Applying Best Practices to Study the Impact of Health Promotion Programs Benefit Costs
The establishment of the health risk appraisal and advent of corporate fitness centers more than 20 years ago was the beginning of worksite health promotion as we know it today. A comprehensive strategy for improving health in the worksite, including a more general whole-person concept across multiple domains in human resource management evolved during the 1990s. Today, we find worksite health promotion migrating as a common strategy from large self insured employer’s into smaller and smaller employer groups.
Today’s worksite health promotion programs are primarily aimed at secondary or tertiary prevention and care management efforts. Health management programs such as disease management, case management are commonly available for individuals with degraded health as a result of a chronic condition or acute issue. The various programs are intended to improve health status, productivity and safety; enhance the appropriate utilization of healthcare services and work loss programs; and control the ever rising expenditures associated with poor health.
A variety of surveys and studies have documented the operation of worksite health promotion programs in the U.S. A fair estimate is that 50-75% of employers with more than 10,000 employee’s offer worksite health promotion programming that includes an HRA and multiple programs across several different delivery media such as seminars, phone based counseling, and Internet applications.
Why Worksite Health Promotion
The health risk profile of U.S. adults poses multiple threats to a high quality of life. Tobacco use in young adults age 18-24 exceeds one in four (25.3%). Almost two out of five (38.0%) adults report virtually zero physical activity in an average week. Approximately one in four adults (23.4%) is obese and almost two-thirds are either overweight or obese. Adherence to recommended self-care and preventive health services can range from 50% – 70%. These lifestyle issues drive degraded functionality, productivity and impact a host of financial factors for U.S. employers .
Health care costs for employers are estimated at a 12% rate of annual inflation. In addition, disability and workers’ compensation costs are increasing by 9% in many states, and Family Medical Leave Act (FMLA) costs are escalating every year. There is often a strong association found when health risks and the previous costs are studied. Health risks tend to accumulate as we age. However, when individuals change lifestyle habits and decrease their health risks, we may observe a change an improvement in health at any age .
The Science of Health Promotion Evaluation
It is estimated that more than 500 studies evaluating health promotion initiatives have been published in peer-reviewed journals . However, there are several issues that challenge a scientific approach to the study of health promotion program impact. Worksite health promotion is an applied field. When a pharmaceutical company studies the effects of nicotine replacement on smoking use, they are able to use a small number of participants and randomize volunteers into a treatment group and a comparison or control group.
When we design an evaluation study of worksite health promotion programs, we typically are not able to perform this type of randomization process. The ability to randomly select participating and non-participating employees is often viewed as denying the opportunity to participate for many who may benefit. In some instances, it may be possible to use lotteries or randomization to tailor the timing when people can gain access to health promotion programs (thus enabling similar comparison groups to be used for evaluation purposes), but this is not always possible either. A solution to this dilemma may be to use rigorous statistical processes in the evaluation to assure that the demographic, health status, and other differences between participating and non-participating subjects are accounted for before making inferences about the impact of the programs they use.
A competent evaluation of a worksite health promotion program can address this issue, and many others. A checklist for consideration when conducting program evaluations is outlined below with additional issues to consider.
Checklist for Sound Worksite Health Promotion Evaluations
- Decide both financial and non-financial questions to address and which hypotheses to test in the impact study.
- Choose an evaluation design that is well suited to testing that addresses these questions and hypotheses. (As noted above, statistical analyses are likely to be required to estimate the benefits of program participation and find unbiased answers about the impact of the programs of interest.)
- Before conducting any analyses of dollar metrics, adjust for inflation.
- Before generating the final ROI and financial metrics of interest, estimate and discount costs for each programmatic alternative under consideration. (for multi-year evaluations)
- Similarly, estimate and discount monetary and nonmonetary benefits. (The adjustments in items 4 and 5 are necessary because, even after adjusting for inflation, dollars received or spent today are more valuable than the same dollars spent in the future. These adjustments account for the reasons why dollar values differ over time in order to generate useful financial metrics.)
- Perform sensitivity analyses to deal with uncertainties and test assumptions that had to be made in order for the evaluation research to be conducted.
- Present results to aid effective decision-making by senior leaders.
- Recognize and describe the consequences of any limitations in the analysis. 
As the above checklist is executed, other challenges may need to be addressed in the evaluation, including:
- The distribution of financial and utilization data is skewed.
- In the absence of randomization, self selection of program participants may lead to differences in demographics, health status, motivation or other factors that create a potential bias.
- The size of participant or non-participant groups under evaluation is small compared to the extreme variability of medical costs.
- Medical inflation is different between various categories such pharmacy, inpatient services, outpatient surgeries, office visits.
- Health benefit coverage changes occur annually and may shift costs dramatically.
- A small but not insignificant percentage of employees may not enroll in employer sponsored health plan benefits, with subsequent medical expense data missing.
- A small to moderate segment of the employees with health plan coverage will not use any medical services or very few (< $100) in a given year.
- New hires, retirees, and terminated employment create turbulence within the study population.
- A very small segment of the population (< 10%) typically accounts for 40%-60% of all health related expenses.
- Costs associated with pregnancy and birth, or hospital admissions may not be related to lifestyle issues impacted by programs.
Best Practices in Evaluation Methodology
The issues described above can be addressed via competent statistical analyses. For example, two-part regression models can be used to account for the likely 20% of subjects who utilize no health care services in any given year, versus the remaining 80% or so who do. Two-part models therefore help address the skewed data and non-normal distributions of expenditures in the population of interest. These analyses can be conducted twice, once with all sample members and once without those who have extremely high expenditures, to assess the impact that a few “outlier” members will have on the program impact estimates.
Other statistical techniques can be used to account for demographic, health status and other differences between program participants and non-participants before the impact of program participation is estimated. For example, propensity score-based matching or weighting techniques can be used to overcome the challenges posed when participants and comparison groups are not randomly assigned. These methods, and other multiple-regression-based strategies are well known to competent program evaluators and expertise with these techniques can often be found at leading universities and evaluation research firms.
Issues related to small sample sizes can be addressed using permutation testing approaches that create thousands of smaller samples for analysis and thereby increase the statistical power of the analytics. This may enhance the reliability of the findings when sample sizes are small.
Generally speaking, health promotion program impact estimates amount to well-designed cost-benefit or cost-effectiveness analyses. The metrics devised from these analyses include the net present value estimate and the ROI ratio. The net present value (NPV) calculation is the difference between the total inflation-adjusted and discounted set of benefits and costs associated with the program of interest over its useful life . The ROI ratio is the ratio of benefits to costs, rather than the mathematical difference between the two. The statistical analyses mentioned above are typically focused on estimating the benefit values used in these metrics, to make sure apples to apples comparisons are made between program participants and non-participants before benefits are estimated.
The consideration of multiple perspectives when evaluating a health promotion program yields a more accurate appraisal of the total program and will result in a better set of NPV and ROI estimates. Conducting the ROI and NPV analyses several times to address the differences of opinion among key stakeholders will lead to more informed discussions about the best ways to use health promotion programs to increase the health, safety, and productivity of the workforce.
Money Isn’t Everything, Though …
A hallmark of successful programs is the contribution to multiple business objectives. The varying goals of the health promotion program should be established a priori. A common observation in program evaluation is the assumed main goal of medical benefit cost reductions. However, interviews with corporate leaders substantiate that medical benefit cost reductions are not the primary goal in as many as 50% of all programs . If program goals such as lower employee turnover, improved morale, or other desired outcomes are important, the program evaluation must be tailored to gauge progress in reaching these goals. This is not to suggest that the financial evaluations should be minimized, but that a total picture of business objectives must be identified in order to establish the evaluation plan [2, 3, 11].
Worksite health promotion has enjoyed a prosperous journey over the past 30-40 years. What started out in the 1970s and 1980s as predominantly on-site fitness centers and the use of executive physicals has grown into sophisticated programs that engage a very large percentage of employees in some cases. The health risk appraisal is commonly used as a gateway program that helps stimulate participation in higher intensity interventions.
The evaluation of worksite health promotion programs has evolved from relatively straightforward group comparisons to include the use of more sophisticated statistical techniques. Regression analysis, propensity score approaches, randomization tests and other features of good cost-benefit or cost-effectiveness analyses are being used more often than in the past. These evaluation techniques are designed to yield more accurate estimates of the financial and non-financial impacts of program participation.
Evaluation areas must be tethered to the business objectives established by corporate leaders. Continuous assessment of health risk expenditures and other outcomes of interest can help predict these outcomes in future years. While heavy-duty statistical analyses may be required to obtain the most accurate estimates of program impact, engaging in this process will yield the information that program managers need to modify program designs and menu choices to continually enhance the welfare of the workforce and dependents.
- USDHHS, U.S.D.o.H.a.H.S., Healthy People 2010, P.H.S. U. S. Department of Health and Humana Services, Office of the Surgeon General, Editor. 2000, U.S. Government Printing Office: Washington DC. p. 60.
- Bartholomew, L.K., et al., Planning Health Promotion Programs. 2006, San Francisco, California: Jossey-Bass. 724.
- O’Donnell, M.P., Health Promotion in the Workplace. 1984, Albany NY: Delmar Publishers.
- CDC, BRFSS Data 2001-2004. www.cdc.gov, 2005.
- Hartman, M., et al., National health spending in 2007. Health Affairs, 2009. 28(1): p. 16.
- Chapman, L., Meta-Evaluation of worksite health promotion economic return studies. American Journal of Health Promotion, 2003. 6(6): p. 15.
- Pelletier, K.R., A review and analysis of the clinical- and cost-effectiveness studies of comprehensive health promotion and disease management programs at the worksite: 1998-2000 update. American Journal of Health Promotion, 2001. 16(2): p. 10.
- Goetzel, R.Z. and R.J. Ozminkowski, The health and cost benefits of work site health-promotion programs. Annual Review of Public Health, 2008. 29(17): p. 18.
- Ozminkowski, R.J. and R.Z. Goetzel, Getting closer to the truth: overcoming research challenges when estimating the financial impact of worksite health promotion programs. American Journal of Health Promotion, 2001. 15(5): p. 6.
- Musich, S.A., L. Adams, and D.W. Edington, Effectiveness of health promotion programs in moderating medical costs in the USA. Health Promotion International, 2000. 15(1): p. 11.
- Chenoweth, D.H., Evaluating Worksite Health Promotion. 2002, Champaign IL: Human Kinetics.
About the Authors
Philip A. Smeltzer, M.S.
Philip A. Smeltzer serves as a Senior Consultant in the Ingenix, Employer Advisory Services Group. In his current position Philip capitalizes on more than 20 years of experience in the health promotion and health management field. He has focused on worksite health promotion activities and evaluation throughout his career.
Ron Ozminkowski, Ph.D.
Ron Ozminkowski, Ph.D., is Vice President of Research and Development in the Healthcare Innovation and Information Group at Ingenix and also serves as Vice President, Research and Policy, at UnitedHealth Group® Alliances. In his current role, he developed and now leads a major research and reporting unit that helps monitor and evaluate health promotion and care management programs.
Shirley Musich, Ph.D.
As Senior Research Director, Shirley Musich is responsible for providing decision support to lead employers and other Ingenix clients through health evaluation, program design and measurement and evaluation processes. Her recent experience includes serving as a Senior Data Analyst at the University of Michigan involved with Integrated database systems and the decision support teams for corporate clients in the administration of their health promotion programs.