Patients undergo more than 600,000 knee replacements every year in the United States. But not all knees are created equal. A 35-year-old athlete may need a simpler, less costly surgery than an 82-year-old with severe arthritis. The hospitals that treat both patients often are paid by the number of services they deliver—not necessarily the outcome and quality.
The first step to improving patient care, the critical challenge in healthcare analytics, is to make it more measurable and factual across a single “episode of care,” such as a knee replacement or pregnancy or heart attack.
Explains Alex Beauvais, vice president and general manager of our healthcare analytics group: “We can map diverse patient journeys across thousands of cases to make measurable, apple-to-apple comparisons that are clinically and technically reliable. For each episode of care, we have physicians and medical coders who can design these patient journeys and identify the sources of value within them—for example, using an outpatient setting rather than a hospital. Then our advanced-analytics and technology staff guide the data inputs and modeling from start to finish. We take into account the many variations, including outlier cases within a single episode of care.”
A key focus area for Alex’s team is eventually using these episodes of care and tangible facts to develop payment models that incent providers to shift from delivering “more individual services” to focusing on “overall better patient outcomes.”
With an MBA and extensive technology experience, Alex joined McKinsey seven years ago to help build our big data capabilities and eventually focused exclusively on healthcare. Today, our North America–based team counts some 100 physicians, medical coders, data scientists, statisticians, and software developers among their ranks—working alongside strategists, consultants, and project managers.
They design and develop algorithms and run analytics on a big data platform that integrates 100 terabytes of proprietary and third-party claim, encounter, and clinical data. McKinsey also has a healthcare-analytics team in Europe.
“We can help a client define a strategy and also design a comprehensive model loaded with production-grade algorithms to forecast the impact of that strategy,” Alex says. The team is also developing models related to primary care, provider performance and growth, payment integrity, and special-needs populations.
Modeling data is an art as well as a science. Gerold Scheunemann, a senior data scientist based in Dusseldorf, has over 15 years of experience in analytics. He explains: “Success isn’t about just having the most powerful or complicated model; it’s about having a powerful model and getting the right information into that model. After all, data are always neutral. They tell you the truth only when they include all the relevant information. A highly sophisticated model with insufficient data is much less useful than a good model with a comprehensive database.”
One other area where analytics can play a significant role is improving the quality of care for the long-term population suffering from mental illness or behavioral-health conditions. “This is a very nebulous area in healthcare, and often a patient will have several interrelated conditions,” says Alex. With analytics, we can help payors and providers better understand such a population, with answers to questions like these: How many patients have multiple conditions? What is the optimum care-delivery mechanism for each condition? What factors explain why some people stick to their meds and others do not?
“With analytics, for the first time we can put facts on the table,” notes Alex, “and that completely changes the conversation.”