A medical crystal ball? Featured

7:00pm EDT December 31, 2006

For years, meteorologists, air traffic controllers and financial professionals have used predictive modeling to peek into the future. Mathematical modeling tools, ranging from relatively simple linear equations to sophisticated intricate software, harness the power of statistical analysis to predict real-world situations. As the financial stakes increase in the health insurance world, anticipating future risks becomes more important.

“In very simple terms, predictive modeling helps people grapple with mountains of data and make decisions about what to do, or not to do, to maximize outcomes,” says Sally Stephens, founder, owner and president of Spectrum Health Systems. “Although predictive modeling is rather late in coming to the health care industry, it has become widely popular.”

Smart Business spoke with Stephens about the benefits of predictive modeling in forecasting and preventing health care expense.

Why is predictive modeling a powerful tool in managing health care costs?

In the past, the health insurance industry focused on disease management programs. These initiatives offer special assistance to patients already diagnosed with specific high-cost conditions such as heart failure, diabetes and asthma. Disease management will continue to be a critical component of health care, but to manage costs more efficiently, insurers now need to analyze whole populations.

Predictive modeling can identify high-risk patients that might otherwise fall through the cracks. New tools go beyond the chronically ill with comorbid conditions and the chronically ill who have yet to develop comorbid conditions. Predictive modeling looks at outcomes for patient populations. Its broader approach encompasses plan participants who are at risk for chronic illness as well as those who are not currently at risk.

How can predictive modeling facilitate the underwriting process?

Traditional underwriters have used data such as age, sex, geographical location, industry and prior medical costs to project future expenses and set premiums. But as the amount of underwriting data available to health plans has exploded, effective underwriting has become increasingly challenging. Predictive modeling tools can be far more accurate than a simple analysis of claims outlays. For example: a $500 claim for a broken arm is not a good indicator of future claims, while $500 for an initial cancer workup could indicate significant claims in the future.

How can predictive modeling assist companies by educating employees?

Predictive modeling enhances employee education initiatives because it

  • Isolates the high-impact individuals for whom it is possible to effect the greatest change.

  • Pinpoints the diseases and conditions that need to be managed.

  • Identifies the population in need of education programs or telephonic support.

  • Determines what risk factors influence the utilization patterns.

  • Demonstrates the value-add of care management and disease management interventions.

In what ways does predictive modeling help individuals with chronic diseases?

Medical societies and disease-based associations, such as the American Diabetes Association, have established standards of care for the major chronic diseases. These standards list the tests, maintenance drugs and care frequency for each illness. Failure to comply with these standards generally results in patients with uncontrolled disease, worsening health and rising costs.

The medical community underserves plan members with chronic conditions. Compliance with national standards of care ranges from a low of 28 percent for members with asthma to a high of 69 percent for members with coronary artery disease.

Predictive modeling can assist care managers by identifying members who are noncompliant with the standards of care and provide appropriate level interventions to improve outcomes.

How can businesses obtain predictive modeling services?

Since predictive modeling has become increasingly popular, more employers will have access to this tool through a variety of sources. Third-party administrators, insurance companies and case or disease management providers are offering this system as a standard or additional service offering. Employers can seek the advice of their benefits consultants or insurance brokers to determine the best approach for utilizing predictive modeling as part of their cost management tool box.

SALLY STEPHENS is the founder, owner and president of Spectrum Health Systems. Reach her at (317) 573-7600 or sally.stephens@spectrumhs.com.