Background: The number of older adults discharged to post-acute care (PAC) facilities (such as skilled nursing facilities) after hospitalization is increasing rapidly, but their clinical course in PAC is uncertain. More than 25% will be readmitted, and some may not successfully rehabilitate and return to the community. Clinicians and patients currently lack data to inform their decision about the risks and benefits of PAC at the time of hospital discharge, when the decision to pursue PAC is made. We sought to identify important prognostic factors that influence outcomes of older adults discharged to PAC.
Methods: This was a retrospective analysis of the 2003-2009 Medicare Current Beneficiary Survey (MCBS), a nationally-representative survey of Medicare recipients matched with claims data. Community-dwelling adults age 65 and older who were hospitalized and discharged to a PAC facility were included. The primary outcome was a composite of events representing failure to return to the community, including death, readmission to the hospital, or remaining in a PAC facility 100 days post-discharge, when Medicare benefits for this care expire. We used survey data and the PAC admission Minimum Data Set to evaluate the influence of multiple domains, including patient demographics, health status (including functional and cognitive status), social supports, and active symptoms and treatments at time of PAC admission. We selected variables significant at the p≤0.05 level in univariable analysis for multivariable logistic regression, then measured the discriminative ability of these factors using a c-statistic.
Results: Of 1421 eligible patients, 510 (35.9%) were readmitted, died, or did not return to the community by 100 days post-discharge. In univariable analysis, these patients significantly differed from those who returned to the community in all domains. In multivariable analysis, the most important factors associated with the primary outcome included the presence of dyspnea (OR 1.46; 95% CI 1.09-1.96), cognitive impairment (1.12; 1.02-1.24), use of antipsychotics (1.10; 1.04-1.17), number of physician visits in the PAC facility (1.09; 1.03-1.14), index hospital length of stay (1.02; 1.01-1.03), PAC facility length of stay (0.99; 0.98-0.99), and functional status (0.80; 0.75-0.85). The c-statistic incorporating these seven factors was 0.694.
Conclusions: More than one-third of older adults discharged to PAC facilities are readmitted, die, or remain in the PAC facility 100 days post-discharge. Several factors that influence these outcomes may be modifiable. Their predictive value is similar to most readmission prediction models, which have been successfully used to target interventions to high-risk groups. These findings may serve as a starting point for better informing provider and patient decision-making and improving outcomes during the increasingly common transition of care from hospital to PAC facility.