The goal of posture assessment in GPR is to determine muscles that are responsible for the posture alterations and to plan treatment consisting of stretching postures (for anterior or posterior muscles) to increase muscle flexibility and sensorimotor integration exercises to correct posture. In our study, we investigated inter-rater reliability of the muscular chain evaluation associated with posture alterations among physical therapists and assessed agreement with experts in GPR.
We found a moderate to substantial level of reliability for 12 out of 23 posture indices and a good level of agreement with the two experts in GPR for 18 out of 21 posture indices for the muscular chain assessment. The muscular chain impairment associated with posture could be determined for the majority of indices. However, there was a low level of reliability among the PTs for muscular chain assessment associated with head lateral bending, head rotation, increased cervical lordosis, protracted shoulder and knee flexum. In line with these findings, muscular chain impairment associated with posture alterations was not determined for head lateral bending, head rotation and knee flexum. These results are corroborated by the low level of reliability of these indices in our study and in previous reliability studies on visual observation of posture
]. Except for knee flexum index, our experts (before consensus) had similar percentage of agreement results as those reported by Watson and MacDonncha
] for qualitative observation of posture indices. To our knowledge, no previous study has reported the psychometric properties of muscular chains evaluation associated with posture alterations. Moreover, no true “gold standard” criterion could be used to assess the validity of this concept. However, the good level of agreement between the PTs and experts may reflect the uniform standards taught in GPR.
The poor levels of inter-rater reliability may be attributable to the importance of the 3D component needed to assess these posture indices which is not really possible from photographs (even when using different views like in this study). The magnitude of the posture alterations may be another factor. It is possible that some posture alterations were too discrete to be visually identified. For example, head lateral bending is always associated with some degrees of flexion or extension and rotation
]. It might be confusing to determine which component (lateral bending, flexion, extension or rotation) is present when alteration of head position is small and therefore difficult to select the responsible muscular chain. Head lateral bending and rotation can be attributed to anterior muscles such as scalenius, SCM or posterior muscles such as upper trapezius, levator scapulae and cervical erector spinae
]. PTs trained in GPR can compare posture alterations in the standing and “long sitting” positions to determine whether anterior or posterior muscles are implicated in the posture alteration
For knee flexum, the two experts could not reach a consensus and a third expert had to make the final decision. One expert drew a line to assess this posture index while the other estimated it visually. Differences in their methods combined with discrete posture alterations among the youths for this index may explain the discrepancy between the experts as well as the low and negative kappa coefficients reported respectively for muscular chain and posture evaluation. It may also explain the difficulty to determine inherent muscular chain impairment.
The lack of a clear definition of protracted and rounded shoulder may also explain poor inter-rater reliability among the PTs and the experts. This inconsistency is confirmed by the negative level of agreement between PTs and experts found in our study for rounded shoulder evaluation
]. In both posture alterations, the shoulder (acromion) can appear forward but muscle implication is different
]. The protracted shoulder is associated with shorter pectoralis minor muscle whereas rounded shoulder is caused by retraction of pectoralis major muscle and/or serratus anterior muscle
]. These two different concepts are often used in an interchangeable way or are not well defined in the literature
We found that more experienced physical therapists (Group 3) had better level of reliability for muscular chain evaluation. This is in contrast with previous studies on visual observation of posture or of gait analysis who reported that inexperienced raters achieve a comparable level of reliability than more experienced raters
]. The PTs in Group 1 and Group 2 had similar results. Group 2 was the most heterogeneous group in terms of nationality: it is possible that more discrepancies exist between physical therapists trained in different countries.
We used photographs to assess posture. As already mentioned, photographs are a 2D perspective of a person and PTs are used to doing bony palpation when they want to validate their visual observations. However, our main goal was to assess muscular chain impairment associated with the posture alterations. Thus agreement for muscular evaluation was verified for PTs that identified the same specific posture alteration. Another limitation concerns the number of categories for each posture index and for muscular chain evaluation in our assessment scale. Increasing the number of categories in a measurement scale decreases the Kappa coefficients
]. The analysis for the posture indices was done by combining two to four elements for each of the 23 indices. For example, in the case of knee flexum index, the PT had four choices on the grid (right knee flexum – R, left knee flexum – L, R
L or R
L; see Table
). In order to have perfect agreement, all the physical therapists would have had to choose all of the same choices (2 to 4 choices) for each of the indices. When we compare the grid for the two experts, there are only a few differences. For the 79 items (representing the 23 posture indices) the two experts are in perfect agreement for 59 of the 79 items; there is a difference for one youth on 17 items and on 3 items for two youths. Thus we have been much more conservative by choosing to analyze the grid by 23 posture indices instead of 79 individual items, and our coefficients are considerably lower as a result. Moreover, Kappa coefficients are less favourably influenced by a large number of raters than are ICCs. The Kappa coefficient is a conservative measure since it eliminates agreement by chance. This explains a Kappa coefficient of zero even if the two experts agreed for three youths out of five and had a percentage of agreement of 60% (Table
). The small sample of youths included in this study is also a limitation since Kappa coefficients are more favourably influenced by sample size magnitude than by large number of raters
]. Moreover, some posture alterations were not present among these youths and thus muscular chain impairment could not be determined.
Clinical applications and recommendations
We found that the muscular chain evaluation (done by PTs trained in GPR) is reliable for most posture indices among youths with idiopathic scoliosis and for the most part, there was good agreement with experts in GPR. This kind of assessment may have considerable diagnostic and therapeutic utility in physical therapy practice as it guides the understanding of muscular impairment associated with abnormal posture patterns. It may assist the physical therapist in the selection of anterior and/or posterior muscular chain stretching exercises to improve posture and increase quality of life
Some adjustments in the teaching of muscular chain evaluation are necessary to improve agreement between physical therapists for less reliable posture indices. We suggest a better definition of certain concepts such as protracted and rounded shoulder because muscle impairment and treatment will be different in these cases. We also recommend standardizing the teaching method of posture and muscular chain assessment between instructors in different countries. Future studies are still needed to document if these qualitative observations can be verified with standardized quantitative tests to assess muscle flexibility.