The California Report on Coronary Artery Bypass Graft Surgery
The California Report on Coronary Artery Bypass Graft Surgery
The California Report on Coronary Artery Bypass Graft Surgery
- No tags were found...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
THE CALIFORNIA CABG MORTALITY REPORTING PROGRAMIV. RISK MODEL FOR ADJUSTING HOSPITAL MORTALITY RATES,2000-2002Patients treated at different hospitals often vary in the severity of their pre-operative clinicalc<strong>on</strong>diti<strong>on</strong>. To fairly compare outcomes at different hospitals, it is necessary to adjust fordifferences in the case mix of patients across hospitals. CCMRP "levels the playing field" byaccounting for the pre-operative c<strong>on</strong>diti<strong>on</strong> of each patient. Hospitals that routinely handlecomplex cases (e.g., sicker prior to surgery) get a larger risk-adjustment weighting in the riskmodel, while hospitals that handle less complex cases get a smaller weighting.CCMRP used a multivariable logistic regressi<strong>on</strong> model to determine the relati<strong>on</strong>ship betweeneach of the demographic and pre-operative risk variables and the likelihood of in-hospitalmortality. Multivariable logistic regressi<strong>on</strong> models relate the probability of death to theexplanatory factor (e.g., Patient Age, Creatinine Level, Type of Arrhythmia), while c<strong>on</strong>trolling forall other explanatory factors in the model.In model development, the three-year dataset was divided into two parts: Data for 2000 and2001 were used as a “training set” to develop the model, and data for 2002 were used as a “testset” to validate the model. After a final model was chosen and tested, the coefficients were reestimatedusing the entire three-year dataset.Table 3 presents the final model based <strong>on</strong> the 2000-2002 dataset. Although the risk adjustmentmodel is based <strong>on</strong> data from 83 hospitals, a risk-adjusted mortality rate is reported for <strong>on</strong>ly 77hospitals: Six hospitals provided data but did not want their results published.<str<strong>on</strong>g>The</str<strong>on</strong>g> final risk model included all variables used in the 1999 CCMRP risk model with theexcepti<strong>on</strong> of angina. <str<strong>on</strong>g>The</str<strong>on</strong>g> 1999 audit and subsequent analyses revealed uneven coding of thatrisk factor across hospitals, so it was dropped. In additi<strong>on</strong>, Cardiogenic Shock (Yes/No), NYHA(Class IV), and BMI were added to the model.17