Predictive Modeling: Finding the High-Cost Employee

By Dr. Julie A. Meek of The Haelan Group Apr 12, 2005

last reviewed October 2013

Health care coverage is seen by many employees as an entitlement guaranteed by employment. Yet, each time an employee comes on board, an employer must add another hefty line item to the business expense column and struggle with the decision of how much of the financial burden to put on the company's or the employees' shoulders.

Unfortunately, the promise of health care cost containment is still a work in progress": Claims costs continue to increase between 10 to 30 percent per year, while premiums keep rising. All the while, a health care tug of war between employers and employees continues over who will bear the burden of these increased costs.

That's why more and more third-party administrators, professional employer organizations and other HR consultants are offering predictive modeling as a valuable benefit program. Employers can inquire or encourage either their insurance company or benefits consultant/broker to provide predictive modeling, if they already don't.

Disease Management: Just a Start

In recent years, health plan administrators have adopted disease management (DM) techniques. DM focuses on finding plan participants who meet a certain disease criteria and encouraging them to proactively treat their condition. Patients with diabetes, asthma, coronary artery disease or low back pain are prime targets.

But, after more than a decade, DM has not been able to bring costs under control and, as evidenced by a report by the Congressional Budget Office, the return on investment has been less than compelling. Why haven't DM programs created the cost containment its proponents predicted?

People become sick and seek costly care for symptom relief, which may not be related to the presence of a specific disease. A study by the Institute for Health and Productivity in 2000 showed that two-thirds of patients and three-fourths of health care costs are not attributed to the top 10 most costly conditions meaning that conditions are than those most commonly targeted by DM programs is causing the majority of the costs.

Treating employees after a diagnosed disease doesn't bring down costs.Finding the sickest employees before they reach a crisis is the key to reducing costs.

So, past attempts to control costs have not been very accurate in prospectively targeting the future, high-cost health care employee. Fortunately, new science and technology can help employers find the costly group of employees. Using a predictive modeling/health management firm to implement the program and gather employee information (which is treated as highly confidential following HIPAA rules and never shared with employers) allows health coaches to intervene early to help employees avoid high-cost care in an effective and supportive way.

What Is Predictive Modeling?

Predictive modeling is a set of tools used to stratify, or sort out, a population according to its risk of nearly any outcome. This means finding employees that will seek high-cost, near-term health care (in the next six to 12 months). As a result, this subset of employees can be helped, via health coaches, to better manage their health and reduce their reliance on avoidable, costly care.

Combining survey-based health assessments with predictive modeling presents an even better picture of those that will generate future, as opposed to past, high claims and a way to help employees avert this eventuality.

That is, with the right type of confidential survey and predictive modeling tool, 10 percent of employees that will account for 70 to 80 percent of employer costs can be quickly and effectively found. And, as a by-product, employees get a level of coaching and targeted health resources that help them make better health decisions and become savvy health care consumers.

A predictive modeling health program requires little effort from employers other than providing incentives to boost voluntary participation and helping communicate the programs goals to employees.

Ensuring Participation

There are three key ingredients to high participation:

Privacy protection.



It's imperative to help employees understand that no one except the service-delivery personnel for the health management firm will have access to the employees' protected health information (PHI). The predictive modeling/health management firm's communication plans should include every media outlet the employer can offer in addition to phased and extensive communication process that includes face-to-face meetings with employees. Finally, the firm should help the employer to design an incentive plan that rewards employees for participation.


Although employers are limited in their ability to require employees to submit to a medical examination, the Americans with Disability Act (ADA) allows employers to establish health promotion programs for employees, so long as those programs meet the following requirements:

Participation by employees is voluntary.

Information obtained from the examination remains confidential.

Information obtained from the examination is not used to discriminate against any employee.

Survey-based predictive modeling adheres to HIPAA privacy standards. The firm collecting the data and providing health coaching services only knows what the employees (or their spouses) self-report. Data is made available only to those with a need to access an employee's health information, such as the employee's health coach.

The employer only has access to summary-level data, which doesn't reveal individual information. Except for the eligibility feed (limited to demographic information), there are no other data transfers between the predictive modeling/health management firm and the third-party administrator and health plan.

Focused Effort, Real Results

Survey questions are based on new "perceived health" principles. These principles define people as at-risk when their sense of feeling or doing is below their own specific level of tolerance because care-seeking behavior is so directly tied to the desire for symptom relief. People are not at-risk if they only have a specific disease. (See graphic: Traditional vs. New View of Health.

For most people, there are a variety of factors that contribute to their sense of feeling ill, in addition to specific medical conditions: stress emotions, how they are functioning in everyday life, compliance with treatments, the relationship with and trust of their physician(s) and personal beliefs -- all these can greatly impact people's perceptions of their health.

The survey results produce a very specific score showing which employees are high-risk (most apt to consume a lot of health care dollars near term) and which are low-risk. Health coaching by telephone can then be focused in a more strategic way.

Once the results are in, both high- and low-risk employees receive a very personalized Health Action Guide. This resource, based on each persons own survey results, gives employees specific step-by-step instructions on how to prepare for an upcoming physician's appointment (what questions to ask), manage stress, and improve or maintain healthy lifestyle behaviors. In effect, this individual report provides a self-guided health coaching session resulting in a specific health improvement plan for each employee who responds to the survey.

Because their health concerns are more complex, the high-risk employee group receives more intensive, one-on-one health coaching over a 6-12 month timeframe. A health coach is trained to interact with high-risk individuals from a motivational perspective rather than from a reminding/telling/advice-giving perspective. The goal is to provide the correct balance of support, encouragement, and information to motivate employees to seek assistance, learn new skills and make lifestyle/behavior changes that improve their perceived health.

This type of personalized contact leads to overall improved health and a decreased use of health care resources.

During the course of a health coaching session, the high-risk employee will discover one of five pathways that, if successful, would most impact and improve how they're feeling and functioning. These five pathways are integral to improving the persons level of perceived health, thus influencing their health care seeking behavior:

Stress emotion management (depression, anger, anxiety, etc.)

Lifestyle behavior change (smoking, overeating, lack of exercise, etc.)

Improved interventions with the health care system

Physical symptom management

Chronic disease condition management

As part of the intervention, the health coach encourages the high-risk individual to describe their situation and challenges in tackling their unique health concerns. A health coach becomes a catalyst for change by helping a patient identify problem areas, set achievable goals for self-improvement, find and utilize health plan and community resources and set up additional, follow-up calls. Per high-risk member, a health coach spends an average of 2 to 2 hours on the phone over a 2-4 month (or longer) period of time.

Making the Program Work

With a high level of predictive accuracy and focused health coaching, survey-based predictive modeling makes a real difference on an employer's bottom line. Results to-date show that in addition to lower claims, predictive modeling can show a first-year return on investment of 30 percent.

Finding the high-cost employee and directing specialized resources at the right people helps employers meet what is sometimes seen as incompatible goals: keep expenses down to boost company performance and take care of employees to keep them happy, healthy and productive.

Dr. Julie A. Meek, president and CEO of The Haelan Group, an Indianapolis-based health care consultancy, has been designing, testing and evaluating systems for effective population health management for nearly two decades. You can contact Dr. Meek at

Job Finder

Find an HR Job Near You
Post a Job


Find the Right Vendor for Your HR Needs

SHRM’s HR Vendor Directory contains over 10,000 companies

Search & Connect