Three years after wellness was hailed as perhaps the only truly bipartisan component of the Affordable Care Act, both lay and trade commentators have begun observing that the assumptions behind it were incorrect while downsides were overlooked. As a predictable result, savings have proven elusive even in seemingly ideal baseline circumstances for health improvement. For example, a wellness program at BJC HealthCare in St. Louis reduced hospitalizations for wellness-sensitive medical events, but the savings were limited (and offset by other cost increases) by the fact that older employees there on average were hospitalized for a wellness-sensitive medical event only once every 12 years to begin with. (See Note 1.)
Consistent with that finding, commentators (including the authors) have noted that every vendor claiming savings from what the Affordable Care Act (ACA) terms “health contingent” wellness programs has employed obviously flawed study design (like comparing the results from active motivated participants to non-motivated non-participants, and crediting the program, rather than the obvious difference in motivation, for the savings) and/or has simply made up or misinterpreted their own outcomes .
One reason for the absence of savings is that the biometric screenings themselves on which wellness economics are based cost far more money than they can conceivably save, due to both the likelihood of overdiagnosis and the marginal benefit of taking frequent measurements in generally healthy adults. Routine screening lacks an evidence basis and is eschewed by the medical community. For example, the federal government recommends lipid screening only once every five years.
Simple arithmetic validates the government and medical societies’ positions. Consider the typical screen, intended to find undiagnosed diabetics, pre-diabetics and people at risk for heart events who don’t know it. Since no one thinks screening, testing, and medicating for slowly progressive diseases like diabetes (and especially pre-diabetes) pay for themselves in other avoided complications over all but the longest term, let’s focus on heart attack avoidance both in diabetics and non-diabetics. Heart events are sudden, serious and theoretically largely preventable, making them ideal screening candidates.
Our arithmetical assumptions are straightforward and conservative. Call the cost of a screen $40, assume a $200 incentive versus the $521 national average, and throw in $20 of lost work time to participate. (Some employers allow, encourage or insist that employees visit a doctor to do these things, dramatically increasing both the lost work time and direct expenses, though possibly getting a more accurate result.)
Next, the heart attack rate in the employed working-age population is about 1 in 500. Of people suffering those attacks, assume three-quarters already knew they were at high risk. This is a conservative assumption because 90 percent of all major CAD events (through age 75) happen to people known to be at high risk. Maybe they know this because they smoke, have angina or a coronary artery disease (CAD) diagnosis already, and/or have been to the doctor on their own.
For the other 25 percent, let’s generously assume the screens do their job and accurately identify people who actually have elevated risk from which that 25 percent would be drawn. This assumption is very generous—a comprehensive new study suggests screens may miss 23 percent to 59 percent of people at high risk. Further, the average LDL among people hospitalized for heart issues was 105, which until recently was not viewed as high-risk. But let’s assume all that away and rather say that risk factors can be identified in screens.
Identifying risk factors isn’t enough. Risk needs to be reduced, so let’s also generously assume that fully half the people at risk change behavior enough to prevent their heart attacks. This implies smoking cessation rates and/or other lifestyle changes and/or medication compliance rates of 50 percent or greater.
Even with all these generous effectiveness assumptions, you would actually avoid only about 1/8 of all the heart attacks you otherwise would have had. Put another way, you’d have to screen 4000 people to prevent a single heart attack. Multiplying that figure by the conservative financial assumptions noted above means possibly preventing a heart attack –the total expense for which rarely reaches six figures even including lost work time — costs more than $1 million.
Consistently Wrongheaded Recommendations
Yet apparently insensitive or oblivious to both the recommendations of the medical community, the arithmetic, and the poor predictive power of cholesterol testing, benefits consultants and wellness vendors increasingly recommend the opposite: screening as many people as possible at least annually, and tying as much premium as allowable to participation in those screens. Using one well-publicized recent company whose incentive is much closer to the national average figure, CVS has tied $600 of insurance premiums to participation in screening. Increase the incentive from $200 to CVS levels, and the cost per avoided heart attack starts approaching $3 million.
Even after you’ve spent all this money, you still don’t know which one of those 4,000 people whose heart attack you’re avoiding, and sometimes more intervention on a small high-risk subset of those folks is necessary. But this $3-million estimate is before the cost of that follow-up, which could include primary care physician visits, maintenance medication, cardiologist referrals, cardiologist testing and, in a few cases, possibly unnecessary angioplasty. (Invasive testing/procedure rates outnumber infarction rates by almost 3-to-1. One would have to imagine that a screening program designed to find and follow up on risk factors would increase that ratio further.) While those latter numbers would vary widely enough to make modeling difficult, these follow-ups –even if only on a small subset — could easily add six or even seven figures more to the cost per avoided heart attack.
The assumptions underlying these figures are just that — assumptions. However, it would be virtually impossible to massage any realistic assumptions into a conclusion that company-wide screens are cost-effective. (Among other things, the economic impossibility of actually saving money helps explain why vendors feel compelled to make up outcomes instead.)
Yet companies proceed with screening programs even in the face of these real-dollar hurdles while ignoring more obvious worthwhile steps. Consider, for example, a CEO of a 20,000-employee organization proudly boasting of 80 percent employee participation in their annual biometrics bacchanal. At $40 per employee, we observed that this “health fair” costs the company $640,000 plus incentives annually, a calculation the CEO admitted never seeing. We also observed that after completing their biometric screens, employees could saunter into company-subsidized cafeterias, which encourage staff to stay onsite for lunch and hence get back to work quicker, to get their low-cost sugar sweetened soft drinks. Surely whatever good the screen did (if any) was negated by this mixed message.
Health assessments should always be made available to those who actively want them. For example, blood pressure testing is easy, almost free, and multiple readings over a prolonged period could help guide treatment decisions. So why not make an automated blood pressure cuff available to any employee who wants to use it? And in general, wouldn’t the large sums spent on these semi-involuntary companywide screens be more productively deployed by facilitating or subsidizing healthy behaviors?
What the ACA terms “participatory wellness strategies” — subsidized healthy choices in the cafeteria, gym memberships or on-site fitness, flexible scheduling to facilitate physical activity, automated health kiosks, nutrition education, and free smoking cessation programs — would cost much less than what is being spent now on biometrics and incentives. They might also positively impact risk factors enough to eventually shave a small sum off total health spending (wellness-sensitive medical events in total don’t account for enough spending in the working age population to shave a large sum), while creating a positive wellness culture that employees support rather than resent.
Note 1. This estimate is based on the assumption that 90 percent of the wellness-sensitive inpatient medical events in the study occurred in people more than 45 years old, and that age category accounts for 25 percent of all employees and dependents.