Insumos e produtos na avaliação da eficiência técnica de hospitais: um estudo de seleção de variáveis baseado na entropia

Abstract

In recent decades, interest in measuring efficiency has emerged in the most diverse areas and health is no different. A sector of extreme relevance for human beings. There are several efficiency measurement techniques, but Data Envelopment Analysis (DEA) has been a preferred technique in many studies due to several advantages. However, it is important to highlight that the result of the analysis through DEA can be influenced by changing variables (Inputs and Outputs). With the aim of reducing such distortions and seeking to identify the ideal variables to compose the model, the study aims to propose a scheme for selecting inputs (inputs) and outputs (products) for evaluating technical efficiency in hospitals. Initially, two systematic literature reviews were carried out. The research converted two literature reviews. The first objective analysis of the variables (Inputs and Outputs) of technical efficiency in hospitals. In 115 studies, 404 input restrictions and 320 output restrictions were identified. The second review sought to identify the variable selection analyzes in the measurement of the technical efficiency of hospitals, based on the study developed by Fernandez-Palacin, Lopez-Sanchez and Munoz-Márquez (2017). Initially, 320 studies were identified, of which 11 were considered eligible, using various methods based on efficiency, classical statistical methods, among others. Finally, three approaches were selected that resulted in 4 distinct adjustments in the composition of inputs and products in the model adjustments: the method with all available variables, the method adopting a prior reduction of the dimensionality of the input and product matrices through protection of factors and, the variable selection model using a genetic algorithm seeking to minimize the entropy of the input and output matrices and maximize the entropy of the efficiency vector. The results show that the strategy of maximizing the entropy of the efficiency vector presents greater discretion among hospitals, providing a more appropriate interpretation of the results.