Q&A with Shay Bess, M.D.
Shay Bess, M.D., believes the key to the future of medicine is data—and he works tirelessly to spread the message of its power to transform the quality of healthcare. As director of surgical quality and resource utilization for spine services at the Denver International Spine Center at HealthONE’s Presbyterian/St. Luke’s Medical Center, and founder and president of the International Spine Study Group (ISSG), a nonprofit research foundation, he has merged his love of hands-on clinical practice with a desire to improve patient care through academic research. He spoke with The Source about the future of predictive analytics and its promise for improving patient outcomes and reducing costs and complications, particularly for at-risk populations.
What led you to your present role balancing clinical and academic duties?
After medical school and residency training, I worked in orthopedic spine surgery at several places including the University of Utah and San Diego Center for Spinal Disorders. For about eight years, I was director of the pediatric scoliosis services at the Denver International Spine Center at Presbyterian/St. Luke’s Medical Center, and then I was recruited to be the director of the scoliosis and spine research programs at New York University (NYU). I was there for a year before I realized that one does not need to be in a university setting to be productive in research and academics. The research we were doing in Denver and with ISSG was more productive and influential than some of the work I had done in a university setting. Consequently, I returned to Presbyterian/St. Luke’s Medical Center in 2016 to serve as director of surgical quality and resource utilization for spine services and to focus my efforts on the ISSG, HCA Healthcare and HealthTrust.
My practice is devoted to pediatric and adult patients with spine problems, the majority of whom have scoliosis. Adults with spine deformities are one of the most expensive patient populations. They are also one of the most at-risk for complications. So, much of our research seeks to identify and mitigate those risks and improve care for this patient population.
What is the main focus of the International Spine Study Group?
The current incarnation of ISSG (issgf.org) was founded in 2009. Dedicated to advancing the treatment of adults with spinal deformity and complex spine surgery, the ISSG has presented the model for multi-center research efficiency as well as over 1,000 abstracts worldwide and is now the most productive spine study group in the world. With the goal of achieving the best possible outcomes, we have 20 participating sites nationwide sharing ideas and producing research studies on spine treatments and techniques. We collaborate with the European Spine Study Group (ESSG) based out of Barcelona, Spain, and we also conduct research in Japan and Korea.
Working with the ISSG sparked my interest in predictive analytics, or using data and algorithms to try to predict which patients will benefit from certain treatments, who will likely have complications, what the costs will be, and the duration of hospital stay. The ISSG initiated an online database in 2009 to allow for seamless data entry and centralized data quality assurance.
Our research has demonstrated that, based upon a patient’s profile, we can predict outcomes—not just how they’re going to do in surgery, but also how they’re going to respond to post-surgery questions, such as, Do you feel healthier? Can you now do the things you want to do? Do you like your posture and the way you look? Not only does the data we’re collecting hold enormous potential for us to assess who is a good candidate for spine surgery and who is at risk for complications, but it also enables us to counsel patients and identify their post-surgery goals. This kind of analysis has positive ramifications for the patient and physician, as well as for the hospital and payer systems.
How would you describe the evolution of analytic research?
There has been an evolution in research from descriptive analytics (what happened) to diagnostic analytics (why did it happen) to predictive analytics (what will happen). There are numerous examples of predictive analysis in finance and business. Stock analysts assemble your retirement portfolio based upon your wishes and capacity to take risk. Investment firms use data to know when to buy, sell and trade. Companies like IBM can use data to predict if an employee is going to quit after six months, be promoted or stay in the job for 30 years. Academic institutions use data to improve student retention and optimize their resources.
So, why can’t those of us in medicine make the same kind of predictions when it comes to outcomes, length of hospital stay, types of complications, catastrophic costs, and good and bad spine investments? We should be optimizing data and using predictive analysis to tailor our care for patients just as other industries do.
How do you determine which predictive questions to ask your spine patients?
Machine learning, a subfield of computer science, is only as smart as the questions asked. The algorithms you develop can quickly take you in a wrong direction if you’re not asking the right questions. As with most forms of artificial intelligence, good data begets good data.
That’s why we plan our questions depending on our goals: Are we trying to reduce hospital stay or reduce duration in the ICU? Are we trying to preoperatively identify cases that might trigger catastrophic costs? We discovered that within roughly a 95th percentile we can predict what the cost of a patient’s care is going to be at the outset. Data becomes an extremely powerful tool, allowing us to pinpoint outliers and then change variables that are driving up costs so that the care becomes affordable.
Since August 2016, Shay Bess, M.D., has served as director of surgical quality and resource utilization for spine services at the Denver International Spine Center at HealthONE’s Presbyterian/St. Luke’s Medical Center. Previously he served the hospital as director of pediatric scoliosis services from 2008 to 2015. In 2015, he served as chief of adult spine deformity service and chief of spine research at the New York University School of Medicine and Hospital for Joint Diseases. From 2007 to 2008, he was a fellowship faculty member in pediatric and adult spinal surgery at the San Diego Center for Spinal Disorders in La Jolla, California. From 2005 to 2007, Bess was assistant professor of orthopedic surgery at the University of Utah School of Medicine.
Bess earned his undergraduate degree at Columbia University and medical degree from Johns Hopkins. He completed an internship and orthopedic surgery residency at Case Western and completed a fellowship in pediatric and adult spine surgery at Washington University in St. Louis. Bess is the president of the International Spine Study Group Foundation, which was founded in 2009. He’s active with the Scoliosis Research Society and the North American Spine Society, among other organizations.
As we looked further into our data set, we found that we could predict patient responses to different questions on a validated, standardized questionnaire within 70 to 80 percent accuracy. The questions we asked targeting pain level, activity level and attractiveness proved to be the most accurate. These questions included: How much pain have you experienced during the past six months? What is your current level of activity? Do you feel attractive with your current back condition? We use activity markers to see not only if patients are more active physically and socially, but if they’re also more satisfied with their lives. We are looking at how effective certain treatments are from a cost standpoint and in improving the quality of life.
We start by finding out what the patient’s expectations are at the outset and changing the way we assess outcomes. Patients want answers to their issues: Will I walk better? Will I be able to return to work? Will my pain improve? Will my mood improve? Will I feel better about the way I look? Will I be satisfied with surgical treatment? Because of predictive modeling’s ability to put incredibly powerful data in your hands, we have the ability to tell patients they will improve by roughly X amount or their pain scores will improve by X amount. The data can also indicate potential complications, so that you can more accurately counsel a patient on their best surgical options as well as make strategic budget decisions.
How do you handle resistance to change when it comes to data analysis? How do you convince people that even if there’s a little pain to get through the start-up phase, it will ultimately result in something better?
Everybody wants what’s best for patients, but it takes time for some people to be open to the power of data analysis to change patient lives on a daily basis. Once you show surgeons the capacity of data to improve their outcomes, they tend to come around and adopt meaningful change in how their care is delivered.
It’s hard to get people outside of their comfort zone, but change is going to happen anyway, so why shouldn’t we be at the forefront of it? Look at the latest disruption to the healthcare system from Amazon, Berkshire Hathaway and JP Morgan. By banding together and sharing information, these companies are using data to reduce healthcare waste and build new efficiencies for their employees.
If you use data effectively, you can identify areas where waste can be reduced safely and efficiently. These more streamlined ways of practicing medicine are not only faster and easier, but they make the whole system run so much more efficiently. Data can be utilized to improve cost efficiencies, reduce wait times and lessen the amount of unnecessary labor that goes into providing care for patients.
How can data help reduce the redundancy and variance that exists in the supply chain?
When you look at the amount of goods delivered to a hospital at a single point in time, it can be staggering. Some of it is unnecessary. If we could predict what we’re going to use in a certain case—whether it’s screws or biologics or a rod of a particular shape or length for a scoliosis surgery—we can better manage our hospital inventory and eliminate a huge amount of redundancy. Data can also help us reduce shipping costs and improve efficiencies in supply processing.
We are going to have to change the way we look at our routine practices. Healthcare professionals are somewhat reluctant to adopt these efficiencies because they’re used to having the comfort of excess inventory waiting in the back room just in case. We have found that the data to help us predict exactly what was needed for a specific procedure is sitting there, but it’s not being used. When data is used to predict specific patient behaviors and outcomes, facilities can better plan surgeries ahead of time and markedly reduce variance and redundancy.
What are some of your goals as a HealthTrust Physician Advisor?
When I was asked to be part of the physician advisors program at the beginning of 2018, I was excited about helping others develop algorithms and use data platforms in a significant way to improve care delivery. There’s so much data available to us through HealthTrust, and I would like to help analyze it on a national and global level to improve supply chain and reduce waste to a nominal level. Healthcare costs are astronomical and non-sustainable because of the variance and redundancy within the system. The good news is: The solutions are right here in front of us. We just need to apply the data we have to optimize patient situations.
In what ways are you promoting the message of the power of data analytics?
Since 2009, our ISSG team has presented more than 1,000 abstracts across the world. I’m fortunate that I’ve not only been part of the research team, but I also get the opportunity to share those results with others. That’s critical: You need to be able to effectively enunciate your concepts or they won’t help anybody. You have to put your ideas together distinctively and open people’s eyes to new methodologies for improving the way they practice—further enhancing the good points and improving on the bad. I want to be active in the decision-making process for how healthcare is delivered. If it isn’t going the direction that we want it to go, I want to be a driver for change, not just a passive observer.
What do you enjoy most about the career you’ve forged so far?
We have to be passionate in our lives; otherwise, what’s the point? We have to enjoy what we do. Luckily, I get to work as a physician providing patient care in a clinical setting, but I also get to work in academics, developing algorithms and trying to improve the way we deliver care at a system level. That’s the fun of medicine for me. I have been able to wear these different hats, have a diverse practice, and continue to grow and learn.