Mortality Rates From COVID-19 Are Lower in Unionized Nursing Homes

The quick and dirty DL: Unionized nursing homes had better COVID-19 mortality rates than non-unionized nursing homes. It appears this may have been driven by better access to PPE and higher pay (resulting in less staff members having multiple jobs at multiple facilities) among unionized nursing homes.   

The unpacked DL:

When we think about the factors that impact our health, we typically think of things like: diet, exercise, genetics, whether or not we take the medications our doctor prescribes, etc. These are some very common, tried and true determinants of health.

But we often forget that there are a whole host of other health determinants; things like racism, health policy, environmental factors (air quality, forest fires, etc.), built environment, access to/ use of health services, healthcare quality, etc. I would argue that health care market features should also be added to this list of health determinants, and today’s article is just one of the many examples we will explore that demonstrates how the way our health system is organized impacts our health outcomes.

I was “raised” (academically speaking) in an Economics department, and for that reason I always come back to incentives—what outcomes do our actions incentivize? A mentor of mine always said, “the outcomes we observe are the outcomes the process was designed to produce—for better or worse”. If we observe terrible outcomes—it is because the process is designed to produce terrible outcomes. If we observe incredible outcomes, it is because the process is designed to produce incredible outcomes. Where we really get into trouble is with unintended consequences—the thing that “no one could have predicted”, or the thing that blindsides everyone and disrupts our intended results. 

Health care worker unions in the U.S. date back to the 1930’s—and like most unions at the time—focused on fair wages and proper working conditions. They were not “designed” to make patients safer or healthier—they were designed primarily to protect workers. Perhaps to no surprise—there’s a body of evidence that suggests that health care worker unions are actually pretty good for patients too (see Dube et al 2016 for starters).

Dean et al 2020 demonstrate that nursing homes in New York with health care worker labor unions had 30% lower COVID-related mortality rates and a 42% relative decrease in COVID infections than nursing homes without health care worker unions, even after adjustment for a wide range of covariates and other robustness checks. There’s no way any scientist can control for every factor in a retrospective analysis—but in this particular analysis, several covariates are included: age of nursing home residents, overall resident illness acuity, presence of Medicaid insurance among residents, surrounding community COVID-19 infection rate, etc.

While I don’t think these results are that surprising, I think questions about the mechanism are really interesting. The authors hypothesized that improved access to PPE among unionized staff relative to non-unionized staff may explain some of the variation—and indeed, they did find that unionized nursing homes did report better access to PPE. But still- perhaps other mechanisms were at play—greater adherence to standardized infection prevention protocols, more full-time workers making a living wage and workers moonlighting at other facilities to make ends meet, etc. What is interesting is that, it appears that there is are positive spillover effects into other categories of labor that occur when other health care workers have some level of labor protection. Is this replicable in a non-union setting with similar positions (e.g., nurses, nursing assistants, etc.)? Is this replicable in traditionally non-union positions (e.g, physicians)? If paying workers a little better, and providing a basic level of labor support can help a nursing home avoid 42% of COVID-19 infections and 30% of COVID-19 related deaths, it would certainly suggest that this may quite possibly be a cost-effective intervention.

In reference to Health Affairs article by Adam Dean, Atheendar Venkataramani, and Simeon Kimmel 9/10/20. https://www.healthaffairs.org/doi/full/10.1377/hlthaff.2020.01011