Processes predictive of CABG complications

Processes of care are what care providers do to and for patients on a one-to-one basis during a care episode. The association of processes of care with risk-adjusted perioperative complications has not been assessed for patients undergoing coronary artery bypass graft (CABG) surgery. The primary objectives of this thesis were to (1) develop a prediction model for perioperative morbidity using preoperative patient risk factors, (2) determine the association between preoperative, (3) intraoperative, and (4) supervisory processes of care and risk-adjusted perioperative morbidity after CABG surgery. Perioperative morbidity is defined as the occurrence of any one of eight perioperative complications or death within 30 days of surgery. Study populations were from two VA datasets: (1) 20,267 CABG patients from Continuous Improvement in Cardiac Surgery Program (CICSP), and (2) 3,597 primary CABG patients from Process, Structures, and Outcomes of Care for Cardiac Surgery Study (PSOCS). A risk model was built on CICSP and applied to PSOCS, calculating a risk estimate for each patient. Mixed model logistic regression was used to assess the three process of care groups, while controlling for risk and hospital effect. The overall model fit was significant (p < 0.05) for intraoperative processes and non-significant for preoperative and supervisory processes. Two individual preoperative processes (right heart catheterization and not documenting left ventricular end-diastolic pressure) and six intraoperative processes (longer ischemic time, high complexity of monitors, inotropic support, use of blood products, higher systemic temperature, and more complete documentation of the primary surgeons record) were found to be significant. Intraoperative processes were comparatively more significant than preoperative processes, indicating that what happens during the operation may be more important than what happens before surgery. However, in an attempt to clinically interpret significant processes, concerns were raised that some processes may be confounded by severity of preoperative illness not captured in risk models or may occur as a response to an intraoperative complication. It may be impossible to accurately distinguish processes of care from responses to outcomes in an observational study design. For future research, randomized trials may be necessary to adequately assess the effect of processes of care on CABG surgery outcomes.