He concluded by saying that "the results should improve the quality of care for patients with coronary disease, and it should clarify the indications for intervention in the subgroups that we’ve presented here."
Prediction Models Gleaned From Data
During the same presentation, Dr. David M. Shahian reported on long-term prediction models of death and nonfatal events for both CABG and PCI. "Longer term outcomes are clearly going to be necessary if we’re really going to determine the true comparative effectiveness of these various strategies." he said.
The researchers looked at all isolated CABG patients at STS-participating hospitals who were discharged between the beginning of 2002 and the end of 2007. STS procedural records were linked to CMS claims and denominator files.
The final cohort included 348,341 CABG patients at 917 sites. Follow-up was through 2008 (median follow-up, 4 years). Long-term variables were based on those from short-term CABG models and clinical experience. Separate hazard ratios were estimated for each of these variables for four time intervals: 0-30 days, 31-80 days, 181-730 days, and more than 2 years.
Kaplan-Meier estimated mortality rates for CABG were 3% at 30 days, 6% at 180 days, 8% at 1 year, 11% at 2 years, and 23% at 3 years. Predicted mortality rates were superimposable with observed mortality rates, said Dr. Shahian, who is a cardiothoracic surgeon at Harvard Medical School in Boston. Dr. Shahian is also the chair of the STS Adult Cardiac Surgery Database and the STS Quality Measurement Task Force.
"We did observe the obesity paradox here. It’s the frail, almost cachectic individuals, who do the worst, while the more obese individuals tend to do better over time," he said.
The impact of some predictors changed over time. For example, patients with an acute MI have an increased initial mortality risk, which becomes generally insignificant over 1-2 years. In addition, early reoperation, shock, and emergency status have high up-front risks that decrease over time. However, preoperative atrial fibrillation progressively increases risk over time, he said.
"Among hospital survivors, higher ejection fraction and higher [body mass index] are protective at all time periods. A past history of stroke ... [and] chronic lung-disease immunosuppression have a persistent and negative impact on survival. Smoking, diabetes, dialysis-dependent renal failure – their negative impact increases over time. ... Some early important risk factors, like shock, emergency status, and reoperation are not predictors of late outcomes."
Dr. Edwards and. Dr. Shahian reported no relevant financial relationships.