DEVELOPMENT AND EVALUATION OF PREDICTIVE MODELS FOR COMMUNITY-BASED INTERVENTIONS TO IMPROVE NUTRITION IN NIGERIA

Abstract

The nutritional status indicators in Nigeria, clearly demonstrate that the magnitude of malnutrition is of public health concern. These indicators show that 'stunting' is above 40%, 'underweight' is over 30% and 'wasting" is over 10% amongst the under-five children. In addition, low birth-weight is over 10%, infant mortality rate is 114 per thousand while under-five mortality is 191 per thousand. Several attempts have been made to identify and evaluate factors responsible for success or failure of nutritional intervention programmes and projects in many countries of the world where malnutrition had been endemic. However the success of these programmes and projects In reducing malnutrition has not been consistent. This study was carried out to review the lessons learnt from successful community-based programmes and projects to develop predictive model equations which can be used in Nigeria and elsewhere in sub-Saharan Africa to evaluate various planned interventions and predict which one will have a good impact in reducing malnutrition within the community. A systematic content analysis of twenty-five nutrition-relevant programmes and projects around the world was carried out from which thirty (30) variables that contribute to programme success were identified . Using the statistical package for the social sciences (SPSS): seven clusters of variables (factors) that are crucial, and their contribution to programme success were identified: 'awareness of the nature of the problem and its ramifications or Perception', 'counterpart funding', "community participation", 'felt ownership", "motivation", "control", and "tangible benefit". Using stepwise multiple regression analysis, four (4) variables; "management", "empowered women", "initiation, promotion, support", and "targeting" were isolated as good predictors. These variables were then employed to build predictive equations. Only the equation/model which explains 99% of the variation and incorporates all four variables is recommended by this study. The predictive model was tested and validated by using seven pairs of completed projects with known

outcomes. This study therefore provides: a compendium of variables as well as the identity and contribution of factors associated with programme success, an instrument to assess and hence predict programme success, and a method of testing other similar nutrition-relevant data sets. The timely application of this predictive model, paying attention also to the crucial factors will reduce resource wastage associated with failed projects.