Can Data Help You Find the Best Heart Surgeon?
By Jeff Craven
Heart surgery has come a long way from when Dr. F. John Lewis performed one of the first open heart surgeries on a 5-year-old girl in 1952 to correct a birth defect that left a hole in the wall of the upper chamber in her heart.
Today, cardiac surgeons can operate to repair or replace the valves of the heart, fix irregular heart rhythms, remove enlarged portions of the heart, bypass blocked coronary arteries with graft surgery, and transplant new hearts into patients with advanced heart failure. But, even among patients within the same hospital, differences in outcome can occur - which can depend on the health of the patient or the type of surgery. Random variation can also affect how outcomes are tracked and published.
But what can these reporting outcomes tell patients looking into heart surgery? How do cardiac surgical societies track outcomes, what do they track, and how do they measure complications like disability, stroke, and heart attack? How are centers using this data? How can patients interpret these outcomes?
Tracking three decades of post-surgery data
In 1989, the Society of Thoracic Surgeons (STS), an organization of cardiothoracic surgeons, created a voluntary National Database to track quality improvement and safety for their patients. The database is considered the gold standard in cardiac outcomes measurement, and, as of January 2018, includes components with long-term outcomes data for Adult Cardiac Surgery, General Thoracic Surgery, Congenital Heart Surgery and the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) for patients with heart failure.
“Cardiac surgery has really led the profession in being very proactive in trying to figure out what constitutes quality,” says Paul A. Kurlansky, MD, Associate Professor of Surgery and Director of Research of Recruitment and Continuous Quality Improvement at Columbia University. “We were also quite fortunate in the way that we have discretely measurable outcomes that we could look at.”
In 2010, the STS National Database began their public reporting initiative, ranking institutions that perform cardiac surgery for isolated coronary artery bypass grafting (CABG), isolated aortic valve replacement (AVR), and combination AVR+CABG surgeries. CABG is the most common cardiac procedure performed, and is a surgery that improves blood flow to the heart for patients with coronary heart disease. AVR is a procedure performed to replace a damaged valve that involves placing a new valve into the heart.
Patients and their families can use the public reporting website to search through centers and filter by each center’s respective score for CABG, AVR and combination CABG + AVR surgeries, which include a center’s overall composite score, and performance in terms of major surgical complications and survival.
The same year the STS National Database was founded, the Department of Health in New York also founded their own database, which tracks adult cardiac surgery outcomes from centers within the state. However, unlike the STS database, which compiles data four times per year and plans to soon release data on a real-time basis, the New York state database reports data that are approximately three years behind, which can paint a less accurate picture for patients using the state database as a decision-making tool on where to receive care.
Risk adjusting heart surgery outcomes data
Outcomes data is straightforward in cardiac surgery because the effects of surgery are usually evident: surgeons see a stroke, wound infection, or other complication within 24 hours of the procedure, and all this data is recorded by data managers at the hospital.
But interpreting how different centers perform over others when they have different case volumes, surgical teams, and patient populations would be nearly impossible without cardiac outcomes databases using what is known as risk adjustment. By studying cardiac outcomes carefully, statisticians can create a risk adjustment model and rate each metric with a confidence interval, or probability that the model will accurately predict the outcome within a certain range of values.
“What the risk model attempts to do is it tends to correct for differences in patients and to account for random variation so that you not only find out the actual mortality, but you find out whether or not there is a significant difference between a surgeon, or a hospital and what the performance is generally in the state,” says Dr. Kurlansky.
The most studied outcome in cardiac surgery is mortality or risk of death. The STS has created a mortality risk model based on the following factors:
- Primary procedure
- Prior cardiothoracic procedure
- Presence of any non-cardiac anatomic abnormality since birth
- Chromosomal abnormalities, syndromes or neurological issues
- Preoperative factors, such as the use of an intra-aortic balloon pump or ventricular assist device
- Mechanical ventilation for cardiorespiratory failure; and
- Kidney failure.
In pediatric cardiac surgery, the STS also adjusts mortality based on weight and prematurity.
Reading and interpreting heart surgery data
In the STS National Database, data is made patient-friendly through institutions being ranked in a star system “using a combination of quality measures for specific procedures performed by an STS Adult Cardiac Surgery Database participant.” A three-star designation is reserved for between 3% and 10% of centers in the database.
However, reading and interpreting outcomes is not always straightforward, Dr. Kurlansky admitted. For example, a 10% risk of mortality at a center may look very different based on the volume at that center. If the center has 100 cases with ten deaths, it would look the same on paper as a center where 100 out of 1,000 patients die from a procedure.
In addition, while the STS database can be easily interpreted by patients, reading data from a registry like the New York state database and understanding it will likely require a conversation with a patient’s surgeon to fully understand his or her own individualized risk, Dr. Kurlansky notes. A truly nuanced discussion, he said, is challenging because it will define what expectations, goals, and “red lines” a patient would have and base the conversation of outcomes around those points.
“Our job is to know our own data, but tailor our remarks to the patient in order to provide information in a context that is meaningful to him or her,” Dr. Kurlansky says.
Even still, people in New York are pretty sophisticated. “They come here and they've already looked you up,” says Dr. Kurlansky. “They've already gone to the internet and read about the procedure, but you have to provide context for them.”
Using outcomes to drive improvement
Some institutions, like Columbia, partner with vendors that let them view and analyze their STS data in real time, which allows surgeons to act on the most up-to-date data. Since Columbia is a high-volume surgical center and their staff is able to access their outcomes data quickly, Columbia can provide their cardiac surgeons with their group’s outcomes data as well as data for each individual surgeon. Dr. Kurlansky says Columbia surgeons use the data to see how they match up with national benchmarks, but they use it to improve the surgical and patient experience as well.
In one instance, surgeons at Columbia noticed their ventilation time—the amount of time patients stay on a breathing machine after heart surgery—was longer than expected. Most heart patients who wake up after surgery have the breathing tube removed, but in certain cases, like for patients who are sick, unstable, or have a respiratory condition, the breathing tube may need to stay in.
After digging into the data and looking at factors like how long patients were on the breathing machine, and what type of patients were staying on the breathing machine longer, they worked together with the ICU team to shorten the ventilation time for patients by over half.
“We use that data to answer the big questions: how we are doing [with] mortality, how are we doing [with] major complications, but also to help drive process improvement projects,” says Dr. Kurlansky.
Long-term outcomes remain elusive
While the STS and other cardiac surgery outcomes databases provide a year-to-year snapshot of how a center performs, one measurement has remained elusive: long-term patient outcomes.
There have been single-center studies that have shown good long-term results, while some institutions have matched their patients up to public death records to measure mortality and hospital records to find which patients had been re-hospitalized or needed reintervention after cardiac surgery.
“You have a tremendous conundrum in outcomes research because you do something to affect somebody's life if you really want to know not only did they make it, but how are they doing now? How are they doing five years from now? How are they doing ten years from now?” says Dr. Kurlansky. “You really need that kind of information in order to really understand the impact, the importance of what you're doing.”