Comparison of Strategies for Detection of Conditions Acquired in a Public Hospital According to the System of Related Diagnostic Groups (DRG)

Abstract

The system of Related Diagnostic Groups (DRG) allows to evaluate the institutional performance regarding the services of the health team. Acquired conditions (AC) are associated with complications of the actions of this team. The objective of the study was to analyze strategies to identify hospital CAs using DRG system records. We have adopted the Centers for Medicare & Medicaid Service (CMS) Hospital Acquired Conditions (HACs) list. HACs were compared to the CAs identified in the database (DB) of the institutional DRG system, applying Relative Risk (RR) and 95% confidence interval. The concordance between observations (HACs and CAs) was analyzed using Kappa statistics. Randomly selected DRG system records were classified as to the probability for HAC. We recorded 550 CAs among 31,472 participants (1.7%) and 1,594 HACs (5.0%), with 123 events (7.7%) common in both. RR of HACs compared to CAs was 5.4 [95% CI: 4.5-6.6, p value <0.01]. The agreement was poor (0.09 kappa, 95% CI 0.07-0.1 and p value <0.01). Analyzing subgroups of HAC 5 (Secondary diagnosis of falls and traumas) and 6 (Secondary diagnosis of catheter-associated urinary tract infection) in 91 hospitalizations, we classified 33 as low risk for HAC, 34 as high risk and 24 as undetermined risk. We conclude that the analysis of Institutional BD underestimates the incidence of CAs, although the HACs method may include false positives. We recommend comparative studies with other HAC identification models.