H.M. Engebretson1, S.J. Lowe1, J. Schmidt1, L. Khan2, M. Zielinski2, C.T. Wilson2 1Baylor College Of Medicine, School Of Medicine, Houston, TX, USA 2Baylor College Of Medicine, Division Of Trauma And Acute Care Surgery, Houston, TX, USA
Introduction: Timely and appropriate discharge planning is crucial for reducing hospital stays and enhancing patient satisfaction. Discharging patients to post-acute care facilities (PACFs) necessitates interdisciplinary planning, as there is no standardized method to assess factors influencing this decision. Our study aimed to identify factors predisposing trauma patients to PACF discharge and develop a model to quantify the likelihood of discharge to PACF throughout their hospital stay.
Methods: We retrospectively reviewed 9,786 trauma patients admitted to our institution injured in Harris County from 01/2017-12/2023. Patients were stratified by discharge location: to home (n=8816, 90.9%) or a PACF (n=970, 9.91%). PACF included discharge to skilled nursing facility (SNF), hospice, and rehabilitation centers. Refined analysis utilized propensity score matching (PSM) with a 1:n nearest neighbor scheme, matching on age, gender, and injury severity score to create a refined cohort (n=1,591) consisting of patients discharged home (n=656) or to a PACF (n=930; rehab n=537, 33.8%; SNF n=254, 15.9%; hospice n=144, 9.1%). The primary outcome was odds of being discharged to a PACF calculated using univariate and multivariate analysis in the PSM cohort. We then used the odds ratio (OR) of each significant variable (p < 0.05) to create a weighted index called the Predictive Index of Post-Acute Care (PIPAC). To calculate the PIPAC, the natural log of the significant variable’s OR was added to the initial PIPAC score of 1.0 (PMID: 37903063). The PIPAC model was then internally validated with the original cohort (n= 9,786) with a receiver operator curve (ROC).
Results: Nine variables were significant predictors of PACF discharge in the PSM cohort. Factors associated with a higher likelihood of PACF discharge included Asian race (OR=3.5, p=0.022), fall-related injury (OR=2.0, p=0.004), Medicare insurance (OR=2.3, p <0.001), staying in the ICU less than one day (OR=2.6, p < 0.001) or greater than 3 days (OR=2.7, p<0.001), and hospital LOS >1 week (OR=5.01, p <0.0001). Factors associated with a lower likelihood of PACF discharge were self-pay (OR=0.4, p <0.001), Medicaid insurance (OR=0.4 p=0.002), and if on a ventilator for < 1 day (OR=0.3, p<0.001). These ratios were used to create the PIPAC model with a C-Statistic of 0.73 (Equation 1) after ROC analysis in the PSM cohort. We internally validated the model by applying it to the initial cohort, which yielded a C-statistic of 0.83 after ROC analysis (Figure 1).
Conclusion: Various medical and social risk factors influence the likelihood of a patient being discharged to a PACF. Our study identified risk factors within our patient cohort and created an objective scoring system for clinicians to assess the likelihood of PACF discharge. The PIPAC model can be used to initiate early interprofessional discharge planning, potentially reducing the length of stay associated with discharge delays.