With penalties for higher-than-expected 30-day readmissions for CABG on the horizon, concerns about how hospitals will be measured have been growing. A method that linked clinical data to administrative data may provide an answer.
The Centers for Medicare & Medicaid Services (CMS) intends to add CABG in 2017 to its Hospital Readmission Reduction program, an initiative designed to increase quality and lower costs for conditions whose readmissions were linked to soaring Medicare spending. To date, the program withholds 2 percent reimbursement from hospitals for what is considered excessive readmissions for heart failure, MI and pneumonia. The penalty will reach 3 percent next fiscal year.
Readmission rates are risk adjusted based on Medicare claims data, but administrative data may not adequately reflect clinical differences. David Shahian, MD, of the surgery department and Center for Quality and Safety at Massachusetts General Hospital in Boston, and colleagues reasoned that the Society of Thoracic Surgeons Adult Cardiac Surgery Database (STS-ACSD) might be a useful tool for risk adjustment for CABG. The database already is used to provide risk-adjusted feedback reports to 1,050 cardiac programs in the U.S.
“The STS-ACSD provides distinct advantages compared with administrative data. These include uniform cohort definitions, standardized risk factor specifications designed by surgical content experts, and rigorous quality control,” they wrote in the study, which was published online June 10 in Circulation. “Eight to ten percent of participating sites undergo extensive annual external audits, and data accuracy is high (97% overall).”
Their goal was to develop an all-cause readmission measure using the database. They identified 265,434 Medicare CMS admissions between 2008 and 2010 to link to the STS database. Of those, 86 percent could be linked, and of those, 61 percent were an isolated CABG procedure. They defined all CABG procedures as unplanned.
The risk-adjusted 30-day readmission rate for the isolated CABG procedures was a median 16.8 percent. Readmissions were attributed to the index hospital, in keeping with a proposed rule by CMS. The agency argues that transfer of care would likely be due to complications from the original procedure or perioperative care.
They identified predictors of readmission, including dialysis or elevated creatinine, preoperative atrial fibrillation, age, severe chronic lung disease, insulin-dependent diabetes, immunosuppressive therapy, MI within six hours preoperatively, obesity in women and low body surface area in men.
They reported excellent agreement between the predicted and observed readmission rate across risk deciles, for a C-index of 0.631 (C-statistic of 0.648 under hierarchical modeling). Based on risk-standardized readmission rates, 6.1 percent of hospitals had unexpected readmission rates, about evenly split between better and worse than expected. Reliability increased with higher case volume.
“The C-index, calibration, and reliability of this measure meet or exceed the performance of previously adopted CMS models,” they determined.