In a Health Affairs Blog article, Dr. Bruce Sherman and several co-authors at IBM Watson Health issued an important call to action titled “Race and Ethnicity Must Be Included In Employee Health Data Analyses.” They not only address the challenges of collecting race and ethnicity data from administrative sources such as healthcare claims, EHRs, and personnel records, but confront these challenges with practical advice.
The authors stress the urgency of addressing the health risks of systemic racism and target their arguments towards the producers of employee health data, including health systems, insurers, and employer plan sponsors. But as data consumers, the community of employee health researchers and analysts bear our own responsibility to create a demand for health data that reflects the diversity of the workforce.
As researchers, we have long understood the trade-offs of using integrated health datasets—populated heavily with data collected for billing or administrative purposes—that lack race and ethnicity information. We accept that while imperfect, analyses based on incomplete data can still yield valuable insights. This is often acknowledged in the “limitations” section of a paper, with reference to potentially mis-specified models or lack of generalizability to some populations.
What rarely gets mentioned is that muddling through available “race-neutral” data comes at the expense of advancing knowledge about the specific needs of historically marginalized populations. As a result, we miss opportunities to improve the evidence base for investments in healthy workers and unwittingly contribute to the invisibility of racial disparities in employee health—and by extension, we fail to confront the harms of systemic racism.
For our part, employee health researchers should consider analyses across race and ethnic groups a basic part of our research design, in the same way we include sex and age (and region, and industry, and many other characteristics that make their way into claims and benefits data). To that end, we must make every effort to obtain research data that permits analyses across race and ethnic groups. In some cases, this could come down to licensing data only from vendors that can provide race and ethnicity information (and letting the others know why their products did not meet our research specifications).
Other cases—for example, when we conduct analyses of a client’s data from their warehouse—may require difficult conversations about sensitive issues of race and employee privacy that our partners may prefer to avoid. The good news is that Sherman et al. provide a practical roadmap for explaining how racial health disparities harm both employees and businesses, why collecting and reporting these data with due concern for privacy and security is feasible and permissible by law, and how knowledge of health disparities can help employers design and implement effective health benefits strategies.
IBI is committed to making sure race and ethnicity is included in our 2021 and future research analyses, with an eye towards highlighting unmet needs and disparities to help fill in the information gaps. We invite the health and productivity community to join us in this effort. We know it will represent a small step towards addressing the multigenerational impact of systemic racism. But if along the way it advances knowledge of employee health while helping businesses benefit from healthier, more productive employees, it could result in great strides.