Validação do modelo de previsão da suficiência nutricional Moçambicano

Abstract

A thorough and rigorous assessment of nutritional adequacy is an expensive and time-consuming activity. Rose and Tschirley (2002), using data from Cabo Delgado and Nampula provinces, devised a relatively low-cost method to assess the nutritional adequacy of rural Mozambican households. The method uses the variety of foods to assess the sufficiency of nutrient intake The purpose of this study was to assess the nutritional adequacy of households in the province of Zambézia, use the same approach used by Rose and Tschirley to obtain a model for the study area and verify the spatial validity, external validation, of the prediction model by these derivative. External model validation is important in that it tests the accuracy or applicability of the model on data from different places, because internal validation is only done to test the model's susceptibility to small differences in the dataset from which the model was derived. . To carry out this study, data collected from August to December 2004 by the collaborative research project between the Ministry of Agriculture and the Ministry of Health called Sustainable Nutritional Improvement in Zambézia province was used, in a sample of 741 households. Data collected include food consumption in the last 24 hours and demographic characteristics. To achieve the outlined objectives, nutrient adequacy ratio, linear regression models, graphical methods were used to assess nutritional adequacy in terms of calories, proteins, vitamin A and iron, nutritional adequacy prediction and to validate the Cabo Delgado model and Nampula respectively. The results of the study showed that of the foods that the majority of households consumed, they had a nutritional intake above 75% of their recommendations for three of the four nutrients studied. In contrast, 66% of households had a low vitamin A intake. The forecasting model for Zambezia province presented a relatively robust set of coefficients that works well in predicting nutrient intake. For the prevalence of low intake, the model predicted that 42.1% of households would have low vitamin A intake against 66% observed. The Cabo Delgado and Nampula model did not perform well on data from Zambézia province, as it showed large differences between predicted values ​​and observed results. It also presented square root error values ​​greater than those of the Zambézia model, showing that they do not fit the analyzed data. The poor performance of the model can probably be explained by the temporal factor, since more than ten years have passed since it was developed, and by eating habits, as the applied data included foods that were not included at the time of its derivation.