A Public Health of Consequence: Review of the August 2016 Issue of AJPH

In this issue of AJPH, Bor’s¹ invited editorial focuses on the role of natural experiments in helping guide inference and thinking in public health. Bor suggests that with the appropriate cautions and caveats, we need not leave the results from nonrandomized trials at the doorstep. We can and should invite them in. In particular, he argues convincingly that natural experiments present an opportunity to study hard-to-randomize exposures, with real potential for translation into policy—a core aim of a public health of consequence. Bor makes the case that this returns public health (to some extent) to its roots, studying hard to assess macro-level changes that stand to influence the health of populations. We find Bor’s argument compelling, and of a piece with the argument made by Westreich et al.² in a recent issue of AJPH about our need to move beyond the traditional “hierarchy of evidence” to a broader, more imaginative use of the potential methodological arsenal that can help us understand the forces that drive the health of populations.

In this issue of AJPH, two articles well illustrate the point made by Bor—how capitalizing on a natural experiment can help guide inference for generating generalizable knowledge.

The first of these articles concern perhaps the most prominent macro-level health policy intervention of the past decade: the implementation of the Affordable Care Act (ACA). Buchmueller et al.³ assess insurance coverage for different racial/ethnic groups with the introduction of the ACA using the American Community Survey from 2008 to 2014. The need for action on this front is in many ways astonishing. In 2013, 41.3% of Hispanics and 25.8% of Blacks were uninsured, compared with 14.8% of Whites. This threefold difference in insurance coverage should have long provoked outrage at the differential access to health care across the United States. Buchmueller et al. show that with the introduction of the ACA, these gaps narrowed as Hispanics saw more gains in insurance coverage than did Whites. Importantly, this difference was more pronounced in states that implemented the Medicaid expansion. This narrowing in racial/ethnic disparities is heartening, although the wide gaps in coverage between these groups remain alarming. The evaluation of these effects of the ACA, though, highlights the long-tail consequences of macro-level policy change, and how their evaluation can guide us to the full range of consequences of legislation.

In an interesting addition to this discussion, Nguyen et al.⁴ tackle another aspect of the ACA. Despite the many evident benefits of the ACA, now supplemented by data about narrowing of racial/ethnic differences in coverage, 19 states, mostly in the US south, have not expanded Medicaid under the ACA. Nguyen et al. show that low-income adults with both public and private insurance have substantially better access to care than do adults without insurance, further building the case for insurance coverage and the compelling logic for participation in the full range of benefits available under the ACA. The limited uptake of the ACA remains unfortunately a data-ill-informed political issue, ill-serving the populations who would benefit from its widespread uptake.

On a different issue, Schmidt et al.⁵ tackle the growing wave of medical marijuana laws around the United States. To date, 23 states and the District of Columbia have legalized marijuana for medical purposes, with more states likely to follow suit. While this expansion has come in large part as a response to the growing acceptance of the benefits of medical marijuana for a range of medical conditions, concerns remain that this shift will result in shifting attitudes about marijuana use (particularly among young people) with subsequent changes in marijuana use and attendant adverse consequences. This, then, represents an unfolding natural experiment nationwide, ripe for analysis. Schmidt et al. assess whether there were associations between living in states with medical marijuana laws and more permissive attitudes toward marijuana. Demonstrating some of the challenges in these types of analyses, Schmidt et al. do find a trend toward more permissive attitudes among states with medical marijuana legislation but note that it is rather difficult to disentangle whether this permissiveness gives rise to a political culture that is likelier to pass marijuana legislation or vice versa. Longer-term longitudinal assessments will be needed to tackle this question more definitively, including, potentially, the adoption of methods that extend beyond the regression approaches adopted in this article. Others in the field⁶ have similarly wrestled with this issue, leaving the result as yet unclear, and fertile ground for future analyses.

We are heartened both by Bor’s commentary and by these articles’ grappling with some of the more challenging natural experiments of our time. We look forward to more innovative work in the area, toward understanding the macrosocial drivers of population health.