How one medical specialty society’s use of measures and reporting dramatically improved patient care

For nearly 20 years The Society of Thoracic Surgeons’ national database has offered insights and improved quality through practice feedback

Professor and Chief of Cardiothoracic Surgery, University of Florida at Jacksonville, Chair, Society of Thoracic Surgeons Workforce on National Databases, Chicago, Illinois

As painful as it might be to hear, when it comes to patient care we’re not as good as we think we are. Not long ago, doctors at several renowned American medical institutions were stunned to learn that, following submission of their hospital data to Medicare, they had repeatedly neglected to implement basic steps in the care of patients—eg, giving a beta-blocker to all heart attack victims within 24 hours of admission and continuing the prescription upon discharge. Jonathan Perlin, MD, then Acting Under Secretary for Health at the Department of Veterans Affairs, summed up the surprising findings: “Absent the data, you think you’re doing a pretty good job.”¹

To put Dr Perlin’s observation another way: Given regular feedback from accurate clinical data, you can do a much better job. Indeed, our experience in The Society of Thoracic Surgeons (STS) is that the value of such feedback is enormous, and it is measurable.

  Data feedback: The perspective we needed to do better

In the early 1980s, advocates of quality took an interest in the performance of coronary artery bypass grafting (CABG). In responding to constant requests for results, we realized we had no national benchmarks for standards of care, no data other than what had been reported in the literature. In 1989, STS rolled out a national database to monitor the quality of care in several areas—initially in adult cardiac surgery, followed by congenital cardiac surgery and general thoracic surgery.

The chief value of the database has been in collecting and analyzing aggregate data from across the country and creating truly meaningful reports with which local practices can evaluate their performance against national benchmarks on a number of important clinical parameters. Recently we have also broken down these data so that participants can compare themselves against similar institutions as well as the national aggregate.

Risk adjustment is critical. Our database reports are fair and meaningful in part because the risk models we’ve developed ensure that the outcomes we’re comparing are “apples to apples.” Records are risk-stratified into several subsets of patients having similar operative risk. With matched subsets, we can then statistically compare actual mortality against the predicted mortality. These models are internally validated and have undergone journal peer review as well as independent review to ensure statistical accuracy.

For years, our risk models were proprietary, to prevent users from substituting inappropriate definitions that could produce errors in the system. However, transparency is important today, and we have divulged the complete content of our risk models. The latest models, including all variables, will be published this year in the Annals of Thoracic Surgery.

  Actual CABG outcomes defied bleaker expectations

When physicians, who are self-starters, can see where they fall short, they do something about it. One of the more compelling testaments to the success of STS’s national database has been improved outcomes with CABG. Given the results of risk models that calculate the net effect of all significant risk factors, the expectation for operative mortality rose over a decade (FIGURE 1A). In the face of this increasing risk, however, actual operative mortality dropped (FIGURE 1B), primarily due to the feedback practitioners received from database reports. The mathematical difference between the 2 curves translates into tens of thousands of lives saved over the decade.

Assessing unintended consequences of surgery

The STS database contains more than 3 million patient records, which allows us to drill down and conduct exhaustive, sophisticated analyses of select populations with the aim of uncovering unintended consequences of surgeries.

In CABG, for example, one of the national measures STS promotes is the use of the internal mammary artery (IMA) for grafts in patients 75 years of age or older. Because the IMA supplies blood to the sternum, it was suggested by some that removing it might devascularize the sternum, causing dehiscence and mediastinitis. This concern proved to be unwarranted (FIGURE 2A). Because the phrenic nerve lies close to the IMA, it was also suggested that removing the IMA could traumatize the nerve and interfere with breathing, causing patients to become reliant on mechanical ventilation. Again, this concern was proved to be unfounded (FIGURE 2B).

Only a clinical database rich in patient data can yield such insights. Administrative databases cannot provide this kind of helpful information.

Feedback is narrowing gaps in care

The STS database also allows us to monitor the percentage of member participants who comply with national measures for high-quality care. Using IMA grafting as the example once again, we plotted the 2002-2006 performance of participating hospitals, from lowest to highest, according to the percentage of their CABG cases in which IMA was used (FIGURE 3). Hospitals either fell below or rose above the national average for meeting this measure, and the change in the Δb/Δa ratio of these 2 groups over time expressed the rate at which the gap was changing. Our ongoing analyses show we are continuing to close gaps in care.

Such objective assessments are possible across other specialties, but significant funding will be needed to make it happen. STS has testified before Congress to urge significant funding for initiatives that will be truly meaningful for quality improvement.

Taking these lessons to heart

Improving care of patients is what all physicians want. The STS experience with its national database has shown we can continually improve care by establishing feedback loops that enable physicians to act on real clinical data.

We must recognize that this is a new day characterized by unprecedented calls for accountability and transparency. Particularly given the national emphasis on measuring quality, it seems prudent to explore the possibility of creating or joining a local or regional database collaborative.

Participating in a database has distinct advantages, including the ability to use the registry as a data-entry platform for national quality reporting and for maintenance-of-certification requirements. The clinical registry can actually diminish administrative burdens by using one central quality report provided to a variety of third-party payers.

Most important, we should recognize that to improve quality, we must be able to measure quality. Clearly the most expedient approach to quality measurement is through the routine use of clinical databases. The transition to this paradigm may not be fully welcome, but it is inevitable and it will almost certainly improve the care we provide to our patients.

References

1. Kolata G. Program coaxes hospitals to see treatments under their noses. New York Times. December 25, 2005. http://www.nytimes.com/2004/12/25/health/25medicine.html. Accessed September 8, 2008.