The aim of this assignment is to critically appraise the scientific research paper “The role of physiotherapy in the treatement of subacromial impingement syndrome” by Dickens, Williams & Bhamra (2005) while detailing the objectives of the paper, research methods used and the outcomes of the research findings.
What is the scientific paper is about?
The aims of the research paper according to Dickens Williams & Bhamara (2005:1) was to investiagte the effectiveness of a physiotherapy programme in patients with subacromial impingement syndrome.
This paper did not outline specific modalities and was reliant on convincing the reader that physiotherapy should be viewed as a first line management for subacromial impingement syndrome against the present orthepedic view which would swing towards operative intervention as the corrective action.
The paper was published by Elsevier for the Chartered Society of Physiotherapy by a combined group of physiotherapists and Orthepedic surgeons and was funded by the Physiotherapy Research Foundation. When combined these groups may lead the paper to be viewed as self servicing and lead the reader to pinpoint a lack of impartiality.
Patients for the research where taken from a waiting list for surgery for subacromial impingement syndrome. Each of these patients was independently reviewed by surgeon James L Williams, a coauthor of the paper, and had “underwent three steroid injections into the subacromial space, given at 6-weekly intervals as part of an exisiting protocol” (Dickens et al, 2005:160).
The study was conducted in a randomised fashion in that the eighty five clients were selected by giving a client on a surgical waiting list an envelope that had within it either control or physiotherapy, 45 joined the physiotherapy group while 40 joined the control group. There were 100 envelopes split 50:50 between the groups which leads the reader to believe this may have been quasi-randomised.
In order to assess physiotherapy without bias clients who had previous physiotherpay treatement were excluded from the study according to Dickens et al. Also clients with signs of “cervical radiculopathy, adhesive capsulitis or clinically obvious rotator cuff tears … or a grade III subacromial spur on their shoulder suprasinatus outlet radiograph” (Dickens et al, 2005:160). The randomised fashion was administered by human administrators and not via a computerised system.
A control group was used and this group had no alternative but to continue on towards surgical intervention, this created a bias for this group since the probability for having surgical intervention was P-1.
Intitially the group of patients numbered eighty five patients from an initial set of 100 envelopes. 9 of the initial 40 patients in the control group refused to attend the repeat assessmemt at the end of the program, 3 of the physiotherapy group dropped out for social reasons leaving 42 partcipating in there group. Due to the substantial number of dropouts weight must be given to the emergence of unfairness to any comparsion portrayed in the paper.
Follow up occurred after a 6 month period and performed by James L Williams, a coauthor of the paper, and may not have been blind since he could have easily discussed whether “they felt they still needed surgery” (Dickens et al, 2005:161).
How the study was designed?
The aims of the research paper according to Dickens Williams & Bhamara (2005:1) was to investigate the effectiveness of a physiotherapy programme in patients with subacromial impingement syndrome. According to the Webster dictionary the word effectiveness means “to produce a desired effect”, the desired effect is not established in the aim. This ambiguity around the aim of the paper leaves any conclusion open to interpretation by the reader. When the paper is read in its entirety you could potentially interpret the aim as conveying the message that physiotherapy should be first line management for subacromial impingement therefore moving interventive surgery to second line.
There is no published pilot data therefore we cannot correctly establish if the sample size for the scientific paper is justified. We do know that the randomisation process catered for 100 patients, 100 envelopes, of which only 85 were taken up, 72 patients successfully making it to the reassessement stage.
The age ranges within the groups have relevance to the outcome of the scientific research. We only have a mean age of each group without an reference to outliers who could affect the statistical data. It is clear that the more junior males respond better under all conditions. This contradicts the statement “the two groups were well matched for age, sex and initial constant score” (Dickens et al, 2005:161).
The probability values (p-values) have been averaged (p<0.05) which converys a less accurate level of information. Ideally the exact probability value should have been given. Similarly though mentioned not data was provided for the Chi-squared value therefore negating the validity of any imperical data.
This lack of depth in the data provided does not allow the reader to establish if a null hypothesis was proved and may lead them to believe this was pure coincidence. Scientific research should always start from the null hypothesis point of view to ensure impartiality.
The selection process for patients, waiting list for surgery, post steroid injection, exclusion of specific pathologies, clinical history and examination meant the group may have been skewed to fulfill the authors aims. Steriod injections can provide an improvement in subacromial impingement due to it’s anti-inflammatory effect. The selection process did not seem to take into account the duration nor the severity of the impingement syndrome on the individual nor if they were receiving treatment from other practitioners not listed.
There is no detailed information regarding the treatment programme dispensed to the physiotherapy patients. If a specific treatment programme had been documented and applied to all patients in this group more quantative data and allowed the research to be replicated and potentially validated by other authors. This would have also allowed the treatments to be cross referenced with socio demographic data from each patient and establish sub sets within the master data.
We are unsure how the null findings are interpreted since the authors do not detail this. Nor have they given any data around the chi-squared test. Probability values are give in a round format (p<0.05) and exact (p=0.0008), mixing both approaches ensures ambiguity and imprecise outcomes.
The constant score has a low systematic error but is not reliable for clinical follow up in patients.The constant scores taken at the start of the research were based on 85 patients not on the same 72 patients whom allowed themselves to be reassessed at the end of the programme. This lead to a lack of confidence in the method used by the authors to compare pre and post programme data, they may not have itemised which data belonged to each patient and therefore could not remove this anomoly.
The involvement of James L Williams in the reassessment process ensured a lack of blinding and a bias, though the authors clearly did not see this position “the follow up assessments were performed by JLW in a blinded situation” (Dickens et al, 2005:162). All assessments should have been performed by a validated third party reusing the initial assessment criteria.
Since we have no detailed information about the modalities utilised, treatment cycles matched with socio demographic information there is no clinical relevance to the outcomes. To have clinical relevance the process needs to be reproducable which is highly unlikely based on the information presented in the scientific paper.
How was the study conducted?
A quasi-randomised human administer method of 100 envelopes split evenly between the control and physiotherapy groups were handed out to 85 patients. All 85patients were told that participation on the programme would not affect there standing on a waiting list for surgery. This promise may have affected participation since they would have already been convinced of the necessity for surgery by an orthepedic surgeon.
The dropping out of patients in both groups weakened the statistical data which the paper relies upon and imbalanced any findings. Interpreting clinical programmes requires greater participants reducing any potential for the play of chance.
How was the study analysed and were there limitation and errors in the study?
Each treatment group should have been similar based on age, sex, duration of syndrome, decrease in range of movement and similar capacities to perform the home care plan. Based on the information conveyed in the paper we must assume none of these points were established and therefore does not allowed for each group to have a comparable baseline. The quasi-randomised allocation of each patient to a group ensured that the treatment groups were not comparable.
Since all participants stayed within their allocated groups we can establish that the intention was for the results to be analysed by intention to treat. Unfortunately there were patient withdrawals from the programme which would allow to construe that the comparision of treatments would no longer be fair. Also the treatment received within the physiotherapy group as a whole may have differed from individual to individual but no patient moved between groups.
Not enough importance was placed on statistical information like the control group having members who improved, yet were not involved in the physiotherapy programme. Also confounding may have occured due to the pre programme steroid injection. Steroid injections are interrelated to anti inflammatory improvements in a range of impingement syndromes. The report had no reference to any confidence intervals which would have ensured the removal of the chance effect and imporved the significance of any statistics.
The lack of statistical data in table format and the reliance on prose within the paper show the paper to be more a marketing document than a reliable source of data. Quantative data tables would have allowed the reader to view and validate the authors outcomes. This lack of independence in data lead the reader to wonder does the data actually support the outcome. The approach of only conveying ‘results’ taken by the authors, could create suspicion in the mind of the reader, undermines the credibility of the paper.
Side effects are an important factor in all scientific research papers. What if the side effect of the treatment modalities outwayed the alternative approach taken by the control group. Effectiveness of treatment and a lesser set of side effects would need to be established against the control group to ensure there is no bias. Since there is no mention of side effects for either groups we can only assume that the authors wished to purposefully withhold this information. This factor only would ensure that it clinical relevant is negated.
How would you interpret the study and what if any are the implications of the study for your practice?
The main finding “confirms that a physiotherapy programme is of benefit” (Dickens et al, 2005:163) does not confirm the objective of the scientific paper. All the paper conveys is that there are successful alternatives to surgery for a subset of the population. There are too many pervasive factors to rely on the statistical significance of the data put forward by the authors. No true finding can be extracted from the paper and it portrays a message set by the authors who went out to prove it.
Therefore I cannot see any attempt by the authors to perform a null hypothesis test which should have been their approach. The only assumption to an attempt at a null hypothesis is the assumption that at the outset of the programme no difference existed between all patients in each group. Other alternatives could account for the 11 physiotherapy group patients improving, steriod injection, age, change in lifestyle etc.
This overlooking of the steroid injection pre programme participation has a major impact on the validity of the results. The severity and duration of the syndrome on the patient could potentially have an impact of any ‘positive’ results. The research funder being a physiotherapy organisation has may have had an effect on the interpretation of data. The length of gap between the post surgery and final step in the physiotherapy programme and final assessment may have had an effect on the results. Rehabilitation programmes may not have been adhered to during this period. Why not take periodical assessements every fortnight over the final 6 month period? We also do not know how the drop outs from the programme affect the data use to support the outcome. We can only assume if this data was removed the outcome may have not supported the objective and therefore shown the physiotherapy is not or no more effective than surgery.
Alarmingly this scientific research paper references 26 papers published prior to 2000, the oldest being from 1973, while only 5 papers are based between 200 and the time of release of the paper. This points to either a lack of interest in the particular field or a selective extraction of papers to support the authors objective. Normally supporting information referencing in other reports should be relatively up to date and from journals of quality.
Ideally a research paper should look to changing your clinical practice. This paper does not provide me with any rational to change nor if it had a credible case would I understand what I should be changing. Since I cannot replicate the treatment modalities used within the study I cannot change practise nor would I recommend another practioner to perform the same.
Cite This Work
To export a reference to this article please select a referencing style below: