OBJECTIVE: To assess the use of comorbidity measures to predict the risk of death in Brazilian in-patients. METHODS: Data from the Sistema de Informações Hospitalares do Sistema Único de Saúde (Unified Health System Hospital Information System) were used, which enables only one secondary diagnosis to be recorded. A total of 1,607,697 hospitalizations were selected, all of which occurred in Brazil, between 2003 and 2004, and whose main diagnoses were: ischemic heart disease, congestive cardiac failure, stroke and pneumonia. Charlson Index and Elixhauser comorbidities were the comorbidity measures used. In addition, the simple record of a certain secondary diagnosis was also used. Logistic regression was applied to assess the impact of comorbidity measures on the estimate of risk of death. The baseline model included the following variables: age, sex and main diagnosis. Models to predict death were assessed, based on C-statistic and Hosmer-Lemeshow test. RESULTS: Hospital mortality rate was 10.4% and mean length of stay was 5.7 days. The majority (52%) of hospitalizations occurred among men and mean age was 62.6 years. Of all hospitalizations, 5.4% included a recorded secondary diagnosis, although the odds ratio between death and presence of comorbidity was 1.93. The baseline model showed a discriminatory capacity (C-statistic) of 0.685. The improvement in the models, attributed to the introduction of comorbidity indices, was poor, equivalent to zero when C-statistic with only two digits was considered. CONCLUSIONS: Although the introduction of three comorbidity measures in distinct models to predict death improved the predictive capacity of the baseline model, the values obtained are still considered insufficient. The accuracy of this type of measure is influenced by the completeness of the source of information. In this sense, high underreporting of secondary diagnosis, in addition to the well-known lack of space to note down this type of information in the Sistema de Informações Hospitalares, are the main explanatory factors for the results found.
Comorbidity; Hospital Mortality; Risk Adjustment; Health Services Evaluation
