August 26, 2019

What is known about drivers of health: a literature review

By Austin Frakt

While there is widespread understanding that the health system and other factors — social determinants — affect health, we know relatively little about their precise contributions to health differences across a population at a point in time or differences in health of a fixed population over time.

Our just-released literature review provides a historical background on the work done in the last forty years to quantify the contributions of various factors to health. It goes on to summarize a selection of scholarly work on the drivers of health.

The work summarized includes efforts to decompose health outcomes into several key determinants, some of which is included in the chart below.

These studies organize the factors that affect health into, at most, five categories: behaviors, medical care, social circumstances, environment, and genetics (the last three of which we have grouped together in the chart, for convenience).

Though these five categories have become standard, they are not equivalent to one another in several respects. For example, some are reasonably considered proximate, directly affecting health, such as medical care, while others are likely more distal, affecting health through more indirect pathways, like social circumstances. The more distal factors affect health through multiple pathways not conveyed by the categorization framework. For example, social circumstances can degrade health through diminished physical access (distance, transportation), reduced resources (income, health insurance), poor health literacy, or direct discrimination, to name a few possible pathways. Additionally, medical care affects health through purposeful interventions (treatments and preventive care) while social factors are largely static or, in some cases, not modifiable (e.g., though age changes, its pace of change is unalterable).

Studies we surveyed do not all examine the same health outcomes. While we group these studies for the purpose of comparison, their different methodologies, definitions, parameters, and time-spans all influence their estimates. They are arguably not comparable for these reasons.

You’ll find more summaries and analysis in the literature review.