Prevalence rates for nutrition, SES, education and environmental indicators are summarized below and sub-regions are ranked based on nutritional status. The other indicators are used to confirm the ranking, and to see if indicator status is consistent with nutritional status, i.e. are there a few sub-districts that are consistently worse-off in most or all categories than others?

One-Way Associations:

· Does sanitation have an association with weight for age z-score independent of socio-economic and environmental factors?

After considering several factors, it was found that access to sanitation does show an association with respect to weight for age. Even with multiple factors, the coefficient of no latrine stayed relatively high and significant. As expected socio-economic status, as defined by roofing type also shows an association. Unsafe drinking water, low education level, and low literacy were significant. If improving sanitation were to be done, there would an expected maximum improvement of 0.283 for weight for age z-score. Furthermore, this would yield a theoretical reduction of the prevalence by a percentage of 9.76%.

· Does access to safe drinking water have an association with weight for age z-score independent of socio-economic and environmental factors?

Taking several regression models into consideration (see table below), it was found that access to safe drinking water remained significant throughout all models. After controlling for all factors, the regression model suggests that a maximum improvement of 0.212 for weight for age z-score can be achieved. This would mean a reduction in the prevalence by 6.5%, which is a decrease of 31.7% in the number of cases of underweight.

· Does access to healthcare, as defined by place of delivery, have an association with weight for age z-score independent of socio-economic and environmental factors?

After controlling for socio-economic factors (see table below), it was found that the proxy for access to healthcare, place of delivery, showed a strong and significant relationship with weight for age z-score. The statistical analysis further suggests that if access to healthcare was available, the maximum change in weight for age z-score would be 0.212. In terms of prevalence, there would be a maximum decrease of 5.07%, which means that prevalence would drop by approximately 25%.

· Does literacy have an association with weight for age z-score independent of socio-economic and environmental factors?

The effect of literacy on nutritional status is most compelling. It has been shown that, throughout all the regression models (see above tables), literacy has remained significant. This shows that an improvement in literacy will result in a maximum change of .239 in the z-score for underweight. This means that there would be an expected decrease of 7.26% in prevalence, which is relatively the highest proportionate change (35.4%).

In the previous section, it was found that the possible key vectors of change in underweight were access to sanitation, safe drinking water, measles vaccination, place of delivery, and literacy. As there some of these factors are addressed through current programs, it is of interest to assess how well targeted these programs are to the malnourished.

The ovearall coverage was 0.8192 and the coverage for the malnourished was 0.7544. The test for targeting (F/PP=.9209) confirmed that immunizations are not well targeted to those worse-off.

The coverage was 0.3634 and the coverage for malnourished was 0.2775. The test for targeting (F/PP=.7636) confirmed that the worse-off do not have access to delivery in clinics or hospitals, indicating low access to health services in general. If access to clinics and modern health facilities were available, there would be a significant of decreasing the prevalence of underweight.

As can be seen from the child feeding patterns table (see excel chart), the percent of those who breastfeed after 3 to 5 months is low, and the percent of exclusive breastfeeders is extremely low, with only about 27% practicing. Exclusive breastfeeding until the age of 6 months needs to be emphasized through a nutrition education program. The introduction of complementary foods between 4 and 9 months is good, but efforts need to be made to promote continued breastfeeding after introducing liquids and solids into the infant's diet.

Taking the DHS data and the statistical information from this study, there are interventions that can be taken to reduce the occurrence of underweight. These measures can be done at two levels, short-term and long-term.

· Short-term Interventions:

The DHS data has suggested that water and sanitation have an important effect on underweight. For example improvement of 0.212 for weight for age z-score can be achieved by improving water. This would mean a reduction in the prevalence by 6.5%, which is a decrease of 31.7% in the number of cases of underweight. If improving sanitation were to be done, there would an expected maximum improvement of 0.283 for weight for age z-score. Furthermore, this would yield a theoretical reduction of the prevalence by a percentage of 9.76. Taking this into account, I would urge that the government of Kenya invest in improving access to tap water and sanitation.

It is also important and beneficial for the Kenyan government to take into consideration long-term interventions. Data has suggested that access to healthcare, literacy, and income generation are areas where improvements will impact nutritional status. Healthcare access can be increased and be efficiently run if there are money and resources. It is important that the Kenyan government keep in mind that a system that has a strong referral system is a good way to improve efficiency. Another way is to implement viable social safety nets. Two possibilities are rotating drug generation programs (e.g. the Bamako Initiative and the development of community based insurance schemes. There should also be promotion of privatization of healthcare to remove the burden placed on the limited resources of the Kenyan Government.

Literacy should be improved, this analysis shows that improvement in literacy will cause a maximum change of .239 in the z-score for underweight. This means that there would be an expected decrease of 7.26% in prevalence. This can be done through the aid of Non-Governmental Organizations. Literacy is a also a good way to promote healthy and preventive measures. Furthermore, insurance schemes require a literate population to work efficiently. This tool can also enhance the human capital of Kenya and improve the fall back position of women, which in turn can aid in improving the health status of women and children.