Keywords ACL - reliability - universal goniometer - inclinometer - smartphone app
Knee range of motion (ROM) is a common clinical measurement used to monitor the progress of patients after an anterior cruciate ligament (ACL) injury, as one out of three patients will experience loss of knee extension ROM at 12-month follow-up after this surgery.[1 ] This deficit could be attributed to multiple of reasons including, but not limited to, inflammation and swelling, surgical problems (e.g., cyclops lesion, graft impingement of suboptimal placement, intra-articular loose bodies), persistent muscle guarding, and pain inhibition.[1 ]
[2 ] Inadequate knee extension ROM postsurgery is associated with gait defects, altered biomechanics, knee osteoarthritis, impaired quadriceps function, and patient-reported outcomes.[2 ] All the above highlights the importance of identifying knee extension deficits in ACL patients as early as possible during the rehabilitation process.
In daily practice, objective, reproducible, and reliable measurements of clinical outcomes, such as knee extension, are paramount when documenting the treatment efficacy of any rehabilitation intervention. Consequently, the reliability of the tools used should be established so that clinicians can determine whether differences during repeated testing are related to measurement error of the device or represent actual changes in the outcome itself.[2 ] Health care practitioners have used several devices for measuring knee extension ROM, as well as different positions (e.g., supine and prone). The most common device is the universal (long-armed) goniometer.[3 ] Recently, an increasing number of clinicians report the use of inclinometers and smartphone apps to measure ROM.[3 ]
[4 ]
[5 ]
[6 ]
[7 ]
[8 ] Several studies have been conducted investigating reliability and minimal detectable change (MDC) for these devices with the results ranging widely from poor to excellent.[9 ]
[10 ]
[11 ]
[12 ]
[13 ]
[14 ]
[15 ]
[16 ]
[17 ] In these studies, the populations varied, with most recruiting either patients with total knee arthroplasty or healthy participants. The heterogeneity of these results and the absence of studies conducted on ACL patients preclude recommendations from being made in this population for whom knee extension ROM is an important clinical measure.[18 ]
[19 ]
Another key measurement property that has direct clinical implementation is the MDC. The MDC reflects the smallest change in measurement that is beyond the measurement “noise” and therefore is the minimum amount that needs to be overcome before ascribing any “real” change in a measure to a change in patient status.[20 ]
As neither the reliability (inter- and intrarater) nor the MDC is available for these methods of measuring knee extension ROM in a population of ACL-injured patients, the primary aim of this study is to describe the intrarater, interrater, and test–retest reliabilities of the goniometer, inclinometer, and smartphone application for measuring the knee extension ROM in ACL patients. Additionally, a secondary objective of this study is to establish the MDC associated with these methods, which will better inform clinical choices in ACL-injured patients.
Methods
The current observational cohort study was conducted at the assessment unit of Aspetar Orthopaedic and Sports Medicine Hospital, Doha, Qatar. It was a single-visit design study and ethical approval was obtained by the Aspire Zone Foundation Institutional Review Board (E202202031).
Participants
Participants of this study were 92 patients undertaking rehabilitation after ACL injury referred to the assessment unit, as part of routine testing between February 2021 and August 2022. Patients were included if they were treated conservatively or surgically after an ACL injury. A pilot study using 20 participants was performed in February 2021, to estimate the sample size for the current study. Prior to the pilot study, two training and familiarization sessions of 2 hours each were performed by the two testers. This pilot investigation showed an intraclass correlation coefficient (ICC2,1 ) value of ∼0.85, 0.90, and 0.90 for the universal goniometer, inclinometer, and smartphone app, respectively. The subsequent a priori power analysis (testing the null hypothesis H0 : p = 0.8 ICC vs. H1 : p > 0.8 ICC) suggested a minimum of 50 participants were required to have an observed power of 80% (α = 0.05) with two measurements per participant.[21 ]
Procedures
The knee extension measurement is part of the mandatory periodic assessment for patients with ACL injuries undertaking their rehabilitation at our facility. The routine testing protocol includes, in order of execution: clinical testing, instrumented laxity test, functional (movement) testing, and isokinetic strength testing. Participants were informed regarding the nature of the reliability study, and if they consented, we performed the additional knee extension measurements during the assessment visit ([Fig. 1 ]). Two independent, experienced physical therapists (M.P. and R.K.) performed the knee extension measurements in a single session. To prevent carryover effects (i.e., systematic effects of one test on the other), the test sequence was randomized.
Fig. 1 Flow chart of the study design.
At the beginning of the routine testing protocol, the patients undertook three knee extension measurements with three different devices in a randomized order: universal goniometer, inclinometer, and smartphone app. One examiner (M.P.) performed two measurements with each device to determine the intrarater reliability. Between the two measurements, the examiner had to replace the instrument to storage and repalpate the anatomical landmarks. A second examiner (R.K.) performed one measurement with each instrument to determine the interrater reliability. At the end of the entire routine testing protocol, participants were measured again by tester 1, to evaluate the test–retest reliability.
Results were recorded separately by an administrative assistant who ensured each tester was blinded to the results of other; however, the practitioners were not blinded to the measurement method.
Knee Extension Measurements
Participants were supine with their heels resting on a 10-cm thick box and asked to relax such that their knee fell into passive knee extension without any overpressure.
For the inclinometer measurement, the Empire Magnetic Polycast Protractor with 1 degree markings was used. This device operates with the assistance of gravity, eliminating the need for resets between measurements. The edge of the base of the device was placed 5 cm inferior to the tibial tuberosity, at the edge between the muscle belly of the peroneal muscles and the tibia. Gentle pressure was applied, so the base of the device would be stable and as parallel as possible to the anterior edge of the tibia ([Fig. 2 ]).
Fig. 2 Demonstration of the three measuring devices for knee extension deficit. (A) Inclinometer, (B) smartphone app, and (C) universal goniometer measurements.
For the smartphone app measurement, we used the “iHandy Level Free” app (Android and iOS versions were in common use in the rehabilitation department). The application was calibrated at a stable and even surface prior to each measurement using a spirit level. During testing, the phone was placed on the anterior border of the tibia, medial to the peronei, 5 cm inferior to the tibial tuberosity, in the same position as the inclinometer. The phone was always facing the same direction and the same side of the phone was touching the limb of the patient to minimize any differences in measurement due to the orientation of the gyroscope of the device ([Fig. 2 ]).
For the universal goniometer measurement, a Baseline 360-degree plastic goniometer was used, with 1-degree markings. Two transparent plastic 30 cm rulers were aligned and attached to the arms of the goniometer to extend its length and facilitate accurate positioning. The proximal arm was placed at the greater trochanter of the hip and parallel to the longitudinal axis of the femur. The axis was placed at the lateral femoral condyle. The distal arm of the goniometer was placed with its end over the lateral malleolus ([Fig. 2 ]).
Statistical Analyses
Descriptive statistics of the participants were calculated for age, height, weight, gender, and sport performed. Intrarater and interrater ICC2,1 and their 95% confidence intervals (CIs) were calculated. Two-way mixed approach with absolute agreement was used for the intrarater, interrater, and test–retest ICC analyses. An ICC ≥0.90 was described as “excellent,” between 0.75 and 0.89 as being “good,” between 0.50 and 0.74 as “moderate,” and below 0.50 as “poor.”[22 ] Additionally, MDC with 95% CI (MDC95 ) was calculated for the three different methods. To examine systematic variability between the measurements, Bland–Altman plots and joint plots were constructed and analyzed for each of the possible comparisons ([Supplementary Material ]). All analyses were performed using SPSS v.28 (IBM Corporation, Armonk, NY) and JMP v.16 (SAS Institute Inc., Cary, NC, 1989–2023).
Results
There were 84 male and 8 female participants. The mean age of the participants was 29.5 years (standard deviation [SD]: 9.4 years, and range 17–52 years). Their average height was 174 cm (SD: 8.3) and the average weight was 81.8 kg (SD: 16.5). Forty athletes were participating at a competitive level and 52 recreational athletes were included.
Seventy-two patients were evaluated at an average of 6.0 months (SD: 4.7) following ACL reconstruction, while 20 conservative/preoperative patients were assessed at an average of 4.4 months (SD: 3.9) postinjury. Out of the 72 patients who were treated surgically, there were 48 with hamstring graft, 18 with bone-patellar tendon-bone graft and 6 with quadriceps graft. Twenty-five patients underwent meniscal repair, 17 underwent meniscectomy, and 30 did not require any meniscal intervention. Among the six patients with cartilage injuries, only one underwent surgical treatment as determined by the treating surgeon.
Intrarater ICC
Intrarater reliability was excellent for all three methods ([Table 1 ]) ranging from 0.92 to 0.94.
Table 1
Intrarater, interrater, test–retest ICC2,1 and MDC results
Universal goniometer
Inclinometer
Smartphone app
Intrarater ICC (95% CI)
0.93 (0.90–0.95)
0.94 (0.91–0.96)
0.92 (0.89–0.95)
Interrater ICC
0.36 (0.17–0.53)
0.80 (0.71–0.86)
0.79 (0.70–0.86)
Test–retest ICC (n = 75)
0.83 (0.74–0.89)
0.89 (0.84–0.93)
0.86 (0.79–0.91)
MDC95 intrarater (deg)
3.5
2.0
2.2
MDC95 interrater (deg)
10.4
3.7
4.0
MDC95 test–retest (deg) (n = 75)
5.4
2.6
2.9
Abbreviations: CI, confidence interval; ICC, intraclass correlation coefficient; MDC, minimal detectable change.
Interrater ICC
The interrater results differed between the three devices ([Table 1 ]). The inclinometer and the smartphone app had good interrater reliability with an ICC2,1 of 0.80 and 0.79, respectively. The reproducibility for the universal goniometer was poor with an ICC2,1 of 0.36.
Test–Retest ICC
Only 75 of the total 92 participants were able to perform the final (retest) measurement at the end of the assessment procedure due to personal time constraints. The test–retest reproducibility was good for the three devices, ranging from 0.83 to 0.89 ([Table 1 ]).
MDC95
The intrarater, interrater, and test–retest MDC95 values are shown in [Table 1 ]. The inclinometer had the smallest intrarater, interrater, and test–rest MDC values, followed by the smartphone app, and the universal goniometer. The latter displayed the worst results of these three approaches.
Mean Differences, Bland–Altman Plots, and Joint Plots
The mean differences between measurements for each of the nine conditions are presented in [Table 2 ] which shows no systematic error except for interrater reliability for the app (1.145-degree bias). Bland–Altman plots and joint plots revealed no variation in error across the ranges of motion observed ([Supplementary Figs. S1–S18 ]).
Table 2
Mean differences (and 95% CIs) for each of the comparisons along with the standard error and p -value for the difference
Comparison
Mean difference (deg) (95% CI, standard error of the difference)
p -Value (mean difference)
Intrarater inclinometer
−0.054 (0.096 to −0.205, 0.076)
0.4765
Interrater inclinometer
−0.036 (0.238 to −0.310, 0.139)
0.7965
Test–retest inclinometer
0.201 (0.450 to −0.047, 0.126)
0.1119
Intrarater app
0.001 (0.152 to −0.149, 0.076)
0.9886
Interrater app
−1.145 (−0.849 to −1.442, 0.150)
<0.0001
Test–retest app
−0.125 (0.135 to −0.385, 0.132)
0.3441
Intrarater goniometer
0.177 (0.484 to −0.129, 0.155)
0.2557
Interrater goniometer
0.511 (1.240 to −0.218, 0.369)
0.1685
Test–retest goniometer
0.387 (0.915 to −0.141, 0.267)
0.1500
Abbreviation: CI, confidence interval.
Discussion
In this study, the first to examine the reliability of measuring knee extension ROM in patients after ACL injury, we demonstrated excellent intrarater reliability for all three devices, suggesting that they can be deemed appropriate for clinical use when measurements are performed by a single individual. Conversely for the interrater and the test–retest, the reliability varied markedly, with the inclinometer and the smartphone app demonstrating good reliability, while the universal goniometer exhibited poor reliability. These findings should inform clinical practice.
Inclinometer
The inclinometer displayed the best intrarater, interrater and test–retest results among the three devices. Regarding the intrarater reproducibility, the results were excellent as well, similar to the other two devices. We hypothesize that the main reason for that was the minimal palpation that is required when using this device and the lack of any calibration requirement, as opposed to the smartphone app. Additionally, both testers in this study reported that the inclinometer was the easiest device to operate of the three.
This standard inclinometer has been used by several studies.[23 ]
[24 ]
[25 ] Unfortunately, direct comparisons of their results with our study could not be made since the methodology used, the statistical approach, and the target populations were different. Specifically, Maltais et al (2019) conducted a study on children with cerebral palsy, Reurink et al (2013) with acute hamstring injured patients, and Piva et al (2006) used patellofemoral pain patients. All the previously mentioned studies included measurements of the knee joint, yet none of them assessed knee extension in supine.
The intrarater results of Maltais et al (2019) were slightly lower than ours (ICC: 0.87). This can be attributed to the longer period between the two measurements of the same tester, on average, 5.4 days, whereas in the current study, the second measurement was repeated with the patient in the same position only 30–60 seconds after the first measurement. The interrater ICC was slightly higher compared with our study, with an ICC of 0.86. Short time intervals between the measurements (5 minutes between the two testers) and standardization of the patient's position could be a possible explanation since a second physiotherapist was helping to stabilize the pelvis of the subjects. It has to be mentioned from a methodological perspective that the shorter the time interval between the measurements and the similar patients' position across the measurements can positively affect the ICC results.[14 ]
[26 ] In the study of Reurink et al (2013), the number of participants and the population characteristics were similar to our research, albeit hamstring rather than ACL-injured patients. Only interrater ICC values were calculated, 0.77 for the injured limb and 0.69 for the uninjured limb. The time intervals between the measurements were not mentioned, and the testing position of the patient was different compared with our study though, consequently direct result comparisons could not be made. Finally, Piva et al (2006) performed four different lower limb measurements with an inclinometer in patients with patellofemoral pain syndrome. Apart from the Craig's test, where the interrater ICC was rated as “poor,” 0.45 (95% CI: 0.10–0.70), for the other three tests, the interrater ICC scores were excellent, ranging from 0.91 to 0.97, slightly higher than the results observed in the current study, perhaps attributable to the patient's position and the different statistical approach used.
Smartphone App
The second approach used in this study was the smartphone app “iHandy.” The intrarater, interrater and test–retest results were slightly poorer compared to the inclinometer. After using the smartphone app extensively internally, we concluded that the sensors of the smartphone devices are sensitive to movements in cardinal planes.[5 ] Consequently, even the slightest movement when transferring the device from the table where we calibrated the device to the patient's limb could affect the results, likely explaining the slightly worse results compared with the inclinometer. Future research may examine whether improved training or modifications to the apps could mitigate this. Additionally, during the study, recalibrating the device after each measurement was time consuming, and aligning it to the tibia proved challenging due to the protective phone cases used by the testers, unlike the inclinometer. Moreover, the smartphone app was the only device where we observed a statistically significant interrater mean difference (1.4 degrees). We noticed several measurements which were almost exactly the same in absolute terms but opposite sign, for example, 4.5 degrees tester 1, −4.3 degrees tester 2. We suspect that operator error may be to blame as the smartphone app displays a numeric result only which makes it relatively easy to mistakenly record a small flexion value for slight hyperextension or vice versa. Finally, while this was the least expensive option (if the physiotherapist already owns a smartphone) with the app being free, these results should be weighed when considering clinical implementation.
Knee extension was measured using a smartphone app in two studies.[11 ]
[13 ] Different instrument placement, smartphone apps, patients' position, and active rather than passive knee extension were used across those studies, so direct comparisons with our results could not be made. The intrarater ICC results ranged from 0.94 to 0.96, being similar or slightly higher compared with our study.
Pereira et al (2017) reported an interrater ICC of 0.10 which is markedly worse than the results reported here. The authors of this study note that the presence of pain during the measurements could be an explanation for the poorer reliability observed. In our study, the patient was lying supine, and the knee was extended passively without any overpressure. Pain was not reported by any patients in this study, nor have we observed this in routine clinical practice.
Universal Goniometer
For the universal goniometer, while the intrarater reproducibility was excellent, the interrater ICC results for the goniometer were the lowest of the three devices and were characterized as poor. This can be attributed to the difficulty of accurately palpating the anatomical landmarks, especially in patients with a large soft tissue mass at the lateral hip. Such difficulties in palpation can influence the reliability scores, as even the slightest misplacement of the instrument on the anatomical landmarks can lead to measurement discrepancies.[11 ]
[13 ]
[14 ] Finally, both testers expressed a preference for the other two devices due to less time required to perform measurements and easier handling compared with the goniometer.
Previous studies measuring the knee extension showed similar results for the intrarater reproducibility with ICC scores ranging between 0.82 and 0.99.[9 ]
[11 ]
[13 ]
[16 ]
[27 ]
[28 ]
[29 ] Only in the study of dos Santos et al (2012), the intrarater reliability was significantly lower (ICC 0.49) which was likely a result of the time between the two measurements being the longest at 7 days apart.
The interrater ICC results of the study of Pereira et al (2017) and Peters et al (2011) were lower than ours with reported ICC of 0.05 and 0.21, respectively. In the study of Pereira et al (2017), the measurements were performed only a few days after the surgery (8.5 ± 7.4 days, PO group) which could explain the lower interrater scores since pain was reported during the measurements. The results of the healthy (likely pain-free) population of their study (HS group) were similar to our results with an interrater ICC of 0.40. Finally, the studies of Brosseau et al (2001), Lenssen et al (2007), Pandya et al (1985), dos Santos et al (2012), and Verhaegen et al (2010) used similar patient position and methodology to our study. Still, they reported better results with an interrater ICC ranging between 0.55 and 0.93. For the studies of Lenssen et al (2007), dos Santos et al (2012), and Verhaegen et al (2010), we are not aware if the patient changed position between the measurements of the two testers or remained at the same position which could have affected positively the ICC values. Finally, in the study of Pandya et al (1985), the measurements were recorded to the nearest 5 degrees, affecting the precision of the measurement and the ICC results consequently.
MDC95
MDC95 values are perhaps more clinically important than correlation coefficients as they represent the minimal level of change in the measurements that can be attributed to a “real” change in ROM beyond measurement error of the device. In this study, the inclinometer exhibited the smallest MDC95 for all three measurements. The smartphone app exhibited the second smallest intrarater, interrater and test-retest MDC95 values, 0.2, 0.3 and 0.3 degrees less than the inclinometer, respectively. The results of the universal goniometer were lower, especially for the interrater value which was 10.4°. Clinically, this device is only of use in patients with severe knee extension deficits.
Test–Retest—No “Warming-up” Effect Seen
Another purpose of this study was to investigate if performing the measurements at the end of the assessment routine would affect the knee extension measurements, that is, was there any “warming-up” effect on knee extension flexibility. For this reason, a third measurement was performed at the end of the assessment. The number of the participants was smaller (n = 75) as some patients were unable to remain after all the standard testing was completed due to personal time constraints. Overall, the test–retest reliability was good but slightly lower than the intrarater results. Importantly, there appeared to be no bias from the tests done at the start of the assessment routine compared with the end. We had postulated that the warming-up effect of the testing process (which includes a formal cycling warm-up, isokinetic knee strength testing, jump testing, etc.) would result in increased knee flexibility at the end of the session, which was not the case. From a clinical perspective, the MDC values were 0.6 and 0.7 degrees higher for the inclinometer and the smartphone app and 1.9 degrees for the universal goniometer. This deviation, especially for the inclinometer and the smartphone app, are relatively small which suggests that measurements of knee extension can be performed at any time during an assessment session with relatively minor effects on the ultimate result. Where accuracy is critical, it is suggested that the order of tests should be standardized.
Limitations
The recruited patients were mostly adult males which reflects our clinical caseload in this consecutive series, but we should be cautious in extrapolating these results to females and adolescents. Additionally, this study was a single-visit study and the time intervals between the measurements were short, therefore tests done at longer intervals apart cannot be compared. It is crucial to consider this aspect when extrapolating the findings to clinical setting where multiple measurements on the same patient may be taken days or weeks apart. It should be noted that we modified the universal goniometer with the addition of plastic extensions to facilitate placement over the proximal and distal landmarks, yet this was our least accurate measure. We suspect that without this addition, aligning the goniometer arms visually would result in worse reliability again. Furthermore, it has to be mentioned that the testers used two different smartphone devices with different operating systems (Android and iOS) which could influence these results.[5 ]
[30 ] Finally, it was infeasible to blind the participants and testers to the measurement methods, but this may have influenced the results.
Conclusion
The inclinometer and the smartphone app displayed the best intrarater and interrater reliability. Although the smartphone app achieved similar results, it is suggested the device of choice is the inclinometer as it is inexpensive and was reported to be easier to use than the universal goniometer or the smartphone app. In a clinical setting where only one practitioner will conduct the measurements, all three devices had satisfactory results.
