Contents - Previous - Next

This is the old United Nations University website. Visit the new site at http://unu.edu

Nutritional profile of the population in a food-for-work project area: A case study from Samburu District, Kenya

A. A. Kielmann, N. S. Kielmann, A. A. J. Jansen, D. N. Njama, G. K. Maritim, R. Mwadime, and K. Saidi

Results

Of the total population potentially available in the five locations (3,874), 894 (23.1%) were surveyed. As expected from the self-selected nature of the sampling, participation in the survey differed from location to location (table 1). It was highest (55%) in the control manyattas (Ngutuko'ngron) and lowest (12%) in Lodungokwe, the trading centre. Preschoolers were the best-represented of the six age and social groups, with 34% of all preschoolers in the five locations participating. Twenty-six per cent of the women and 25% of the school-age children were included, with relatively fewer morani (16%) and elders (13%). Participation among the school-age group was highest in L'Kisin (59%), since a community school was next to the survey site, and lowest in Oro Modei (10%). Of all 270 school-age children involved in the survey, 136 of 177 boys (76.8%) and 35 of 93 girls (37.6%) actually attended school.

TABLE 1. Total population and survey sample by age and social group in the input and control locations

a. Approximately < 5 years old.
b. Potentially school-going (approximately 5-15 years old).
c. Circumcised females (approximately > 15 years old).
d. Circumcised males, approximately 15-25 years old.
e. Males approximately > 25 years old.
f. Approximately 17% of the total population of Wamba Division (23,000).

Among all the individuals surveyed, 49% belonged to the socio-economically stronger ("able") and 47% to the weaker ("unable") group; socio-economic status could not be determined for 4% (table 2). The distribution differed considerably between the input and control groups—with a statistically significantly larger proportion (p<.0001) of unable to able families among those surveyed in the control location (61% to 35%) than in the input locations (41% to 55%)—as well as among the five locations in general.

TABLE 2. Distribution of the survey sample by socioeconomic status

See text for definitions of "able" and "unable".

Sex distribution was practically equal, with overall 433 males (48%) and 462 females (52%) examined. Male-to-female ratios among those examined ranged from 40:60 in the control location to 60:40 in L'Kisin.

Distribution of food and money for work

Participation in the programme was highest in Oro Modei and lowest in Lodungokwe. Among the three age groups potentially available for work, the morani scored best, with a participation index of 2.5, even though they participated in only two of the four project sites surveyed. On average, each woman took part in the programme 1.7 times per year and each elder once.

Table 3 provides an overview of the total nutrients per day and cash per week made available per capita of the total population as a result of participation in the work programme [2].

Contribution of the rations to recommended daily allowances

From the data in table 3 we can calculate that the rations provided the equivalent, on average, of 8% of the recommended daily allowance (RDA) of calories for each member of the community (based on WHO recommendations, adjusted for the observed age distribution), 14% of that for protein, almost 3% for calcium, and 10% for iron (see Figure. 1. Contribution of the rations to recommended daily allowances of calories, protein, calcium, and iron per capita for the community). Overall, 11.8% of the calories provided by the rations came from protein.

TABLE 3. Nutrients and cash per capita provided through the food-for-work programme, December 1986-November 1987

Values are averages assuming equal distribution of the food and money payments among the total populations of each location.
a. 1 Kenyan shilling (Ksh) then approximated US$0.054.

On average, 87% of the maize meal, 88% of the beans, and 51% of the fat in each ration received were consumed by the family of the participant. The remainder—corresponding to approximately 2.5 kg of maize meal, 0.5 kg of beans, and 0.5 kg of fat was shared with others from the same community who were not participating at that particular time.

Use of cash

Two-thirds of those interviewed used the cash part of the payment to buy more food and/or clothes. Only 16% bought other things.

Sources of major food items

When asked to identify the sources of the six most common and important foods—maize meal, tea, milk, meat, beans, and whole maize—consumed by their households within the preceding 24 hours, the respondents listed the WFSP for 8% overall, shops for 43%, and other sources for 47%. By specific items, the WFSP was reported as the source of 26% of the beans, 12% of the maize meal, 6% of the whole maize, 2.5% of the tea, and 2% of the milk consumed. Tea, milk, and whole maize were not distributed by the project, but apparently they were bought with the cash payments.

Nutritional effects

Since the differences in socio-economic status varied between the survey locations and so might have accounted for observed nutritional differences, a first analysis compared the nutrition status of the two socio-economic groups. No statistically significant difference was found. (Using goodness-of-fit statistics for 80% of the reference weight-for-height median for the "able" group, the coefficient was 0.074, the Z value was 1.829, and the lower and upper 95% confidence intervals were -0.053 and 0.153 respectively.) Further analyses could therefore be done by age and social group only without regard to socioeconomic status.

Nutrition status by age and social group

As shown in figure 2 (see Figure. 2. Distribution of wasting (weight at or below 900X0 and 80% of the reference median weights for height) by age and social group in the study population), nutrition status was relatively best in the preschool age group and was considerably poorer for all the other groups. Among the preschool population, 46% were at least mildly malnourished (wasted; <=90% of the reference median) and 7% were moderately to severely so (<=80%); these proportions increased to 86% and 35% respectively for school-age children, 74% and 33% for women, and 88% and 64% for elders. The proportion of morani who were at least mildly wasted was essentially the same as for women; however, the proportion who were moderately to severely wasted was only about two-thirds as high as for women.

The combined input locations had better nutrition overall than the control location, with lower proportions of both mild and moderate to severe wasting for every age and social group except for school-age children at or below the 80% threshold (see FIG 3. Wasting, by age and social group, at the input and control locations). In three of the four individual input locations, after adjustment for socio-economic status and sex, the probability of being at or below the 80% threshold was consistently lower for all the groups than in the control location, but in L'Kisin it was higher for all the groups (table 4).

TABLE 4. Probability of being at or below 80% of the weight-for-height median by location and age and social group

Values are probability P for age and social group (ageci) location (loci), P = 1/(1 + e ml), where mi = 2 (log odds constant + fog odds of ageci + log odds of loci).

Preschool children: In the combined input locations, 40% of the children under about five years of age were at or below the 90% threshold, compared with 54% in the control location. By individual input location, 7% in Oro Modei, 39% in L'Kisin, 46% in Niri Mirimo, and 50% in Lodungakwe were at or below the threshold. Though the difference between the combined input locations and the control location is highly statistically significant (p <.02), when the locations are disaggregated only Oro Modei was significantly different from the control.

Using the 80% threshold, 3% of the preschool children in the input locations were frankly wasted, versus 12% in the control location. The small numbers involved (3 and 9 respectively) preclude statistical analysis by individual location. However, the fact that in the two locations with the highest participation (Oro Modei and L'Kisin) not a single child was at or below 80% of the median and in the other two input locations (Niri Mirimo and Lodungokwe) the proportion was half that in the control location lends credence to a beneficial effect of the project's inputs.

On the other side of the scale, that of being overweight, 43% of the preschool children in Oro Modei, 29% in L'Kisin, 9% in Lodungokwe, and 8% in Niri Mirimo were above 1 10% of the NCHS weight-for-height median, compared with 4 % in Ngutuko'ngron, the control location.

Remainder of the population: The nutritional effect of the food rations could not be determined for morani, because not a single moran was available for examination in the control location. All the other groups except school-age children tended to be nutritionally better off in the input locations than in the control location. There were proportionately more school-age children at or below the 80% threshold in the input locations, but this difference did not reach statistical significance. Using the 90% threshold, the differences in nutrition status between the input and control locations reached statistical significance only for school-age children (p<.02); using the 80% threshold, the differences were significant for women and preschool children (p < .03) (table 5).

TABLE 5. Chi-square values and statistical significance of observed nutritional differences between the input and control locations, using 90% and 80% of the reference weight-for-height median as thresholds, by age and social group

NS = not significant

Mid-upper arm circumference

Mean MUACs for the various age and social groups were as follows: preschoolers 17.3 ± 2.94, school-age 14.6 ± 2.72, women 23.7 ± 2.86, morani 25.4 ± 1.45, and elders 24.4 ± 2.21. Since precise ages could not be determined, the data were analysed only for preschool children; no sizeable (nor statistically significant) differences were found among the five localities.

Haematocrit

Haematocrits were available for 895 individuals (table 6). The distribution by age and social group is shown in figure 4 (see Figure. 4. Haematocrit levels, by age and social group). Preschool children had the lowest average haematocrit (26%), followed by women (29%), school-age children (30%), elders (33%), and morani (40%). The differences between all the groups except that between school-age children and women were statistically significant (p<.05). Other than preschool children (and morani, for whom no control values were available) all the groups in the control location had significantly lower values (p<.00). On a location-by-location basis, adjusted for age and social distribution as above, Niri Mirimo had the highest mean value (38%), followed by Lodungokwe (37%), Oro Modei (31%), L'Kisin (27%), and Ngutuko'ngron (22%).

TABLE 6. Mean haematocrits by location and age and social group

Hct = mean haematocrit percentage ± SD.

A forced multiple linear regression, with haematocrit as the dependent variable and location (with the control as reference), age or social group (with elders as reference), socio-economic status (with unable as reference), and weight for height ( <= 80% of the reference median) as independent variables, showed location to have the overall most significant effect, followed by preschool age (table 7). Of the adjusted R2 of 13%, 12.6% was explained by location alone. When the equation was rerun with weight for height <=90% of the median as an independent variable, no significance resulted.

TABLE 7. Multiple linear regression using haematocrit percentage as the dependent variable and age and social group, location, nutritional status, and socioeconomic status as independent variables

All the independent variables are dummies: Having the indicated characteristic = 1; not having it = 0.
Adjusted R² = .133, of which location alone accounts for .126.

Clinical findings

All 895 subjects were examined for eye problems, goitre, and caries or missing teeth. Overall, 11% (99 individuals) had eye problems, and 4.2% (38) had palpable and 0.8% (7) had visible goitres. Twenty-two per cent (197 individuals) had some tooth defects: 10% (90) had missing teeth because of the intentional removal of incisors (among most of Kenya's nomadic groups, either one or both of upper or lower incisors may be removed preventively to permit feeding during tetanic seizures); 8% (71) had posttraumatic broken teeth; and 4% (36, of whom 19 were from Lodungokwe) showed signs of caries.

As expected, elders had the most eye problems (25.4%)—due to unilateral post-traumatic scarification of the cornea as a result of "having run into sticks or thorns" when young (12.7%), cataracts (7.6%), or conjunctivitis (5.1%). They were closely followed by women (23.8%), most of them elderly, for essentially the same reasons, and by preschool children and morani (11% each). Among preschool children, twenty (9.2%) had conjunctivitis, three (1.4%) had xerophthalmia, and one (0.5%) had post-measles opacification of the cornea of one eye. Among morani, three (8.1%) had conjunctivitis, and one (2.7%) had post-traumatic scarification or opacification of the cornea of one eye. Among school-age boys, fourteen (7.9%) had eye lesions, nine (5.1%) had conjunctivitis, four (2.2%) had unilateral scarification or opacification of the cornea reportedly because of trauma such as running into branches at night, and one (0.6%) had unilateral opacification that arose shortly after measles. Among school-age girls, two (2%) suffered from conjunctivitis.

The prevalence of goitre was about equally divided among three groups, elderly men and elderly women (9% each) and women of child-bearing age (10.2%). It was present to a lesser extent among school-age children (3.0%) and morani (5.4%).

The prevalence of dental abnormalities was highest among elders (42.3 % ), elderly women (32.8%), women of reproductive age (30.0%), morani (24.3%), school-age girls (21.5%), and school-age boys (14.1%). The majority of the abnormalities were limited to missing multiple teeth among the adults and missing one or two of the incisors among children and morani.

The distribution of the clinical symptoms was generally unrelated to socio-economic status and location, except for eye problems, which were significantly more prevalent (p<.001) in the input locations (13.5%) than in the control location (5.7%), partly because of conjunctivitis (7.0% input, 1.8% control). Of the input locations, only L'Kisin had a relatively low prevalence (9.2%), of which less than a fourth (2.3%) was due to conjunctivitis.

Contents - Previous - Next