In France and in the US, predictions for 2030 include an increased number of total hip arthroplasty (THA) and total knee arthroplasty (TKA) procedures together with an overall trend towards shorter hospital stays. Predictors of hospital length of stay (LOS) include the day of surgery, discharge destination, and patient comorbidities. Available data are conflicting, however, and to our knowledge predictors of LOS after THA or TKA have not been evaluated in France. Improved knowledge of these predictors would be expected to increase patient care efficiency. The objectives of this study were: (1) to determine whether the above-listed factors predict LOS after THA or TKA, (2) to identify predictors of discharge to a rehabilitation unit and of readmission within 30 days after surgery.

Both patient-related factors unamenable to modification and modifiable organisational factors are associated with LOS after THA or TKA.

This large single-centre retrospective cohort study included all adults who underwent primary THA or TKA at our university hospital between 1 January 2015 and 31 December 2016. Non-inclusion criteria were revision arthroplasty, THA with femoral or acetabular reconstruction, TKA using a constrained hinged implant, and fracture as the reason for arthroplasty. Preoperative parameters, type of arthroplasty, and postoperative care were recorded.

We included 938 patients with THA and 725 patients with TKA. By multivariate analysis, the likelihood of being discharged by day 5 decreased with older age (HR, 0.986; 95%CI: 0.98–0.99) and was lower by 13% in females (HR, 0,871; 95%CI: 0.77–0.986), by 39% in patients with diabetes (HR, 0.606; 95%CI: 0.5–0.73), by 68% in patients discharged to rehabilitation units (HR, 0.322; 95%CI: 0.267–0.389), and by 27% in patients who had arthroplasty on a Friday (HR, 0.733; 95%CI: 0.631–0.852). Factors predicting discharge to rehabilitation unit were older age, female gender, chronic obstructive pulmonary disease, anxiety-depressive disorder, and a history of stroke. Risk factors for 30-day readmission were male gender, obesity, and discharge to rehabilitation unit.

In this study, predictors of LOS were identified using a survival model that considered age as a continuous variable, separate comorbidities, and the discharge destination. Our findings are consistent with earlier reports and confirm the strong associations linking LOS to diabetes, day of surgery, and discharge destination in France. We also identified predictors of discharge to rehabilitation and of readmission within 30 days.

IV, retrospective observational cohort study.