According to a recent Cochrane Review, surgical removal of mandibular third molars (M3M) is one of the most common operations undertaken in oral and maxillofacial surgery (Coulthard et al. 2014). There is concern from the authors of the paper under review that this procedure may result in persistent or developing new periodontal defect distal to the mandibular second molar (M2M). To mitigate this risk various techniques have been proposed including root surface debridement, flap design and guided bone regeneration.

The purpose of this review and network meta-analysis (NMA) was to assess:

• The effect of third molar removal on periodontal health in the second mandibular molar.
• Rank the efficacy of the treatment options
• Identify the best approach in terms of clinical attachment length gain and probing depth reduction distal to the second mandibular molar.

Methods

The review followed the PRISMA statement (Moher et al. 2009) and the protocol was registered with PROSPERO (CRD42012003059). Searches were conducted in PubMed MEDLINE, Cochrane Central Register of Controlled Trials (CENTRAL), and EMBASE (1974 to 2014) via OVID. Grey literature was also searched, with no language restrictions. Eligibility criteria were randomized controlled trials (RCTs) of parallel or split-mouth design, investigating surgical techniques of M3M removal and reporting CAL and/or PD at the distal aspect of the M2M at baseline and at least 6 months of follow-up.

Two independent reviewers evaluated the selected studies for inclusion, abstracted data and assessed risk of bias using the  Cochrane Collaboration tool for assessing risk of bias (Higgins JPT et al. 2011). The primary outcome measures were; clinical attachment level gain (CALg) and probing depth reduction (PDr) with a minimum follow-up of 6 months. The secondary outcome was a Bayesian network meta-analysis model to estimate direct and indirect effects and establish a ranking of treatments.

Results

• 16 RCTs are included in this review and sub-grouped according to the type of M3M removal techniques:
• 10 studies on regenerative/grafting procedures, 3 studies on flap design, 1 study on type of suturing, 2 studies on periodontal care of M2M.
• The overall quality of the studies were low. There were 3 studies at low risk of bias, 8 at unclear risk of bias, and 5 at high risk of bias.
• A total of 317 participants completed the follow-up in these studies, with singular studies reporting complete data for seven to 45 participants.
• In the regenerative/grafting procedures group 9 studies compared the test intervention to a standard control, whereas one study compared two different guided tissue regeneration (GTR) techniques.

Primary outcomes

 Clinical attachment length gain (CALg)

Guided tissue regeneration (resorbable) + Anorganic Xenograft treatment (GTRr+Ax)	2.40 mm, (90% Credible Interval [CrI] = -0.51 to 5.36)
Guided tissue regeneration (non-resorbable) (GTRnr)	1.79 mm, (90% CrI = -0.59 to 4.17)
Guided tissue regeneration (resorbable) (GTRr)	1.59 mm, (90% CrI = -0.23 to 3.55)

 Probing depth reduction (PDr)

Guided tissue regeneration (non-resorbable)	1.65 mm, (90% CrI = -0.69 to 4.00)
Guided tissue regeneration (resorbable) + Anorganic Xenograft treatment	1.50 mm, (90% CrI = -1.41 to 4.38)
Guided tissue regeneration (resorbable)	1.33 mm, (90% CrI = -0.54 to 3.17)

Secondary outcomes

The treatments with the highest ranking were the following:

	CALg	PDr
	Ranking             Best	Ranking           Best
Treatment	Est.     90% CrI      Pr	Est.     90% CrI    Pr
GTRr + AX	2.29     1 to 6        0.45	2.77     1 to 6      0.32
GTRnr	2.80     1 to 6        0.22	2.52     1 to 5      0.30
GTRr	2.99     1 to 5        0.12	2.83     1 to 5      0.15

Conclusion

The authors concluded:-

1) GTR therapies with non-resorbable and resorbable membranes and their grafting-combined therapies achieved some additional clinical benefit compared to standard non-regenerative/non-grafting procedures in terms of CALg and PDr. However, the overall low quality of evidence suggests a low degree of confidence and certainty in treatment effects.

2) The body of evidence for therapies consisting of alternative access flap design, type of suturing, and periodontal care for M2Ms indicates that none of these aspects is decisive on M2M periodontal healing.

3) In view of the importance of M3M extraction as a common dental surgical intervention, more well-designed and well conducted RCTs are necessary to suggest robust evidence-based guidelines.

Comments

This is a very interesting systematic review for a multitude of reasons. Possibly the most important issue that is mentioned, but not answered is the magnitude of periodontal problems suffered by patients following M3M extraction leaving the reader uncertain if CALg and PDr are surrogate outcomes for resolution of future periodontal pathology that may or may-not be symptomatic.

The second issue is whether a Bayesian Network Meta-Analysis (NMA) is an appropriate tool for analysing the data compared to a standard pairwise meta-analysis?(Puhan et al. 2014; Li et al. 2011; Mills et al. 2013). The purpose of NMA is to compare multiple treatment interventions and outcomes for a specific population/treatment. This allows the reader to interpret graphically, wider interactions both directly between treatments A-B for example, and indirectly, via A-C, B-C. This process can then be analyse using a Bayesian approach of prior and posterior probabilities using direct and indirect evidence to create a ranking of treatments. This review question however is not suitable for NMA due to a lack of studies in terms of number and quality to create a viable network, since the majority of the nodes are not well connected with the exception of GBRr and GBRnr. There is also no clear indirect evidence and this leads to poor network geometry and hides the high heterogeneity in the studies. NMA can also distort ranking of treatment effect, because if one critically appraises the quality of the highest ranked paper here (Hassan et al. 2012), GBRr + Ax, one finds it to be severely underpowered and the outcome assessment was not blinded, thus placing it at a ‘high-risk of bias’, plus the wide credible intervals (Bayesian version of confidence intervals) have been reduced to 90% from the more usual 95%.

In fairness, the author points some of these quality issues out and perhaps a head-to-head comparison of NMA and standard meta-analysis on this data-set would have been interesting.

Links

Primary paper

Barbato L, Kalemaj Z, Buti J, Baccini M, La Marca M, Duvina M, Tonelli P. Effect of Surgical Intervention for Removal of Mandibular Third Molar on Periodontal Healing of Adjacent Mandibular Second Molar: A Systematic Review and  Bayesian Network Meta-Analysis. J Periodontol. 2016 Mar;87(3):291-302.

Other references

Coulthard P, Bailey E, Esposito M, Furness S, Renton TF, Worthington HV. Surgical techniques for the removal of mandibular wisdom teeth. Cochrane Database of Systematic Reviews 2014, Issue 7. Art. No.: CD004345. DOI: 10.1002/14651858.CD004345.pub2.

Hassan, K.S., Marei, H.F. & Alagl, A.S., 2012. Does grafting of third molar extraction sockets enhance periodontal measures in 30- to 35-year-old patients? Journal of Oral and Maxillofacial Surgery, 70(4), pp.757–764.

Higgins JPT et al., 2011. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. British Medical Journal, 343, pp.889–893.

Li, T. et al., 2011. Network meta-analysis-highly attractive but more methodological research is needed. BMC Medicine, 9(1), p.79. Available at: http://www.biomedcentral.com/1741-7015/9/79.

Mills, E.J., Thorlund, K. & Ioannidis, J.P. a, 2013. Demystifying trial networks and network meta-analysis. British Medical Journal, 2914(May), pp.10–15.

Moher, D. et al., 2009. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ, 339(jul21 1), pp.332–336.

Puhan, M. a. et al., 2014. A GRADE Working Group approach for rating the quality of treatment effect estimates from network meta-analysis. British Medical Journal, 349, pp.g5630–g5630.