Hospitals are now increasingly admitting large numbers of elderly patients, a subset of which are at risk of nursing home entrance. Clinicians are often now able to identify correctly those patients who will need admission to a nursing home. This delay in identification can cause inappropriate or inadequate discharge planning to occur. The purpose of this study is to investigate the potential for improving hospital discharge planning for the elderly through the application of event history analysis methods to acute hospital stays. Delays in discharge planning may result in increased time spent in the hospital waiting for nursing home admission. These days are termed administratively necessary days (ANDs). Since hospitals are not fully reimbursed for ANDs, financial losses are associated with these days. With the advent of the prospective payment system (PPS) the necessity of identifying patients at high risk of nursing home admission, early on during hospitalization, has intensified. Targeting patients at risk of nursing home entrance has the potential, not only to reduce ANDs, it also has the potential to benefit the elder by preventing premature or unnecessary nursing home entrance. This study uses data from the Medicus Systems Productivity and Quality of Patient Care Classification system (NPAQ: Medicus Systems Corporation, 1978) to assess how the risk of nursing home admission differs across subgroups of disabled elderly. The intended purpose of NPAQ, used in over 400 hospitals nationally, is to improve the efficiency of staffing patterns by quantifying the nursing personnel needs of hospitals. It has never been utilized to predict discharge to a nursing home. NPAQ contains indicators of patient care needs that have been demonstrated in the literature to be predictive of nursing home admission, for example, function and cognitive information. A survival analysis model is chosen because it can overcome two problems that will occur if OLS regression is utilized to fit the model. The two problems have been termed """"""""right censoring"""""""" and the problem of """"""""competing risk"""""""". The predictive accuracy of the model will also be tested, on a subset of the sample, once it is developed.
