Today’s blog is a response to benefits managers and corporate medical directors who have (and will in the future) exclaimed “these programs were supposed to save us money, so why are your data saying they don’t? Is there something wrong with your analysis?”
While I am formally trained in Evaluation Methodology, I recognize that only a few kindred spirits share my passion for this field. This blog doesn’t require that you LOVE evaluation, but it is a little more detailed than usual, because it seems important to explain why our direct evaluations so infrequently show (expected) measurable savings from health improvement programs. There are three overarching reasons, each of which I will highlight briefly.
1. Healthcare costs are not a good measure of health status.
2. We use real numbers rather than projections.
3. Too often, companies ignore the powerful influence of business practices.
While in general healthier people cost less, healthcare costs at the individual level are a very imprecise measure of health status. They are an even worse measure of risky health behaviors (which indicate a future likelihood of health problems). In the 1990s, research into an area of “demand management” documented that level of morbidity (illness) was only one determinant of how people use healthcare (1). The other three areas were:
a. Perceived need for care. This includes all the beliefs people have about what might be wrong with them, how serious the illness might be, what they perceive about the value of medical expertise, and how uncomfortable they are with uncertainty. Two people with the same exact symptoms may react in vastly different ways, affecting the amount of care they use (1, 2, 3).
b. Patient preferences. This includes differences in a person’s primary goals and preferences regarding what he or she wants to achieve by getting treatment. Again, a person seeking a cure at any cost will consume healthcare differently than a person wanting to manage pain and maximize functionality, for example(1, 2, 3)
c. Non-health motives. As we often discuss in this blog (and describe in more detail below), economic, social and occupational factors influence how and why people use health care (1, 2, 3).
So, if you are not seeing a correlation between health factors and health costs, there are many potential reasons why:
Reason Two. We use actual data rather than the usual method of projecting savings from other sources.
When we conduct evaluations in our firm, we analyze actual data. Conversely, many vendors project estimated savings using equations that rely on other metrics. Most projections used by health vendors are attributed to a few studies that show either how the number of current risks correlate to the current costs, or, in some cases, how changes in the number of risks are correlated with changes in cost. However, there are flaws in this logic:
a. Many equations presume that all risks are of equal value. When a company’s employees eats a few more servings of fruits and vegetables, or wears their seatbelts more often, that will not have the same effect as other factors such as significant weight loss. We have known for years that some risks have more value than others (4).
b. Some health “risks” in the most popular equations are actually indicators of current health status. The most concrete examples are “number of sick days,” and “number of chronic conditions.” In reality, these factors are not “risks,” they only tell us that these folks are sicker today; if you are sicker today you will cost more today. Of course those are predictive of costs.
c. Risks that require medical treatments today (i.e., high blood pressure and high cholesterol) are associated with HIGHER costs. This means that health care costs will go up when these risks are treated.
d. When we look at health-program participation, more often than not, those who sign up are less costly than those who don’t (5). Because average costs are predominantly driven by the most expensive few (6), it is difficult to achieve an overall effect with small changes in people who are already low cost.
So again, if a company has documented some changes in risks, it is important to look carefully at the types of risks that are being measured before presuming those changes in risk will translate into cost savings.
Reason Three: Too often, companies ignore the powerful influence of business practices.
Any misalignment in incentives to use benefits will overshadow small health status changes. Examples we have seen:
a. People in a health plan with a very small deductible will use a lot more care than people in less-rich plans (7). This excess use may create too much “noise” to find any small changes in cost.
b. Employees experiencing financial pressure and job uncertainty will experience a spike in medical claims, disability leave and workers’ compensation events as employees use services in advance of a concern about layoffs (8).
c. Employees with low job satisfaction/engagement will exhibit higher levels of absenteeism and use of medical services unrelated to traditional health risks and health status (9).
d. As we showed recently, realignment of business practices can have as much as three or four times as much of an impact on healthcare and absence costs than interventions focused on health status (10). As we mentioned above, the relationship between health risks and costs can be found when “other things are equal;” in most cases they are not.
Add to all these: a steady inflation in the cost of healthcare, changes in health plan design, the lack of a true comparison group, and many other factors…. and we find ourselves looking for the proverbial needle-in-a-haystack.
Just to be clear about our position, it will be very unlikely that real data will match projected savings unless all other factors are accounted for. And employers must be very cautious about over-projecting what effect health status or risks can have on medical claims costs.
The reality is that our industry has overstated the ability to create and detect medical cost savings from health-improvement efforts.
Some may misinterpret this blog to mean that I believe the underlying science regarding health risks is wrong. Not so. I have no doubt that if a large portion of people in our society achieved and maintained a healthy weight, did not use tobacco, and exercised regularly, among other things, there would be a dramatic reduction in many chronic illnesses and the costs associated with treating them.
However, that connection does not translate into a foregone conclusion that employers making an investment in health programs, while ignoring all other business practices, will see it translate into lower healthcare and absence costs; nor that all changes in any defined health “risk” will result in lower healthcare costs; nor that all health improvement programs improve health. There are simply too many other factors involved.
So, what should employers do with projections and data that don’t match? In my humble opinion, there needs to be an entirely different conversation:
1. If your goal is to reduce healthcare costs, the top priority should be to realign all business practices to reward high performance and good health, and encourage people to become better healthcare consumers.
2. If your goal is to encourage healthier behaviors, FIRST realign incentives (to make health matter more), SECOND examine your company’s practices to take an honest look at whether the work environment supports people’s efforts to be healthy. If the company cannot honestly support a reasonable work schedule (are healthy sleep habits, exercise, and family time even possible?), flexibility and access to preventive services, and a competitive wage, then health improvement efforts are disingenuous. THEN FINALLY—if your incentives and your business practices and environment are supportive—make investments and track improvement in key behaviors themselves, not estimated projections in savings those behavior changes MIGHT produce. The big three risks that matter: have people achieved a healthy weight, started exercising, and reduced their use of tobacco? Hold vendors accountable (or better yet, reward individuals!) for improving those three things. If people achieve these, you are truly making a difference.
Why this matters: Employers invest a great deal in health programs for their employees and expect to see cost savings as a result. Yet, there are so many reasons changes in health risks WON’T affect health/absence costs, it is very unlikely that those savings will be detected in actual data. Instead of relying on “estimated” savings, it would be wise for employers to reassess their goals and choose interventions and metrics that have a higher likelihood of solving their cost problems and measuring actual health improvement.
1. Self-Care Institute. Demand Management Handbook. Washington, DC: Partnership for Prevention; 1996.
2. Lynch, W. D., D. W. Edington, and A. Johnson. 1996. Predicting the demand for healthcare. Healthc Forum J 39(1): 20-24.
3. Lynch, W. D., and D. M. Vickery. 1993. The potential impact of health promotion on health care utilization: An introduction to demand management. Am J Health Promot 8(2): 87-92.
4. Milliman & Robertson, Inc., and Chrysler Corporation. 1995. Health risks and behavior; the impact on medical costs.
5. Lynch, Wendy D., Lynn A. Gilfillan, Carol Jennett, and Joe McGloin. 1993. Health risks and health insurance claims costs: Results for health hazard appraisal responders and nonresponders. J Occup Med 35(1): 28-33.
7. J. P. Newhouse’s Free for all? Lessons from the RAND Health Insurance Experiment. Cambridge, Mass.: Harvard University Press, 1993.
8. Getting value from health benefits: Use them or lose them. Health as Human Capital Foundation Blog 2007
9. Unpublished data, Health as Human Capital Foundation
10. Part IV: Business Practices—A major, modifiable driver of healthcare costs. http://hhcf.blogspot.com/2009/10/part-iv-business-practicesa-major.html