OBJECTIVE: To analyze the use of the Brazilian Hospital Database (SIH/SUS) on risk adjustment of hospital mortality, and to evaluate the usefulness of the Charlson comorbidity index (CCI) for risk adjustment of indicators calculated with the available data from the SIH/SUS. METHODS: The comorbidity index was applied on 40,299 patients admitted in hospital in Rio de Janeiro, Brazil. CCI determines specific values to 17 clinical conditions to measure the burden of the patient's comorbidity, not taking into consideration the main diagnosis. Multiple logistic regression was applied to assess the impact of CCI in estimating the probability of dying. RESULTS: CCI was greater than zero in only 5.7% admissions. When combined with age (combined CCI), the percentage of cases with a value greater than zero increased considerably. These models, however, showed to have a low sensitivity. CONCLUSIONS: Despite comorbidity is an important predictor for the risk of dying, it was observed that this is not a good discriminatory variable of case severity in the studied database. This maybe due to incomplete diagnostic information in the database. In the SIH/SUS data, age is the most important predictor of the risk of dying. However, despite the limited quality of diagnostic information in SIH/SUS, the use of CCI combined with age for adjustment of the risk of dying is recommended in measures using this database.
Risk adjustment; Outcome assessment (health care); Hospital mortality; Severity of illness index; Hospital information systems; Risk assessment; Comorbidity; Mortality rate
