Abstract
Range of motion (ROM) is a well recognised outcome measure following total knee arthroplasty (TKA). Reduced knee flexion can lead to poor outcome after TKA and therefore identification at an early stage is important as it may provide a window for intervention with targeted physiotherapy, closer follow-up and in resistant cases possible manipulation or arthrolysis. ROM combines both flexion and extension and in contrast to flexion, fewer studies have recognised the importance of a lack of full extension or fixed flexion deformity (FFD) following TKA. A residual FFD can increase energy cost, decrease velocity during ambulation and result in pain with knee scores more likely to be diminished than if knee extension was normal. Recognition and early detection of FFD is therefore important. Methods of assessment include by visual estimation or goniometric measurement of knee flexion angle. While goniometers are inexpensive, easy to use and provide more accurate than visual estimates of angles, they have been shown to exhibit poor inter-observer reliability. Therefore they may not be sensitive enough to consistently identify FFD and therefore distinguish between grading systems based on absolute angular limits. The aim of this study was to investigate the accuracy of standard clinical ROM measurement techniques following TKA and determine their reliability for recognising FFD.
Ethical approval was obtained for this study. Thirty patients who were six weeks following TKA had their knee ROM measured. An infrared (IR) tracking system (±1°accuracy) that had been validated against an electro-goniometer was used to give a “true” measurement of the lower limb sagittal alignment with the knee fully extended and maximally flexed while the patient was supine. The patients were also assessed independently by experienced arthroplasty practitioners using a standardised goniometric measurement technique. For goniometric clinically-measured flexion (Clinflex) and extension (Clinext) linear models were generated using IR-measured flexion and extension (IRflex and IRext), BMI and gender as covariables. Data for extension were categorised in none, moderate and severe postoperative FFD as per Ritter et al. 2007 and agreement in classification between the two methods was assessed using the Kappa statistic.
For the linear models for Clinflex and Clinext neither BMI nor gender were significant variables. Therefore the final models were:
Clinflex = 0.54 + 0.66∗IRflex (r2adj = 0.521)
Clinext = 0.23 + 0.50∗IRext (r2adj = 0.247)
The model for Clinflex showed that the IR and clinical measurements coincided at approximately 90° so that for every 10° increase in flexion above 90° clinical measurement only increased by 7° but for every 10° decrease in flexion below 90° clinical measurement only decreased by 7°. The model for Clinext showed that the IR and clinical measurements coincided at approximately 0° so that for every 10° increase in FFD angle, clinical measurement only increased by 5° but if the knee went into hyperextension this would be underestimated by the clinical measure. In identifying FFD there was moderate agreement between the two measurements (κ = 0.44). Clinically nine patients were assessed as having FFD but the IR measurements showed 18 patients having FFD, of which nine patients were not identified clinically.
When assessing knee ROM following joint arthroplasty manual goniometric measurements provided a poor estimate of the range when compared to the “true” angle as measured with a validated IR measurement tool. When the knee was held in maximum flexion there was a tendency to both underestimate and overestimate the true angle. However when the knee was held in extension there was a tendency to underestimate which we believe is important as it would underreport both the frequency and magnitude of FFD. In our study, 18 patients had a moderate FFD as identified by the IR system, only half of which were identified by goniometer measurement alone. Studies of comparisons of both visual and manual goniometry measurements of the knee in maximum flexion with lateral radiographs have shown most errors involved an underestimate of true flexion. It has been concluded that it was safer to underestimate knee flexion angle as it would result in higher pick up rate of cases being performing less well. In contrast however, underestimation while in extension is less desirable as it fails to pick-up FFD which may have benefited from intervention had they been identified. It is known that residual FFD can increase energy cost and decrease velocity during ambulation with pain and functional knee scores more likely to be reduced. Recognition and early detection is therefore important. With the use of more accurate systems to identify and measure FFD, such as the one used for this study may in turn allow more timely treatment and therefore hopefully improved outcomes.
