RESUMO
Effective resource allocation policies relating to the long-term effects of complex surgical procedures require accurate prediction of the likelihood of future hospitalization. By approximating clinical conditions with administrative data and controlling for complex case-mix scenarios, we provide evidence of a trade-off between costs and outcome in cardiac surgery. We modelled administrative data to account for clinical conditions in a population of patients admitted for cardiac surgery and their readmissions for complications. Costs were calculated at first admission, the outcome variable was defined as time to readmission within six months post-discharge. Risk factors for readmission were defined as comorbidities and postoperative complications, derived by clinical judgement from the International Classification of Diseases. We predicted health outcome as a function of costs and other patient- and hospital-level features using a two-stage residual inclusion estimation method to tackle endogenous relationships applied to Cox proportional hazard models. We confirmed the trade-off and negative association between costs and hazard of readmission when controlling for all complex risk factors. Accurate matching of standard codes for diseases and procedures with clinical conditions may be a reliable methodology to assess time to readmissions and costs on a large population scale.
