July 06, 2020

An excellent practice-based research paper on the effectiveness of Invisalign.

There has been limited high-quality research into the effectiveness of Invisalign.  A specialist orthodontist did this new prospective study. It provides us with some clinically meaningful information and I thought that it was great.

A few weeks ago, I posted about a study done in a private practice setting.  I felt that this was an important step. This is because investigators should do trials in the setting that an intervention is most frequently used.  As a result,  it was good to see another study done in a practice setting. This Virginia based team did this study to find out the accuracy of Invisalign in moving teeth.

The AJO-DDO published the paper.

Has Invisalign improved? A prospective follow-up study on the efficacy of tooth movement with Invisalign

Nada Haouili, Neal D. Kravitz, Nikhilesh R. Vaid, Donald J. Ferguson, and Laith Makkia

AJO Online advanced access:  https://doi.org/10.1016/j.ajodo.2019.12.015


I was also pleased to see that this paper was open access and the AJO has started an advanced access section, so we do not have to wait for papers once they are accepted.

What did they ask?

They did the study to:

“provide an update on the accuracy of Invisalign with new technology”.

In their introduction, they pointed out that since 2008 Invisalign have made several improvements. The most notable have been introducing the SmartForce features and the SmartTrack material. Furthermore, they have replaced impressions with scans.

What did they do?

They did a prospective cohort study of 38 patients with a mean age of 36 years.  He treated 29 the patients with Invisalign Full and 9 had Invisalign Teen treatment.

The mean number of aligners for each arch was 21 maxillary and 20 mandibular. The average time between the start and final scans was 8.5 months.

One specialist orthodontist treated all the patients.  Importantly, he overengineered the tooth movements when he felt that this was necessary for good treatment results.  He also asked the patients to honestly record their compliance with the 22 hours per day wearing advice.

The inclusion criteria for the patients were that they had completed treatment with scans, and their compliance was good. He excluded six patients because 3 of them did not complete in time, and 3 had errors in their scans.

They evaluated the tooth movement superimposing the individual teeth from the initial ClinCheck model over the final ClinCheck model.  They measured the M-D crown tip, B-L crown tip, intrusion extrusion and rotation.  Finally, they calculated the per cent accuracy of the tooth movement. This was the primary outcome of the study.

Finally, they randomly selected half of the sample and evaluated the ABO cast evaluation system scores.

What did they find?

They found that the mean accuracy of Invisalign for all tooth movement was 50%. The most accurate movement was bucco-lingual crown tip of the upper incisors (56%), and the least accurate was the mesial rotation of the lower first molars (28%).

When they looked at the ABO scores, they found that 74% were allocated as passing.  The average length of treatment was 8.5 months.

Importantly, they pointed out that the ClinCheck is a graphic depiction of force systems; it is not necessarily the decided final tooth position.  As a result, if the tooth movements are over-engineered, it does not mean that the treatment is only 50% effective.

Their overall conclusions were:

  • The mean effectiveness of Invisalign was 50%
  • The per cent accuracy may underestimate the products overall accuracy.
  • Invisalign appliances are improving but they still struggle with specific types of tooth movement.
What did I think?

This was another excellent study done by a specialist practitioner, and I want to compliment the authors. We need more of this type of research.

Their findings were fascinating and clinically relevant.  I found that one of the most interesting points was that Dr Kravitz “over-engineered” the tooth movements to compensate for any mechanical inefficiencies in the system.  This will have contributed to the low level of effectiveness of Invisalign of 50%.  As an experienced orthodontist, I can see that this is necessary, even though I have never done an Invisalign treatment.

It was also important to see that the overall quality of the treatment was high, as most of the cases were ABO “passes”.  Nevertheless, the mean duration of treatment of only 8.5 months suggests that these cases are likely to be mild.

However, this does bring out an important question. A highly experienced orthodontist did the treatment.  I wonder if those who are not so experienced, for example, general dental practitioners, provide their treatment with the same level of expertise and care with respect to over-engineering?  This will have implications for the treatment that they provide. In many ways, this is similar to issues with the effects of specialist training and experience when operators use fixed appliances.

The way forward?

Finally, in 2005 David Turpin suggested that clinical trials were needed to evaluate the effectiveness of Invisalign.  I am disappointed that there are so few trials that compare Invisalign with fixed appliances.

Ironically, the best research into Invisalign seems to be studies like this one. It is clear that this study is a step in the right direction.  It would be great to see a trial of Invisalign and fixed appliances in a practice setting. I hope that someone can do this soon, perhaps with funding from Invisalign?  Or is this a “dead question” because so many people are entirely confident that it just works?

Related Posts

Have your say!

  1. Excelente, Kevin.

  2. Hi Dr. Kevin. Congrats for your blog (for which I am a diligent reader), and greettings from Brazil. Do you mean “over-engineering” as being an over-correction? I work with Invisalign system for more than 8y, being a critic user since I do not agree with their comercial policy. I guess we have clinical evidences for a time that some movements are not effective and my point is that clinical practioners and even recent graduated orthodontists are not prepared for dealing with treatment dificulties, because the orthodontic biomechanics is the same for fixed appliances. Dealing with virtual planning like clincheck needs expertise in knowing that the VTO are not precise in the software, demanding overcorrection settings.

  3. It’s great to see a study with some relevance to practice on the modern Invisalign system. The experienced orthodontist involved in this study has employed some ‘over-engineering’ to the ClinCheck to improve the outcome. In my experience, General Dentists struggle to apply that sort of modification without extra training and experience – as with any orthodontics.
    The ClinCheck has a Tooth Movement Assessment tool which applies a level of predictability for individual movements, based on an extensive database created from a comparison of predicted and actual results. This tool can help the less experienced orthodontic practitioner with their treatment planning and consent process. Inevitably, it applies to the ‘average’ tooth and the practitioner needs to evaluate how close to the average their patient’s tooth is (particularly in relation to crown size and morphology).
    However, if the first series of aligners fails to precisely produce the expected result, all is not lost. Dentists are able to provide additional aligners to refine their case. This does add to the treatment time and costs of delivery, but does not detract from the quality of the final result.

  4. With respect,yes it is a dead question.
    To be proficient requires many dedicated hrs .of additional training.

  5. Thank you for posting this, love your blog!
    The future……..I believe the mindset needs to change to, “it” doesn’t do anything and there is no such thing as “doing Invisalign”.
    “It” (irrespective of the brand) is just a tool that an orthodontist can use to provide CLEAR ALIGNER ORTHODONTICS.
    Like with braces (irrespective of the brand) or any other tool, the doctor you choose is the most important factor!

    It begins and ends with us and our perpetual training and caring.

    Great blog, thanks for sharing!

  6. I think the question that is “dead” is whether or not certain types of movements are effective when using any clear aligner system. That question has been asked and answered over and over. Clear aligners (like Invisalign and a host of others) can be effective in the right situation. Lots of axial inclination (tipping) is virtually impossible for clear aligners, whereas bucco-lingual movements happen readily. Rotations of flat teeth like incisors is predictable, round teeth like cuspids, bicuspids and molars, not so much. It is incumbent on the practitioner to choose the right tool for the job. Sometimes, that tool is a clear aligner system, sometimes brackets and wires. The combo is an exciting alternative that many of the modern systems allow for. Being able to print your own models, and more importantly, make your own trays, is the key to making the combo protocol work. And for me, making my own trays also makes the basic clear aligner system exponentially more effective. For easy and hard cases. The cost for me to treat a patient with clear aligners is reduced dramatically, and the movements are much more predictable for me. We utilize different plastics as well. An .030 and .040 tray made from the same model is used alternately. This protocol has several benefits including loss or breakage of a tray. We can custom trim the trays. Easily merge into a combo plan. So many benefits to producing aligners in house!

  7. “The mean number of aligners for each arch was 21 maxillary and 20 mandibular. The average time between the start and final scans was 8.5 months.”

    I think this is saying the clinical result did not match a computer simulation that can not account for individual biology. I think this will always be a problem with a ClinCheck. The solution is a new scan and additional aligners as Dr. Toy pointed out. I assume this for every case and inform my patients about biology, simulations etc. prior to initiating treatment to expect another scan with a new grouping of aligners to complete the case. Without doing this, the cases in the study were not really completed but only compared to the initial ClinCheck. The result was that the average 8.5 months of treatment was inadequate though it seems the result was pretty good for the first ClinCheck..

    IMHO to really evaluate the Invisalign system requires recognition that additional ClinChecks and treatment are required in most cases. For me, this strategy has proven more predictable than trying to “over-engineer” the ClinChecks.

  8. I did a lot of Invisalign cases, it can be a great tool to treat a malocclusion. But apart from being an experienced orthodontist you need experience with Invisalign too.
    I hardly ever over engineer (I think if you do this a lot you are not treating efficiently). If a study reports ‘50% effectiveness’ that does not sound well, if the cases were reasonably treated the parameters measured are not the right ones to asses treatment result with Invisalign.
    Another problem is that in the hands of ‘uneducated’ or ‘untrained’ doctors things will not work out/go wrong.
    But that is a problem I see also with fixed appliances too, from time to time I get completely mistreated cases (where even some brackets may be put on upside down). Initially (and still…) the straight wire systems led some to believe orthodontics was easier now.
    The Invisalign system may seem easy to some, if you don’t have a clue and just send in impressions the company will provide you with aligners…..

  9. Thank you so much, Prof. O Brien, for this wonderful blog!
    I wanted to share this systematic review that was published last year. Only two RCTs were included and all the studies were performed in a university setting. https://doi.org/10.1186/s12903-018-0695-z

  10. “50% effective in the hands of really experienced orthodontists, trained to select suitable mild cases and to know how to over-engineer selected tooth movements. After 25 years of product development”

    Great slogan for the product advertising banner!

    How did it become the largest orthodontic system?

    1. Human hope & faith in new untested technology
    2. Disinterest in truth describing efficiency & efficacy
    3. Commercial Ambition, leveraging 1.& 2.

  11. This is a very interesting paper that really just confirms what many if us know/suspect. As mentioned it would seem more fair ( and useful) to compare aligners to fixed appliances as this is the choice that patients ( and clinicians) face. I am old school and use mainly fixed appliances, am I wrong? That would be useful information to have.

  12. Clear aligners ( Invisalign), Invisalign in this instance are an orthodontic appliances like any other appliances . Their efficacy is determined largely by the experience and education of the users. It’s tempting for a novice to move digitally in vain hope that it would be transferred in reality . More often than not , after few attempts , reality hits .

  13. Hi, an interesting paper which confirms Invisalign can move teeth. I wonder if you have any comment on the measurements taken to determine how much movement had occurred? It seems that some complex software (Grundheid et al, Angle Orthod 2009) was used, which included algorithms, to superimpose the teeth and arch-forms of the pre and post treatment position of the teeth. The clincheck only includes hard dental tissue (the crowns of the teeth) in the 3D model. Is it possible to make accurate measurements of this type when there is no stable reference point in the clincheck?
    I use Invisalign extensively and achieve results which I am happy with. The results from this study would appear to be of a high quality, since they passed the ABO. Have the authors under or, possibly, over-estimated the amount of tooth movement achieved by the Invisalign system?

    • Yes, Peter, you are correct – conclusion #5: “The percent accuracy determined by a best-fit
      analysis on a predicted ClinCheck digital model may underestimate the product’s overall clinical
      And correct again – any best fit software methodology is inherently inaccurate in measuring change where everything has the ability to move, and just like a ceph super result will vary depending where the software chooses (or researcher) to super! In our studies over the last decade at Melbourne uni, we (Dr James Newby et al) used Geomagic and we program only the single movement we are measuring, with no programmed movement on remaining teeth. This is currently best method I am aware of for measuring degree of expression of single movements with aligners; as presence and position of attachments, age of patients, degree and type of neighbouring movements, anatomy , programming of overcorrection, how many teeth programmed in the vector being measured (e.g. if you intrude a single tooth the degree of expression is greater than a whole anterior segment being intruded) and number of aligners will all effect the clinical expression of the programmed activation (You still have movement in the first adjacent non-programmed tooth so we exclude this tooth- this engineering software is extremely sensitive). Once you identify the potential for expression of a single movement in each tooth type, you may begin to appreciate the impact of the variable mix above, amongst others.

      Too many (confounding) variables measured here to be statistically significant in order to isolate single movement types on individual tooth type, combine the 2 arches (we record statistically significant difference between upper and lower arch movement expression, Table II shows interestingly they did not), with of course the largest variable being the patients – no daily record of aligner wear, when the new aligners were inserted, use of aligner seating devices, and we only have mean age presented here. In Table I we have the degree of IPR and mean number of attachments, but no Physiological parameters: (pregnancy, meds, age range)- …..and then combine this with the clinician who “over engineered” some movements on ClinCheck software, making it now impossible to claim this is a test of the system, but a test of this single clinician’s ability to manipulate the current system to his best ability – great as I am sure it is- as there is no evidence (perhaps anecdotal) of what they did and how that actually effects clinical expression. By “overengineering” we assume that certain movements are not expressing and to what degree, so we dirty the data we are trying to measure. BTW this is such poor terminology – “over engineering” by definition results in less efficiency and results in something being unnecessarily complex and inefficient. We also can’t “over engineer” an aligner by extending the range of movement on ClinCheck software. Impossible. Each aligner has a maximum activation. Only the company may possibly “over engineer” a single aligner or the system via excessive attachment placement, bells and whistles etc. More commonly, for better or worse we use the term “over correction” to describe extending the programmed movement beyond the clinician’s “ideal”, thus increasing aligner numbers that would tend to bias towards lower “mean” expression.
      Next we have the use of “mean” movement measurement, combined data from both arches, with no idea from the paper what measures are statistically significant (I see how they calculated for it but not what qualified, except 5 of 128 data measures published that I am sure cant represent their only significance). Mean % expression – please clarify – that can that mean that 99% of a sample score 0 movement and 1 scores 99.9% and still results in a fair mean accuracy? The standard deviation next to each result in Table II would be useful.
      To be fair, I did think this was an improvement on the first paper and the authors did recognize and state most of the limitations of a most difficult investigation. I’m being critical, much easier than conducting the research; but isn’t this the purpose of this blog, to really look at what is published and what it actually means?
      Finally, someone please tell me “what does 50% clinical accuracy mean?” Or more simply, “clinical accuracy?” I fail to see how this is a valid (especially when we read how the data is combined and measured), clinically helpful statement. Does it mean that I need to repeat the same program once more and I will reach 100%? Does it mean that 50% of cases would fail to pass the ABO? What if I am programming highly expressed movement in a 12-month treatment period? What if I am now combining anterior, posterior movements in every direction over a period of 2 years? No, it means none of the above, may not be applied to any of the above and I don’t know how to utilize the hard-won information! What is the clinical accuracy of fixed appliances? We don’t know because we can’t measure it. Even if we could, what does it mean and how do I use it? We know we may reach ABO or PAR passing grades if the clinician is skilled, just as we may and as this paper shows with aligners. Yes, I do research and speak for Align Technology. Sorry Kevin just couldn’t help myself.

Leave a Reply

Your email address will not be published. Required fields are marked *