Professionals interested in advancing best practices in worksite wellness have been calling for more organizational and environmental approaches that will advance a culture of health. We seem to be getting the message across because a recent national survey shows that “supportive social and physical environment” is one of the five elements used to define a comprehensive program that has increased the most from 29.2% naming this as part of their approach in 2004 to 56% of companies in 2017. Yet two worksite health promotion studies that garnered media attention this past year offer telling examples of what occurs when researchers conflate or confuse the effectiveness of a health education program with a socio-ecological approach.
The full version of this editorial can be found open access HERE at the American Journal of Health Promotion.
Workplace health promotion continues to be a growing movement in America with 46% of companies offering some form of employee wellness. We at the American Journal of Health Promotion were delighted to publish the long awaited study reporting the results from the Workplace Health in America Survey. For this research, a comprehensive program was defined as having the five elements recommended by the Healthy People 2010 report: 1) health education, 2) supportive social and physical environment, 3) integration of the program into the organization’s structure, 4) linkage to related programs such as employee assistance programs, 5) worksite screening and education. When researchers without health promotion experience parachute into the worksite wellness field, it is understandable that they would consider the word “program” to refer to things like classes or coaching. Accordingly, well intended researchers have been apt to study two elements, a health education program and worksite screening, as if just these two elements are what constitutes a “comprehensive program.” It is these cases that have me feeling that the term “comprehensive program” ought to be outed as an oxymoron. That is, if you’re focused on studying health education “programs” one element, you’re not studying a comprehensive approach. Seeing that the health promotion field is finally embracing tenets of the socio-ecological framework that most health promotion professionals were trained in, I’d suggest we replace the term “comprehensive programs” with “socio-ecological approach” or a “culture of health approach” or, at least, “a comprehensive approach;” anything but a “comprehensive program.”
Missed Research Opportunities
Two worksite health promotion studies that garnered media attention this past year offer telling examples of what occurs when researchers conflate or confuse the effectiveness of a health education program with socio-ecological approach. In a recent paper by Song and Baicker, they write that they studied a “comprehensive workplace wellness program” at BJ’s Wholesale Club, a warehouse retail company. However, the manuscript describes only two of the five elements, programs and screenings, indicated by the Healthy People 2010 definition of a comprehensive approach listed above. The program consisted of 8 webinar-based modules lasting 4-7 weeks. Of those eligible to participate, 35.2% took advantage of 1 module and 21.4% completed at least three modules (an average of 1.3 learning modules) for which they could receive a $25 gift card. That the evaluation methods in Song’s study were so impeccably conducted makes their extensive investment in studying such a low dose, program-only intervention all the more disappointing. Indeed, of the 57 pages in supplement 1 offering extraordinarily impressive details behind this program’s evaluation and measures, only one page is provided to describe the intervention.
The BJ’s program intervention period was 18 months, and after the first 6 months of the study, registered dietitians visited sites once a week to support fitness activities or offer cooking demonstrations. Other than this, no detail is offered concerning socio-ecological variables such as leadership, participant planning input, champion networks, environmental supports, communications campaigns, worksite policies or integration with other services or alignments with the organization’s culture and business strategies. Furthermore, although there were 78 metrics reported on relating to individual health behaviors, clinical markers, employment outcomes and medical and drug spending, there were not measures of any of the above named organizational or socio-ecological determinants of health. The New York Times headline for the study read: “Employee Wellness Programs Yield Little Benefits, Study Shows,” and Song’s quote suggests their findings “may temper expectations about the financial return on investment that wellness programs will deliver in the short term.” It’s a curious conclusion given I’ve yet to meet a company leader who expected an ROI after having a self-selecting group of employees attend 1.3 learning modules.
Another study that offered an impressive battery of measures to evaluate a modest program-only intervention was led by health economists and conducted at the University of Illinois. Though their program’s intervention was another example of a basic individual program to a sample offered between health screenings, their outcomes analysis was anything but basic. Like the journalist’s review of the BJ’s program, in the popular press there was no reference to the effects of the actual program on those who attended the program or that the benefits for the few participants were obscured by the health measurements of the large group of non-participants. Also getting no mention was that “the program” results were based on one-year and that no one familiar with a socio-ecological approach would have deemed this intervention to qualify as such. It’s a study that confirms my long standing observation that employers naively diminish modest wellness budgets by over-assessing and under-intervening.
Worrisome Confirmation Bias
What I found more troubling about the University of Illinois study published last year than how another retro, individually targeted program was errantly cast as a contemporary population level approach was what I construed to be the confirmation bias of these economics trained researchers. Though the authors note in their methods section that their study was “not designed to examine recruitment efforts,” the authors conclude nevertheless that a “primary benefit of these programs to employers may be their potential to attract and retain healthy workers with low medical spending.” This conclusion suggests a presumption that employers would design and employ wellness programs as a form of regressive taxation. It’s a disturbing premise on its face but more so given their results refute the very idea. First, though the authors base their differential recruitment notion on a $116 per month medical cost difference between participants and non-participants in the treatment group, the study also showed a $4 average monthly health-care spending difference between the treatment and the control group. Would an employer be motivated to “differentially recruit” employees, disregarding factors like skill, talent, or company need, to save $4 or even $116 a month? Second, as much as non-participants spent more “on average” than participants, the study also found that the very low spending employees were, in fact, more likely to participate in the programs as were the highest health care spenders. Conversely, non-participants actually had a higher probability of health care spending on any given month. Given these findings, the authors could have as readily argued that wellness programs attract the highest health care spenders and, therefore, represent a form of progressive taxation designed to benefit the costliest employee population.
Third, and most importantly, the authors offer no evidence that the University of Illinois, or any employer, could or would intentionally design a program that preferentially advantages healthy, low spending employees and discourages less healthy, higher spending employees. The authors suggest that “wellness programs may act as a screening device by encouraging employees who benefit most from these programs to join or remain at the firm,” however, they offer no logic model or pathway by which this preferential recruitment would occur. This observation did not factor in to a Washington Post editorial which bit fully into the pernicious conjecture of these economists and suggests: “If selection bias really is driving the apparent benefits, then the incentives often offered for participation essentially redistribute wealth and resources upward, from the sick to the healthy and, by extension, from lower-earners to the more affluent.” I wonder if the Post would change their view by reviewing the Song study findings showing that “neither mean health care spending nor the probability of having any spending during the year before the program was significantly different between participants and nonparticipants.”
By definition health education is voluntary, and offering effective programs for those who self-select into them is the surest way to catalyze at least one segment of the population into healthy actions. In the case of the Song study, getting 35% of the employees learning together is not a bad start, albeit of those who showed some readiness to change most elected a small dose of a limited program offering. Still, Song’s study reported “a sizeable and robust improvement in some self-reported health behaviors.” One hopes the champions of the future culture of health can be found among these early adopters. Indeed, well-designed programs would be organized to intentionally enable and reinforce such. Again, where the oxymoron problem comes in is when the results of this small sample of volunteers attending individual-level programs are applied to the health status of the entire population.
Had the studies reviewed here taken a comprehensive approach with all five elements, and had the researchers allowed for a time horizon where such socio-ecological approaches have been shown to take hold, who knows what results they may have achieved? The researchers cited here would certainly have been capable of leading such a study but they, and we, need to drop the oxymoron in order to help future researchers get the questions right.
The full version of this editorial, along with references can be found open access HERE at the American Journal of Health Promotion.
