Research Project One (Complete)

CREATION OF THE HERO DATABASE- HERO has facilitated the creation of a large, retrospective, multi-employer health promotion research database. This was accomplished by collaboration among HERO, the StayWell Company, the MEDSTAT Group, and six large employers: Chevron Corporation, Health Trust, Inc., Hoffmann La Roche, Marriott Corporation and the states of Michigan and Tennessee. All employers are clients of The StayWell Company and The MEDSTAT Group. A top priority in the creation of the HERO database was the ability to examine the impact of risk factors, risk factor combinations and risk factor change on individual medical expenditures.

The HERO health promotion research database includes 47,500 employees, all of whom completed a common health risk appraisal (HRA), the StayWell Health Path®, and were enrolled in a fee-for-service, self-insured health care plan for the study period of 1990 to 1996. Approximately 12,000 of the employees completed two or more HRAs during this time. The HERO database was created by connecting the HRA data set with the medical claims data set along with the eligibility data set. Including the eligibility data permitted the inclusion of study subjects that had no medical claims. The confidentiality of individuals was maintained by scrambling personal identifiers across all data sources. The merging of these data sets yielded 113,963 person years experience. The previous largest research database of this kind is the Control Data - Milliman Robertson database which includes about 13,000 study subjects and provides approximately 40,000 person years experience.

Creation of the HERO database could have a major impact on the future of health promotion and disease management research. It is amenable to the design of numerous longitudinal research studies that examine the association or impact of single risk factors, risk factor combinations, risk factor change, selected chronic diseases and demographics on: medical costs, diagnosis, treatments, procedures, outcomes, hospitalization or any other parameter usually recorded in a typical fee-for-service medical claims database.

A consortium of 20 HERO sustaining partners funded the creation of the database. HERO is willing to facilitate research, using the database, for outside clients.

Research Project Two (Complete and Published)

THE ASSOCIATION BETWEEN TEN MODIFIABLE RISK FACTORS AND HEALTH CARE EXPENDITURES - A medical economics study using the HERO database, that addresses two questions:

• Do those at high risk have greater health care costs than those at lower risk? If so, what is the amount over time? What is the most expensive, what is number two, number three, and so on?

• Do those with specific risk factor combinations have greater medical expenditures than those without these risks? If so, what is the amount over time?

This research project is complete, peer-reviewed and published in the October 1998 issue of the JOURNAL of OCCUPATIONAL and ENVIRONMENTAL MEDICINE, (Goetzel, Anderson, Whitmer,, JOEM, (40) (10). October 1998; 1-12). It was funded by a consortium of 22 HERO sustaining partners. Research design involves a retrospective, two-stage, multi-variate analysis, including logistic and linear regression models. This permits the examination of specific risk factors as independent variables, thus eliminating the impact of other risks. In addition to adjusting for specific risk factors, other confounding factors adjusted for were: gender, age, educational level, race, type of job, employer and number of months employees were followed after the first HRA was completed. Ten risk factors were evaluated, six self-reported and four biometric. The self-reported were: physical activity, alcohol consumption, nutrition, tobacco use, stress and depression. The biometric measurements were: cholesterol, blood pressure, blood glucose and weight.

Using the HERO database, research inclusion criteria were: active employees age 18 to 64 at the time of the first HRA and those who could be followed for at least six months after the completion of the first HRA. Based on this, there were 46,026 study subjects, all of whom completed a common HRA and were enrolled in fee-for-service health care plans. They were followed for up to three years after the completion of the first HRA.

It was found that those with self-reported, persistent depression (n=997, 2.2% of the study sample) had adjusted annual health care expenditures 70% greater than those who reported not being depressed. Number two was uncontrolled stress (n=8,641, 18%). These individuals had annual adjusted medical costs 46% greater than those who were not stressed. The third most costly risk was high blood glucose (n=2,271, 5%), with adjusted medical expenses 35% greater than those with normal blood glucose. The other most costly risks in descending adjusted order were: obesity (+21%), tobacco use (former +20%, current +15%), high blood pressure (+12%), and poor exercise habits (+10%). There was a dichotomy between the adjusted and unadjusted data relative to high cholesterol levels (n=8,641, 18%). Based on unadjusted data, health care costs were 17% greater than those with normal cholesterol levels, however, when adjusted, health care costs were 0.8% lower. Those at high risk for health problems due to excessive alcohol consumption (n=1,723, 4%) had adjusted health care expenditures 3% lower than those at lower risk. This is not unexpected, as those with drinking problems often avoid the health care system. In the case of nutrition, those who reported poor nutritional habits (n=9,278, 20%) had adjusted health care expenditures 9% lower than those who reported good nutritional habits. This finding was perplexing, because it is in contrary to the body of published nutrition research. It may be explained by the fact that the impact of all other risks usually associated with poor nutrition (obesity, hypertension, high cholesterol, high blood glucose) have been eliminated through the adjustment process.

The finding that psychosocial risks were the most costly was unexpected and medically newsworthy. This study suggests that sufficient attention should be directed toward worksite depression and stress screening along with the opportunity for adequate diagnosis and treatment.

Research Project Three (Complete and Published)

THE GENDER SPECIFIC, EFFECTS OF MODIFIABLE HEALTH RISK FACTORS ON CORONARY HEART DISEASE AND RELATED HEALTH CARE EXPENDITURES, (Wasserman, J, Whitmer, R, Bazzarre, T, et. al., Jour Occup Env Med, (42)(11), November 2000; 973-985). This research project uses the HERO database. Of the 46,026 employees in the database 2,459 were diagnosed as having CAD through CPT and ICD-9-CM codes. This represents the study group.

A variety of risk factors, many of which are controllable through lifestyle changes, contribute to the probability of CAD and health care costs. The risks include: stress, smoking, obesity, hypertension, diabetes, lack of regular exercise, high cholesterol and family history.

While there is minimal data available on the comparative economical impact of risk factors on health care cost, there is even less when investigating this question based on gender. Because of this, the following questions are addressed:

• Based on gender, what are the absolute and relative costs associated with each risk factor when treating patients with CAD? What are the differences across risk factors? Are there differences in costs when comparing males and females? If so, what are the differences?

• Based on gender, what is the occurrence of CAD and hospitalization due to CAD? Are there differences when comparing males and females? If so, what are the differences?


• Based on odds ratios, which risk factors are most prevalent among a group of employees with CAD? What is the rank order?

Among this large, multi-employer group of workers who completed a health risk appraisal (HRA), the difference between the occurrence of CAD between males (6.3%) and females (5.7%) was only 0.6%. Among males, smoking was the number one predictor of heart disease, while with women, profound obesity and uncontrolled stress were the prime predictors. There was no level of consistency between men and women relative to the association between health risks and costs. For example, men reporting to be depressed most of the time had total health care costs 91% more than men who reported not being depressed. Among women, those reporting to be depressed most of the time had health care costs only 5% more than those reporting not being depressed.

Behavioral change intervention application has usually been the same for men and women. If the intent is to provide interventions based on the potential for maximum reduction in medical costs, occurrence of CAD and hospitalization due to CAD, this study suggests different intervention goals between males and females may be appropriate.

This study was funded through grant number NAG-6218 from the National Aeronautics and Space Administration (NASA).

Research Project Four (Complete and Published)

THE RELATIONSHIP BETWEEN MODIFIABLE HEALTH RISKS AND GROUP-LEVEL HEALTH CARE EXPENDITURES, (Anderson, D, Whitmer, R, Goetzel, R, et. al., Am J Health Promot, (15)(1), Sept/Oct 2000; 45-52). Uses the HERO database in which 46,026 employees met all inclusion criteria for the analysis. The purpose was to assess the relationship between modifiable health risks and total health care expenditures for a large group of employees. This study is different than Research Project Two which examined individual health care expenditures as the outcome. Here the outcome is the total cost impact of a given health risk.

Risk data were collected through voluntary participation in health risk assessments (HRA) and workplace biometric screening. These data were linked to health care plan enrollment and employee health care expenditures from employer’s fee-for-service health care plans over a six year period.

Several research questions were addressed:

• What is the association between each of the eleven modifiable health risks and health care expenditures?

• What percent of total health care expenditures are associated with each of the eleven modifiable risks?

It was found that employees with modifiable health risks were responsible for 25% total expenditures. Those employees who reported being under constant stress with no methods for coping were responsible for 7.9% of total health care costs. Being a former smoker was associated with 5.6% total medical expenditures followed by obesity at 4.1%. The association between risks and expenditures was estimated using a two-part regression model, controlling for demographics and other confounders. Risk prevalence data were used to estimated group-level impact of risks on expenditures.

This study was co-funded through an unrestricted grant from HERO and The StayWell Company.

Research Project Five (Complete and Published)

PROJECTING FUTURE MEDICAL CARE COSTS USING FOUR SCENARIOS OF LIFESTYLE RISK RATES, Leutzinger, J, Ozminkowski, R, Dunn, R, et. al. Am J Health Promot, (15)(1), Sept/Oct 2000; 35-44. This study uses the HERO database. Union Pacific Railroad (UPRR), like many others, has an aging workforce. Prudent financial planning mandates that future medical expenses be understood.

UPRR has a long history of providing and promoting aggressive health promotion programs. For this reason, they have a rich database of employee health risks and demographics. The objective was to adjust the HERO database to accurately reflect the demographic characteristics of the UPRR employee population. Multivariate statistical techniques were used to create models predicting health risk prevalence and expenditures based on information contained in the HERO database plus demographic characteristics, risk values and cost data provided by the UPRR database. Risk factors examined are: 1) alcohol consumption, 2) blood glucose, 3) blood pressure, 4) cholesterol, 5) nutrition, 6) fitness, 7) mental health, 8) tobacco use, 9) stress and 10) weight. Demographics included are: 1) age, 2) gender, 3) ethnicity and 4) job classification. These models will be used to estimate future health risks and expenditures.

In summary, the study indicates:

This study paper describes the development of an economic forecasting model to predict medical care expenditures assuming four different scenarios of population risk. The variables used to predict medical care expenditures are employee demographics and health risk profiles. Intermediate outcomes include health risk measures related to exercise patterns, body weight, eating habits, smoking, alcohol consumption, total cholesterol, blood glucose, blood pressure, stress and depression. Major outcome measures included projected total annual payments by UPRR for medical care services, for the decade following 1998. The UPRR work force is projected to grow by 500 employees per year over the ten-year study period. The average age is expected to increase form 44 to 48. The study reports that without further health promotion intervention, seven of the 11 risk factors assessed would likely worsen among UPRR’s work force. Medical care cost increases are projected to range from $22.2 million to $99.6 million in constant 1998 dollars over the next decade, depending on the effectiveness of risk factor modification programs. If UPRR is successful in reducing modifiable health risks 1% per year over a ten year period, the aggregate reduction in health care costs are projected to be 77.4 million With an expected health promotion budget averaging $1.9 million annually over ten years, health risks must decline at least 0.09% per year for the program to pay for itself.

This study was funded by and unrestricted grant from the Union Pacific Railroad.

Research Project Six (Complete and Published)

CARDIOVASCULAR RISK REDUCTIONS ASSOCIATED WITH AGGRESSIVE LIFESTYLE MODIFICATION AND CARDIAC REHABILITATION. (Aldana S., Whitmer R., Greenlaw R., Heart & Lung (32) (6), Nov/Dec 2003; 374-381. Patients who have been treated for coronary heart disease can enroll in traditional cardiac rehabilitation, the Ornish program, or no rehabilitation. No study has compared the impact of each on cardiovascular disease (CVD) risk factors.

This study compares CVD risk changes in post coronary bypass graft or percutaneous coronary intervention procedure patients who participated in the Ornish Heart Disease Reversing Program, a traditional cardiac rehabilitation program, and a control group that received no formal cardiac risk-reduction program. This was a longitudinal, observational study of 84 patients receiving CVD standard of care who elected to participate in one of the three study groups. Assessments of CVD risk factors and anginal severity were obtained at baseline, 3 months and 6 months.

Those patients participating in the Ornish program had significantly greater reductions in original frequency, body weight, body mass index, systolic blood pressure, total cholesterol, low-density lipoprotein cholesterol, glucose, dietary fat and increases in complex carbohydrates than were experienced in the traditional or control groups.

Editorial Project (Complete and Published)

A WAKE UP CALL FOR CORPORATE AMERICA. (Whitmer R., Pelletier K., Anderson D., Journal of Occupational and Environmental Medicine (45) (9), Sept. 2003; 916-925. In 2003, the cost per employee, for family health care coverage, when averaged across all plan designs was $9,068. Of this amount, the employer paid $6,656. If increases average a conservative 10% per year over the next several years, then in 2008, the cost per employee will be $14,601, with the employer paying $10,659 per employee. This assumes the current percent contribution between employer and employee remains constant. For most employers, this is a crisis situation, with little relief in site.

This invited editorial provides a detailed over-view of employer reactions to routine double digit annual increases in health care costs based on numerous published employer surveys. Commentary explores why there is little hope for moderation in health care costs: the baby boomers, the graying of America, escalating hospital charges and the obesity epidemic. The editorial documents that 50% - 70% of all diseases and medical problems are caused by life style choice: smoking, obesity, excess stress, lack of fitness, poor nutrition, lack of compliance in managing diabetes, hypertension, etc. The dichotomy of the health care cost crisis is that of the $1.8 trillion annual budget, less than 6% is devoted to prevention of all kinds, including attempts to influence lifestyle choices.

The editorial concludes, “A Wake-Up Call for Corporate America is that an employee health care cost crisis is here. If the employer assumes the responsibility to pay for the diagnosis and treatment of employee/dependent illness, then serious consideration must be given to the reallocation of existing investment in human capital funds. This redirection of funding should be toward health enhancement programs and services that optimize employee/dependent health, which can reduce health care use, moderate cost increases, reduce illness absence and improve work performance”.

The following research is complete but not yet submitted for publication or in progress

A PROSPECTIVE AND RANDOMIZED RESEARCH PROEJCT TO COMPARE THE EFFECTIVENESS OF THE ORNISH PROGRAM WITH TRADITIONAL CARDIAC REHABILITATION IN REDUCING CAROTID ARTERY MEDIA AND INTIMA THICKNESS AND A VARIETY OF THE OTHER HEALTH PARAMETERS. A prospective, controlled and randomized evaluation of heart disease patients who agree to be randomized into either the Ornish or traditional cardiac rehabilitation programs. In addition to all the outcomes evaluated in the above research project, these study subjects also have carotid artery ultrafast-sonography to determine media and intima thickness (a direct measure of vascular disease.) This test will provide hard copy pictures of atherosclerotic (plaque) build up in the carotid artery. There is good evidence that the amount of plaque in the carotid artery corresponds to the plaque in coronary (heart) arteries. In addition, there will be blood analysis for homosystene (HCY), C-reactive protein, ferritin, and fibrinogen. The hypothesis states that those who are randomized and remain in the Ornish program will have less thickening or actual reduction in carotid artery thickness (reversal of cardiovascular disease) over time compared to the traditional rehabilitation group. 

The last research project was funded by the Midwest Center for Health and Healing in Rockford, IL.