It's no secret employers are at a crossroads when it comes to health care costs. Many are weighing several different approaches in an effort to reduce costs, including financing options, plan design, employee contributions and vendor management.

Predictive modeling is one concept that is quickly gaining traction in the health care debate. It involves providing a risk assessment and adjustment process to the employee population to determine if the workforce is susceptible to contracting particular illnesses or disease states. This could help employers better predict their future medical costs and determine which health and wellness programs would suit them best.

"Employers should want to understand their employee population," says Russell Robbins, principal and senior clinical consultant with Mercer. "Predictive modeling is a key toward improving productivity and health care quality and also making sure the services are in place to help those in need."

According to Robbins, who presented at The National Predictive Modeling Summit in Washington, D.C., in December, there is a list of key predictive modeling principles, including: a focus on total population and addressing the entire health care continuum, emphasizing long-term behavior change, supporting health plan designs with strong communication and incentives, and creating data-driven programs tailored to individual risk, health status and learning.

To create these data-rich programs, Robbins and other experts urge employers to use health risk assessment surveys to get a snapshot of their employees' overall health status.

Employers should "complement and expand opportunities to address further domains of health that they may not be concentrating on," advises Dan Dunn, senior vice president of research and development at Ingenix, a health care IT solutions company based in Eden Prairie, Minn. The hope is that comprehensive personal health records will be created, which can "integrate information from a number of data sources to provide a multi-dimensional profile of an individual's health," Dunn adds.

Working models

One employer has created a new pharmacy model by using predictive modeling, Robbins noted. It looked at data regarding employee illness, predicted costs, and then waived copays on generic drugs and halved copays on brand-name drugs treating diabetes, asthma and heart disease. The result: first-year savings from reduced nonpharmacy medical costs were equal to the cost of the copay reductions.

"Employers should expect predictive models to provide them the ability to understand the current workforce and trends," said Robbins, "so they can make more informed business decisions on future health care costs."

Another Mercer case study involved a university testing a new diabetes pilot program. Modeling showed that nearly half of certain diabetic populations did not follow an appropriate pharmacy treatment regimen, and that those at risk could develop severe symptoms if they continued to be noncompliant.

Under the university pilot program, copays were eliminated for any medication treating diabetes, including ACE inhibitors (pharmaceuticals used primarily to treat hypertension and congestive heart failure), antidepressants and blood-sugar control drugs. The program also includes educational material and focused outreach efforts to improve workers' health.

Robbins did not have data results at the time of the conference, but wanted to present a working model that could possibly serve as a blueprint for employers one day. According to the Centers for Disease Control and Prevention, roughly 20.8 million people in the United States have diagnosed or undiagnosed diabetes. The university is tracking costs, risks and absenteeism.

Robbins stresses that the information employers receive is de-identified data, meaning they cannot tell what a particular employee is afflicted with.

Evidence-based medicine

The basis and landscape for predictive modeling is evidence-based medicine, which involves assessing clinical data to develop better quality health care procedures.

"Evidence-based medicine is about finding the right course of treatment and the right diagnostic method to improve our risk profiles," remarks Paul Keckley, executive director of the Deloitte Center for Health Solutions, based in Washington, D.C.

According to Keckley, there are some common misconceptions about EBM that should be dismissed. For example, some label it "cookbook" medicine, he says, when it's really based on population-based guidelines, and not applicable to every patient.

Another misconception is that EBM is about changing physician behavior, when it's really about increasing adherence among clinicians and patients, Keckley observes.

For predictive modeling to work effectively, Keckley continues, greater access to clinical data from provider and patient sources is needed. Programs should intervene early, to "positively influence patient habits and match to useful tools and resources."

The April 1 EBN will feature expanded coverage on predictive modeling.