January 12, 2016

Is there a relationship between orthodontic extractions and Obstructive Sleep Apnoea?

Is there a relationship between orthodontic extractions and Obstructive Sleep Apnoea?

Following last week’s blog on extractions I’m going to continue with the same subject by reviewing a paper that was sent to me by Dr Dave Turpin. This outlines a study that investigated whether orthodontic premolar extractions were associated with obstructive sleep apnoea.

Evidence support no relationship between obstructive sleep apnea and premolar extraction: An electronic health records review.

Ann Larsen et al
Journal of Clinical Sleep Medicine 2015: Vol 11 1443-8. DOI: 10.5664/jcsm.5284.

This was a very well-written paper. In the introduction, the authors outlined several points about obstructive sleep apnoea. These are:

  • It affects 10 to 20% of middle aged and older adults
  • Individuals with obstructive sleep apnoea are more likely to die from cardiovascular disease than those without OSA.
  • Daytime sleepiness caused by sleep apnoea can increase the probability of motor vehicle and work-related accidents
  • There may be an association between craniofacial morphology and obstructive sleep apnoea.

They pointed out  that changes in airway may contribute to obstructive sleep apnoea. This has then led to the proposal that orthodontic treatment, involving premolar extractions, may reduce the airway size and increases the probability of a person having obstructive sleep apnoea. The aims of this study were to examine this hypothesis.

What did they do?

They carried out that a study that used medical and dental data extracted from the claims repository and electric health record of Health Partners. I think that this is a USA based health insurance agency?

  • They extracted the following data from the patients data;
  • Age
  • Body Mass Index (BMI)
  • Presence of obstructive sleep apnoea (OSA)
  • Whether the person had one premolar missing in each dental quadrant. They assumed that this group had previously received orthodontic treatment.

They matched 2792 patient records of people who had missing premolars with a similar number who had not. This resulted in a final sample of 5584 subjects.

As this study used data that had been previously collected they carried out several validations to assess the quality of the data. I will not go into these into much detail as this was very extensive. But, they seem very logical and suggested that the data had validity.

They provided descriptive data and carried out data analysis using logistic regression which controlled for the relevant cofounders. This was entirely appropriate.

What did they find?

They found the following;

  • 10% of the sample had obstructive sleep apnoea
  • A diagnosis of obstructive sleep apnoea was more likely if patients were older, where males, and had a high BMI.
  • The prevalence of obstructive sleep apnoea in subjects with and without premolar extractions was almost identical.

In their discussion they provided detailed descriptions of the limitations of the study. They pointed out that there may be sample bias in this study because of the following;

  • 80 to 90% of patients with obstructive sleep apnoea are undiagnosed.
  • Patients in their study may be a unique group because they were able to afford all were insured for their health care
  • This is a retrospective data set and there is minimal control of the data.

They also suggested that their findings were consistent with other studies and this provided some validation of their methodology and conclusions.

What did I think?

I feel that this was a very interesting paper that used a method that is relatively new to orthodontic research. My last blog generated considerable discussion on the relative usefulness of retrospective and prospective studies. This study used a different methodology as it is was a retrospective study of a large dataset. This research method is gaining popularity as it makes use of ‘big data”. In short, this involves the analysis of components of computerised data that is collected as part of the patient record in a health service or insurance company. It has been suggested that this will provide evidence that it has a higher level of a poor retrospective study, but not as strong as a randomised controlled trial. Importantly, this approach may be more cost-effective than running an expensive randomised controlled trial. In many ways this is similar to the approach used by Dr Lysle Johnston several years ago. While the theory is compelling we still need to consider the possibility of selection bias but this is an interesting way forwards.

I thought that it was good to see that the authors drew attention to the potential issues with their findings. I also felt it was important that they carefully validated their data and illustrated that their data was somewhat robust. Furthermore, their findings in terms of body mass index, gender, age and obstructive sleep apnoea was similar to other studies. As a result I cannot help feeling that their findings on premolar extraction not being related to obstructive sleep apnoea were also logical.

This study and the comments made following my blog last week and made me think more about study methods. I will cover this in in another blog post in the next couple of weeks. Having to think takes time!

Related Posts

Have your say!

  1. Thank you, Kevin. Given the emphasis, interest and claims related to sleep apnea this study and your critical review are timely.

  2. Thanks for sharing with us your study analysis Kevin.

    As in last week’s discussion the question at the end is if extractions were done to just relieve crowding (no changes in “tongue space”) or to reduce protrusion (some changes in “tongue space”) or indeed the fact that no extractions were done (maybe increase in “tongue space”). This is an oversimplification of a complex problem, but part of the hypothesis hanging around. In any event it can be seen that a definitive is unlikely to be extrapolated from this type of retrospective database. Proper selection of cases/controls is cornerstone. Sleep disorder breathing problems are multifactorial and requiere multiple professionals to diagnose and manage them. Just because we have imagining tools to statically measure areas or volumes does not really make a lot of sense when sleep is body function that is reflected through continuous cycles. A snapshot is only a very tiny part of a large picture.


    • Your comment of “oversimplification” suggests that we were looking for the CAUSE of OSA. In fact, the opposite is true. We were studying whether extractions are NOT a cause of OSA. You are correct that OSA is multifactorial and complicated and many more studies need to be done to help determine the actual causes(s) of OSA.

  3. ‘Whether the person had one premolar missing in each dental quadrant. They assumed that this group had previously received orthodontic treatment.’

    assumed they had ortho?…well that’s no good


  4. ‘Whether the person had one premolar missing in each dental quadrant. They assumed that this group had previously received orthodontic treatment.’

    this analysis is flawed. You can’t MAKE that assumption.

  5. Science, in the search for explanations , requires methods. But the methods do not overlap the facts. However, without an appropriate methodology , it is not possible to practice evidence-based dentistry. Follow this blog with interest because I value your critical interpretation of papers and the importance given to statistical results . Maybe the sample size is the way to validate retrospective research. This post( your response to last discussion) was very adequate . Thanks for your commitment and dedication.

  6. Dear Prof O’Brien

    Once again you have opened the door to what many consider to be a critical conversation per its pertinence to long term airway health, morbidity…..and even mortality; I commend you sir (once again) for raising awareness about how well-positioned your specialty is to impact not only healthcare professionals’, but also the general public’s, knowledge of how certain malocclusion phenotypes…..and certain Tx remedy options, might or mightn’t be a risk factor for systemic illnesses like SDB/OSA, and related co-morbidities such as ADD/ADHD, type 2 diabetes, CVD, etc. However, one point the authors mentioned that I most definitely think needs clarification, was their rationale for why they felt confident that likey only a ” ….limited percentage of patients….” were participating in this retrospective trial with ‘undiagnosed apnea’. Missed Dx of SDB/OSA was very likely at play here Dr. O’Brien and probably at play in both of the groups. I think the authors made a huge and unreferenced hunch that because the data sets for both groups were derived from insured/gainfully employed individuals, they all had been assumed (by the authors) to have faithfully attended regular health checkups with their physicians, and that the said physicians were all assumed to have been competent in sleep-airway health assessment, and how to appropriately Tx, refer, etc. I’m not sure how this might have played out in the UK, but in the US where this retrospective analysis was conducted, the following are true (i.e., evidence supported): 1. one’s owning of employer-supported medical insurance coverage does not directly correlate with compliance with respect to recommendations for attending periodic physician wellness visits; and 2. evidence (refs upon request) suggests that didactic and/or clinical competence in Sleep Medicine is not a significant component of undergraduate and/or post-grad medical education in the US. But for argument sake, lets assume most all of their subjects did indeed buck this trend by actually being compliant for recommended periodic well visit attendance with their physicians, because of the afore-mentioned paucity in medical education pertaining to the discipline of Sleep Medicine, many/most American docs would likely not have even picked up physical (high/narrow-vaulted palates, anterior open bites, high angle, crowding, poor lip seal, low hyoid b. position, forward head tilt, etc.) and/or behavioral (bruxism, snoring, open mouth posture, etc.) signs/symptoms evidenced to be associated with OSA risk.

  7. Great to read your blog post and become aware of this interesting study.

    I think a question that could be more meaningful and future research should be directed towards is does retractive mechanics to close extraction spaces contribute to sleep-disturbed breathing?

    It would be great to see that specific subgroup of patients with retractive orthodontics to close extraction spaces studied further in the future.

    It was useful to see the study was limited to patients aged between 40-70 since OSA often does not appear in the later decades of life. However, snoring or upper airway resistance syndrome are also prevalent forms of sleep-disturbed breathing that could also be relevant to investigate in this type of study in the future rather than limiting focus on OSA.

    The retrospective study of a large data set is an interesting alternative to RCTs and hopefully with the increasing awareness of potential links, we can start collecting more comprehensive and relevant information for future analysis.

  8. I’m happy to see more and more of this research making it’s way into the mainstream. Dr. O’Brien, your insight is certainly helpful in guiding the process towards higher quality research and thus conclusions that can ultimately benefit the patients. I worry that lack of credible evidence opens the door to manipulation of less than credible evidence for the purpose of financial gain. The more knowledge we have the better off our patients will be.

Leave a Reply

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