Background
Passive range of motion (PROM) measurement is an essential tool for pediatric orthopedists to document disease progression, to plan treatment and to evaluate its results. While many methods are available for ROM evaluation, most physicians consider goniometry as the gold standard and its validity is currently widely accepted especially when measurements are taken by the same examiner within the same session and on the same day [
1‐
16]. Passive ROM measures are generally thought to be more reliable in individuals who have normal tone than in individuals with hypertonicity [
11‐
15]. Kilgour et al [
15] found that sagittal plane ROM goniometric measures have similar levels of reliability in children who have spastic diplegia and in control children, both within and between sessions. Validity and reliability of goniometric measurements were well studied in the paper of Gajdosik [
16], concluding that the universal full-circle goniometer is the preferred instrument for measuring ROM.
However, in regular daily practice, the majority of clinicians rely on visual estimations (VE) of angles, the reliability of which is yet to be established. In a wide search of the current literature, we found few studies assessing the validity of this method in evaluating the PROM of one or two joints of the lower limb of normal subjects [
6,
9].
We studied PROM of hips, knees and ankles in a healthy pediatric population.
The aim of this study is threefold:
-
To assess the validity of visual estimation
-
To study the reproducibility of the results (intra-tester reliability)
-
To evaluate the influence of experience (inter-tester reliability)
Discussion
This study brings up the advantages and limitations of visual estimation of PROMs in the pediatric lower limb.
This is to our knowledge the first study evaluating reliability of VE for hip PROMs. We found high level of accuracy of VE for hip flexion and rotations, and good accuracy for hip abduction, but a lack of reproducibility of measurements performed by a less experienced examiner. Concordance was poor for hip adduction (ρc ≈ 0.5). Many factors may have contributed to these poor results of VE of hip adduction: the examination technique with the hip flexed and the low absolute values of hip adduction compared to flexion, abduction and rotations. The level of experience was found to be important in estimating hip PROMs other than flexion.
Marks et al [
26] investigated the reliability of VE of knee ROM taken by three physicians on patients with rheumatoid arthritis, who examined the patients independently. One could assume that each examiner may have applied a different force to move the joint, theoretically leading to bias in interpretation of VE. Despite this possible bias, the authors found good intra- and interexaminer reliability. In order to avoid a similar bias in our study, PROMs for each joint were performed once by the same examiner.
Watkins et al [
10] studied passive flexion and extension of 50 knees. For each tested knee, two physical therapists were chosen randomly from a list of 14 examiners. In this study as well, estimations were done separately by each examiner in a position of his choice. Visual estimations were performed before goniometric measurements for each joint. We believe that this may influence subsequent visual assessment by adjustment of visual estimations. In our study, the two examiners who did visual estimation never knew goniometric values of PROMs. Although the ICC (Intraclass Correlation Coefficient used by Watkins et al [
10] is less reliable than the CCC (Concordance Correlation Coefficient) [
21] used in our study, we found similar good concordance for knee flexion and high concordance for popliteal angle. Therefore, our results are similar to those reported by previous studies [
10,
27] which found good accuracy for visual estimation of knee PROMs, with a tendency to slightly underestimate real values. Experience does not seem to play a major role in VE of knee PROMs. Even unexperienced examiners may satisfactorily estimate knee motions without using a goniometer.
Youdas et al [
7] evaluated active range of motion (AROM) of 45 ankles (dorsiflexion and plantar flexion). The study included 10 physical therapists who performed measurements with their own preferences regarding position and method. They assessed interobserver reliability for visual estimation and intraobserver reliability for visual estimation and goniometry. AROMs are much more subject to variations than PROMs. The authors found low concordance between VE of different examiners (ICC = 0,34 for dorsiflexion and ICC = 0,48 for plantarflexion) and a fair concordance between visual estimation and goniometry (0,58 for dorsiflexion and 0,625 for plantarflexion). We believe that these low values are mainly due to the fact that AROMs are much more subject to variations than PROMs and to the absence of standardization of goniometric measurements as recognized by the authors themselves.
Allington et al [
2] assessed intra- and interobserver reliability and reproducibility of goniometry and visual estimation of ankle PROMs in 24 children with spastic cerebral palsy (46 ankles). Two physical therapists performed all the measurements. They found very good correlations between the goniometer and the naked eye (r > 0,94) for ankle dorsiflexion and plantarflexion. They also found a mean error (ME) of 5° with a SD of 5° in the inter- and intraobserver measurements. Thus, even with very good reliability, there is a significant error to be taken into account when performing visual estimates. In our study, we found good level of concordance between the specialist's VE and goniometry for ankle PROMs (ρ
c ≈ 0.7). We do not have an explanation for the very high concordance observed in the study of Allington et al [
2] but we believe that this may be partly due to the small number of patients examined.
Disparity in results of VE of ankle PROMs between the different studies is due to many factors: the number of examined ankles, the maneuver technique and the method of goniometric measurement.
One of the major limitations of our study was the small number of examiners. It is obvious that by increasing their number with different levels of experience, more conclusions could be stated concerning the role of experience. Even though there were only two examiners performing visual estimation, we believe that results can be extrapolated to other experienced and less experienced examiners.
Validity and reliability of goniometric measurements were not verified in our study. This could have been done for example by an additional examiner doing another set of goniometric measurements, but we really think that this problem was thoroughly discussed and verified in the literature [
1‐
16] and we've deliberately chosen goniometric measurements as a reference to study more precisely visual estimation and the role of experience. Considering all measurements as variables to be verified would have certainly complicated the statistical analysis and weakened the conclusions. Based on literature statements and on our standardized technique for goniometric measurements, we can assume with some caution that these measurements were valid and reliable.
While performing the examination, we sometimes encountered lack of compliance especially with younger children, making measurements difficult to obtain. In addition, the force applied on the limb may vary for the same range of motion not only from one child to another, but also for the same child when repeated measurements are done. Different results could have been then obtained if the same motion was repeated by each examiner. This problem was discussed by Amis and Miller [
28] who explained that passive motions are difficult to reproduce because stretching of soft tissues at the limit of the motion depends on the force applied on the limb. Wagner [
29] found a greater variability when measuring passively motions of pronation and supination of the forearm. Kilgour (15) demonstrated that while some measurement error arises during the placement and reading of the goniometer, the majority of it is most likely related to difficulties in determining end-range joint positioning. Perhaps force dynamometers (including those that are hand-held) could be used to standardize the amount of passive force applied and thereby decrease the potential for error [
30]. We tried to limit the bias relative to these problems by excluding children of three years or less, and by maintaining the same position while doing the three measurements (two visual estimations and one goniometric measurement). But while this may enhance reliability by reducing error, it does not reflect clinical practice where examiners perform ROMs separately. For this reason, we think that a patient should be followed by the same therapist to document disease progression or to evaluate treatment results.
While comparing the charts of measurements (visual and goniometric) of the two parts of our study, we noticed that for a given child, respective values of PROMs are different and this difference is sometimes up to 20 degrees for both measurements (visual and goniometric). This can be hardly explained by the sole visual or goniometric error and we are sure that a large part of variation is due to change in PROM itself. To avoid such an important bias, we tested intra-rater reliability by comparing respective concordance coefficients instead of direct comparison of visual estimations. We can also understand in this setting why it is misleading to study the reliability of goniometric measurements by comparing the two sets of measurements.
We had some difficulties with goniometric measurements. Short limbs were easier to measure because landmarks were closer to the goniometer's arms. This was especially true for motions around the hip and the knee. In retrospect, a longer armed goniometer would have decreased some of the problems. On the other hand, some bony landmarks were moving under the skin. Therefore, the greater trochanter was marked while flexing the hip to minimize variations.
We had some overweight children. Obesity can make estimations difficult to obtain. It may also hide some bony landmarks especially the greater trochanter. This is added to the fact that, naturally, identification of bony landmarks is more difficult in the lower limb [
31]. The AAOS [
32] and Rowe [
27] have suggested that visual estimation is more reliable than goniometric measurements when bony landmarks are not easily seen or palpated. However, overweight children were not excluded from our study to avoid the bias of selection.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
IG conceived of the study. RR, IG and IK performed the study and participated in its design. SEH participated in the design of the study and in the statistical analysis. KK and FD reviewed the paper and made corrections. All authors read and approved the final manuscript.