237100 Time-Dependent Covariates in Modeling Rehospitalization Medicare Data

Monday, October 31, 2011: 3:10 PM

Anh Nguyen, BSC; MS , Quality Management Department, Banner Good Samaritan Medical Center, Phoenix, AZ
Jeffrey Wilson, MS; PhD , NIH Center/Department of ECN, Arizona State University, Tempe, AZ
Objectives: We revisited factors, while adjusting for the feedback if any between covariates process and response process, which were found significant in modeling times to rehospitalization, but have been overlooked or not fully explored in prior models. Methods: Using Medicare data taken from the State of Arizona Hospital Discharge Database 2003-2005 we identified significant factors contributing to the probability of rehospitalization within 30 days while including time-dependent covariates (after classifying them into three types) through the generalized method of moments estimation for coefficient. In particular, we took into account the correlation among responses inherent over time and the correlation due to feedback from time-dependent covariates process. We fitted logistic regression models while classifying the time-dependent covariates accordingly, which if misclassified may result in different findings about their significance. Results: Length of previous stay in hospital, age of the patient, total numbers of diseases, diagnosed with diabetes mellitus and pneumonia have significant impact on the probability of rehospitalization within 30 days after discharge. These significant time-dependent covariates were investigated while adjusting for the feedback from the covariates process and being hospitalized within 30 days. Gender, race, total number of procedures and complication of device have little or no impact on whether or not a patient will be rehospitalized within 30 days after discharge. Conclusions: Older patients, those who had longer previous stays, those with more total diseases, those with pneumonia or diabetes mellitus, had a better chance of rehospitalization within 30 days.

Learning Areas:
Biostatistics, economics
Systems thinking models (conceptual and theoretical models), applications related to public health

Learning Objectives:
Describe the outcomes or actions participants can expect to demonstrate as a result of the educational experiences

Keywords: Statistics, Hospitals

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: Graduate Degree in Statistics. Part of my graduate work
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.