Zambia is among the countries with the highest burden of undernutrition in children under five globally.

Transcription

1 1

2 In most developing countries, farming remains important for rural households as their source of food. According to the latest Food and Agriculture Organization (FAO) estimates, 805 million are exposed to food insecurity or hunger, with 50% of the global population exposed to hunger living in smallholder subsistence farming communities. Meanwhile, child undernutrition continues to be a major public health problem in most developing countries. Lack of adequate nutrition is also well documented among children, with 99 million children younger than 5 years classified as underweight and 161 million classified as stunted. 2

3 Zambia is among the countries with the highest burden of undernutrition in children under five globally. According to the 2013 Zambia DHS, 40% of children are stunted, 15% are underweight and 6% are wasted. The National Food and Nutrition Commission estimates that 54% of women and 13% of children are vitamin A deficient, and that anemia rates among children and women are 53% and 30%, respectively. 3

4 Eighty percent of Zambia s population depends on agriculture for food, income, and employment. Most farmers are small-scale, cultivating less than 2 hectares (ha) of land with generally low levels of productivity. If you farm in Zambia with 2 hectares, the most you can get (if you are as efficient as a US farmer which is very unlikely) is something like US$ 2000 of harvest output; some of it you need to repay loans for fertilizer and other inputs; some for schooling fees leaving for sure less than US$ 1 per person and day all of these guys are poor. The major food crop and main food staple grown is maize which is harvested in March or April. Other staples include cassava, sweet potatoes, millets and sorghum, as well as legumes, groundnuts, beans, cowpeas and Bambara nuts. 4

5 Agriculture is rainfall dependent. In the absence of irrigation, Zambia s climate allows only for one harvest per year. The availability of food reserves peak after the typical harvest month April and thereafter gradually declines. Even in years of good harvests, many household are unable to consistently meet their basic nutritional needs, with particularly large and common food shortages in the months preceding the next harvest. According to the National Food and Nutrition Commission, In typical agricultural seasons, 60 percent of farmers face a hungry season of several months, which is particularly acute during November, December, January and February. New crops usually become available after March, and have to provide farms with the reserves needed throughout the subsequent year. 5

6 We hypothesize that seasonal variations in agricultural yields and food reserves affect the quantity and diversity of food intake during pregnancy, and that pregnancies during periods with limited food reserves are associated with poorer child health outcomes. 6

7 7

8 A growing body of literature suggests that in utero exposure to hunger negatively affects children s health later in life. Building on David Barker s original work on fetal programming Several recent studies have shown that undernutrition before and during pregnancy is associated with low birth weight, growth faltering and by growth faltering meant, poor cognitive ability, reduced income-earning capacity in adult life, and chronic diseases in adult life. 8

9 A good amount of literature suggests that these effects are dependent on the gestational stage. Susser et al. summarize the findings of earlier studies, by stage of gestation. 9

10 We use the Post-Harvest Survey (PHS) administered by the Zambia Central Statistics Office for the years 2001/02, 2002/03, 2003/04, 2004/05, 2005/06 and 2006/07 as the source of agricultural production and crop reserves data. PHS provides annual agricultural data since 1992/93 which covers area planted with individual crops, production quantities, sales of produce and income realized, purchase and use of agricultural inputs, capital formation and other operational expenses, demographic characteristics of heads of rural households, farming practices and soil conservation methods used and access to agricultural loans. The PHS tracks households reserves of all produced crops over time, and thus allows us to compute measures of food reserve availability at the household and district level for each month in the period. In order to focus on subsistence and semi-subsistence households, we restricted our analysis to small and medium farms. Farms with <1 and >5 hectares were excluded (52,361 observations or about 39.67% of the original sample). For the maternal and child health and nutrition information, we use the data from the Demographic Health Survey (DHS) for the year DHS interviews were conducted between November 2001 and June The DHS survey collects information about children 0-5 years, including their weight, height, birth size and survival status. It also collects information about the parents and the socioeconomic characteristics of the 10

11 household. To analyze the associations of food reserves seasonality during pregnancy and children s survival and nutritional status young children, this study retrospectively merges the PHS to the DHS data. Month and year of harvests from PHS were matched with month and year of birthdates of children from the DHS. 10

12 We classify our data into three categories

13 Our primary outcome measures are child survival, birth size, weight-for-age (WAZ), height-for-age (HAZ) and weight-for-height (WHZ) Z-scores. Height (or length) and weight were standardized based on WHO child growth standards of 2006(26) to create height-for-age (HAZ) and weight-for-age (WAZ) z-scores. We eliminated biologically implausible values based on WHO recommended cutoffs at +/- 6SD. 12

14 Our main independent variables are food scarcity variables: food reserve availability index variable (FRAI) and low reserve trimesters (LRT). For the PHS survey, each farm reports the total production from each crop for each agricultural calendar year (i.e. April-March of each year pairs). Farmers are asked if they still have the crop from last harvest in storage and if not, which month they ran out of stock. We used the responses to these questions to construct a food reserve availability index variable (FRAI). The FRAI is defined as the proportion of households without crop reserves from the previous harvest in a given month and district. We also generated a binary variable identifying pregnancy-trimesters occurring in particularly food-scarce periods and call it low reserve trimesters (LRT). We characterize pregnancy-trimester as low reserve if the FRAI>33%, that is if at least one third of households in the district did not have any food-reserves left during the respective trimester period. The main idea of this approach is to aggregate the generally detailed production data into a summary score predicting the nutritional intakes of farming populations, which would facilitate the identification of vulnerable populations. 13

15 Month fixed effects seasonality Dummies for region - potential compositional differences related to regional food prices and diets Dummy whether the mother works form home - peak of labor during the hungry season Age Groups and Gender - potential physiologic differences vaccination, whether the household has access to safe water and improved sanitation and whether the mother smokes tobacco health coping Finally, we control for mother s education in single years, mother s marital status, and household s asset in quintiles. 14

16 For the regression analysis, we estimate the following multi-variable models. Where Outcome i is the health outcome measure (i.e. child still alive, small in birth size, WAZ, HAZ, WHZ) observed for child i during the DHS interview. X i is a vector of child and household characteristics. And R i and M i are region and month indicators, respectively. Recall that each child observation is matched to the monthly FRAI for when the child was conceived and in utero. FRAI i is derived as the average of food reserve availability index in the first, second, third trimesters while LRT i is the binary measure for low reserve trimester. For the binary outcomes of child survival and small birth size, we use logistic models. While for the z-score outcomes, we use ordinary least squares to perform linear regression. We first estimated the model with only the food scarcity variables as independent variable, then included all of maternal controls, and then added regional and month controls. This procedure allowed us to gauge the extent to which the bivariate correlations can be explained by maternal and child characteristics as well as general country-wide seasonal patterns. For robustness check, we also run alternative linear 15

17 regression models with FRAI for each trimester separately and one with all the trimesters which yielded very similar results. 15

18 The proportion of households that ran out of food reserves during the cropping season is presented in Figure 1. The common trend is that food scarcity gradually increases after the harvest in May and June and peaks in the January to March period. 16

19 Figure 2 shows the FRAI index for February of 2002, 2004 and The darker areas represent areas with higher predicted probability of food scarcity by December and/or February. 17

20 Figures 3 and 4 show child mortality (deaths by 1000) and stunting by month. Child mortality is highest for months of January and June. Note that children born in June are those children whose first trimester falls in typical hungry season, and January is a concurrent hunger season. On the other hand, the highest rates of stunting are around June through September. The children born in those months (June-September) are those children whose first trimester falls in typical hungry season. 18

21 The associations of food scarcity during pregnancy on children s nutritional outcomes are presented in Table 2. For WAZ and HAZ, we find negative and significant associations with food scarcity in all trimesters. These beta coefficients are larger when the mother is exposed during the first and third trimesters. An increase in probability of exposure to food scarcity during pregnancy by 1% is associated with a weight-for-age z-score reduction of about in the first trimester, 0.35 in the second trimester and 0.60 in the third trimester compared to those not exposed at all. Similarly, children exposed to food scarcity in gestation have significantly lower height-for-age z scores compared to those not exposed at all. We did not find any significant associations on the birth size. The non-significance of birth size in any of the result can be possibly explained by the data. The reported birth weights and birth sizes are based on mother s recall which might be causing insignificance. On the other hand, as would expected the odds of child s death were higher for mothers who were exposed to food scarcity during pregnancy in their second trimester than those not exposed at all. However, exposure during the first trimester yielded protective result. A plausible reason on child s survival is selection. If the mother is really malnourished, pregnancy will not just live and might result to early 19

22 miscarriage. Pregnancy selection is tough and children who survive will do better on. Hart showed that the Dutch famine at conception was associated with a substantial increase in the risk of stillbirth and perinatal mortality in the famine area. Dutch famine conceptions were subject to 100 percent excess fetal death. This experience is also similar to the Chinese Great Leap forward Famine (28-29). Both Song (28) and Huang et al. (29) found that the exposure to famine increased the risk of stillbirth. As a robustness check, we generate run a separate regression including FRAI during conception and find protective results as well (see Appendix Table 1). Thus, we posit that low food reserve in conception period and first trimester results in pregnancy selection. 19

23 Associations of food scarcity during pregnancy on children s health outcomes using binary measures for food scarcity (LRT) are presented in Table 3. We find that the trends of results are similar to the results from using continuous measure of food scarcity (FRAI). Again, increased exposure to food scarcity during pregnancy had no significant effects on the births size. Similarly, children exposed to food scarcity in gestation have significantly lower weight-for-age z scores and height-for-age z scores compared to those not exposed at all. And Odds ratio of child s death for first trimester exposure to food scarcity was protective as well. 20

24 This study expands on a limited body of research examining the links between agriculture and nutritional as well as health outcomes. Because agricultural production is seasonal, it also follows that post-harvest food availability depends on the size of production. Seasonal changes in food supplies often place severe stress on the ability of households to maintain nutritionally adequate food intakes. Although food insecurity is harmful to any individual, it can be particularly devastating among pregnant women as the consequences stretches out from birth complications to developmental issues in the first two years of life to illnesses and quality of life in the long run. In summary, the effects on children s growth measures are significant all throughout pregnancy but more pronounced in late gestation and that the effect on child s survival is protective in early gestation with the reason of selection. If the mother is really malnourished, pregnancy will not just live and might result to early miscarriage. Pregnancy selection is tough and children who survive will do better on. So what? Timing matters. The results also suggest proper timing of food aid and government programs to aid in proper food storage to secure food reserves. 21

25 22

26 23