July Literature Review - Physiotherapy Vs surgical repair for rotator cuff tear

surgery or physiotherapy for a rotator cuff tear Jul 27, 2019

The great debate: how does physiotherapy compare to surgery for full thickness tear of the rotator cuff?

The shoulder physio July literature review will analyse 3, now infamous, papers by Stefan Moosmayer et al comparing conservative management (physiotherapy) with surgery in the treatment of rotator cuff tears of small to medium size. There is a 1, 5 and 10 year follow-up, which is great!

Background and methodology

There have been 3 papers published by this group concerning physiotherapy Vs surgical tendon repair of the rotator cuff:

1 year follow-up - published in 2010

5 year follow-up - published in 201410 year follow-up - published in 2019

The 2019 paper has garnered a lot of attention on the socials - mainly from those keen to dispute the effectiveness of conservative Mx for rotator cuff tear. Below is a brief overview of them methods.

52 participants were allocated to surgery 51 participants were allocated to physiotherapy  The allocation was performed with appropriate randomisation.

Inclusion criteria:

  • Lateral shoulder pain at rest or with movement
  • A painful arc
  • Positive impingement signs
  • Passive ROM of >140 degrees flexion and abduction
  • MRI confirmed full-thickness rotator cuff tear not more than 3cm
  • Traumatic and atraumatic tears were INCLUDED

Exclusion criteria

  • Age <18 years
  • Tears with an absolute indication for surgery, for example those involving the subscapularis
  • Presence of local or systemic disease affecting shoulder function
  • Previous surgery on affected shoulder
  • Medical co-morbidities
  • An inability to comply with follow-up

Surgical repair was performed by mini-open or open approach.

Physiotherapy was directed towards correction of upper quarter posture, centralising the humeral head and the restoration of scapulothoracic and glenohumeral muscular control and stability (more on this later).

Outcome measures: Constant score, ASES score, VAS, and SF-36 (QoL measure).

NULL HYPOTHESIS: There would be no significant difference in treatment effect between the 2 groups.


Image 1: Physiotherapy Vs surgical tendon repair at 6 month, 1, 5, and 10 year follow-up. Surgery out-performed physiotherapy at all time points, except 6 months.


Image 2: Surgical tendon repair Vs physiotherapy Vs secondary tendon repair (for those who 'failed' physiotherapy).

Let's discuss

Is there a clear winner? It appears it is a resounding win for surgical repair, especially from years 2-10, leaving physiotherapy substantially in it's wake. End of story? Not quite ;)

Let's delve deeper.

97% of participants were followed up at 10 years! This is a great result and credit to the authors for achieving this.

The final constant score at 10 year follow-up was 80.5 for surgical repair group and 71.8 for physiotherapy group. A maximal score is 100. 71% of the surgical repair group achieved a good to excellent Constant score (>81 points) while this was achieved by only 42% of the physiotherapy group.

The number needed to treat (NNT) was 4. This means 4 patients would have to undergo surgical repair instead of physiotherapy in order for 1 additional person to achieve a good to excellent outcome. This result was portrayed as being entirely positive for the surgical group in the discussion of this paper, but I'm not convinced. If you were a patient and you were informed that 4 people would have to undergo surgery in order for 1 to get better more than physiotherapy, would you like these odds?

27% of participants treated with physiotherapy crossed over to the surgical repair group and received a secondary tendon repair. Of importance here, 64% (of the 27%) did so in the first 12 months! 21% did so in the second year and 14% from 2 - 10 years. I interpret this as if you can make it through the first 2 years, there is an 86% chance you will not require surgery in the next 8 years. You could also look at it from the perspective that 73% of people who underwent physiotherapy were satisfied with their result and did not wish to escalate their management to surgery. Again, this is not a bad result for physiotherapy and is actually in alignment with other research that suggests 75% of people with a full thickness cuff tear can be managed successfully without surgery (Kuhn et al 2013). This result is really not that transformative, the number (75%) is repeated often in the literature when assessing whether physiotherapy is effective for atraumatic full thickness rotator cuff tears.

Are there any non-physical factors that predict who will fail conservative management? This study did not specifically investigate any such factors. We know from other research that expectations really matter in outcomes for physiotherapy (Chester et al 2018) but this was not measured. Could some of those who failed physiotherapy had poor expectations regarding it's effectiveness? It is very possible. Of note, 41% of those in the physiotherapy group had tried and "failed" physio before, could this have negatively influenced their expectations? What about self-efficacy, beliefs, kiesiophobia, or pain catastrophising? These emotional, behavioural and cognitive factors can influence the outcome of physiotherapy and were not considered in this study. Future research MUST consider these not so obscure treatment influencers.

The physiotherapy group also had 20% of people undergo a previous corticosteroid injection, Vs 10% in surgical group. We know that corticosteroid injections can negatively affect collagen integrity, did this play a role?

Mechanism of injury (MOI) is also of immense importance. This study did not delineate whether a traumatic MOI performed any differently to an atraumatic MOI. It is widely accepted that a traumatic full thickness tear does better with surgical repair, and an atraumatic small full thickness tear can do well with physiotherapy. The physiotherapy group had 57% of participants report some form of trauma that precipitated their symptoms. Meaning an acute traumatic MOI was reported in over half of participants in the physio group. Knowing that these people do less well with physiotherapy, did this affect their outcome? In fact, a trial is about to start that will clarify this, (ARC trial), where those with an atraumatic rotator cuff tear will be randomised into either surgery OR physiotherapy. This trial will exclude those with a traumatic MOI. It will also focus in on those over 50 years of age. Should be very interesting and provide information on the optimal Mx of an atraumatic full thickness cuff tear.

Image 3: Re-tear after surgical tendon repair and functional outcome..

What about tendon re-tear after the surgical repair? This study reported a re-tear rate of 34% after 10 years. This is in alignment with reported re-tear rates in the literature. However, the main point of difference in this study is the in-tact repairs were reported to have out-performed those who had sustained a re-tear (82.9 Vs 76.9 Constant score). Although, as you can see in Image 3, those with a partial thickness re-tear did significantly worse that those who sustained a full-thickness re-tear at 5 year follow-up. In fact, a full-thickness re-tear performed exactly the same as those with an in-tact repair. BUT at 10 year follow-up it seems the partial thickness re-tears have been aggregated with the full-thickness re-tears, which could have reduced the Constant score for the re-tear group, and inflated the difference in outcomes between an in-tact repair and a re-tear. Interesting... The relevance of a healed repair Vs failed healing is the topic of another literature review, which I may do next month. Much of the literature actually report no difference in functional outcome regardless of tear healing or not (READ).

Does tear size progression matter? YES. This is a noteworthy point, and a reason why this study may change my practice immediately. Tear size progression was noted in the physiotherapy group, the average being 1cm in width and 0.6cm in length over the 10 years. Significantly, those who sustained a progression of tear size >1cm had a functional outcome that was far inferior to those who had a tear progression of <1cm. In fact, those who had a tear size progression of <1cm achieved a functional outcome that was clinically and statistically the same as surgical repair. Therefore, is tear progression the most important factor influencing outcome of physiotherapy? If so, are there any factors that predict who is vulnerable to progression of tear size? Age? Location of tear? Tear size? Fatty infiltration? Time? Physical activity levels? It's complicated when considering all of these potential influencing features. More so, when you consider it must be >1cm of tear enlargement for it to have a significant effect on outcome. Read this paper for a succinct review of the natural history of cuff tears.

Placebo? Post-operative physiotherapy? What is the exact mechanism of action of surgery??

The surgical group received physiotherapy after their surgery. This included ROM and strengthening exercises, in various forms. What effect did the post-operative physio have on the outcomes of the surgical repair group? Is it possible to determine the specific contribution the essential surgical component had on the outcome Vs the post-operative rehab given there was no placebo-control group? No! However, we are in luck, a study is under way that will answer this very question!!! (ACCURATE trial). Just how valuable the actual surgical component of repairing the rotator cuff tear cannot not be determined from this study (due to no placebo group). We know the non-specific and contextual effects are very powerful with surgical intervention. Watch this speech by Dr. Ian Harris HERE on youtube about the placebo effects of surgery. (Note: Dr. Ian Harris is an orthopaedic surgeon!). 

What about QoL measures?

There were no significant differences in quality of life measures between the surgical group and physiotherapy group. In fact, the physiotherapy group trended towards having BETTER QoL measures, although not reaching statistical significance. You could argue that QoL is the most important factor to measure when assessing the effect of a treatment, and the results of this study suggest no major difference between the groups. Why? Could the physiotherapy group be coping better with the pain? Be more resilient? Be at peace or at ease with their situation? I would like to think that physiotherapy possibly sets a person up well to independently cope with their pain (self-efficacy, right?). Here is nice paper that suggests you can live well with persistent pain - if pain is framed in a certain way (READ HERE).

Ok, let's talk about the physio intervention itself.

The physiotherapy group received advice and exercise that focused on centralising the humeral head to minimise superior translation of the humeral head, thereby reducing "impingement". There was also lots of attention directed towards scapula stability. If you're aware of any of my work, you would know that I don't subscribe to the theory of subacromial impingement and therefore the obsession of centralising the humeral head. This is not opinion, it is merely scientific fact at this point. Scapula stability is contentious, it seems scapula strengthening exercises are GOOD but they often don't change scapula mechanics. So did this group get the most optimal physiotherapy intervention of solid (non-nocebic) advice and education, progressive loading, and brain-rich rehab? Maybe, maybe not. This again could have influenced outcomes, but another experiment would need to be designed to prove this.

Any other research?

There have been two other systematic reviews of the literature that compare physiotherapy Vs surgical repair for a rotator cuff tear. Piper et al 2018 reported a statistically significant but not clinically significant effect favouring surgery. Ryosa et al 2017 reported no difference between physiotherapy and surgery (this study was limited to a short follow-up).

Final thoughts

Disclaimer: I am in no way trying to discredit this study or surgery for full-thickness rotator cuff tear. I am merely trying to illustrate the limitations of this paper and the complexity of the research question.

Will the results of this study change my practice? Yes. I will inform people with a rotator cuff tear of the evidence favouring surgery at longer term follow-ups (>5 years). I will also qualify this statement and say, if surgery is not something you wish to pursue, there is good evidence that 75% of people with your exact pathology (small to medium rotator cuff tear) can have very good quality of life and shoulder function WITHOUT surgery. I will ask the person what their goals are, what their expectations and beliefs are, and screen for behavioural factors such as self-efficacy and kinesiophobia. Then, in concert with the person and hopefully GP/surgeon, come up with a collaborative decision that best suits the individual. This seems a common sense interpretation of this study.

Want to know more about the effectiveness of physiotherapy Vs surgery for rotator cuff tear? Or shoulder pain, more broadly?


Ready to gain the confidence to manage any shoulder pain patient who comes to see you?

My comprehensive course – developed over 10 years of treating shoulder patients and researching and educating professionals about shoulder joint function – covers all the above areas and more. In over 16 hours of training delivered in an engaging, self-paced, online format, you’ll learn everything you need to know to feel at ease treating people with shoulder pain.