The price per total knee replacement (TKR) performed is fixed but the subsequent length of hospital stay (LOS) is variable. The current national average for LOS following TKR is six days. LOS is an important marker of resource consumption, has implications in patient satisfaction, and is used as a marker of hospital quality. The aim of this study was to describe the temporal change in demographics between 2004 and 2009, and to identify intra-operative factors and patient characteristics associated with a prolonged LOS that could be addressed to improve clinical practice. We performed a retrospective cohort review of 184 patients (2004 n=88, 2009 n=96) who underwent primary TKRs at Chorley District General Hospital. The median LOS in 2009 was eight days compared to ten days in 2004, an average of 3.5 days less (p < 0.001). Patients were significantly younger (p < 0.001) in 2009 (median 66 years) compared to 2004 (median 74 years), with both years having a similar female predominance. There was no significant change in the BMI or American Society of Anesthesiologists score between 2004 and 2009. This data suggests that block contracts with the private sector has not influenced the demographics of patients being treated in the NHS. Intra-operative factors including the use of a peripheral nerve block, the surgeon grade, the day of the week the operation was performed, the operation length, and the change in pre- to post-operative haemoglobin were not found to significantly increase the LOS (p = 0.058, p = 0.40, p = 0.092, p = 0.50, p = 0.43 respectively). Cemented TKRs had a median LOS of nine days compared to eight for uncemented implants (p = 0.015). However, patients with a cemented implant were on average 6.2 years older than patients with an uncemented implant (p < 0.001). Using Cox proportional hazard regression modelling, the occurrence of a post-operative complication (p < 0.001), female sex (p = 0.024), advancing age (p = 0.036), and the need for a blood transfusion (p = 0.0056) were the most significant factors for prolonging the LOS. Patients who were given a transfusion stayed a median of 13 days compared to nine for those who did not (p < 0.001). The median pre-operative haemoglobin for those who required a transfusion was 11.85g/dl compared to 13.6g/dl for those who did not (p < 0.001). Being obese or morbidly obese did not significantly prolong the LOS (p = 0.95).

In conclusion, this study highlights significant patient characteristics which are associated with a prolonged LOS following TKR. The relatively low pre-operative haemoglobin in patients requiring a blood transfusion is a potential target for reducing the LOS.