Predictive Analytics Continue to Elude Healthcare

Though analytics can be descriptive and predictive, today most healthcare organizations are users of descriptive analytics, leveraging reporting tools and applications to understand what has already happened in the past to classify and categorize historical data.

What is needed by providers and payers alike is an analytics capability that “allows for analysis, allows for stratification and most importantly allows for prediction,” according to Kristin Landry, vice president of market analytics at Optum. Landry told an audience July 15 at Health Data Management’s Healthcare Analytics Symposium in Chicago that prediction is the key.

“The true value lies in the predictive analytical capability across the cost and clinical spectrum,” she says. “When you get to that capability of prediction, that’s when you can truly manage the population you are responsible for in the value-based world. Until then, you are playing catch-up. You’re looking at everything that’s already happened and you’re trying to avoid it from happening again.”

Data can be a powerful, valuable tool. Analyzing disparate data from across a health organization is vital to improving quality outcomes, increasing patient satisfaction and reducing costs. However, Landry argues that not all data is created equal. The real opportunity lies in the leveraging of clinical data in order to identify those patients that are high risk for the onset of chronic diseases and actively intervening before their medical conditions take a turn for the worse.

While claims data is generally available and defines cost and utilization trends, Landry characterizes claims data as “insensitive, not specific, and not timely.” On the other hand, she says that clinical data looks at clinical performance in a way that is “very sensitive, very specific, and very timely but it is not always generally available.”

Nevertheless, the ability to integrate and use both clinical data and claims data collectively as well as independently is crucial, according to Landry. “You need both to be successful,” she argues. “Because of their strengths and weaknesses and because of their focus, they answer different questions...the sweet spot is where financial opportunity meets clinical impact.”

Analytics technology is valuable but “good data” is the foundation for predictive models to drive better decision making and to take more effective actions in the future. “If the data is not good, the technology is meaningless,” asserts Landry. “What I mean by really good data, whether clinical data or claims data, is that it must be clean, normalized and takes in every component of data that is available to it.”