South Kingstown, RI – June 27, 2013. A growing number of sources are suggesting that wellness programs do not work. Can we answer in the affirmative that wellness programs do work? The Rand Report
(http://www.rand.org/pubs/research_reports/RR254.html) clearly indicates that such a sweeping claim cannot be supported. But why would one assume that such a heterogeneous field – filled all too often with fads, fashions, and far too few facts – would be found to have much impact?
The more critical question is: Can wellness programs work? Here we can respond affirmatively and with confidence. A national Task Force for Community Preventive Services (CHES) has been developing bodies of evidence for recommendations, predictions, evaluations, and innovations for population-based health promotion or wellness programs (Zaa et al., 2005; Soler et al., 2010). The Task Force relies on an impressive range of criteria on which confidence can be based. These include the number of studies, the consistency of results in the same direction, the magnitude of the effects and the consistency of the magnitude, the type of study design, the quality of the study implementation, and length of follow-up. To these criteria, we have added the number of replications across different types of populations, diverse sub-populations and multiple health risk behaviors, and the percentage of participants in programs. We also added strength of outcome measures. Simply measuring improvement is not as strong as the percentage of participants who progress from being at-risk at baseline to not-at-risk at long-term follow-up (e.g., from smoking to not smoking).
A Task Force team applied their original criteria to employee health promotion programs that included HRA assessments with feedback reported at least twice plus at least one additional health promotion program. Across 51 studies, they found average participation rates of about 50% and average follow-ups of about 2 ¾ years. They had confidence in concluding that these wellness programs can produce important effects with exercise (a median success rate of 15.3% from 15 arms from 12 studies) and smoking cessation (a median abstinence rate of 17.8% from 30 arms from 24 studies). Smaller effects were found with high-fat diets (median decrease of 5.4%) and no effects for fruit and vegetable consumption).
Building on the Task Force’s body of evidence, we found much higher impacts for all four of these behaviors with results from 26 arms of population-based RCTs. These results are included in a paper that will be published soon
(http://online.liebertpub.com/doi/abs/10.1089/pop.2012.0094). Unfortunately, the Rand Study did not build on the CHES body of evidence nor did it include the range or rigor of criteria that were used in their review of wellness programs. It is well known that the easiest way to prove the null hypothesis (e.g., wellness programs do not work) is to rely on weak studies. The poorer the studies, the more error and noise is produced, preventing the likelihood of detecting adequate signals of significant effects. The companies reviewed by Rand were as small as 50 employees. As with any program, product or service, there are various levels of quality and one cannot conclude that wellness programs don’t work because the “average” results aren’t impressive. Employers need to use best practices to increase the probability their efforts will reap positive outcomes. One best practice is to use programs that have been scientifically tested and shown to work and not fall to marketing claims that lack peer-reviewed evidence of their offerings.
The Rand review reports that “employers are using incentives” to increase employee engagement in wellness programs. We have been using plan design incentives with our clients for more than a decade and have been able to consistently achieve over 80% of participation rates.
What we can conclude from the Rand Report is that employers need such growing bodies of evidence, based on a broad range of rigorous criteria, in order to benchmark alternative wellness programs. With such benchmarking tools, employers will be able to predict whether they are providing high impact programs, average impact programs, or programs that will produce minor impacts and major disappointments.