Commentary

This investigation is interesting because it attempts to provide data that could be helpful to clinical decision-making. Furcation-involved teeth are harder to treat than nonfurcation-involved ones. Should we offer such treatment to patients and, if so, what can be expected? Furthermore, with tooth replacement options such as dental implants available, how can informed choices be made?

So, how good is the systematic review? First, it is important to note the difficulty of reviewing this topic, resulting from a number of factors: there are no trials, and the available studies are mostly retrospective (with much greater potential for bias) and conducted over a 35-year span. Also, there are substantial differences between the included studies, which limits the potential to generalise conclusions. Despite the problems with the literature, however, the authors were correct to carry out a systematic review in an attempt to produce 'best evidence'.

The strengths of the review are that it brings together a well-established team with appropriate skills for this research. They decided not to conduct meta-analysis on the basis of the heterogeneity of the studies (as well as their potential for bias) and this seems sensible. Carrying out a meta-analysis is possible but would most likely produce both a false sense of precision and a questionable estimate of how well treatments worked.

The limitations of the review are potentially important and require good judgement when interpreting the possible impact of the authors' conclusions. The main issues are the electronic search and its risk of bias/ quality assessment. The search was limited to one electronic database, Medline. It should be routine to search other databases for systematic reviews, in particular Embase (around 40% of journals indexed in Embase are not found in Medline) and CENTRAL, the Cochrane Collaboration database. The risk with searching Medline alone is of missing important data, with resulting greater uncertainty accompanying the review's conclusions. The authors did handsearch four main journals of periodontology from 1975–2008, as well as checking the reference lists of retrieved papers, which is helpful, although unlikely to identify papers published in journals not indexed in Medline. My impression is that they found most of the relevant publications and, in fact, additional studies would not have changed the conclusions.

Quality assessment of included studies is a fundamental property of a systematic review and one that distinguishes it from a traditional review. As a reader, the question that I would like an answer to is, can I trust the findings of the individual studies? For this review, the domains that I would be interested in are risk of bias, use of appropriate statistical methods, were the patients selected (sample) the right ones and was the duration of followup sufficient to detect the outcome (tooth retention)? The authors of this review do not present such an assessment and don't really take the opportunity to interpret the results for us in the light of the quality of the scientific literature. Even without reading the original studies, however, the fact that they are mostly retrospective carries a high risk of bias, which is likely to be in the direction of overly optimistic findings. We cannot comment on appropriateness of statistical methods. The lack of details regarding the types of patients limits our knowledge of the generalisability of the conclusions, although studies represent both university and private specialist clinics. The duration of followup was good: 5 years would seem a sensible minimum period.

Overall, then, this review is valuable both in identifying the literature on the effect of periodontal treatment on retention of furcation involved teeth. The data suggest high levels of tooth retention comparable with implant-supported restorations in the posterior mandible and maxilla. The level of evidence is not high and there may be an overestimation of true retention rates, but similar comments are applicable to much of the relevant implant literature.