Why Predictive Modeling in Healthcare Requires a Data Warehouse (White Paper)
The healthcare industry has begun to adopt predictive analytics for a variety of purposes. Viewed by experts as a prerequisite for population health management, these statistical tools are being used to forecast which patients are likely to be readmitted to the hospital. Some healthcare organizations also apply predictive analytics to large clinical and administrative data sets in an effort to identify and intervene with certain patients before they become seriously ill.
At times, predictive analytics can be valuable. For example, a predictive modeling application that predicts the chances of patients developing a serious chronic condition or having a heart attack was successfully tested in a Kaiser Permanente clinic. As a result of clinical interventions, the risk of patients at that site developing coronary artery disease dropped 22 percent on average, compared to a decline of 9 percent in a clinic that didn’t use the tool.1
An Israeli study showed that using a five-point scoring tool could predict hospital readmissions with 80 percent accuracy.2 Other studies have used different models to identify patients who were at an elevated risk of being readmitted.3-4
Despite these successes, however, the evidence that predictive modeling (also known as “health forecasting”) can improve patient outcomes remains thin. In their article on risk stratification and predictive modeling, “The Promise and Peril of Healthcare Forecasting,” authors Frank Wharam and Jonathan P. Weiner note:
There is little evidence regarding how or whether [health] forecasting improves healthcare value. This is due to both the modest level of research and what is termed the “impactibility” problem. That is, even if prediction algorithms accurately identify at-risk patients, intervening to achieve desired outcomes is often inhibited by limitations of current disease management approaches or the general state of medical science.5
This is a key point in any discussion of predictive analytics. Unless the results of health forecasting can be translated into effective interventions with individual patients, the analytic tools will be useless. So healthcare organizations must develop the infrastructure and the culture required to turn the data into action. That infrastructure must provide the ability to generate timely reports and use automation tools to apply intervention strategies across a patient population.
Using computers to predict risks is not new. The Defense Department has long employed predictive analytics to model nuclear war scenarios or optimize the order of battle. The life insurance and casino gaming industries have also invested heavily in programs that help them calculate their odds of success.
Health insurance companies, similarly, use actuarial risk models to compute the chances that particular individuals will cost the insurers more than they pay in premiums. Until the Affordable Care Act took effect, health insurers utilized this type of analysis to determine whom to exclude from coverage and how much to charge the people they did cover. Some health plans have also been using it to intervene with high-risk patients in disease management programs.6
With the emergence of accountable care organizations (ACOs) and value-based reimbursement, many hospitals and healthcare systems have also begun to recognize that they need predictive analytics and health risk stratification to manage population health and deliver care more cost effectively. At the same time, provider organizations are now focused on reducing readmission rates so they won’t be financially penalized by Medicare.
The current interest in predictive modeling is part of a larger trend to employ business and clinical intelligence (B&CI) applications in healthcare. Until