Back

Improved beta-binomial estimation for reliability of healthcare quality measures

Zhou, G.; Lin, Z.

2023-01-09 health systems and quality improvement
10.1101/2023.01.07.22283371 medRxiv
Show abstract

BackgroundThe popular beta-binomial approach to estimate the reliability of healthcare quality measures (Adams et al. 2010 New England Journal of Medicine) may yield grossly over-estimated reliabilities for providers with event rates equal to 0% or 100%. ObjectiveImprove the beta-binomial approach to yield more reasonable reliability estimates for providers with event rates equal to 0% or 100%. MethodWe revise the beta-binomial approach by substituting Bayesian estimates with various priors for the crude event rates. We evaluate the new reliability estimates using Monte Carlo studies and two real-world measure examples. Results and conclusionThe revised beta-binomial approach based on Jeffreys non-informative prior yields more reasonable reliability estimates for providers with event rates equal to 0% or 100% and statistically outperforms the original beta-binomial approach regarding bias and standard errors.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.