The Impact of Workplace Wellness on Health, Health Care, and Employment Outcomes: A Randomized Controlled Trial

Workplace wellness programs are increasingly popular in the United States. Over 80% of large firms offer such a program, fueling an $8 billion industry. Despite this prevalence and substantial public investment in workplace wellness through the Affordable Care Act, robust evidence on the causal effect of wellness programs on health and economic outcomes is scant.

We conducted a randomized controlled trial of a workplace wellness program at a large U.S. employer across 160 worksites along the Eastern U.S. in partnership with a wellness vendor. Randomization was at the worksite level, with 20 worksites randomly selected to receive the intervention. This design allowed us to capture the effects of an intervention aiming to affect workplace culture and utilizing group as well as individual programming.

The wellness intervention comprised 8 modules fielded over 18 months, focusing on wellness components such as physical activity, healthy eating, and stress reduction, with financial incentives for participation. Registered Dietitians in the intervention sites helped administer the program. Biometric data and survey responses were collected in person at the 20 intervention sites and at 20 randomly selected control sites, and administrative data on employment outcomes and health insurance claims were collected across all 160 worksites. The study sample included 4,038 individuals in the 20 intervention sites, 4,106 individuals in the 20 primary control sites, and 28,938 individuals in the 120 additional control sites.

We assessed the effect of being randomized into the treatment group (intent-to-treat estimates) and the effect of participating in the intervention (local average treatment effects). We estimated the latter using the worksite randomization as an instrument for participation in a two-stage least squares regression. We tested robustness of our main estimates with adjustments for balance and alterations in the functional form, as well as models at the worksite level. Heterogeneity analyses were pre-specified by sex and age.

On average, employees in the intervention worksites participated in 2.4 modules of the wellness program, with 56.6% of employees participating in at least 1 module. Results on program impact will be presented for outcomes across a wide range of domains. Employment outcomes include absenteeism, tenure, and job performance. Health care outcomes include spending and utilization across categories such as primary care, medications, and emergency department visits. Health behaviors include exercise, nutrition, and stress management. Health outcomes include blood pressure, weight, and blood sugar. Collectively, these estimates allow for a much more nuanced assessment of the program’s costs, benefits, and return on investment.

This randomized controlled trial produces rare, causal estimates of the effects of an increasingly prevalent but poorly assessed workplace intervention that has the potential to inform both public policy and private investments.