Layman’s abstract for paper on an extension of the median odds ratio for measuring variability between clusters by subgroup

Every few days, we will be publishing layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.

The article featured today is from Statistics in Medicine and the full article, published in issue 38.22, is available to read online here.

Yarnell, C, Pinto, R, Fowler, R. Measuring variability between clusters by subgroup: An extension of the median odds ratio. Statistics in Medicine. 2019; 38: 4253– 4263. doi: 10.1002/sim.8286

In most clinical research we are interested in the relationship between patient characteristics (such as age) and outcome (such as mortality). Patients are often grouped by hospital, and the groupings are known as clusters. Outcomes for patients within the same cluster may be similar, so special statistics are needed to measure variability in outcomes across clusters. In this paper we derive and demonstrate a new method for investigating equity in clustered data using the example of end-of-life care for patients of different ethnicities in the United States. We hypothesized that patients of non-Caucasian ethnicities may have a different prevalence of care limitations (eg. no CPR orders) than patients of Caucasian ethnicity. We used our new method to investigate variation in care limitations by hospital, hypothesizing that non-Caucasian patients experience a higher variation in care limitation prevalence because hospitals will vary in their ability to provide equitable care. We found differences in care limitation prevalence by ethnicity but no differences in variation across hospitals, suggesting that differences in care limitation by ethnicity are not likely explained by hospital-level factors and instead more likely attributable to broad social, economic and cultural forces.