A classic paper that informed us, years ago, that orthodontic extractions do not harm facial profiles!
Orthodontic extractions do not harm facial profiles
I am continuing with the theme of orthodontic extractions again in this post. This is in response to comments made by Lysle Johnston and Jay Bowman who reminded me that the effects of extractions had been addressed in the literature. I have, therefore, decided to base this blog post on Lysle Johnson’s classic paper from 1992. I highlighted this work when I discussed the top 10 papers that have influenced my career.
D Paquette, J Beattie, L Johnston.
AJO-DDO 1992; 102:1-14
Now that I have read this paper again, I think that it is still a classic. The authors start with a really good introduction, in which they discuss the dilemma of whether to extract or not. They point out that if we want to compare extraction and non-extraction treatments this can only be done for borderline cases. They also stated that it would be difficult to randomly allocate patients to extraction/non-extraction treatment in a trial because of consent issues. For example, as part of obtaining consent for a patient to take part in a trial, a clinician needs to inform the patient that one reason for doing the trial is that they do not know the “best” treatment. In other words, we do not know whether it is best to extract teeth or not. As they would clearly choose the least stressful option for themselves, they are more likely to decide that they do not want to take part in the randomisation and they would like non-extraction treatment! This would affect trial recruitment.
They addressed the difficulties of running a trial by adopting an approach that used retrospective data and yet reduced bias.
What did they do?
They carried out the study in several stages;
- They made at least five attempts to contact Class II Division I first premolar extraction and non-extraction cases who had been treated at St Louis from 1969 to 1980 and asked them to take part in the study. One out of nine possible patients accepted this invitation.
- They then collected data from the patient’s initial cephalograms and study models. They then used this in a statistical technique called discriminant analysis. This grouped samples of patients records together according to their characteristics.
- They then used this data to identify a sample of patients who could have been treated with or without extractions.
- Finally, they contacted 48 extraction and 48 non-extraction patients and obtained a sample of 30 non-extraction and 33 extraction patients. These patients returned for cephalograms, study models and self-assessment of the aesthetics of their profile.
- They carried out an extensive analysis of this data with individual comparisons across multiple cephalometric variables. This was the standard approach at the time.
What did they find?
They found the following;
- 73% of the extraction and 57% of the non-extraction patients had less than 3.5 mm crowding
- The patients did not prefer either the non-extraction or extraction profile
- The lips of the non-extraction patients were 2 mm more procumbent than those who had had extraction treatment.
Their discussion was extensive and logical and I recommend that anyone interested should read it. They emphasised that in their sample there were no differences between the effects of extraction and non-extraction treatments. They also came to the great conclusion that the pressure at that time on the promotion of non-extraction treatment had resulted in the specialty “ being bullied into fixing something that was not broken”. I’m not sure times have changed!
What did I think?
Firstly we need to remember that they published this paper 24 years ago and our knowledge of research methods and statistical analysis have changed. My feeling is that this is a classic paper and all orthodontists should revisit it several times throughout their career.
We should consider how this work has stood the test of time and there are several important points;
- The initial sample of patients that were analysed was 238 out of approximately 2142 patients. As a result, they are a selected group. We need to consider if the people who took part in the study were different from those who did not respond to the invitation. There is no information on this in the paper and this must increases the uncertainty of the conclusions. Importantly, if there is selection bias we do not know its direction.
- There were further dropouts from the final sample and 33 could not attend for records
- Nevertheless, they did identify a group of patients, who were identical apart from the extraction/non-extraction decision.
When I consider all these factors, I can conclude that for these patients, of which there was maximum uncertainty in the treatment decision, the findings were valid.
Overall this paper still provides us with evidence, that for the borderline extraction/non-extraction patient, the decision to extract or not did not have a a clinically significant effect.
My posts this month have being devoted to a theme and I hope readers are finding it useful. I will attempt to summarise the evidence about orthodontic extractions next week and consider a way forwards for clinicians and researchers.