DEVELOPMENT OF COMPOSITE INDICATORS FOR HEALTH SYSTEMS EVALUATION: A NEW APPROACH BASED ON LINEAR ENTROPY

This paper proposes a new weighting approach for the construction of composite indicators called Linear Entropy. This method aims to mitigate the impact of extreme values and increase the discriminatory power of the final composite indicator. The approach calculates the dispersion of the sub-indicators in relation to an optimal uniform distribution for assigning their weights. The method is applied in the evaluation of public health systems in 57 municipalities in Minas Gerais, with a population greater than or equal to 30 thousand inhabitants, in 2019, which offered health services in their locality. The results obtained by the proposed approach are compared with the results obtained through weighting based on Shannon entropy and reveal important differences in the weights attributed to the sub-indicators, such as a significant reduction of 21.3% in the Hospital Production sub-indicator and an increase of 8.8% in the Family Health sub-indicator.