CABG Readmissions: Not Your Garden Variety Measure

The Centers for Medicare & Medicaid Services (CMS) will kick off its second round of Medicare penalties for 30-day readmissions soon, this time with CABG in its crosshairs. But the CABG initiative differs in several ways from the current three measures. 

The Reduction to Hospital Payments for Excess Readmissions program allows CMS to withhold Medicare reimbursement to hospitals with higher than expected 30-day readmission rates for three costly procedures: heart failure, acute MI and pneumonia. In 2015, the bite will total 3 percent of aggregate payments. Based on 2007 recommendations from the Medicare Payment Advisory Commission, Medicare is poised to add the first procedure to the list with CABG for fiscal year 2017.

While the first three measures penalized discharging hospitals, CMS proposed that the CABG penalty should focus on the hospital involved in the index admission. It argued that if a patient is transferred after a CABG procedure, then it most likely was because of a complication from the procedure itself or the perioperative care the patient received. “The traditional thought is if you develop a complication, then that is something with the surgeon’s quality or the hospital’s quality,” explains Terry Shih, MD, a general surgeon at the University of Michigan in Ann Arbor and a fellow at the Center for Healthcare Outcomes and Policy.

Looking for clues

In some ways, the CABG penalty ushers in potential improvements over the initial three measures in efforts to reduce costly readmissions and improve care. Two databases, the New York Cardiac Surgery Reporting System and the Society of Thoracic Surgeons (STS) Adult Cardiac Surgery database, are a treasure trove for researchers to build evidence about 30-day readmissions for CABG. 

“CABG in general is a good target because we have so much data, such a high volume procedure and it is so impactful economically, it is appropriate,” says Todd K. Rosengart, MD, chair of the Michael E. DeBakey Department of Surgery at Baylor College of Medicine in Houston. While at Stony Brook University Medical Center in New York, he and colleagues plumbed data from 1,205 CABG patients between 2006 and 2011 using the two databases and their EHR to identify risk factors for readmissions. Although it was not their intention, they generally confirmed previously identified variables.  

“That probably means that this is less improper discharges based on any one signal as opposed to building a better system to prevent readmissions,” Rosengart observes. In other words, the ball is in the hospital’s court.

Shih and colleagues question whether case volumes for CABG are sufficient to reliably measure differences in quality. CABG discharges dropped from 15 to 9.9 per 10,000 population between 2000 and 2009-2010, according to the Centers for Disease Control and Prevention. Mimicking CMS policy to use a three-year spread to offset sample size limitations, they calculated the reliability of risk-adjusted readmission rates based on Medicare data on 244,874 patients who underwent CABG between 2006 and 2008 at 1,210 hospitals (Ann Thorac Surg 2014;97:1214-1219).

Statistical “noise” accounted for 55 percent of variation and true differences explained 41 percent of the variation. Reliability of rates increased with volume, but only 4.4 percent of the hospitals had a sufficient number of cases to achieve excellent reliability.

“Over the last 10 to 15 years the volume for CABG has been steadily decreasing because the volume for PCI has been increasing and the expanded indications for that procedure,” Shih says. “As the volume for CABG goes down nationally the overall reliability has continued to drop.” Nonetheless, he foresees CABG volumes reaching a steady state or even rising, but the trend may not be uniform across all hospitals.

Building a better measure

CMS typically uses quality measures that rely on its claims-based data, an approach that critics say misses clinical nuances not captured in administrative data. Statistical methods such as hierarchical models can address some of the problems with sample sizes but they cannot fill in the clinical gaps. Robust databases such as the STS registry, however, might do the trick.

David Shahian, MD, a surgeon and vice president of the Center for Quality and Safety at Massachusetts General Hospital in Boston, and colleagues linked STS clinical data to Medicare administrative data to develop a risk-adjusted readmissions measure for CABG (Circulation 2014;130:399-409). The measure had excellent agreement between the predicted and observed readmission rate across risk deciles.

CMS likely will not introduce this approach in the inaugural CABG measure, but the agency is keen on the possibility. “At least for the present time, it looks like CMS is going to go with the claims-based measure, but they told us they would like to continue to move closer toward greater use of clinical measures such as those based on STS data because everybody believes clinical data are better,” Shahian says.

Rosengart challenges an approach that punishes readmissions rather than rewards best practices but adds, “CMS has done us a benefit by putting [CABG readmissions] on our radar screen.” The Stony Brook team realized through its study that while its 13 percent 30-day readmission rate was below the state’s average, it was still higher than its 8.7 percent predicted rate. About 60 percent of the readmissions occurred within the first week of discharge and a third of those within two days of discharge.

“That raises two or three questions,” Rosengart says. “One is appropriate education, two is appropriate evaluation at discharge and three is appropriate evaluation and triage of the patients when they re-present.”

With its eye-opening data in hand, Stony Brook initiated a quality improvement program. Rosengart’s move to Houston precluded participation but he took many of the lessons learned with him. Those include use of an observation unit, properly educating patients about symptoms such as shortness of breath or signs of infection to pre-empt more serious events, and postdischarge follow-up with well-educated and well-instructed visiting nurses and phone calls.     

Shahian adds that widening the readmission reduction program with a second round of targets will create a more balanced picture of a hospital’s overall quality. “What you would like to do as a hospital is not have your entire performance measured by two or three conditions,” he says. “You would like to develop an expanding portfolio of conditions and procedures and other metrics that characterize your performance.”