Inclusão de etapa de pós-processamento determinístico para aumento de performance do relacionamento (linkage) probabilístico

Abstract

The objective of the present study was to demonstrate the application of a deterministic post-processing step, based on similarity measures, to increase the performance of the probabilistic relationship with and without the clerical review. The databases used in the study were the Information System of Notifiable Diseases and the Mortality Information System in the period from 2007 to 2015 of the municipality of Palmas, Tocantins, Brazil. The probabilistic software used was OpenRecLink; a deterministic post-processing step was developed and applied to the data obtained by three different probabilistic matching strategies. The three strategies were compared to each other and added to the deterministic post-processing step. The sensitivity of the probabilistic strategies without manual revision varied between 69.1% and 77.8%, while the same strategies, added to the deterministic post-processing step, ranged from 92.9% to 96.3%. The sensitivity of two probabilistic strategies with manual revision was similar to those obtained by the deterministic post-processing step. However, the number of pairs destined for manual revision by the two probabilistic strategies varied between 1,177 and 1,132 registers, against 149 and 145 after the post-processing step. Our results suggest that the deterministic postprocessing step is a promising option both to increase sensitivity and to reduce the number of pairs that need to be revised manually or even to eliminate their need.