More than a year ago, Medicare agreed to release data to ProPublica that allowed us to publicly compare the performance of surgeons who do common elective procedures. We viewed the handling of this information as a profound responsibility to both surgeons and their patients.
These days, consumers can review ratings on everything from plumbers to hair salons to the latest digital cameras. The process of undergoing surgery includes some of the most consequential decisions any of us ever make. So we began with the view that the taxpayers who pay the costs of Medicare should be able to use its data to make the best possible decisions about their healthcare.
We were aware from the beginning that release of this information would affect the reputations of hardworking women and men who have spent years honing their craft. Any effort to create a “Surgeon Scorecard” would have to be carefully designed in conjunction with leading experts. It would also have to be transparent with readers about its limitations and methodology.
The ProPublica reporters, data journalists and editors working on this project consulted with some of the leading figures in the field of patient safety. Their names appear on the detailed paper outlining our approach. We hired Sebastien Haneuse, a professor of biostatistics at the Harvard School of Public Health, to help craft a methodology that would be both useful to patients and fair to surgeons.
We were also guided by our shoe-leather reporting on this subject. For nearly three years, ProPublica has been interviewing patients and doctors and scholars about medical mishaps. Our analysis of the data was informed by our patient safety Facebook group where we heard from ordinary patients who provided valuable insights and a constant reminder of what’s at stake. Again and again, we were told that medical errors – identified by one recent study as the third-leading cause of death in the United States – can and should happen far less frequently.
A number of patient-safety advocates told us greater transparency about surgeons’ performance could spur that goal. They noted that New York’s 1989 decision to publish complication rates for individual cardiac surgeons spurred significant improvement.
To get the clearest picture of the surgeon effect, we focused on elective procedures done thousands of times a day, mostly without incident, in hospitals across the United States. These non-emergency operations are scheduled in advance and generally performed on patients who are in stable health. To be fair to surgeons, we excluded patients who came in through the emergency room or were transferred to the hospital from facilities like nursing homes. After consulting with our experts, we picked eight elective procedures to study.
Experts and previous academic research pointed us towards two aspects of Medicare records that could be used to compare surgeon performance – deaths while in the hospital and readmissions to the hospital within 30 days.
Deaths are easy to count but readmissions posed a more complicated question. Medicare uses “all cause” readmissions to measure hospital quality. This means a patient who has elective knee surgery and returns within 30 days with cancer is counted as a readmission. Working with a panel of doctors, including surgeons who do these procedures, we identified a narrower list of readmissions that were likely the result of a post-operative complication such as blood clots or infection. It’s reasonable to question whether this is fair, since it’s often not clear who, if anyone, is directly at fault when patients develop complications. In the end, we were guided by the view shared by the experts we interviewed and the American College of Surgeons – that surgeons are responsible for the full range of care arising from a patient’s operation.
Have You Been Harmed in a Medical Facility?
Help inform and guide our reporting on patient safety by telling us about your own experiences. Complete the questionnaire.
We assembled five years of Medicare data to make sure we had a sufficiently large sample. We applied standard statistical techniques to account for differences among patients, such as their health and age. These sorts of adjustments are crucial to any comparison of physicians since those who work on healthier populations are likely to have better raw results. Our methodology was reviewed by outside experts whose views can be read here. Here again, we tried to be conservative. Our formula assumes that surgeons with the highest complication rates were victims of a certain amount of random bad luck. A surgeon with raw complication rates as high as 14 percent could have an adjusted rate as a low as 7.6 percent in Surgeon Scorecard. Conversely, we assumed that surgeons with no complications benefitted from at least some good luck. Therefore, the lowest adjusted complication rate any doctor can reach is 1.1 percent.
Our model produced a range of possible values for each surgeon’s complication rate. The number we’re reporting is the most likely one, near the middle of the range. We designed our online presentation of each surgeon’s scorecard to make clear that while higher and lower values are possible, they are increasingly less probable as one moves away from the rate in either direction.
Critics will fairly question why we are publishing rates that are probabilities, not certainties. The answer goes to the heart of why we believe Surgeon Scorecard will serve patients, doctors and the medical profession.
Statistics are ultimately about using less-than-perfect data to help make decisions. Many of us grab an umbrella when the weather forecast says there’s a high probability of rain. Mostly, we’re better off as a result, though sometimes we end up needlessly carrying extra weight on our commute. We see the complication rates as the starting point for patients as they talk to their primary care doctors or surgeon about a possible operation.
Some might say that a 5 percent complication rate is actually pretty low. After all, that means 95 percent are done without incident. But these are operations on otherwise healthy patients which some surgeons in our data performed over and over without any complications.
The data we’re making public today has long been in the hands of the Centers for Medicare and Medicaid Services, the agency that oversees the Medicare program and pays for two in five U.S. hospital stays. CMS and most of the nation’s hospitals have taken the view that it’s better not to calculate or act on this sort of information, in part because it is so controversial. Though no one says it aloud, the attitude has been that the risk of unfairly tarring any doctor trumps all other considerations.
Our reporting suggests that this reluctance to focus on individuals is one reason that patient harm has persisted in the face of considerable effort by the medical establishment. The landmark 1999 study by the Institute of Medicine, “To Err is Human,’’ called on Congress to create a nationwide mandatory reporting system for patient deaths and serious harm so researchers could detect patterns. That has never happened. Surgeon Scorecard is a crucial step in the larger process of spurring accountability for breakdowns in patient care. We know that many in the medical community will have ideas for improving it and we welcome them. We believe it’s an important tool for improving patient safety, one that incorporates the rights of patients as well as of doctors.