Small-Scale Livestock Production in Nepal Is Directly Associated with Children’s Increased Intakes of Eggs and Dairy, But Not Meat

Animal source foods (ASF) provide nutrients essential to child growth and development yet remain infrequently consumed in rural Nepal. Agriculture and nutrition programs aim to increase ASF intake among children through small-scale animal husbandry projects. The relationship between livestock ownership and children’s consumption of ASF, however, is not well established. This study examined associations between livestock ownership and the frequency with which Nepali children consume eggs, dairy, and meat. We analyzed longitudinal 7-day food frequency data from sentinel surveillance sites of the Policy and Science of Health, Agriculture and Nutrition (PoSHAN) study. Data consisted of surveys from 485 Nepali farming households conducted twice per year for two years (a total of 1449 surveys). We used negative binomial regression analysis to examine the association between the number of cattle, poultry, and meat animals (small livestock) owned and children’s weekly dairy, egg, and meat intakes, respectively, adjusting for household expenditure on each food type, mother’s education level, caste/ethnicity, agroecological region, season, and child age and sex. We calculated predicted marginal values based on model estimates. Children consumed dairy 1.4 (95% CI 1.1–2.0), 2.3 (1.7–3.0) and 3.0 (2.1–4.2) more times per week in households owning 1, 2–4 and >4 cattle, respectively, compared to children in households without cattle. Children consumed eggs 2.8 (2.1–3.7) more times per week in households owning 1 or 2 chickens compared to children in households without chickens. Child intake of meat was higher only in households owning more than seven meat animals. Children’s intakes of dairy, eggs, and meat rose with household expenditure on these foods. Small-scale animal production may be an effective strategy for increasing children’s consumption of eggs and dairy, but not meat. Increasing household ability to access ASF via purchasing appears to be an important approach for raising children’s intakes of all three food types.


Introduction
Animal source foods (ASF) provide essential nutrients for healthy cognitive and physical development of young children [1][2][3][4]. However, in subsistence agriculture settings like rural Nepal,

Materials and Methods
Data were collected in three rural Nepali communities (one each in the mountains, hills, and plains) as part of the larger Policy and Science of Health, Agriculture, and Nutrition (PoSHAN) study, described in detail elsewhere [26][27][28][29][30][31]. Within each village development committee (VDC), three wards (the smallest administrative unit in Nepal) were randomly selected and, within those wards, all households that contained a child under the age of six years, or a pregnant or newly married woman, were invited to participate in the study. Households were surveyed thrice yearly, to account for seasons, over a period of two years. Two survey rounds conducted between September and February each year comprised unique seasonal assessments, while one survey round conducted between May and July each year was undertaken as part of nationally scheduled PoSHAN surveys during those months. Data collected included agricultural practices and livestock ownership, household economic and expenditure information, and women's and children's dietary intake measured using a 7-day food frequency questionnaire (FFQ) of 51 commonly consumed foods. The FFQ was adopted from a previous tool used specifically within the Nepali context to reflect usual intake, which was further pre-tested and adapted for use in this study [32,33]. Informed consent was obtained from all participants prior to administering each survey. Ethical approval for the PoSHAN study was granted by the Institutional Review Board at The Johns Hopkins Bloomberg School of Public Health and from the Nepal Health Research Council.
The analysis presented here used data from four timepoints: the 2013 and 2014 rainy seasons (May-July) and the 2014 and 2015 dry seasons (January-February). Criteria for inclusion in this analysis included being a farming household (defined as a household that reported growing at least one field crop in the previous year) and having a child between 6 and 72 months in the household. One child was randomly selected as the index child in cases where the household contained two or more 6-72-month-old children. This yielded an analysis sample of 485 households surveyed up to 4 times each for a total of 1449 observations. Three different regression analyses were conducted to assess: (1) the relationship between poultry ownership and children's egg consumption frequency; (2) the relationship between meat animal ownership and children's meat consumption frequency; and (3) the relationship between cattle ownership and children's dairy consumption frequency. Outcome variables were the number of times that a child consumed foods falling into each of these three food types in the seven days prior to the survey. These variables were generated using the FFQ described above; for dairy and meat the frequencies of consumption for all relevant foods on the FFQ were summed to generate the outcome variables. The main predictor variables of interest were the number of poultry, meat animals, and cattle that a household owned categorized as none, low, medium, or high for each livestock type (using cut-points based on the inter-quartile range for households with any livestock ownership). Our definition of meat animals included animals commonly raised specifically for consumption in Nepal (i.e., not cows, buffalo, yak, or horses): poultry, goats/sheep, guinea fowl/pigeons, and pigs. Our definition of cattle included cows, buffalo, and yak-animals which are commonly used for milk production in Nepal, in addition to serving a variety of other agricultural purposes. In order to adjust for household expenditure on eggs, meat, and dairy, 30-day household expenditures for foods within each category were summed and then categorized as none, low, medium or high for each food type (again, using cut-points based on the inter-quartile range for households with any expenditure). Each analysis also adjusted for child-specific, demographic, and geographic factors known to influence household food access and/or child diet, including child age, child sex, mother's education level, caste/ethnicity, region (mountains, hills, or plains (Tarai)), and season. Missing values were negligible (less than 0.5% of observations) and a complete-case analysis was conducted.
Summary statistics of the participant characteristics and of the outcome variables were generated and negative binomial generalized-estimating equation (GEE) models were used to generate unadjusted and adjusted population-averaged incident rate ratios (IRR), accounting for the correlation among repeated measures on the same household. All analyses were conducted in STATA 14 [34]. Analysis results were interpreted using a significance cut-off of p < 0.05. To ease interpretation, final model estimates were used to generate and plot predicted marginal values of each outcome at varying levels of livestock ownership and ASF expenditure.

Results
Children's consumption of ASF was very low, with most not consuming any eggs or dairy within the seven days prior to each survey date, and median meat consumption per week was 1.0 (Table 1). For the majority of observations, households owned cattle (71%) and meat animals (67%), and for 36% of observations households owned poultry. Expenditure on meat was highest compared to the other food groups, with 83% of observations indicating that households purchased meat in the 30 days prior to the survey; these households spent a median amount of 1000 NPR on meat (equivalent to approximately 10 US dollars). Far fewer households purchased eggs or milk (35% of observations and 30% of observations, respectively), with median monthly expenditure of 400 NPR for milk and 180 NPR for eggs. Adjusted GEE models ( Table 2) indicated that low, medium, or high poultry ownership were associated with higher children's egg intake compared to no ownership. Similarly, low, medium, and high levels of cattle ownership were associated with higher levels of dairy intake compared to no ownership. Meat intake was associated with only a high level of livestock ownership compared to no ownership. As the estimated marginal values plotted in Figure 1 show, the relationship between increasing levels of livestock ownership and children's intakes differed across the three food types. Ownership of just one or two poultry was associated with children consuming eggs a predicted average of 2.2 (95% CI 1.7-2.8) times per week, while children in households without poultry were predicted to consume eggs less than once per week. However, at higher levels of poultry ownership, findings indicated a leveling off-i.e., children in households owning three or more poultry were not predicted to consume significantly more eggs compared to those in households owning just one or two poultry. For dairy, findings suggested a monotonic relationship between level of cattle ownership and children's dairy intakes. Children in households owning zero, one, two to four, or more than four cattle were predicted to consume dairy an average of 3.1 (95% CI 2.4-3.9), 4.5 (95% CI 3.6-5.4), 7.1 (95% CI 6.1-8.1), and 9.3 (95% CI 7.0-11.5) times per week, respectively. For meat, only the highest level of livestock ownership (more than seven) was associated with an increase in children's meat intake, and the estimated effect was small-children in households owning more than seven meat animals were predicted to consume meat an average of 1.6 (95% CI 1.3-1.8) times per week, while children in households without any meat animals were predicted to consume meat an average of 1.2 (95% CI 1.0-1.4) times per week.
Household expenditures on each type of ASF were strongly associated with intake of those foods. As shown in Table 2 and Figure 1, the multivariable-adjusted analyses and marginal plots indicate that greater levels of expenditure were associated with higher consumption frequencies for all food types. The only exception was low monthly expenditures on dairy (NPR 1-180 = USD 0.01-1.8), which were not associated with an increase in children's intake compared to children in households with no expenditure on dairy in the past month. However, both medium and high levels of expenditure were associated with significantly higher intake levels.  foods. As shown in Table 2 and Figure 1, the multivariable-adjusted analyses and marginal plots indicate that greater levels of expenditure were associated with higher consumption frequencies for all food types. The only exception was low monthly expenditures on dairy (NPR 1-180 = USD 0.01-1.8), which were not associated with an increase in children's intake compared to children in households with no expenditure on dairy in the past month. However, both medium and high levels of expenditure were associated with significantly higher intake levels. Figure 1. Predicted average weekly consumption frequency (with 95% confidence intervals) of eggs, meat, and dairy at different levels of animal production and expenditure on animal-source foods, adjusting for all other covariates (mother's education, caste/ethnicity, region, season, child sex, child age). Predictions were generated using negative binomial model estimates presented in Table 2.

Discussion
Our study suggests that in rural Nepal, poultry and cattle ownership, even at low levels, is associated with higher intakes of eggs and dairy, respectively, among young children. These findings provide further empirical support for programs that promote small-scale animal production as a potentially effective strategy for increasing children's consumption of eggs and dairy [12][13][14]17]. They are broadly consistent with recent observational studies of the positive relationship between cattle Figure 1. Predicted average weekly consumption frequency (with 95% confidence intervals) of eggs, meat, and dairy at different levels of animal production and expenditure on animal-source foods, adjusting for all other covariates (mother's education, caste/ethnicity, region, season, child sex, child age). Predictions were generated using negative binomial model estimates presented in Table 2.

Discussion
Our study suggests that in rural Nepal, poultry and cattle ownership, even at low levels, is associated with higher intakes of eggs and dairy, respectively, among young children. These findings provide further empirical support for programs that promote small-scale animal production as a potentially effective strategy for increasing children's consumption of eggs and dairy [12][13][14]17]. They are broadly consistent with recent observational studies of the positive relationship between cattle ownership and children's dairy intake conducted in other rural low-income countries [19,20,35], and provide a valuable counterpoint to several studies indicating little relationship between poultry ownership and egg consumption in such settings [16,18,20,21]. These differences in findings point towards the complexity of the relationship between poultry ownership, egg production, and children's egg consumption and the important role of myriad context-specific factors in determining how households utilize poultry and eggs. Indeed, our findings also indicate a complex relationship. Further research is needed to understand the leveling-off that we observed between poultry ownership and children's egg intakes-i.e., that higher levels of poultry ownership (three or more birds) did not seem to increase egg intake more than a low level of ownership (one to two birds). It may be that households owning more than two birds tend to sell the additional eggs produced or may be raising poultry for meat purposes rather than egg production, whereas households with just one or two birds raise them primarily for in-home egg consumption.
In contrast to the relationship we found between even low levels of livestock ownership and increased egg and dairy intake, increases in meat intake only seem to occur when a household raises a large number (seven or more animals) of livestock commonly used for meat purposes (i.e., poultry, goats/sheep, guinea fowl/pigeons, or pigs). Even then, the predicted effect is quite small-equivalent to less than one additional time consuming meat per week. This is unsurprising given that only those families owning a large number of animals (likely substantially more than seven) will be able to consume meat from their own livestock regularly; instead most families purchase the meat they consume. Indeed, the survey results showed that the majority of households had purchased meat in the past month, while only one-third purchased eggs and dairy. However, raising meat animals may indirectly increase children's meat consumption if households sell the animals that they raise and then use the money earned to purchase meat [14]. A recent review of studies examining the relationship between production diversity (including livestock ownership) and dietary diversity among smallholder farming households suggests that the relationship depends largely on the household's degree of market orientation, and that increasing affordable market access to a diverse range of foods may be more important for improving diets than diversifying home-produced foods [22]. For the households in this study, income generated by raising and then periodically selling meat animals may enable higher levels of expenditure on a variety of foods and other important goods and services. Further analyses are needed to understand how these households utilize meat animals and any income earned through their sale. However, to increase meat consumption, small-scale animal production programs would likely need to have a substantial income generation emphasis. The strong association between increasing household expenditure and children's ASF intakes suggests that enabling households to increase their monthly ASF expenditure or subsidizing households to access such foods may be important strategies for increasing children's consumption of eggs and dairy, and would be the primary pathway for increasing children's consumption of meat.
Our findings also reflect differences in children's ASF intake by caste/ethnicity in the cases of eggs and dairy, and also, by region and season in the case of dairy. These findings are expected and concur with the previous literature on dietary variation in Nepal [36][37][38].
While this study focused on ASF intake as the outcome, further research is needed to determine the levels of ASF consumption by children that may have an impact on different aspects of their nutritional status. This threshold remains uncertain and is likely to vary by indicator and by context. Trials testing the effects of daily egg provision to young children have found mixed results, with one study in Ecuador initially finding a marked impact on linear growth but with little evidence of lasting benefit at two years post-intervention, and another from Malawi finding no effect on linear growth [39][40][41]. Study authors note that this may point towards a need for longer intervention periods in the case of the Ecuador trial, and that in Malawi, the intervention's potential impact may have been limited by already-ASF-rich diets and low stunting prevalence at baseline. In Nepal, the likely potential benefit of increased ASF consumption is substantial (given high stunting prevalence and low ASF intake). However, questions remain as to how much additional consumption of eggs or milk by children is needed to improve child growth, development, or micronutrient status, and what degree of animal ownership by households or by communities is needed to assure that level of consumption.
Limitations of this analysis include potential error in food frequency data due to inaccurate recall, and lack of data on ASF obtained via sharing, bartering, and hunting. However, a similar 7-day FFQ approach has been used in the Tarai of Nepal and results of maternal dietary intake were found to reflect seasonal patterns and correlate with socioeconomic status [32,33]. Another limitation is that the recall period for dietary intake data was 7 days, while that for household food expenditures was 30 days, providing an imperfect alignment and potential for error in adjustment for expenditures. Additionally, the analysis approach taken (specifically, adjusting for ASF expenditure levels) means that we were unable to investigate the income generation pathways noted above that undoubtedly are also important routes through which livestock ownership influences children's ASF intakes. However, by adjusting for ASF expenditure, we were able to examine the direct association between livestock ownership and children's ASF consumption. These findings make a valuable contribution to the existing literature on livestock production and diet in low-income rural settings. They provide empirical support for small-scale livestock husbandry programs as a strategy to improve children's consumption of eggs and dairy products in Nepal.