Background: Despite a sevenfold decline in the number of postoperative nights (21-3) after a total knee arthroplasty (TKA) over the last four decades, predictors of length of stay (LOS) are not fully understood. We attempted to ascertain these factors by analyzing a large cohort of patients.
Methods: Prospectively collected data between January 2016 and March 2017 were retrospectively analyzed at our institute. Charts of 1663 consecutive, simple primary unilateral and bilateral TKAs were reviewed for the LOS excluding staggered bilateral, complex primary and revision knees.
Statistical analysis: Preoperative variables [demographics, cash/credit status, historical, clinical, laboratory findings, Knee Society Function Scores and Oxford Knee Scores (OKSs)] were scrutinized by multivariate regression to identify significant factors affecting LOS and formulate model equations for patients and health caregivers. Results were incorporated into an iOS application, which was tested for accuracy.
Results: Among 1524 unilateral and 139 bilateral TKAs, mean LOS was 4.4 and 5.2 days, respectively. Five factors, namely insurance, flexion/hyperextension deformity, preoperative OKS and a rheumatoid etiology, were significantly associated with prolonged LOS in unilateral knees. The impact of these independent variables on LOS could be calculated by: [Formula: see text]For bilateral cases, the only significant variable extending LOS was a low preoperative OKS and the equation is given as follows: [Formula: see text]The iOS-app-predicted LOS and actual LOS were similar (p > 0.05) for 115 prospectively operated knees.
Conclusion: Poor preoperative OKS, rheumatoid etiology, flexion and hyperextension deformity and delays in insurance affected unilateral TKR LOS, while poor preoperative OKS alone affected LOS in bilateral cases.
Keywords: Discharge delays; Equations predicting stay; Insurance delays; Length of stay; Primary total knee arthroplasty; Severe deformities.