Risk Prediction Model for 6-Month Mortality for Patients Discharged to Skilled Nursing Facilities.

Area of Science:

• Gerontology and Geriatric Medicine
• Health Services Research
• Biostatistics and Epidemiology

Background:

• Hospitalized patients discharged to skilled nursing facilities (SNFs) face high risks of adverse outcomes.
• A lack of effective prognostic tools complicates care planning and decision-making for this vulnerable group.
• Accurate mortality prediction is crucial for optimizing post-acute care strategies in SNFs.

Purpose of the Study:

• To develop and validate a predictive model for 6-month all-cause mortality.
• To identify key predictors of death among patients discharged from hospitals to SNFs.
• To provide a tool for improved risk stratification and clinical decision support in post-acute care settings.

Main Methods:

• Retrospective cohort study involving 11,647 hospital-to-SNF discharges between 2009 and 2016.
• Gradient-boosting machine modeling used to predict 180-day all-cause mortality.
• Model predictors included demographics, comorbidities, prior healthcare utilization, and hospitalization clinical parameters; validated using 10-fold cross-validation.

Main Results:

• The study identified 9803 unique patients, with 1844 deaths occurring within 180 days of discharge.
• Key predictors of 6-month mortality included age, comorbid burden, prior healthcare use, abnormal labs, and in-hospital mobility status.
• The developed prediction model demonstrated strong performance with an area under the receiver operating characteristic curve (AUC) of 0.82.

Conclusions:

• A robust risk prediction model was derived using parameters available at discharge to SNFs.
• This model can accurately estimate the risk of 6-month mortality for patients in SNFs.
• The findings offer a valuable tool for clinicians to develop mortality prediction instruments for post-acute SNF populations.