Advances and Applications in Statistics
Volume 12, Issue 1, Pages 67 - 84
(June 2009)
|
|
ANALYZING PRESSURE ULCER DEVELOPMENT IN 36 NURSING HOMES USING BAYESIAN HIERARCHICAL MODELING
Jing Zhang (USA), Chong Z. He (USA), David R. Mehr (USA) and Robin L. Kruse (USA)
|
Abstract: Pressure ulcer development is an important consideration for judging the quality of nursing service provided by different nursing homes. Several methods have been used to analyze the rate of pressure ulcer development, including traditional logistic regression, Bayesian hierarchical modeling and semi-parametric approaches. In this paper, we use Bayesian hierarchical models to estimate the pressure ulcer development rates in 36 nursing homes. The Bayesian approach provides posterior distributions of model parameters and predictive distributions in addition to parameter estimates and predictions, therefore we could develop different criteria for detecting the nursing homes that have unusually high rates of pressure ulcer development. We computed the Deviance Information Criterion (DIC) to compare the hierarchical models in which different sets of predictors were incorporated. Cross-validation was used to evaluate the predictive ability of the Bayesian hierarchical model proposed. The results showed that the credible intervals of pressure ulcer development rates provided by Bayesian hierarchical models are a powerful tool to identify the nursing homes with unusually high rates; the Z-score (or p-value) provided by the logistic regression model is not suitable when the sample size is small and the large sample assumption is not satisfied. |
Keywords and phrases: Bayesian hierarchical modeling, logistic regression, Markov Chain Monte Carlo, nursing home, quality of service. |
|
Number of Downloads: 346 | Number of Views: 1270 |
|