Clinical assessment of subacromial shoulder impingement – Which factors differ from the asymptomatic population? – Land, Gordan and Watt (2017) – A Critique
An interesting paper has just been published in the Musculoskeletal Science and Practice Journal (formally Manual Therapy) which looked at the clinical examination of people diagnosed with ‘subacromial shoulder impingement’ (SIS) to see whether there were any factors that differed within a matched asymptomatic population.
The authors should be congratulated on the successful completion and publication of what looks to be an extensive project; research is never easy and neither is publication! The execution of their chosen methodology was to a high standard resulting in a well-powered study.
However, I have a few issues with this paper:
1 – The methodology chosen of a case-control study answers their research question, but I would propose the wrong question is being asked.
2 – The use of the term ‘subacromial shoulder impingement’ – please can we as a community stop using this term!? Please!
3 – The population utilised may not be reflective of all patients diagnosed with such a condition.
4 – The assessment procedures utilised are not wholly reproducible in clinical practice and may not necessarily represent best or contemporary practice e.g. the calculation of postural angle utilising computer software.
And most importantly;
5 – As a clinician, has it told me anything that hasn’t already been shown extensively before?
The methodology chosen of a case-control study answers their research question, but I would propose the wrong question is being asked.
The authors report that those patients with SIS, when compared to the asymptomatic matched population, had significantly increased:
- thoracic flexion
- forward head posture
- upper thoracic active motions
- passive internal rotation range
- posterior shoulder range
Within a case-control study, these factors are associations and as is well documented, correlation does not equal causation; essentially meaning that this study cannot determine whether these factors are a cause of SIS, or are an effect of SIS – the classic chicken and egg scenario. To determine causation in its broadest sense, there is a need for a prospective cohort study i.e. a population of asymptomatic people are followed over time, those that develop symptoms are then compared to those that don’t, for identification of causative factors for the development of symptoms. The authors in this study try and elicit cause and effect through the use of conditional logistic regression analysis which showed no independent predictors of SIS; they hypothesised that this was due to the study being underpowered – I would suggest it’s more likely due to measuring the wrong variables!
The use of the term ‘subacromial shoulder impingement’ – please can we as a community stop using this term!? Please!
The validity of the acromial irritation theory has been questioned extensively (Lewis et al. 2011) and probably does not best explain the cause of the problem in this population of people with Shoulder pain. It is important that the terms we use reflect our understanding and that we adapt as new evidence emerges – recently the term ‘Rotator Cuff-Related Shoulder Pain’ has been proposed (Lewis 2016) and probably is most accurate at this time. Not only is it important that we use terms that best reflect our understanding, but we need to ‘mind our language’ and use terms that do not induce harm or avoidance behaviour; I’ve just written a paper with Dr Chris Littlewood that is currently under review that I think will shed further light on this within this population of patients when published.
The population utilised may not be reflective of all patients diagnosed with such a condition.
A small point but one that I find intriguing, part of the exclusion criteria in this study was that of any individual that had participated in shoulder strength training in the six months prior to entering the study – defined as high load upper body weight training two or more times per week. I agree that for the small majority of patients that I see with SIS or Rotator-Cuff Related Shoulder Pain (from this point forward I will only refer to Rotator-Cuff Related Shoulder Pain) they have not been loading their shoulder sufficiently to build strength or tendon capacity that may have prevented the onset of symptoms; those patients that may be deemed to have a ‘weak and painful shoulder’.
However, what about those people that see me with Rotator-Cuff Related Shoulder Pain (RCRSP) that have recently returned to the gym or recently started the gym for the first time and have done ‘too much, too soon’ and overloaded their shoulder? What about the regularly gym goer who may have increased their weight ‘too much, too soon’ or may have introduced a new exercise or pushed themselves too hard in a session? What about those rugby players or climbers that I see who could often bench press their body weight or at least hang their body weight off one arm; those patients that may be deemed to have a ‘strong and painful shoulder’. What does this study tell me about these people that I see? I suspect my point here is related to my second critique point and is a reflection of outdated reasoning as to what maybe the cause of the problem in this population of people with Shoulder pain.
The assessment procedures utilised are not wholly reproducible in clinical practice and may not necessarily represent best or contemporary practice e.g. the calculation of postural angle utilising computer software.
In order to measure cervicothoracic posture, thoracic posture and active motions of the upper thoracic spine, the authors utilised a tripod mounted camcorder to take photographs as well as placing skin markers upon the participants before measuring such variables using digital software. Whilst the authors took lengths to calculate and demonstrate that this was a reliable assessment procedure, I question the clinical utility and therefore application of such an approach.
The authors here have stated that they have measured both ‘posterior shoulder range’ and ‘passive internal rotation range’, whilst I understand that these can be seen as two separate measures, I would suggest that these are in fact measurements of the same variable and thus expect that if a patient has reduced ‘posterior shoulder range’ as measured by the Tyler Method that they would also have reduced passive internal rotation range and vice versa (Walton and Russell 2015).
It has previously been considered that a deficit in glenohumeral internal rotation (GIRD) needs addressing (Cools et al. 2012) however, more recently best practice would dictate that internal rotation measurement needs to be interpreted in relation to external rotation available and compared to the opposite side. This is due to measurement of internal rotation being influenced by humeral torsion, but also in certain populations (for example throwers), the shoulder adapts by increasing the amount of external rotation range available leading to a shift in the arc of rotation and naturally an apparent ‘reduction’ in the amount of internal rotation movement available when measured; in such a situation, the total range of motion is maintained. If there isn’t a concomitant increase in external rotation with an apparent reduction in the amount of internal rotation movement available, this can be deemed a total range of motion deficit or ‘TROMD’ (Manske and Ellenbecker 2013; Wilk, Macrina and Arrigo 2012; Wilk, Hooks and Macrina 2013).
What new information has this study told me as a clinician that hasn’t already been shown extensively before?
It has told me that in RCRSP, static scapula postures probably don’t play a role in the development of symptoms. However, stronger, (including prospective data) has already told me this with greater confidence as we know already that static posture doesn’t tell you how the scapula is going to move, and how the scapula moves doesn’t appear to correlate with the development of symptoms (Morais et al. 2013; Ratcliffe et al. 2013); I suppose this paper adds further support to this notion but it is hardly a hammer blow to the scapula!
“Currently, there is insufficient evidence to support a clinical belief that the scapula adopts a common and consistent posture in SIS. This may reflect the complex, multifactorial nature of the syndrome.” – Ratcliffe et al. (2013).
It has told me that RCRSP is associated with a thoracic kyphosis or forward head posture; it has not told me that it is a causative factor. However, stronger, systematic review data has already told me that a thoracic kyphosis does not appear to be an important contributor to shoulder pain (Bartlett et al. 2016) but an increased kyphosis may limit the amount of elevation available at the shoulder which in turn may have implications for certain occupations that require prolonged use of their arms overhead; although the systematic review authors do suggest further research is required.
It has told me that RCRSP is associated with posterior shoulder tightness; it has not told me whether this is a cause of RCRSP or an effect of RCRSP; this association has already been widely documented (Dashottar and Borstad 2012; Gates et al. 2012; Bach and Goldberg 2006). Due to limitations in the measurement method (Manske and Ellenbecher 2013), I do not know whether there was in fact posterior shoulder tightness as the amount of passive external rotation available is not documented to allow me to determine if indeed a TROMD exists.
To conclude, I’m not sure what this study adds in terms of informing clinical practice. It is a nice study, one that is well conducted but to what end? RCRSP is a clinical presentation that is multifactorial, and individual in nature. It is apparent that there is a complex interplay between structural, psychological, sociological and lifestyle factors that ultimately lead to each presentation being unique and individual in nature. This needs further investigating within research not more reductionist papers trying to identify the magic bullet.
Please do let me know what you think!