Relapse of extraction and non-extraction treatment. A long term study.
One of the most controversial areas of orthodontic treatment is the decision on whether to extract teeth or not. Research has been sparse on this crucial subject. The AJO published this new paper and I thought that it was helpful.
I do not often provide a comment on retrospective studies. However, there are specific questions that cannot be answered by prospective cohorts or trials. One of these is the long term effects of extraction or non-extraction treatment. This is still a very controversial area of orthodontic treatment. In general, the opinion of clinicians and researchers is that the extraction of teeth does not cause any harm. Nevertheless, the choice of treatment mechanics may lead to some problems. The counterpoint is that extractions result in over retraction of the upper teeth and cause a myriad of problems. Indeed, some have suggested that extractions are the amputation of body parts!
This debate is never-ending, and I doubt that it will ever end. As a result, I was very interested in this new paper. A team from Brazil did the study. The AJO-DDO published the article.
What did they ask?
They asked this simple question.
“What are the long term outcomes of orthodontic treatment for extraction and non-extraction patients”?
What did they do?
They did a retrospective study to look at relapse. The authors obtained the records of patients from the orthodontic department of Bauru dental school. The inclusion criteria for the study were:
- Class I or Class II malocclusion at the start of treatment
- Non-extraction or extraction treatment (4 first premolars)
- Complete fixed appliance treatment
- Retention with Hawley retainer (and lower bonded retainer for at least three years
They obtained dental casts and panoramic radiographs at pre-treatment (T1), post-treatment (T2) and long term follow up (T3). This period was a mean of 37 years after treatment.
They selected the records of 57 patients. Group 1 was 16 subjects who they had treated without extractions, and Group 2 included 41 subjects treated with extractions.
The primary outcome measure was occlusal change. They measured this with the Peer Assessment Rating and the ABO OGS. They also asked the patients to complete a satisfaction questionnaire that they sent out via WhatsApp.
I could not find in the paper any mention of how they identified the patients so long after their treatment. I think that this is important, and I will discuss this later.
What did they find?
The mean length of the extraction treatment was 2.3 (0.5) years, and for the non-extraction treatment, it was shorter at 1.8 (0.8) years. This difference was statistically significant.
I am only going to include the occlusal index data for the PAR index. This is because I understand it more than the OGS, and it does not rely on recording the technical aspects of treatment such as angulation of the teeth. Furthermore, we can apply PAR at the start of treatment. This information gives us data on the pre-treatment dental morphology of the patients.
Here is the relevant data.
| Group/Time | T1 | T2 | T3 |
|---|---|---|---|
| Extraction | 23.7 (19.5-27.8) | 3.1 (2.2-3.9) | 7.5 (5.3-9.6) |
| Non-extraction | 25.7 (22.7-28.6) | 2.8 (2.0-3.5) | 8.6 (5.6-11.6) |
They concluded:
“There was significant post-treatment relapse in the long-term for both groups of patients”.
I thought that their survey data was perhaps more important than the occlusal index scores. This is because they collected information that was relevant to our patients. When they looked at this data, they found that 72% of the sample responded. Unfortunately, when I looked closely at their data, I found that only 13 of the non-extraction and 28 of the extraction group responded. This means that we must be uncertain of these findings. Their main finding was that there were minimal differences between the groups. However, the non-extraction group reported more changes in alignment than the extraction group.
What did I think?
When I interpreted this paper, I had to decide whether the scientific method was sufficiently robust. Firstly, this was a retrospective study. As a result, we must suspect that there was selection bias. Furthermore, the authors did not state the response rate for long-term recall. This information is crucial for a study of this nature, as it gives us information on response bias. Importantly, we need to consider if the responders are different from non-responders. In the absence of this information, it is difficult to come to any conclusions, and I have to assume that there is significant uncertainty in this data.
I was also concerned that they did not provide any information on how they identified and approached the patients to obtain long-term data.
However, we also need to consider the difficulty of obtaining long-term samples of orthodontic patients. In this respect, I respect the work and the success that the investigators had in obtaining their sample. In many ways, I suggest that this is the best that we can get. As a result, I think that we should accept their findings. Nevertheless, there is a degree of uncertainty in the data. This is reflected by the relatively wide 95% confidence intervals.
We also need to consider that there must be something, aside from random decision making, that persuaded the operators to extract teeth or not. This concept means that we cannot approach this data as being similar to a randomised trial. Again, this adds to potential bias.
Final thoughts
It was interesting to see that they felt that there was significant relapse with both the non-extraction and extraction approaches. I was not sure that I agreed with this conclusion. This is because the amount of relapse was in the order of 4 PAR points. This is not clinically important.
It was interesting that there was no difference in relapse between the groups. This finding suggests that in this sample, the extraction of teeth did not influence relapse. Nevertheless, there was a shorter treatment time with non-extraction. As a result, there may be some advantages to the non-extraction approach.
I also need to point out that they did not provide any information on the aesthetics of the final treatment result, breathing, life performance, bedwetting, arch width or any other magical effect of non-extraction treatment. I think that the non-extraction/extraction wheel will keep turning.
Emeritus Professor of Orthodontics, University of Manchester, UK.
That last sentence though 😂🤟
Loved the closing statement 😀
I’m finding it hard to know whether this is useful or not. For a start the numbers in each group appear quite low. Also there were class 1 and 2 patients mixed together. I think there is something useful in examining patients years later but not sure this method is really helpful. What needs to be done is find a clinician (ideally multiple clinicians) who treats everyone without extns and then examine their patients years later to see if any and which have relapsed (although the type of retainer would muddy the waters, maybe just removable retainers only in the study group). Since most of us treat with a combination of extracting and not extracting it becomes tricky to determine what’s going on since the decision process is involved. If there is no decision except the one decision to not take teeth out and that is the same for every patient I guess that would eliminate that variable.
Not that it would probably have made a great deal of difference to the conclusions of this study, but in Orthodontic practice I rarely asked for the extraction of four 1st premolars! Mill’s old adage..”four fours, if not why not”. There are plenty of reasons why not!
I frequently ask for the extraction of four fours. Helps the airway, the face and the gingiva. Also distinguishes the real orthos from the quacks. Why not?
Like so many of these articles, the conclusion (or lack of it) leaves us with more more questions than answers….but they do make us think!! Why is it that I feel many of my non-extraction cases take longer than the extraction ones as, clearly, that is not the case here. Is this related to the severity of the crowding?
The Abstract is misleading both in terms of sample size and statistical findings. The first line of its conclusion makes no sense in commonly spoken English. Not sure how this got past the Reviewers? The paper hints of selection bias in reporting IMO. As for “magic”, it helps to read some medical (not dental/orthodontic) journals to get a glimpse of what lies beneath the tip of the iceberg, e.g. Guilleminault C et al. Missing teeth and pediatric obstructive sleep apnea. Sleep Breath 2016.
Thank you for mentioning the article in Sleep Breath but I believe it is also suffers from selection bias. The cases were obtained by a ‘Search of the sleep medicine clinic database (which) identified 31 children with known dental agenesis’. Of course they will have a higher prevalence of OSA (and malocclusion) as that is what was suspected and resulted in referral. This would be equivalent to searching the records of children referred to the orthodontic clinic and finding the majority have crowding or a malocclusion!
From our limited dental vision the extractions are incorrect.
But from a broader point of view, from a biological and physiological point of view, extractions are correct and necessary when indicated.
Dentally, perhaps 90% of the cases do not require extractions, but from a more comprehensive point of view perhaps more than 80% require extractions.
The continual vagueness of these various studies makes me believe neither side has definitely one the debate over extraction/non-extraction. More research into both avenues is definitely needed.
Thanks for the paper. Every paper is useful and research, especially of this nature is difficult. As long as we are aware of the limitations, I enjoy these projects. As Kevin mentioned, the info gleaned form the response rate in itself is interesting.
The conflict in my mind is duo: where is the control, and if we identify a definitive answer what do we do with it? The title suggests that we are looking at long term occlusal change, great; and the experiment suggests that extraction intervention is a significant variable- not sure yet. If we accept that growth and development (change, reflected mainly by decrease in arch length) of the dental and surrounding tissues continue until we are dead, even in retainer wearing patients albeit smaller, then what are we measuring here? Is it relapse with a pattern influenced by extraction/non, relapse with a pattern not influenced by extraction/non , continued physiologic change, or a combo? Perhaps what we are seeing is that any orthodontic intervention for a relatively small period of time in a growing and developing living being has its effect made murky, especially when evaluating at a future point in time, unless held “permanently,” as Little et al. concluded. I don’t know. Does anyone?
Secondly bouncing around in my skull is our expectation and potential altered action – can we ever expect again to have a singular research project or series that is rigorous enough to change our decision making? I don’t believe so; the ego / KOL status of the author is a diluted or negative factor, we ask better questions, have more blogs. Additionally, if the day came when we could actually measure long term occlusal change with a perfect ruler (and I love the PAR), the study passed Kevin’s robust assessment and we did could extract – ha ha – a statistically valid conclusion as to whether extraction or lack thereof, influenced long term arch stability; is it enough to change our clinical behavior with respect to the treatment planning of borderline patients (assuming that we would not extract singularly based on fewer arch changes throughout life)? Any significant research findings based on a mean will not necessarily be kind to the individual, and thank goodness for this or we would be out of work – , as I believe that we will always need to diagnose and treatment plan the best we can for every patient based on their presentation, and not what group they fall into with respect to a finding.
On that note Kevin, would love to have a blog, (you have sort of done this with landmark papers), where we get to suggest 1 or more papers that have been so definitive as to change our clinical behavior.
Thx all
VV
“There was significant post-treatment relapse in the long-term for both groups of patients”.
Unless we use control groups with non-intervention, and we can identify and differentiate relapse from inevitable physiologic change, are we not measuring “post treatment change” rather than “post treatment relapse?”
Thank you Prof. for posting this paper.
Given the average age in this retrospective trial of the combined ext. vs non-ext. cohorts at T1 was 12.1 years old, and the class I:class II ratio was close to 1:1, I’d like to know your thoughts regarding the class II subjects. Specifically Dr. O’Brien, according to McNamara, Bishara and others (refs upon request), while of course unreported, most of these Cl II pre-teens were very likely not only mandibular retrusive (MR), but also maxillary retrusive and maxillary transverse deficient (MTD) at Tx onset, please opine for us if you would, at what age do you think that many of these kids’ deficient transverse and sagittal skeletal phenotypes might have been initially detectable; and also, might you agree sir with published studies from Angle Orthod., AJO/DO and elsewhere, that when MR and/or MTD are evident whilst in the primary/early mixed-dentitions, they will most often persist beyond, usually worsen and often be/become co-morbid with impaired naso-respiratory competence during wakefulness and sleep?
I am hoping to be in error Dr. O’Brien, but after reading your mention that, ‘….they did not provide any information on the aesthetics of the final treatment result, breathing, life performance, bedwetting, arch width or any other magical effect of non-extraction treatment.’, I am presuming you are not yet aware of(?), or maybe just not impressed by(?), myriad published papers regarding Dx: malocclusion and naso-respiratory incompetence co-morbidity? As there are many well-qualified scientists and clinicians who think these aforementioned papers should be taken seriously, or at least arouse some curiosity, this seems to suggest that there is no ‘magic’ getting past the peer-reviewers here. And while you’ve procured several chuckles from some of your most loyal and passionate fans, your condescending remark might indeed result in many young children worldwide NOT getting assessed for risk, definitively diagnosed and appropriately treated for Early Childhood Malocclusion* (ECM)-Sleep Related Breathing Disorder(SRBD) co-morbidity.
Per your vast reach and responsibility that goes with it Professor, I am hopeful you will take this more seriously.
*ECM-maxillary and/or mandibular skeletal transverse deficiency, mandibular and/or maxillary skeletal sagittal deficiency, excessive or deficient vertical skeletal development or deep/narrow palatal vault, before the age of 59 months of age).
hi . loving all of the debate. would have expected going by Bob Little s work that the non extraction group would relapse a lot more and sooner. so the selection bias here may have distorted the comparison. I find it increasingly difficult to get dentists to perform the extractions nowadays as they often wont accept the treatment plan as if they practice orthodontics they tend to be non extractionists. on the issue of the medical journals cited the methodology can be particularly poor. worst thing is patients who perform dr google are none the wiser. loving all of this. 😀
More research required? I think we need to agree on the questions we want to answer in a 30+ year follow up. Otherwise it’s like the Hitch Hiker’s Guide to the Galaxy where you wait millions of years for a computer to tell you the answer is 42 but you don’t really know the question the computer was asked.
I agree with VV – for that length of time, the results should be in the context of what would happen to untreated people.
retention “for at least 3 years”? Are we asking a question about stability with no retention? Then ask that and research that. I wouldn’t expect some really twisty teeth to be straight 4 years out if they stop wearing their retainers after 3 years.
But the big thing here for me is….retrospective study with 57 patients and it still was mixed malocclusions. Can someone help me with the stats? That seems ridiculously small number for so much variability on where you start and vagueness over what happens in the intervening 4 decades.
Data protection permitting, there are hospital departments around the world with a stack of records that should be mined every year for this sort of thing if anyone is interested in this sort of thing. If data protection is a problem retrospectively then start getting those boxes ticked for use of ortho records for long term research and in 30 years why can’t we have cohorts of thousands in tightly matched groups?
Stephen Murray
Swords Orthodontics
Nothing new to see here. All teeth move over time if not retained. The extraction/non extraction debate is stupid. I extract when I need to extract for a proper, healthy finish, and if am able to achieve that finish without extractions, I don’t extract. Regardless, I retain, and I tell them if they want their teeth to stay perfect, they always need retainers. Is that a reasonable request? Probably not, but it is the truth, and only the patient can decide if they want to keep straight teeth throughout their life. If not, then they can live with crooked teeth down the road, or again have orthodontic treatment.
Nothing on the human body is stable. I wish it was. I can certainly attest (and my wife will vouch for it), I don’t look better today than I did when I was 20. We degrade, we shift, we slide, we just get old. Our mouths and teeth do too. Little showed this years ago with a much better study (and subsequent followup studies).
Treat to what you feel is best (extract or not, we all should be good at making that decision and there is certainly grey areas in some cases that can work well either way), and then educate your patients on life long retention and hope they listen.