What do the Oakland A’s have to do with improving delivery and outcomes in healthcare? Potentially, quite a bit according to Dr. Neil Kudler, CMIO at Baystate Medical Center in Springfield, MA. Kudler teamed up with Larry Schor, senior vice president of corporate development and analytics at Medecision, to see if the evidence-based approach to drive strategy — employed by the Oakland A’s and made famous in the Oscar-nominated movie, “Moneyball” — could be adopted to assess and improve physicians’ performance and patients’ outcomes.
Compiled By Scott Westcott, Contributing Writer
With all the talk of Big Data, there are still big questions as to how to most effectively leverage information and data to make a positive impact on healthcare delivery, cost, and outcomes. One health system leader thinks an approach developed by a Major League baseball team might be a game changer.
What do the Oakland A’s have to do with improving delivery and outcomes in healthcare? Potentially, quite a bit according to Dr. Neil Kudler, CMIO at Baystate Medical Center in Springfield, MA. Kudler teamed up with Larry Schor, senior vice president of corporate development and analytics at Medecision, to see if the evidence-based approach to drive strategy — employed by the Oakland A’s and made famous in the Oscar-nominated movie, “Moneyball” — could be adopted to assess and improve physicians’ performance and patients’ outcomes.
In “Moneyball,” management adjusted its game plans based on analysis, real-time information, and experience. Similarly, Kudler and Schor proposed population health management (PHM) is also a team effort, with each teammate playing a specific position, working in a system, anticipating and preventing risk, and adjusting the plan with new data. Kudler recently shared his perspective on the potential of Big Data to transform healthcare as well as insights gleaned from the “Moneyball” experiment.
Q: How do you define Big Data, and what role do you see it playing in healthcare?
A: I view Big Data as consisting of large data sets coming from disparate sources which allow an analytics approach to understanding populations both large and small, while also providing the opportunity to drill down to the individual patient. For years the data we worked with has largely been claims-based, and there is not a lot of punch behind that type of data because of the lack of aggregation to give us a longitudinal look. Big Data gives us a lot of hope in being able to have that broad view that can funnel down to a very specific view. One key benefit is that we can now ask questions in ways we never have been able to do before. We can ask questions regarding variations in care, gaps in care, and effectiveness of delivery, then turn to the data to give us some direction and clarity for both identifying the problems and developing solutions. Big Data can assist in identifying risk of medical error or identifying how we can prevent excessive expense by figuring out when and where to intervene with certain patient populations. Having these data sets allows us to view certain populations so we can compare patients across the board and create algorithms that can help us predict future outcomes.
Q: How are you leveraging Big Data to manage the shift to increased attention on managing risk, gaining efficiency, and improving quality?
A: Right now we are mostly focusing on building adequate data sets into data warehouses. Never before have we been able to join so many different types of data. We can view claims data along with real-time clinical data, financial data, and patient experience data with even demographic data such as credit reports. We can look at clusters of populations, zeroing in on zip codes or neighborhoods to help determine what a patient might be at risk for. We know adherence to medication regimens and recommendations of treatment is critical in creating success, yet we have never really been able to program toward that to predict who has an adequate understanding of what is available to them. Now we can turn to Big Data to try to better determine medical literacy rates in ways which will allow us to focus our resources on driving better outcomes.
Q: What stands in the way of using Big Data to improve care?
A: First, there is the overwhelming change management that needs to occur in order for the healthcare system to buy into the value of a true Big Data platform. Second, there are the barriers created by the interoperability of disparate data sets. Third, in the reimbursement reform landscape, there needs to be a means of improving ROI in order to facilitate investment in Big Data platforms. Too often, what we are trying to achieve with Big Data is in conflict with the reality of our current healthcare system. For instance, one important goal is to reduce in-patient hospitalization; however, for most healthcare systems that is the most significant revenue-producing venue. We have to somehow figure out how, if we reduce that revenue, we can come out on the other end with a return on investment. I can speak from personal experience that it is largely a philosophical debate at this point, and it is only the health plans that are pushing the envelope by presenting sophisticated risk contracts that do encourage this kind of risk-taking.
Q: How are you extracting actionable intelligence from Big Data, and how are your data findings and analysis driving action?
A: We’re in the midst of creating an enterprise data warehouse that will be able to provide a greater number of data points for analytics. We are just about to kick off a project whereby we are not only going to be able to take our clinical data but also, through the use of our health information exchange, be able to partner with other health organizations and aggregate the data so we can derive benchmarks for performance, which is another method of driving physician engagement and improved practice performance. From there, we are also trying to look at smaller questions to identify not just gaps in care but also the strategies to address those gaps in care. At Baystate, one of the ways we have chosen to do that is to engage with vendors to come into partnership. One unique aspect of Baystate is that the population we serve is terrifically diverse and highly representative of the entire country, so we have come together with partners to whom we off er our data set, our political expertise, and our IT ingenuity and work with them to be something of a late stage accelerator for some of their products. Th at is where this idea emerged for the idea of taking a “Moneyball” approach.
Q: Can you summarize what you aimed to achieve with the “Moneyball” experiment?
A: What we were trying to achieve with the “Moneyball” approach was to look not only at past results but also at the dynamism of past results and how they may lead to future performance or future improved results. So it goes beyond assessing patient test results, but what a particular physician’s success has been with patients of a particular subset — in this case diabetes. Th e data helped identify which physicians need more support as well as which patients would benefit from a more focused effort on care management to promote better healthcare literacy. It is really hard for individual doctors to assess how they are doing and where they need to invest resources to improve the care of their entire set of patients. It’s one thing to know which patient’s diabetes is out of control but another to be able to know what is going to be the right strategy to bring that individual back into control, as well as identify subsets of patients that would benefit most from an investment of financial and human resources. Th is approach may off er a practical way to achieve that.
Q: What is your key takeaway from the experiment?
A: Data can’t be static — it needs to be instructive and actionable. Data needs to be presented in intuitive form and used in ways that incentivize individual doctors and groups of providers to perform better and use resources more effectively to lead to better outcomes. No one wants to be static in the care they are delivering. Physicians ultimately do want to help their patients and perform at a high level, and this type of approach could help make that happen.
Q: What needs to be done to extract more actionable intelligence from the data?
A: Interoperability and normalization of data would take us a long way toward having intuitive tools that do not require data scientists to leverage. It speaks to moving from a volume-based strategy to a value-based strategy. CMS has accepted applications for participation in the Next Generation ACO (NextGen ACO). Approximately 35,000 of our Medicare patients will be in a program that will provide the same care they have always received but with aspirations to provide the right care at the right time for the right patient. Th is will surely engender greater sensitivity to the population’s needs as well as allow us to be more tactical in our approach to the broader population in general.
Q: What aspects of healthcare will be most impacted by embracing Big Data analytics?
A: What I hope is that Big Data is able to level the playing field in how healthcare dollars are spent. We know that if you look at populations in, say, Miami, Los Angeles, New York, and Boston, healthcare dollars are spent at among the highest rates in the country. Patients often go from one doctor to the next, and without the right information, the same expensive test might be ordered, or the incentive to do the test is based on creating financial reward for the physician. Th at is something that needs to change, and I am very hopeful that is what new models will do — reward healthcare systems and physician groups that have a sense of awareness about providing the right care at the right time for the right patient, and doing it once with a high degree of efficacy. It has a lot to do with changing incentives so an organization is being judicious instead of voluminous. It’s about using data to make sure we deliver the right tests at the right time for the right patient so they are generating a clinical benefit and real positive change.