Knee Replacement Recovery

Knee Replacement Recovery

Medical Research Council (MRC)

Associate Professor Melvyn Hillsdon from the Sport and Health Sciences department was awarded an MRC CiC Award in order to utilise accelerometer data from the Alternative Knee Alignment Study to provide novel insights into the effect of specific physical activity (PA) behaviours and their patterns, pre-surgery, immediately post-surgery and 12-weeks post-surgery, on patient related outcome measures (PROMs).

The challenge

Total knee replacement (TKR) is a common surgical procedure for people suffering severe osteoarthritis (OA) leading to improved physical function and reduced pain/disability. In theUK, nearly 100,000 primary knee procedures are performed per year. Even though TKR is one of the most successful operations performed in orthopaedics, for a significant number of patients, recovery is prolonged and improvements in pain/function are unsatisfactory. The additional treatment costs of patients presenting with pain after TKR is on average £5,136 per patient. Therefore, there is an imperative need to identify ways to improve recovery times and PROMs.

A central concern in OA patients awaiting surgery is that their low levels of PA and function increase their risk of comorbidities and poor PROMs. Pre-surgery PA/function and incremental changes post-surgery should intuitively improve PROMs, though the evidence for this is currently unclear. This may be due to a reliance on crude, self-report measures of PA that are subject to recall and social desirability bias. Such measures fail to capture the trajectory of changes in PA during the first few days and weeks after surgery that may reasonably be expected to be associated with short term PROMs. Body worn accelerometers are becoming ubiquitous in research and provide objective, repeatable and accurate data on movement behaviours. Translating raw acceleration data into estimates of specific PA behaviours that may be important to PROMs requires the development of algorithms that have analytic and clinical validity. Only when this is achieved can the true association between the level and pattern of specific behaviours with PROMS be examined.

What was done to help

• New algorithms were developed that translate accelerometer outputs into appropriate metrics for use in TKR research and patient care/monitoring to evaluate patterns of specific physical activities and functions that are important for successful prehabilitation and rehabilitation after TKR.
• The association between pre-surgery levels of physical activity and function on post-surgery levels of physical activity and function was examined.
• The association between pre-surgery levels of physical activity and function on 12-week post-surgery PROMs including physical function, quality of life and the patient assessed Knee Injury and Osteoarthritis Outcome Score was examined.
• The association between the trajectory of change in physical activity and function (21-days from day of discharge) on PROMs at 12-weeks post-surgery was examined.
• New, condition-specific measures were created that are meaningful and interpretable for new and existing users.
• Software applications were created that allow easy distribution and plug-and-play functionality for users.
• The data collection protocol was managed by Activity Informatics, a collaboration between the University of Exeter and Activinsights, providing data analytics services for human activity research.

The Results

For PA metrics, there was a consistent pattern between pre-surgery, post discharge and 12 weeks post surgery. Values were worse immediately post-surgery compared to pre-surgery, gradually recovering over the next 3 weeks. At 12-weeks post-surgery, values mostly continued to improve but never exceeded pre-surgery levels. This is consistent with the existing literature. The findings suggest that pre-surgery levels of PA are an important correlate of recovery.

Overall steps per day and low cadence steps per day showed an improving trajectory across the 21 days following discharge, steeper during the first week and continuing to improve at the 12 weeks post-discharge point. Again, levels never quite returned to pre-surgery levels by 12 weeks.

However, high cadence stepping falls considerably from pre-surgery levels and does not improve during the first 7-days post discharge. For the next two weeks there is a gradual increase with a further marked increase at 12-weeks. This slow increase in the number of steps, compared to other metrics, which reach a point of stability after 7-9 days, suggests that the mechanism behind the lower number of steps is different to the mechanism behind the other metrics which stabilise quicker.

All sleep metrics worsened immediately after discharge from hospital and were relatively unchanged during the first 3 weeks of recovery. All of the sleep metrics improved at 12-weeks post discharge, closely returning to pre-surgery levels – more so than the activity metrics.

Conclusion

The project generated a large number of digital biomarkers of recovery from total knee replacement, based on wrist worn accelerometery. It demonstrated that remote, continuous monitoring of behaviour, over an extended period, is feasible, can achieve very high levels of compliance and represents a low burden for both clinicians and patients. The Activity Informatics collaboration between Activinsights and the University of Exeter is a successful model of data collection, management and analysis for clinical trials.

The study provided a number of novel insights into the early recovery of people undergoing total knee replacement. In particular, the measurement of day-to-day changes in behaviour for the first 21-days following discharge from hospital has highlighted why simple self-report measures of PA or average day/week metrics hide important patterns of behaviour that could be beneficial for clinicians and patients in managing expectations as well as in personalised care plans. The fact that 21-day monitoring is feasible and can be achieved with high compliance strengthens the case for remote, continuous monitoring post surgery has been shown.