A validated preoperative risk prediction tool for extended inpatient length of stay following anatomic or reverse total shoulder arthroplasty.

BACKGROUND: Recent work has shown inpatient length of stay (LOS) following shoulder arthroplasty to hold the second strongest association with overall cost (after implant cost itself). In particular, a preoperative understanding for the patients at risk of extended inpatient stays (≥3 days) can allow for counseling, optimization, and anticipating postoperative adverse events. METHODS: A multicenter retrospective review was performed of 5410 anatomic (52%) and reverse (48%) total shoulder arthroplasties done at 2 large, tertiary referral health systems. The primary outcome was extended inpatient LOS of at least 3 days, and over 40 preoperative sociodemographic and comorbidity factors were tested for their predictive ability in a multivariable logistic regression model based on the patient cohort from institution 1 (derivation, N = 1773). External validation was performed using the patient cohort from institution 2 (validation, N = 3637), including area under the receiver operator characteristic curve (AUC), sensitivity, specificity, and positive and negative predictive values. RESULTS: A total of 814 patients, including 318 patients (18%) in the derivation cohort and 496 patients (14%) in the validation cohort, experienced an extended inpatient LOS of at least 3 days. Four hundred forty-five (55%) were discharged to a skilled nursing or rehabilitation facility. Following parameter selection, a multivariable logistic regression model based on the derivation cohort (institution 1) demonstrated excellent preliminary accuracy (AUC: 0.826), with minimal decrease in accuracy under external validation when tested against the patients from institution 2 (AUC: 0.816). The predictive model was composed of only preoperative factors, in descending predictive importance as follows: age, marital status, fracture case, ASA (American Society of Anesthesiologists) score, paralysis, electrolyte disorder, body mass index, gender, neurologic disease, coagulation deficiency, diabetes, chronic pulmonary disease, peripheral vascular disease, alcohol dependence, psychoses, smoking status, and revision case. CONCLUSION: A freely-available, preoperative online clinical decision tool for extended inpatient LOS (≥ 3 days) after shoulder arthroplasty reaches excellent predictive accuracy under external validation. As a result, this tool merits consideration for clinical implementation, as many risk factors are potentially modifiable as part of a preoperative optimization strategy.

Duke Scholars

Author Tally Edward Lassiter Orthopaedic Surgery

Author Zoe Hinton  

Author Cynthia Lea Green Biostatistics & Bioinformatics, Division of Biostatistics

Author Oke Adrian Anakwenze Orthopaedic Surgery

Author Christopher Scott Klifto Orthopaedic Surgery