Anticipated Impacts on Veteran's Healthcare: Reducing hospital readmissions, especially those resulting from poor inpatient or outpatient care, represents an excellent opportunity to improve quality of care. Focusing on specific subgroups in the VA who are at relatively high risk of readmission, i.e., patients with heart failure (HF), acute myocardial infarction (AMI), and pneumonia, may yield a higher reward in terms of improved patient care and successful quality improvement (QI) initiatives, such as improving coordination of care and discharge planning. Project Background/Rationale: Readmissions are common, costly, and can lead to further functional decline of the patient. For many patients, hospitals are a revolving door, as they yo-yo back and forth between the hospital and community. Many of these readmissions may be potentially preventable, the risk being modifiable by the quality and type of care provided. Therefore, identifying this subset will help direct QI efforts more effectively in reducing readmission rates. Objectives: The primary goal of this study is to examine the rate of readmissions for patients discharged with HF, AMI, or pneumonia, determine the proportion of readmissions that is potentially preventable using software designed to identify potentially preventable readmissions, and assess the extent to which this software accurately classifies readmissions as potentially preventable. Specific objectives include: 1) estimate risk-adjusted models to predict 30-day readmissions for patients discharged with HF, AMI, or pneumonia;2) investigate rates of potentially preventable readmissions for HF, AMI, or pneumonia using readmission classification software based on administrative data;3) develop chart abstraction tools to identify potentially preventable readmissions for discharges with HF, AMI, or pneumonia;4) apply chart abstraction tools to VA electronic medical records (EMR) to classify HF, AMI, and pneumonia all-cause readmissions and compare results to the administrative data-based classification software;and 5) re-estimate hospital-specific risk-adjusted rates of potentially preventable readmissions in the VA using supplemental automated data. Methods: This is a three-year, retrospective, observational cohort study using FY06 through FY09 national VA data, supplemented by CMS Medicare files. Assessment of the potential preventability of 30-day readmissions for each condition cohort will be compared between EMR-abstracted data and the 3M" Potentially Preventable Readmissions (PPRs), a clinically-based classification system designed for identifying potentially preventable readmissions. Risk-adjusted facility-specific readmission rates will be calculated using the PPR software. Condition-specific, explicit criteria will be developed and refined by clinical expert panels. These criteria will be incorporated into chart abstraction tools to identify potentially preventable readmissions and classify causes of readmission. Explicit criteria will primarily focus on system- and provider-level factors (processes and structures of care) that are associated with increased risk of readmissions, such as premature discharge and failure of follow-up post-discharge. To assess how well the PPR algorithm performs in identifying potentially preventable readmissions, the sensitivity, specificity, and the positive predictive value of the algorithm will be examined against the EMR-abstracted data. Supplemental automated clinical and diagnostic data based on the panel-derived explicit criteria will be added to the administrative data to improve identification of potentially preventable readmissions. Data elements that may provide information on appropriate in-hospital or follow-up care from discharge to readmission will be considered for inclusion. "Reclassified PPRs" (PPRs and supplemental clinical data) will be developed for each condition cohort. Hospital-specific risk-adjusted rates of potentially preventable readmission rates will be re-estimated using the "reclassified PPRs" and compared with hospital-level risk-adjusted rates.
Reducing hospital readmissions represents an excellent opportunity to improve quality, particularly for those subgroups in the VA with relatively high rates of readmissions. Many readmissions are felt to be potentially preventable, although inconsistencies in definitions of readmissions and differences in decision rules hinder accurate measurement. This study will determine the proportion of readmissions that is potentially preventable (i.e., those readmissions that could have been avoided if appropriate care had been provided) by comparing an administrative data-based algorithm that identifies potentially preventable readmissions with explicit criteria abstracted from patients'electronic medical records. Understanding the hospital and system factors that contribute to readmissions will help direct quality initiatives towards specific care processes that may help reduce readmission rates in these high-risk groups. Identifying those readmissions that are potentially preventable should help to focus these quality improvement efforts even more directly.