The AJO has published a new trial on AcceleDent: Let’s have a critical look at it!
I have posted several times on research into methods of making teeth move faster. This new paper provides us with information on total treatment times with AcceleDent.
Orthodontics is constantly evolving. Recent developments have been methods of reducing the length of treatment. These include, localised trauma, pills and potions, laser light and vibrations. Despite the claims of the manufacturers and those with vested interests, most recent research has shown that there is no evidence that these innovations have a clinically meaningful effect.
Recently, I wrote about a study that looked at the effect of vibration on treatment with Invisalign. This posting attracted a fair amount of criticism on social media. In fact, people accused me of bias, because I was not being critical of research that reinforced my opinions on the introduction of new technology without testing. As a result, I have taken this comment seriously and I am going to take a hard critical look at this paper.
DiBiase et al
Am J Orthod Dentofacial Orthop 2018;153:469-80
https://doi.org/10.1016/j.ajodo.2017.10.021
I will declare an interest, as I know several members of the study team well.
A team from the Kent in the beautiful South of England did this study.
In their literature review they pointed out that several studies had been done on AcceleDent. Importantly, the first studies were non-randomised, retrospective studies and suggested that AcceleDent had an effect. These were followed by clinical trials that found no evidence of beneficial effect. All these studies were concerned with an evaluation of the stages of orthodontic treatment, for example, rates of canine retraction, alignment and space closure. No studies have looked at the total process and outcome of treatment. I thought that this study was important because they addressed this by asking the following question;
“What is the effect of the AcceleDent appliance on the outcome of fixed orthodontic appliance treatment”?
What did they do?
They did a randomised clinical trial. The PICO was
Participants: People under 20 years old with malocclusion requiring fixed appliance treatment. They had mandibular crowding requiring the extraction of premolars.
Interventions: These were (i) AcceleDent (ii) AcceleDent sham (iii) treatment as usual. They treated all the patients with MBT Victory brackets using routine protocols with approximately 6 weeks between visits.
Outcome: The primary outcome was rate of mandibular space closure. Secondary outcomes were (i) rate of space closure (ii) treatment duration in months and (iii) PAR scoring.
They used timers embedded in the AcceleDent devices to measure compliance. Unfortunately, these failed.
They did a sample size calculation based on rate of mandibular alignment. This was the outcome in an earlier paper on this trial. As a result, this paper reported on outcomes in the absence of a specific sample size calculation. The authors pointed out that the sample size calculation of a previous study of space closure suggested that this part of their study was adequately powered. I will return to this point later, as this is very important.
They did a remote pre-prepared randomisation. They concealed the allocation from the operators. Unfortunately, it was not possible to blind the operators or the patients to their treatment allocation. However, all data was blinded at data collection. It was good to see that they carried out a per protocol and intention to treat statistical analysis. Importantly, this was a relevant regression analysis that took the effect of potential confounders, such as, operator, initial malocclusion etc.
What did they find?
They reported that 81 patients entered the trial and 61 remained at the end of study. This was a 75% completion rate.
They presented a very detailed analysis of their data and I have extracted the important median outcomes into this table.
| Time | Control Mean (mm)+/-SD | MOP Mean (mm) +/-SD | p | Mean difference (mm) (95% CI) |
|---|---|---|---|---|
| 1 month | 0.67 (0.34) | 0.65 (0.26) | 0.77 | 0.02 (-0.13,0.18) |
| 2 months | 1.28 (0.5) | 1.36 (0.49) | 0.5 | -0.08 (-0.33,0.16) |
| 3 months | 1.88 (0.67) | 1.93 (0.74) | 0.76 | -0.05 (-0.4,0.29) |
Their extensive statistical analysis revealed no difference between the groups for any of the outcomes.
They pointed out that the rates of space closure and duration of treatment was similar to other studies.
Finally, they drew attention to the limitations of this study in the discussion.
What did I think?
As with all trials there are strengths and limitations. We need to remember that the main advantage of trials over other study designs is that bias is minimised. Importantly, it is not possible to remove all bias and do the perfect trial. It is, therefore, up to us to interpret all studies in the light of any limitations.
What were the strengths?
Firstly, I will look at the strengths. The most important was that this was a multi-centre trial done with multiple operators in clinics working to normal protocols. Randomisation, allocation and concealment were all good. They collected the data blind. They carried out a relevant and sophisticated data analysis. The authors were not funded by OrthoAccel. Finally, the AJO-DDO published the paper and this is our leading orthodontic journal with a reputation for high quality.
What were the limitations?
The authors drew attention to the limitations of their study. I felt that the most important was the sample size calculation. They did not base this on the outcomes that they measured in this part of their study. This means that there is a risk that the study did not have sufficient power to identify any statistical significance between the groups. They did use calculations from other studies, to justify their sample and this was reassuring. However, we also need to consider the effect size of the interventions. This was small. As a result, I am fairly confident in their findings.
Another issue is that the compliance timers did not work. In face, some AcceleDent Key Opinion Leaders raised this as a criticism of the study. My feeling is that while it would have been useful to have data on compliance, this is not essential. This is because the study was measuring the effect of giving a patient an intervention to use. This is similar to a drug trial in which investigators prescribe a self-administered drug to participants in the study. We do not know if they take the medicine. Importantly, this is what happens in the real world and these trials have validity.
The other limitations were less important and the randomisation minimised their effects.
Summary
We need to consider the results of this study with respect to other RCTs. There have now been several and they are all coming to the same conclusion. There is an absence of evidence of a clinically meaningful effect of AcceleDent on tooth movement.
Emeritus Professor of Orthodontics, University of Manchester, UK.
Kevin, I would like to point out what I believe is a fundamental flaw in all of these acceleration device studies. The existing literature on rate of tooth movement shows a wide range between patients with the “average” rate more or less a mm per month. The problem as I see it is that we don’t know which patients are fast tooth movers or slow tooth movers beforehand. The pooled data is then confounded into a meaningless number. The very fact that the fastest rate in this study was the traditional treatment would seem to support that notion. Would it not make more sense to treat each patient without acceleration for say 3 months then add acceleration and see if the rate of movement differed? Just curious on your thoughts.
Thanks for the comment. What you are suggesting is a cross over trial. This could be done but we need to be careful about carry over effects of each of the interventions. But, I cannot see the advantage of this. The main reason for carrying out a trial is to measure the effect of an intervention on the average patient. If the sample size is large enough then the effect of differing responses for the patients is negligible. However, we need to be careful of studies that are too small and then the effect of fast and slow movers may be a problem. There are no hard and fast rules on this but I think that this study had sufficient numbers to achieve this. I hope that I have answered your question.
Kevin, Dave, because of this concern, we actually did complete a cross over trial and measured rate of tooth movement. Half the patients started with the sham device and then were switched to the active device midway through the trial. The other half of the patients started with the active device and then were switched to the sham device. This paper is in preparation for submission at this time. Spoiler alert….no differences in rate of tooth movement.
Maybe using a one sided acceledent (if available) would farely compare right and left side for the same patient. I think to add to this significance or non significance of this device, the patients should be asked about their confort and ease to use this device or would they rather keep their braces on for another 2 or 3 months.
Thus is part of a normal physiological range and is accounted for as long as the sample size is large enough to remove it as a confounding factor. The bottom line is that Acceledent does not work and should not be provided to a patient ( especially at a cost!).
Thank you Kevin,
I used AcceleDent when it first came out. Admittedly, I only used it once or twice but discontinued because of cost, compliance and any perceptible difference in tooth movement.
As you stated, even though compliance was not tracked, the study simulates the real world where we also don’t know how much the device is worn.
Certainly, the cost of the appliance negates any supposed advantage in the speed of tooth movement.
We orthodontists love gadgets. They look so cool, they must work. This is one any reasonable orthodontist would pass on.
Thanks kevin for a nice review . i just would like know the effect of using acceldent on the primary stability of the microimplants ,if one is using them for the absolute anchorage or other biomechanical reasons.
Surely we have enough similar studies to now run a meta analysis which will help identify effects. If there is an effect hiding under all the negatives it may be revealed and the confidence intervals shortened.
David’s suggestion is interesting, it could be tackled with a Bayesian approach, I would love to see more of this approach in Orthodontics but then I also understand the logistics and expense of such an approach may be breathtaking.
Just to add to Kevin’s response.
The purpose of a randomized trial is to provide a fair comparison. Regardless of slow/fast tooth movers the goal of the study design is to have in the treatment groups patients with similar baseline characteristics in terms of “tooth moving speed” to allow for the fair comparison.
In such a trial we are mainly interested in the difference between treatment groups. If you have many patients and/or many studies, eventually you are likely to capture the true difference between treatment groups (AcceleDent vs control) across all patients categories (I.e. fast, average and slow tooth movers).
It is possible that an appliance could work differently in slow tooth movers compared to fast tooth movers, but if in the trial at hand, both treatment groups have patients with similar tooth moving speed, we are ok because we will have a fair comparison. The thing we want to avoid is to have in one group fast tooth movers and in the other slow tooth movers because the recorded difference in alignment would be confounded/ confused by the dissimilarity of the groups in terms of tooth moving speed. In other words, we will think that the difference in alignment speed would be due to the use of the appliance but it would be due to the unfair selection of the treatment groups. If we know that we have comparative results between the 2 treatments only on fast or slow tooth movers because our trial included only fast or slow movers then we can say that although our findings are true/unbiased, they are applicable only to fast or slow tooth movers. So, the eligibility criteria determine the applicability (external validity or generalizability) of our findings to a specific patient group or the entire population, whereas the method of treatment allocation (randomization) works toward the fair comparison ( internal validity).
Kevin, Tim, Nikolaos…thank you for the replies. I understand the randomization and that the intent of eliminating patient selection bias, however the concern was primarily based with the fact that the range of difference in rate of normal tooth movement is far greater than the small differences demonstrated and perhaps the sample size was inadequate. Tim’s study was precisely what had occurred to me as a method to account for that. My clinical experience is completely inline with the findings in this study as well as the previous ones, just wanted to mention that concern. Thanks for the input.
While we strain over and over to assess tiny results, the various studies and KOLs ignore a perhaps even more important consideration: for a given patient, how much of an expected effect would it take to make the patient think it would be worth the extra cost? At the level of the individual patient, we need to consider the time value of money and the money value of time.
In other words how do we provide informed consent. What unbiased information we give the patient to make an informed decision. In any given clinical scenario how much is any patient willing to pay to save X months of treatment. In my personal experience my patients are not willing to pay extra for portrayed “saving” of 2-3 months of treatment time.
It is amazing the dichotomy between what the clinician thinks he has properly conveyed to the patient and what the patient really understood in an informed consent process. It is not only about the wording but the difference in comprehension of any given clinical topic between the health professional and the lay person. We think we do a better job that we actually do in general terms.