This project takes quality, safety, and staffing data examined using R and logistic regression to look for explanatory variables that are the best predictors of various clinical quality outcome measures. Multiple Nursing Home Compare datasets from data.medicare.gov are combined and used, requiring extensive data cleaning.
No easy ‘smoking gun’ causality was discovered when it came to characteristics that higher performing nursing homes share. Surprisingly, it appears that the tried and true hours of nursing per patient per day metric that is a predictor of better quality outcomes (and fewer complications) in the hospital side of health care is not an automatic predictor of similar improved outcomes in the nursing home side of things.