Clear aligners are not as good as fixed appliances? A systematic review
This new systematic review suggests that treatment with clear aligners is associated with worse treatment than fixed appliances. This is an important conclusion. But does it stand up to close scrutiny? Read on….
Clear aligner treatment is now a well-established method of delivery of orthodontic care. The aligner companies and their paid Key Opinion Leaders make many claims for its effectiveness. However, for such a popular and costly method of care, there is surprisingly little high-quality research. I was, therefore, very interested to see this new systematic review that the EJO has just published. A team based in Zurich, Athens and Finland did the review.
What did they ask?
They did the review to answer this question.
“Is there a difference in the treatment outcome with clear aligners compared to fixed appliances for comprehensive orthodontic treatment”?
What did they do?
They carried out an extensive and ambitious systematic review. They followed the standard methods of electronic search, screening the titles, paper selection, assessment of bias, data extraction and meta-analysis.
Importantly, they included randomised trials and non-randomised retrospective studies. I will return to this later. The PICO was
Participants: Orthodontic patients needing comprehensive treatment.
Intervention: Clear aligners
Comparator: Fixed appliances
Outcome: Occlusal index scores using the ABO discrepancy index or the Peer Assessment Rating.
They registered the protocol before the study started.
They measured the risk of bias for the RCTs with the Cochrane Risk of Bias tool. The ROBINS-I tool was used to measure the risk of bias for the non-randomised studies.
What did they find?
They found 11 studies that satisfied their inclusion criteria. These were divided into 4 randomised trials and 7 retrospective non-randomised trials.
When they looked at the risk of bias, they decided that two of the trials were at high risk of bias. This was because of problems with randomisation, deviations from protocol, missing outcome data and outcome measurements. They also reported that the non randomised studies were at a higher risk of bias. For example, five were a moderate risk, one of them serious and one of them was at critical risk of bias. However, when I looked at the supplementary data on the EJO website, I was confused because this stated that 6 of the studies were at serious risk of bias, and one was critical. These shortcomings were lack of control of confounders, selection bias, deviations from protocol, outcome measurement and selective reporting.
The authors presented a large number of meta-analyses. I do not have the space to go into these here. As a result, I shall just concentrate on the data concerning the ABO scores. They reported that treatment with aligners was associated with worse ABO scores compared to fixed appliances. The difference was 9.9 (95% CI=3.6-16.2). This was clinically significant.
This meta-analysis reported on a total of 148 patients who had been treated with aligners and 148 fixed appliance treatments. The heterogeneity was 84% which is high. This data was derived from two non-randomised trials that were at serious risk of bias and one RCT with some concerns.
Their overall conclusion was:
“Orthodontic treatment with clear aligners is associated with worse treatment outcomes compared to fixed appliances in adult patients. Current evidence does not support the clinical use of aligners as a treatment modality that is equally effective to the gold standard of braces”.
What did I think?
Firstly, I thought that these conclusions were very bold. I feel that it is essential that when findings of this nature are made that the paper should be robust. I am afraid that I think that this is not the case for this paper. I have the following problems.
On first reading, it appears to be useful. However, the analyses include non-randomised retrospective studies. We must assume that most retrospective studies are subject to selection bias. As a result, when authors include retrospective studies in a systematic review, then its conclusions must be weakened. This is why Cochrane Reviews do not usually include this type of study.
I also thought that the reporting in the paper was not consistent with the data that they provided. All the non-randomised retrospective studies were at serious or critical risk of bias. Again, this reduces the strength of the evidence.
Finally, I had a close look at the papers that they included in the meta-analysis on the ABO scores. They based this meta-analysis on one trial and two retrospective studies. The International Journal of Clinical and Experimental Medicine published the trial. Importantly, the e-century publishing corporation publishes this journal and this is on Beall’s List of predatory journals. Furthermore, the Journal has retracted the paper because of plagiarism. The AJO published one of the NRS studies. The final study was an unpublished Masters thesis from the University of Montreal. These are not high levels of evidence.
Unfortunately, I cannot help feeling that the papers that they included in their paper cannot support the conclusion they published. I would like to point out that this is simply my academic opinion.