focusing on women, we found that the most important limitation in regard to adequate child nutrition was the lack of material resources needed for procuring food, and not a lack of knowledge or time on the part of the women . The feeding of small children was a high priority in the households. Women also had a high degree of knowledge about nutrition, health, and sanitation due to their general high level of education. Furthermore, they were not so overburdened with work that this hampered their ability to cock and care for their children, as had been found in many other societies . One could therefore expect that whenever food supplies available to the household increased, the children would benefit as well as the adults. Thus, child nutrition seems to be a more relevant indicator of the overall nutritional condition of the household in this society than in many other societies, where the limitations on adequate child nutrition are more related to women's poor nutritional knowledge and lack of time.
As mentioned earlier, this research was carried out during a time of dramatic climatic changes in the Kirama area. The results must therefore be interpreted against the background of a process of general recovery from a drought. The fact that the improvement in nutritional status was most prounounced in the households of the lowest socio-economic stratum indicates that this group was probably the hardest hit during the drought and in fact was "buffering" the impact vis-à-vis the higher-income groups.
The nutritional improvements observed in the north were not found in the south. These improvements were also associated with greater increases in real income and in socio-economic scores and higher paddy yields in the north. In the south, contact with urban areas creates better employment opportunities outside the agricultural sector. The northern part of the Kirama area is more rural and isolated, and its people are more dependent on agriculture. Furthermore, the people are poorer than in the south, which also makes them more vulnerable to climatic changes. It is reasonable to conclude that the drought hit people in the north harder than in the south and resulted in more severe declines in income, food consumption, and nutritional status. At least some of the improvement that was seen in the north thus could be explained as recovery from a temporary bad situation over and above that which was found in the south.
There are, however, strong indications that the Kirama projects also had a positive nutritional impact on the people in the north. Through interviews with extension workers and discussions with farmers, it be came clear that one reason for this was that the Kirama Oya Scheme functioned best in the north. In addition, the Paddy Cultivation Project was only carried out in the north, and those Kirama farmers participating in it had almost double the yield of those who did not participate. The Kirama Oya projects must also have had a positive effect on the households in the north that were dependent on casual labour to make a living. Information from the case study revealed that paddy farmers were more likely to employ casual labour when they were reasonably assured of a successful harvest .
The specific impact of the development projects was also indicated by the finding that the nutritional status of the Kirama farmers' children improved significantly, while that of the children of the other paddy farmers remained the same. It should be noted, however, that the nutritional status of the Kirama farmers' children was exceptionally bad at the outset of the evaluation in 1981 and 1982. The general improvement in the food situation could therefore be expected to have a larger positive impact on them since they were more malnourished than the others. On the other hand, the improvement was so pronounced that the nutritional status of the Kirama farmers' children at the last monitoring, in 1984, was better than that in the other group.
Thus the effects of the Kirama Oya projects came in addition to the general impact of the improved climatic conditions in the area. This explanation is supported by the higher paddy yields and the increase in socio-economic score among the Kirama farmers, which was not found among the non-Kirama farmers.
Improved child nutritional status was associated with an increased consumption of the staple foods selected as dietary indicators in the most disadvantaged groups, i.e. the poorest and the ones in the north (table 5). In these groups, however, most households were only indirectly affected by the Kirama Oya projects. In the group directly affected, the Kirama farmers, the change in child nutritional status was not associated with changes in the consumption of these foods. A possible explanation for this is that the Kirama farmers were better off economically than the other groups in the area (table 8) and that an increase in socio-economic status above a certain level is not associated with a higher intake of rice and total energy from these staples (see table 5). In support of this interpretation is also the finding that among the paddy-farmer groups there was no positive linear relationship between the size of paddy harvest and the household consumption of rice and the other staples. The intake of these tended to stay level regardless of the size of the harvest.
Since the dietary indicators used here comprise only three staple foods, the possibility is not excluded that overall dietary improvement actually occurred in the Kirama farmers' group. Observations reported elsewhere  demonstrated that, when family income increases, a greater proportion of the diet comes from other supplementary foods, such as fish and vegetables. Because of the lack of sensitivity of the method for registration of supplementary foods, no conclusive evidence for an increased intake of supplementary foods could be shown in the Kirama farmers' group. However, a positive effect of improved family income was observed .
The positive nutritional effects of the projects in the Kirama Oya basin can be related to the particular socio-cultural setting in the area. Most significantly, the projects were concerned with the cultivation of a food product that is very important nutritionally as well as culturally for the people concerned. Apart from playing an important role in the daily diet, rice is also one of the most important cash crops in the area. Furthermore, it is easy to store, and a surplus can prevent seasonal fluctuations in food availability.
The experience from this study emphasizes the need to find appropriate indicators for nutrition monitoring that are relevant in a particular socio-economic setting to apply in assessing the impact of agricultural projects. The use of children's nutritional status as an indicator of the nutritional conditions of the household may not always be applicable. This variable should if possible be supplemented by measuring the weight of other household members.
The problem of finding relevant dietary indicators is particularly pertinent. Such indicators must be easy to collect but at the same time valid and sensitive enough to register quantitative changes in the diet. In this case the indicators of household food consumption registered changes in the lower socio-economic strata but not among paddy farmers who were better off.
The usefulness of applying a combination of different methods and approaches should be noted. This approach made it possible to interpret the nutritional changes in relation to important factors and processes in the society, and to launch hypotheses concerning the possible mechanisms of nutritional impact caused especially by the two agricultural projects in question.
We are grateful to Dr. Davy Perera, of UNICEF, Colombo, and Wenche Barth Eide, of the Institute for Nutrition Research, University of Oslo, for their contributions to this study. They participated in the planning team of the larger research programme of which this study was a part. This programme was funded by the Norwegian Agency for International Development (NORAD).
Contents - Previous - Next