At a 10-Year
Follow-up, Tendon Repair Is Superior to Physiotherapy in the Treatment of Small
and Medium-Sized Rotator Cuff Tears
To repair or not to repair, that is the question. This recent
paper by Moosmayer et al (2019) appears to provide some much-needed evidence in
relation to this question. The authors of this randomised controlled trial
conclude: ‘At 10 years, the differences in outcome between primary tendon
repair and physiotherapy for small and medium-sized rotator cuff tears had
increased, with better results for primary tendon repair.’
When considering whether to go ahead with a surgical repair
of the torn rotator cuff, both clinicians and patients are rightfully mindful
of the long-term implications of such a decision. Therefore, this 10-year
follow-up study is an important contribution to the evidence base.
However, it can be seductive to think that this study
supports the need for rotator cuff repair surgery in people with small to
medium-sized rotator cuff tears. As always, the devil is in the detail and it
is important to read beyond the headline conclusion reported in the abstract.
There are many points of discussion that this paper raises.
But, in this short blog, I will focus on statistical significance versus
clinical importance and the confidence interval.
The headline figure from Moosmayer et al (2019) is that
there is a statistically significant difference between the two treatment
groups (surgery versus physiotherapy) in favour of the surgical group with
reference to the primary outcome measure, the Constant score. However,
statistical significance tells you nothing about clinical importance. For
example, you might have a statistically significant difference of 0.4 points on
a visual analogue scale but that difference is not recognizable or meaningful
to clinicians or patients. However, with regards to this study, the difference
between groups is reported as 9.6 points on the Constant Score and a clinically
important difference is 10 points. So, this is close and any critique suggesting
this is not relevant based on lacking 0.4 points is open to challenge given the
uncertainty around determining clinically important differences.
But, and there is a big but here, the point estimate, i.e.
9.6 points on the Constant Score, does not tell the whole story. When we do
research, we recruit a sample of patients rather than the whole population and
then we attempt to infer the results from this sample to the wider population.
Given that the smaller sample cannot be fully representative of the whole population,
e.g. everyone with rotator cuff tears, there is always some uncertainty around
the treatment effects observed in the sample. Often we express this uncertainty
in terms of confidence intervals, i.e. the range of values In which the true
population value lies. So, although the estimate of 9.6 points difference on
the Constant Score might represent the sample, it doesn’t necessarily reflect
the difference in the population and so we need to express a degree of
uncertainty.
Despite current convention, it is interesting to note that
Moosmayer et al (2019) do not report such confidence intervals in the abstract.
But, when reading the full paper, they are easily accessible in Table IV where
the confidence interval is reported as 3.6 to 15.7. This means that in the
population of patients with rotator cuff tear, surgical repair is superior to
physiotherapy and the difference in clinical outcomes, measured on the Constant
Score, might range from 3.6 points (clinically meaningless) to 15.7 points
(clinically meaningful).
Where from here? Well, there is still a degree of
uncertainty. All thing being equal (and that is open to question), surgery
offers superior clinical outcomes to physiotherapy at 10-years as measured by
the Constant Score. But, that superiority might not be clinically recognisable
to the patient and clinician and any decision to undergo surgery needs to
consider the risk, burden, e.g. time off work, and patient preference. So, the
clinical discussion might be along the following lines; ‘you have a rotator
cuff tear, we have different treatment options including surgical repair,
physiotherapy, or we could wait and see. Surgical repair seems to offer better
clinical outcomes 10 years after surgery, not so after 1, 2 or 5 years, but the
extent of this benefit is unclear. If you do opt for surgery then you need to
think about the risk and burden of surgery and whether that is acceptable to
you given that the clinical importance of long-term benefits are unclear.
Management of rotator cuff tears remains challenging with
much uncertainty. My own view is that we need to be more balanced in reporting
such research and remain open to the current uncertainty.
Thanks for making it this far – comments welcome and
appreciated.
Chris
