Knee replacement outcome predicted by physiotherapists: a prospective cohort study

Introduction

Knee arthroplasty (KA) is considered a successful orthopaedic procedure to alleviate knee pain and disability in knee osteoarthritis (OA) and the demand for KA is large and growing worldwide (Ackerman et al., 2019; Culliford et al., 2015; Nemes et al., 2015). Despite being considered a generally safe and successful procedure, there is a significant proportion (20%) of patients who endure years of disability and dissatisfaction with their postoperative function (Bourne et al., 2010; Gunaratne et al., 2017; Kahlenberg et al., 2018). Many of these patients do not undergo revision surgery, but all add to the society’s burden of health care as clinicians and allied services strive to remedy their dissatisfaction.

While efforts have been made to predict the best candidates for KA (Birch et al., 2019; Harbourne et al., 2019; Judge et al., 2012) it has proven very difficult to identify robust pre-operative prognostic factors (Gunaratne et al., 2017). It is important to be able to identify inappropriate candidates pre-operatively, but until that is possible, efforts to improve post-operative rehabilitation and identify individuals that may need special attention post-operatively is equally important.

Physiotherapists play an important role in the post-operative treatment of patients that have received a KA. In general patients are mobilised very early, and rehabilitation is commenced during the first days after surgery (or even on the same day) while the patient is still hospitalised.

There are numerous factors that can predict the outcome of KA, of which most are either poor predictors or unsatisfactorily assessed scientifically (Harmelink et al., 2017). In daily clinical physiotherapy practice at the orthopaedic wards, previous experiences, knowledge, and personal interaction with a patient altogether result in the physiotherapist’s intuition about the future course of a patient’s recovery. Previously, it has been shown that physiotherapists’ prognoses during the initial clinical encounter for the projected outcome of patients with low back and neck pain they treated, was associated with the actual clinical outcome (Cook et al., 2015). However, it is unknown if such intuition is a reliable predictor of the outcome of KA.

With this study we aimed to assess if physiotherapists treating inpatients at orthopaedic wards could predict the 6- and 12-months post-operative self-reported pain, functional status and health related quality of life of patients undergoing KA.

Materials & Methods

The study is a prospective, pragmatic, longitudinal cohort study conducted from December 2016 to December 2019 with a 6- and 12-month follow-up. The study was conducted according to a prespecified protocol, which was pre-registered and submitted to the Health Research Ethics Committee of The Capital Region of Denmark (file number: 16039254). The committee deemed the study exempt from approval as the study only uses questionnaire data. Such studies can be implemented without approval from the Health Research Ethics Committee according to Danish legislation. The prespecified study protocol was pre-registered at http://www.clinicaltrials.gov (Identifier: NCT02982785) prior to data collection.

Results

From December 2017 to December 2018 a total of 382 individuals (57% women) were invited to participate (Table 1). Of these 20 were not eligible and 1 did not have a prognostic score. Accordingly, 361 patients were included with a mean age of 69.2 years (SD 7.9). There were 42 patients who did not respond to any survey or reminders, 16 who responded to the 6-month but not the 12-month survey, and 12 who did not respond to the 6-months but responded to the 12-month survey (Fig. 1). Thus, at the 6-month follow-up 307 answered the survey and at 12 months 303 answered the survey.

The prognostic scores ranged from 4 to 10, with a mean of 8.3 (SD 1.2) and the median being 9 (Table 1). Summaries of the outcome variables are presented in Table 2.

At the 6-month follow-up, the incidence of positive answers (‘yes’) to the PASS was 240 (78%) and Treatment Success (OKS > 32.5) was 251 (82%). At the 12-month follow-up the incidence of positive PASS answer was 255 (85%) and of Treatment Success was 251 (83%).

A high prognostic score increased the odds of achieving PASS (‘yes’) at 6 months (OR = 1.47 (95% confidence interval 1.19–1.82)) and at 12 months (OR = 1.45 (95% CI 1.14–1.84)). Similarly, a high prognostic score increased the odds of Treatment Success at 6 months (OR = 1.62 (95% 1.29–2.03)) and at 12 months (OR = 1.92 (95% CI 1.51–2.46)).

The performance of these models in terms of discrimination is presented in Table 3. The prediction of positive answer to the PASS and Treatment Success at 6 months showed statistically significant yet poor discrimination (AUC 0.62 and 0.67, respectively). At 12 months the prediction of positive answer to the PASS and Treatment Success showed poor to acceptable discrimination (AUC 0.64 and 0.73, respectively).

The calibration of the models is presented using calibration curves (Fig. 2) showing the performance of the models for PASS and Treatment Success at 6 and 12 months.

The results of the univariate regression analyses used to assess the prediction of the continuous outcomes are given in Table 4. The prognostic score was statistically significantly associated with all the dependent variables at both 6 and 12 months, although with limited explained variance (R² < 0.12; Table 4).

Discussion

In summary these findings support the hypothesis, that the prognostic scores of future outcomes made by physiotherapists attending patients undergoing KA in the first days postoperatively associate with the 6- and 12-month outcomes of KA. The discriminative performance can be considered poor to acceptable. However, the AUC and ROC curve analyses did not suggest a cut-off value that may be used to screen patients. Although the calibration plots suggest a linear relationship between the predicted and observed probability of PASS and Treatment Success there are significant uncertainties as judged by the 95% confidence intervals. From Fig. 2 is can be seen that the uncertainties (width of the 95% CI) are related to the distribution of the predicted probabilities. This is likely due to the small number of patients with low prognostic scores (Table 1). The linear regressions show that the physiotherapists’ prognostic scores are associated with the 6- and 12-months outcomes. Altogether the data suggest that the physiotherapists’ prognoses associate with the outcomes, but the predictive values of the scores are not convincible. Thus, the physiotherapists’ prognostic scores should be supported by other information and individual assessments (such as age, sex, body weight, post-operative pain and mobility, comorbidities etc.), when the future course of a patient is discussed and decisions about possible extra attention is taken. Further studies on defining patients that may not have successful outcomes of KA are needed.

The present study is the first on physiotherapists’ prognoses of KA. Previously, it has been shown that physiotherapists’ prognoses during the initial clinical encounter for the projected outcome of patients with low back and neck pain they treated, was associated with the actual clinical outcome (Cook et al., 2015).

A strength to the present study is the use of outcomes widely accepted for assessing outcomes of KA (Conner-Spady et al., 2018; Dunbar et al., 2001; Garratt et al., 2004; Jin et al., 2019). The total mean OKS score of the cohort, is comparable with the ones found in other studies (Williams et al., 2013a; Williams et al., 2013b), which support the external validity of the results. Also, the relatively large cohort provides some strength to the study. Further, the fact that the physiotherapists who made the prognoses were not involved in delivery of any post-operative rehabilitation is a significant strength as they could not influence the rehabilitation. Finally, it is a strength that the study was carried out on two independent hospitals in Denmark, which strengthen the generalizability of the results.

The study also has some limitations. Firstly, other types of information (such as pain intensity and functional disability, body weight, comorbidities etc.) from the pre-operative and first post-operative days could have been useful in identification of potentially modifiable factors (such as body weight, muscle strength, functional disability, adjustment in treatment of comorbidities) that associate with the prognoses. Such information could have helped identifying the underlying factors that the physiotherapists based their prognoses on and used to propose future interventions to mitigate negative outcomes. Further, the distribution of the prognostic scores were skewed, with few low scores, which probably reflects the general successfulness of KA and that the physiotherapists in general expect positive outcomes. Nevertheless, about 15% of the patients did not reach PASS or Treatment Success after 12 months and could be candidates for special post-operative attention . In this perspective, the physiotherapist prognosis may be used to inform the identification of patients who may benefit from more intensive post-operative care in order to enhance the outcome of KA. However, the present data suggest that such identification should not rely solely on the physiotherapists prognoses but should incorporate other sources of information. Also, the instruments used to assess treatment success and patient satisfaction (OKS and PASS) does not necessarily cover all aspects, as numerous factors determine these. An OKS score above 32.5 (treatment success) does not necessarily indicate success for the patient and the single yes/no PASS answer is not exhaustive. Several instruments to assess patient satisfaction after KA exist (Kahlenberg et al., 2018), yet there is no consensus on which instrument to use to assess treatment success or patient satisfaction. However, these instruments (OKS and PASS) are widely used in OA and KA clinical and research settings.

Conclusions

In conclusion, physiotherapists prognoses of recovery after KA are associated with 6- and 12-months patient reported outcomes and satisfaction but have weak predictive value. This study suggests that physiotherapists’ prognoses may be useful as an additional source of information when identifying patients in need of additional post-operative care.

Supplemental Information