Social policies might not only improve economic well-being, but also health

"Since the 1960s, a large number of social policies that have been experimentally evaluated include health outcomes, but these were mostly overlooked. By tracking down these studies, we found a unique opportunity to inform evidence-based policymaking. It was sitting right there in front of us this whole time," said Peter Muennig, MD, a professor of Health Policy and Management at the Mailman School of Public Health at Columbia University. "Our goal was to conduct a comprehensive review of experimental studies of social and economic interventions that were not explicitly designed for the purpose of improving the health of participants. Health outcomes were often added as an afterthought and were overlooked." He added, "This is remarkable given the magnitude of expenditures in the nonmedical determinants of health in the United States and the weak evidence base supporting these investments."

The systematic review of all known randomized social experiments in the United States that involved health outcomes included 5,876 papers, reports, and data sources, ultimately encompassing 60 papers, reports, or datasets from 38 randomized social experiments. These experiments spanned the period 1962-2018 and featured a range of policies, analytic approaches, and target groups and measured a wide variety of health outcomes.

The final analysis was based on a sample of 450 unique health estimates across the 38 interventions. Of these, 77% were not able to reliably detect health outcomes because the sample was too small. Among those from which reliable estimates could be obtained, 49% demonstrated a significant health improvement, 44% had no effect on health, and 7% were associated with significant worsening of health. The most commonly reported outcome was mental health.

Eleven of the studies included tested interventions in the domain of early life and education. Nine of the included studies examined the social and health impacts of income maintenance and supplementation programs. Twelve of the included studies focused on employment and welfare-to-work: team-based supported employment, job training programs, employment support services.

The researchers found that early life and education interventions were associated with an 8% reduction in smoking. "Although this is a fairly small percentage, it could have wide implications and may be a proxy for other forms of risk-taking behaviors, such as condom or seatbelt use" noted Emilie Courtin, PhD, a fellow at Harvard University.

Income support and health insurance programs showed the most promise in improving population health. For a very small number of studies, harmful effects of social policies on health were indicated. These were mainly concentrated among time-limited welfare-to-work interventions.

The researchers point out that interventions in the domains of early life and education, income, and health insurance are particularly promising as population health policies. While many experiments were at high risk of bias, the research -- all based on the "gold standard" randomized, controlled trial, gives us confidence that income support programs save lives.

Still, some studies that should have been able to measure health outcomes showed no health impact. This may indicate that they were true negative findings -- and that health care providers and insurers should not randomly invest in social policies and expect health benefits, noted Muennig.

"Our study is exciting because it shows for the first time that it is possible for the government to improve health by making investments outside of the health system. The strong investments made by peer nations in welfare may explain why they have left the United States in the dust with respect to health and longevity. Still, policymakers need to be careful about how they make these investments, because they also have the potential to do harm."