OBJECTIVE: To determine a simple model to calculate the number of deaths which could be expected in a therapeutic feeding centre from the height, weight and oedema of children on admission. DESIGN: Admission weight, height, presence of oedema of the children and outcome were prospectively recorded. SETTING: Data were recorded in 18 feeding centres set-up during emergency operations in Africa. Ten of the feeding centres were selected, a priori, as reference centres and eight centres as test centres. SUBJECTS: Data for 3858 children were recorded. 837 children absconded from the centres and were excluded from the analysis. Analysis was performed on data from 2753 children who left the centre after recovery and 268 children who died during treatment. INTERVENTIONS: The relation between the risk of death and, anthropometric measurements and presence of oedema has been determined in a previous paper. The maximum likelihood estimate of the constant of the model was determined from global analysis of the data of the reference centres. The model was applied to the data of the reference and test centres. RESULTS: The model to predict the individual probability of death was: P(death) = 1/(1 + exp[-(20.63 - 9.99 1n(weight/height1.74) + 1.36 oedema)]) The predicted number of deaths was close to the recorded number of deaths for each reference centre. For three of the eight test centres there was a significant excess of observed deaths over predicted. CONCLUSION: This model can be easily used by the supervisor of a centre to assess the expected number of deaths during treatment of malnutrition from simple measurements on children that are routinely taken on admission and thus help to determine the nature of variation in observed mortality rates.
OBJECTIVE: To determine a simple model to calculate the number of deaths which could be expected in a therapeutic feeding centre from the height, weight and oedema of children on admission. DESIGN: Admission weight, height, presence of oedema of the children and outcome were prospectively recorded. SETTING: Data were recorded in 18 feeding centres set-up during emergency operations in Africa. Ten of the feeding centres were selected, a priori, as reference centres and eight centres as test centres. SUBJECTS: Data for 3858 children were recorded. 837 children absconded from the centres and were excluded from the analysis. Analysis was performed on data from 2753 children who left the centre after recovery and 268 children who died during treatment. INTERVENTIONS: The relation between the risk of death and, anthropometric measurements and presence of oedema has been determined in a previous paper. The maximum likelihood estimate of the constant of the model was determined from global analysis of the data of the reference centres. The model was applied to the data of the reference and test centres. RESULTS: The model to predict the individual probability of death was: P(death) = 1/(1 + exp[-(20.63 - 9.99 1n(weight/height1.74) + 1.36 oedema)]) The predicted number of deaths was close to the recorded number of deaths for each reference centre. For three of the eight test centres there was a significant excess of observed deaths over predicted. CONCLUSION: This model can be easily used by the supervisor of a centre to assess the expected number of deaths during treatment of malnutrition from simple measurements on children that are routinely taken on admission and thus help to determine the nature of variation in observed mortality rates.
Entities:
Keywords:
Africa; Age Factors; Anthropometry; Child; Child Mortality; Demographic Factors; Developing Countries; Diseases; Health; Malnutrition; Mathematical Model; Measurement; Models, Theoretical; Mortality; Nutrition; Nutrition Disorders; Nutrition Indexes; Population; Population Characteristics; Population Dynamics; Research Methodology; Research Report; Validity; Youth
Authors: Nael Lapidus; Andrea Minetti; Ali Djibo; Philippe J Guerin; Sarah Hustache; Valérie Gaboulaud; Rebecca F Grais Journal: PLoS One Date: 2009-01-29 Impact factor: 3.240