Not to be confused with those immortal words often attributed to the great philosopher Yogi Berra, a big joke among economists is: “An economist is someone who upon learning how something works in practice, wonders how it will work in theory.”
That joke morphed into reality last month — though it was a controlled study testing a theory, rather than the theory itself. The “theory” would be that inactive unmotivated non-participants can be used as a control for active motivated participants. Ironically, this study design has never been proposed as legitimate, even in theory. Wellness “researchers” like it because it always show savings, even when nothing happens. For example, even when there was no program for the participants to participate in.
Obviously if this were a legitimate design, the FDA would approve it for clinical trials, saving a ton of time and money vs. having to do controlled trials.
To wit, the National Bureau of Economic Research (NBER) just published a study showing that the participants-vs-non-participants (“par-vs.-nonpar”) study design, used extensively by the very stable geniuses in the wellness industry to do their alt-research to fabricate their alt-findings, is completely invalid.
Highlights of NBER study for the wellness industry: bad news first
First the bad news. Fasten your seat belts and be shocked, shocked to learn that the researchers could identify:
- No noticeable change in health behaviors due to the wellness program
- No noticeable change in health outcomes due to the wellness program
- Clear self-selection bias among participants opting into the wellness program
Lest anyone think we are taking this out of context, here are their exact words:
We do not find any significant effects of treatment on total medical expenditures, employee productivity, health behaviors, or self-reported health measures in the first year following random assignment. We further investigate the effect of our intervention on medical expenditures in greater detail, but fail to find significant effects on different quantiles of the spending distribution or on any major subcategory of medical expenditures (pharmaceutical, office, or hospital). We also do not find any effect of treatment on the number of visits to campus gym facilities or on the probability of participating in a popular annual community running event, two health behaviors that are relatively simple for a motivated employee to change over the course of one year.
This of course merely confirms what we observe in outcomes reports published by wellness vendors, including the two most recent proud recipients of the Deplorables Awards. Actually, in the case of Wellsteps there was indeed a noticeable change in health outcomes among program participants — they got worse.
Also, the authors — no doubt anticipating the objection from the very stable geniuses at the Health Enhancement Research Organization — specifically note that nothing in Year One’s results presage any step-function improvement in Year Two. So the specious “Wait ’til next year” argument is off the table.
You might be thinking, “Another nail in the wellness industry coffin.” True, except that there almost isn’t room for any more nails. Soon the coffin will have enough nails to create its own gravitational field.
Next, the good news
Here are the two pieces of good news for the wellness industry. One finding was:
Our 95% confidence intervals rule out 78 percent of previous estimates on medical spending and absenteeism.
That means that it is possible that 22% of previous estimates may conceivably not be completely invalid. This is not to say that 22% are valid, just that they aren’t automatically invalid. That is great news for the wellness industry, where clearing the bar for not being automatically invalid is cause for celebration. As Dave Chase says, the bar for wellness is so low a snake could jump over it.
However, the 78%-totally-invalid figure specifically invalidates Katherine Baicker, author of the so-called “Harvard Study.” Depending on whether you are a wellness vendor or an oppressed employee, she is either the Johnny Appleseed or the Typhoid Mary of wellness. Her famous 3.27-to-1 ROI was tallied entirely from par-vs.-nonpar studies, exactly the methodology that the NBER just invalidated, citing exactly the studies she cited. (The three examples in my study referenced above were also part of her meta-analysis.)
So perhaps she might now make a formal statement regarding par-vs-non-par as a study design? Either a retraction or a defense. Just something that clarifies her previous statements, which seem to be neither retractions nor defenses but rather more like excuses:
- It’s too early to tell (um, after 30 years of workplace wellness?)
- She has no interest in wellness any more
- People aren’t reading her paper right (we’re only reading the headline, the data, the findings and the conclusion, apparently)
- “There are few studies with reliable data on the costs and the benefits” (um, then how were you able to reach a conclusion with two significant digits?)
The irony is that Kate Baicker has otherwise done outstanding research. Her study on Oregon Medicaid is a classic. In Oregon at the time, Medicaid eligibility was determined by lottery amongst applicants. That meant that — quite the opposite of wellness control groups — the control group of people not picked in the lottery had equal motivation to seek insurance coverage as people who were picked. After following both groups going forward, her finding was that obtaining insurance to access basic medical care did not change outcomes. (Having insurance did bring peace of mind, though.)
And yet somehow in “Workplace Wellness Can Generate Savings,” she was quite comfortable reaching a conclusion that was completely inconsistent with her Oregon finding, not to mention the Law of Diminishing Returns: throwing additional unrequested, generally unwanted, and largely misdirected medical interventions and advice at employees who already have insurance — and recall that most insured Americans are drowning in medical care — could dramatically improve their outcomes enough to calculate not just a massive ROI, but an ROI precise to two significant digits.
What she will hopefully learn through the NBER study is something that I learned 11 years ago: when the data proves you wrong, fess up. Then people like me have to find someone else to blog about. Fortunately, in wellness, that is not a heavy lift.
The second piece of good news for the wellness industry
To quote the study:
…wellness incentives may shift costs onto unhealthy or lower-income employees if these groups are less likely to participate in wellness programs. Furthermore, wellness programs may act as a screening device by encouraging employees who benefit most from these programs to join or remain at the firm…
To be clear, this calculation does not imply that adoption of workplace wellness programs is socially beneficial. But, it does provide a profit-maximizing rationale for firms to adopt wellness programs, even in the absence of any direct effects on health, productivity, or medical spending. [emphasis theirs]
In other words, employers can use wellness programs to subtly discriminate against unhealthy — read, older and poorer — workers. Many of the very stable geniuses in the wellness industry will be happy to hear this. The exceptionally stable genius who will be most thrilled to hear this is Michael O’Donnell. Mr. O’Donnell is the former Prevaricator-in-Chief of the wellness industry trade publication and a current member of the Koop Award cabal. These are excerpts from one of his editorials:
First, he says prospective new hires should be subjected to an intrusive physical exam [editor’s note: notwithstanding the fact that this is totally illegal], and hired only if they are in good shape. OK, not every single prospective new hire needs to be in good shape — only those applying for “blue collar jobs or jobs that require excessive walking, standing, or even sitting.” Hence he would waive the physical exam requirement for mattress-tester, prostitute, or Koop Award Committee member, because those jobs require only excessive lying.
Second, he would fine people for not meeting “outcomes standards.” In an accompanying document, he defines those “outcomes standards.” He specifies fining people who have high BMIs, blood pressure, glucose, or cholesterol.
In other words, Mr. O’Donnell wants to charge for insurance by the pound, as that accompanying document says. Actually, by BMI, which of course is of dubious value as a measure of weight, let alone health.
Here is his actuarial formula:
Although having read his very stable arithmetic elsewhere in this same document, I’d worry about the accreditation status of any actuarial school, or for that matter any school of any kind within the 50 states, that would accept him:
Thirty-one states have no laws that prohibit employers from using smoking status as the reason for not hiring… In the remaining 29 states…smoking status cannot be used as the reason for not hiring.
I’m not waiting around for a retraction from this genius either.