Total knee arthroplasty: does ultra-early physical therapy improve functional outcomes and reduce length of stay? A retrospective cohort study

Early mobilization is an integral part of enhanced recovery after surgery (ERAS) [17]. The ERAS society recommends that patients should be mobilized as soon as possible after surgery - this helps to reduce the length of stay and counteract the adverse physiological effect of prolonged bed rest [6]. In current literature, studies have shown that early mobilization or rehabilitation after total knee arthroplasty improves functional outcomes as well as reduces the length of hospital stay [7, 11, 12, 18, 19].

Our study’s findings correlate with that of previous studies which have found that early mobilization on POD 0 following TKA leads to shorter LOS [13, 15, 20]. Chen et al. found in their prospective cohort study that patients who ambulated with PT on POD 0 had a shorter LOS than those patients who did not [15]. Similarly, Den Hertog et al’s randomized controlled trial showed that patients who were in a POD 0 fast-track rehabilitation program had a shorter LOS compared to standard rehabilitation programs in their randomized control trial [7]. In our study, all patients received PT within POD1. Patients in the ultra-early group received PT on POD0 while patients in the early (12–24 h) group and > 24 h group had PT on POD 1. Although both ultra-early and early physical therapy showed benefits in terms of LOS, our results show that there is diminishing benefits to ultra-early physical therapy (LOS: 4.75 vs. 4.96 days comparing < 12 h and 12–24 h) compared to that of early physical therapy (LOS: 4.87 vs. 5.34 days comparing < 24 h and 24–48 h). Although the ultra-early physical therapy (< 12 h) group showed a lower LOS compared to the 12–24 h group, the small difference (4.75 vs. 4.96 days) is unlikely to be clinically significant. In the healthcare setting where resources and manpower are finite and limited, it would be necessary to weigh the benefits of a slightly reduced LOS from ultra-early physical therapy against manpower and resource limitations.

In addition, the recorded mean LOS in our study is longer than what has been reported in previous literature (Bohl et al.: median of 32 h for POD0 PT and 31 h for POD1 PT; Chen et al.: mean of 2.8 days for POD0 PT and 3.7 days for POD1 PT). There are 2 main factors that could have contributed to our longer LOS, which are: the higher average age of our patients, and the difference in criteria for safe discharge by our physical therapists. The average age of our patients’ population (67.7 years) was several years older than patients in other studies (Bohl et al. 63.7 years, Chen et al. 62.3 years) [8, 15]. This could contribute to slower recovery from the initial surgery as well as a slower rehabilitation process. Secondly, there is a strict criteria for safe discharge from the physical therapists at our institution. These include the ability to ambulate with a walking aid, clear ground level obstacles safely, achieve adequate pain control and a clean and dry surgical site. In the study by Bohl et al. 2019, there was no explicit criteria that was mentioned for safe discharge. An arbitrary term of being “cleared” by the physical therapist was used to denote when a patient was safe to be discharged from the acute hospital ward [8]. Chen et al’s study defined their discharge criteria, which included being able to ambulate 100ft and being able to climb up and down stairs. However, the amount of assistance required and the type of mobility aid required to achieve these discharge criteria were not mentioned. In addition, patients who were discharged to skilled nursing facilities and rehabilitation facilities were also included in their study. We have observed that in our local population, a significant percentage of patients have the expectation of only being discharged when they are close to independent ambulation. This would have a great impact on the LOS as a larger number of PT sessions would be needed to get these patients back to or close to their premorbid function.

Our data finds that at 3 months follow-up, there was no significant difference in change in OKS and KSS functional scores - differing from results from other studies in the current literature. Larsen et al. conducted a randomized controlled trial to investigate the effects of accelerated rehabilitation and physical therapy after arthroplasty. Their study involved a heterogenous group of 87 patients who had undergone either total hip arthroplasty (THA), total knee arthroplasty (TKA) or unicompartmental knee arthroplasty (UKA). They found that at 3 months follow-up, there was a significant gain in quality of life (QOL) using the EQ−5D score compared to baseline [11]. However, a subgroup analysis of TKA patients did not reveal any significant difference in the functional outcomes - it appears that their findings were skewed by significantly better results experienced by the THA patients who had undergone early physical therapy. Furthermore, the intervention group not only received earlier physical therapy, but they were also provided with pre-operative education and longer physical therapy sessions. Patients in the intervention group were also cohorted in the same ward. Thus, while Larsen et al’s study shows better functional outcomes after earlier intervention, it is difficult to isolate these effects to merely early initiation of physical therapy.

While other studies have found an increase in ROM with early initiation of physical therapy, our study did not find similar results [9, 10]. Labraca et al’s randomized controlled trial had patients placed into two groups - those receiving physical therapy either before 24 h or after 24 h postoperatively. Patients who received physical therapy within 24 h showed significantly better LOS, ROM and gait scores. While our study found a similar correlation between earlier physical therapy and a shorter LOS, it did not find that earlier physical therapy led to increased ROM at discharge.

There are several strengths to our study. Firstly, the data was collected prospectively, which reduces the risk of recall bias. Secondly, other studies investigating the early initiation (POD 0 vs. POD 1) of PT mainly use LOS as an outcome factor. The present study not only uses LOS, but also compares the ROM, ambulation distance as well as early functional outcome score (3 months) to determine the difference in outcomes when PT is initiated ultra-early, early or after 24 h from surgery. Thirdly, our sample size (n = 569, ultra-early PT n = 216, early PT n = 262, > 24 h n = 91) is relatively large when compared to other recent studies. Finally, our study isolates time to PT as the independent variable between treatment groups - other than this, there were no differences in intervention across treatment groups.

We recognise several limitations to our study. Firstly, the patients were not randomized and patients who were not well medically, not motivated to undergo PT or underwent surgeries later in the day were self-selected to be in the 12–24 h group or 24–48 h group. This might result in selection bias in the study. Secondly, our study focuses on patients who were discharged home. Over half of our TKA patients are discharged to subacute facilities for further rehabilitation prior to being discharged home - our study does not address the effects of earlier physical therapy on this significant patient population. Thirdly, LOS was defined as the time between admission and discharge at our institution. Besides the effects of earlier physical therapy, other non-related issues such as medical complications, patient preference and administrative discharge paperwork may also influence a patient’s length of stay and obscure the differences due to earlier physical therapy.