Monday, 25 June 2018

Critical Appraisal of a RCT (June 2018)

We return again with a new article to appraise in order to consider usefulness when making decisions with our patients.

When examining research evidence regarding treatment effectiveness we often find contradictory messages and evidence pointing in different directions, this can prove to be a confusing process. This is especially true when considering conditions affecting the shoulder region where inconsistencies in terminology and diagnostic criteria for conditions such as subacromial impingement syndrome (SAIS), can make comparisons between studies even more difficult.

Well, today we bring up a follow-up article with a 10 year follow-up comparing a surgical approach versus a conservative approach in patients diagnosed SAIS:

Subacromial Decompression Yields a Better Clinical Outcome Than Therapy Alone: ​​A Prospective Randomized Study of Patients With a Minimum 10-Year Follow-up. Am J Sports Med. May 2018; 46 (6): 1397-140

This is a prospective RCT with a 10-year follow-up; the objective was to compare the results obtained in patients diagnosed with SAIS treated with subacromial decompression (open or arthroscopic surgery) or with physiotherapy according to the Bohmer protocol.

Patients with subacromial pain that persisted after conservative therapy (unstructured physiotherapy, non-steroidal anti-inflammatory drugs, and local injection of corticosteroids) who were referred to an orthopeadic department from primary care units were recruited. After recruiting 95 consecutive patients with SAIS, 87 patients met the inclusion criteria and gave their written consent. The inclusion criteria were subacromial pain for at least 6 months. The exclusion criteria were diabetes mellitus, neurological or spinal disorder of any nature, radiographic OA, the presence of chronic joint disorders (e.g. rheumatoid arthritis), full thickness rupture of the  rotator cuff, and SAIS stage III.

The patients were randomized into three groups; open acromioplasty (open surgery group [OSG]), arthroscopic acromioplasty (arthroscopic surgery group [ASG]) or non-surgical treatment (physical therapy group [PTG]). The details of the treatments are as follows:

         Open surgery: The procedure was performed according Rockwood and Lyons.

         Arthroscopic Surgery: Arthroscopic subacromial decompression was performed according to Ellman.

         Non-surgical treatment: The PTG received treatment according to the method described by Bohmer.

The clinical evaluation measures were: The Constant score, SF-36, Watson and Sonnabend score,  range of movement in terms of active elevation and internal rotation, and strength in abduction.
In addition,  ultrasound and radiographic examinations were evaluated at the beginning of the study and during follow-up.

The authors hypothesized that after 10 years, patients who had undergone acromioplasty would have a better clinical outcome and a lower risk of developing ruptures of the rotator cuff and osteoarthritis (OA) compared to those treated with the physiotherapy programme.

The results of the study through the 10-year follow-up are reported as statistically significant differences in favour of the surgical interventions in the active range of motion measurements, muscular strength and Constant score. No statistically significant differences were found in the Watson and Sonnabend scores, SF-36, and in the radiographic and ultrasound assessments.

As with many studies, there are limitations that need to be considered. For the purpose of this blog we will focus on the following:

1.       Sample size - statistical power - attrition
2.       Difference in the baseline characteristics between the compared groups

Sample size - statistical power - attrition:
The authors describe in the METHODS section that "The study design was planned to include 40 patients in each treatment group" and that "The estimated sample size was 36 patients per group" but then they add that "the study was closed after recruiting 87 patients (GSO, n = 24, ASG, n = 29, PTG, n = 34)". In other words, the study required 108 patients to to have adequate power to detect a difference. Small sample sizes are generally associated with an increase risk of type II error (not finding a significant different when one exists) due to a lack of statistical power to detect a difference.  Since a significant difference was found between the surgical and physiotherapy groups, this is less of a concern in this study although in the 10-year projection, there were some results that did not reach significance, which could be explained by the small sample size. As discussed in a previous blog ( a small sample size is not only a concern from the perspective of generalisability but also risks false positive findings. Given the size of the three randomised groups in this RCT this is a concern with  this study.   Added to this, during the follow-up, a 38% attrition was observed in the open surgical treatment group, and an overall attrition of 24.2% of the study subjects. This may be expected in a 10-year follow-up study, but these factors should be considered when constructing the research design and consideration given for strategies to account for missing data.

Difference in baseline characteristics between the groups compared:
As we have seen previously in other blog posts ( one of the safest ways to distribute the characteristics of subjects equally between the different treatment groups in intervention studies, both in known and unknown characteristics is through randomization. So through the rules of chance, we can be sure that the groups are comparable to each other, and attribute any differences in the results, to the intervention that is being studied. In the first instance, the authors state that the groups in the baseline were comparable, however, in a more careful analysis, consider the following: when observing the duration of the symptoms of the subjects assigned to each treatment group, 17 individuals ( 55%) belonging to the physiotherapy treatment group showed a duration of their symptoms greater than 36 months, more than double that of the 8 present in the arthroscopic surgery group, and much higher than the open surgery group (11 people). ). This means that the groups that are compared are different even before they are exposed to the treatment groups. Thus we cannot be  sure that the differences found between the surgical intervention groups and the physiotherapy treatment group at 10 years are not influenced by these, and other unknown, differences.

The strengths of the study included a randomised comparison and longer-term follow-up. However, the results should be interpreted with caution because of concerns about the sample size, and risk of chance findings, as well as baseline differences in the treatment groups in relation to important prognostic factors.

These findings should also be considered in the context of the recent CSAW randomised controlled trial ( and a systematic review comparing surgery with programmes incorporating physiotherapist-led exercise ( and also one other RCT with 10-year follow-up that reported no difference between surgery and a programme incorporating physiotherapist-led exercise (

Tomas Parraguez, Brian Cho, Paul Regan, Chris Littlewood,  Sijmen Hacquebord

No comments:

Post a Comment