Claims and denials of clinically relevant effects of, in particular, non-surgical periodontal treatment on markers of diabetic control have not only led to a surge of new randomized clinical trials and systematic reviews thereof. If anybody had hope that the current frenzy has found a happy end with the updated and very comprehensive Cochrane review by Simpson et al. (2015) (s)he has been mistaken. In the June issue of the Australian Dental Journal, Botero et al. (2016) report on an umbrella review in which they systematically reviewed all systematic reviews on the subject, be it with or without meta-analysis, published between 1995 and 2015. The paper has been accepted for publication on January 20, 2016. It has to be emphasized that using the term “umbrella review” is somewhat misleading. In a strict sense, an umbrella review assembles together several systematic reviews on the same condition in the presence of many treatments or many important outcomes.
Botero et al. (2016) claim having considered systematic reviews of studies in which non-surgical periodontal treatments were compared with no (or delayed) treatment, and plasma HbA1c as the primary outcome. They mention (as nuisance?) that the updated Cochrane review by Simpson et al. (2015) also considered experiments in which surgical intervention was compared to no treatment. Botero et al. suggest that the analysis in the systematic review by Simpson et al. “may be biased since the authors included mostly low quality and heterogeneous studies.” Botero’s criticism of, in particular, Simpson’s most comprehensive systematic review is certainly hypocritical. It is appropriate and not a flaw to strive for all published studies in a systematic review (and even unpublished and “grey literature”), what they did when they considered 35 RCTs with altogether 2565 participants. That they also included as yet unpublished information provided by the authors after identifying a related abstract is justified despite the fact that bias assessment of that particular study might be difficult. After all, patient have been bothered in a randomized trial and ethics demand publication.
What is of considerable concern in the paper by Botero is inclusion of an unsuitable systematic review (given the aforementioned PICO), the paper by Wang et al. (2014). These authors mention that they had searched for studies in which different periodontal treatments had been compared as regards primary outcome plasma HbA1c, not one periodontal treatment as compared to no (or delayed) treatment. So, what they found was four studies in which scaling and root planing (SRP) with adjunctive systemic doxycycline was compared with SRP alone. They were able to conduct a meta-analysis of three studies and yielded a non-significant standardized mean difference of -0.238 in favor of SRP plus doxycycline (95% confidence interval -0.616; 0.140, Q-statistic 1.093, I2 0%, p=0.579). They conclude that,
adding doxycycline to periodontal therapy with SRP does not significantly improve metabolic control in patients with T2DM [type 2 diabetes mellitus] and chronic periodontitis. Currently, available evidence is insufficient to support a significant association between periodontal therapy and metabolic control in T2DM patients with PD [periodontal disease], however, evidence suggests that periodontal therapy itself improves metabolic control.
Botero et al. (2016) erroneously claim that Wang et al. (2014) had included a control group with no periodontal treatment.
Now, Faggion et al. have launched another review article of systematic reviews on the same topic which has been accepted for publication on November 28, 2015 in the Journal of Periodontal Research and is still in press. They had exactly the same search strategy as Botero et al. No wonder that they ended up with almost the same number of articles assessed for their systematic review of systematic reviews. It is worth emphasizing that they strangely missed a systematic review by Li et al. which was published on July 3, 2015; and the very comprehensive Cochrane review by Simpson et al. which was published on November 6, 2015. They also included one systematic review (see comments below) which was characterized by Botero et al. as having applied “inadequate methods to combine and analyse the included studies. Inclusion criteria for studies are not clear and several non-RCT [randomized clinical trials] that could have produced bias in the results.”
The apparent main difference as regards Botero’s umbrella review is that Faggion et al. did not further consider articles which did not include meta-analyses. But they did consider Wang et al. (2014), which would not suit their research question (“HbA1c following periodontal treatment compared to those who did not receive any periodontal treatment,” emphasis added). In the studies considered by Wang et al., all patients received periodontal treatment. Faggion et al. quote as “Authors’ [Wang et al.] conclusions”, “Available evidence is insufficient to support a significant association between periodontal therapy and metabolic control in diabetes type 2 patients with periodontitis,” but that quote refers to evidence from trials and systematic reviews (not considered in the systematic review by Wang et al.) on that particular question. Immediately before that quote, Wang et al. actually conclude from their results, namely that “adding doxycycline to periodontal therapy with SRP does not significantly improve metabolic control in patients with T2DM and chronic periodontitis.” Faggion et al. didn’t get the point here (and neither did Botero et al.).
In both instances (Botero et al., Faggion et al.), reviewers appraised quality of systematic reviews by the AMSTAR checklist. It “consists of 11 checklist items. Each item is given a score of 1 if the specific criterion is met, or a score of 0 if the criterion is not met, is unclear, or is not applicable. An overall score relating to review quality is then calculated (the sum of the individual item scores). AMSTAR characterizes quality at three levels: 8 to 11 is high quality, 4 to 7 is medium quality, and 0 to 3 is low quality” (Botero et al. 2016).
Botero’s and Faggion’s AMSTAR scores for all 11 systematic reviews which were assessed by both groups differ largely, which is of further concern. One might calculate Spearman’s rank correlation coefficient which is just 0.28. What is the purpose of AMSTAR when two groups of reviewers would not agree when assigning scores to questions of a simple checklist? In particular, an above mentioned study which, according to Botero et al., had “applied inadequate methods to combine and analyse the included studies” and was thus given a score of 3, yielded a score of 9 in Faggion’s paper. Faggion et al. furthermore consistently denied that any systematic review they had reviewed was not prone to conflict of interest. The respective AMSTAR item (“Was the conflict of interest included?”) states that ,”To get a ‘yes,’ must indicate source of funding or support for the systematic review AND for each of the included studies”. Botero et al. have apparently overlooked the AND as they assigned to several systematic reviews a maximum AMSTAR score of 11.
Faggion et al. calculate, from their 11 systematic reviews with meta-analyses, an average HbA1c reduction of -0.46% three months after periodontal therapy (range -0.24%, the mean value in the paper by Wang et al. which should not have been included in Faggion’s overview; to -1.03%). Given the fact that Simpson’s very comprehensive Cochrane review of 2015 (with a mean reduction of HbA1c of 0.29%) was published just days before Faggion’s paper had been accepted for publication (and thus inevitably had to be ignored) but another systematic review by Lie et al (2015) was missed by Faggion et al. (with mean HbA1c reduction of -0.27% based on a meta-analysis of 9 RCTs; note that a sensitivity analysis yielded no effect at all when only studies were considered with sufficient sample size of 80 patients or more), the current average HbA1c reduction may in fact be much lower.
The question may be allowed, why should anybody trust an average from 11 meta-analyses of smaller and bigger systematic reviews (with numerous RCTs considered in several systematic reviews and the most recent systematic reviews were omitted from the analysis) when a recent Cochrane review had looked at 35 RCTs and authors were able to calculate a robust estimate of the reduction, -0.29% (95% CI .0.48; -0.10)? According to a very revealing table in Botero’s paper, Simpson et al. (2015) actually covered most if not all RCTs assessed in one way or the other in a total of 12 other systematic reviews. So, what’s the meaning of continued interest in older systematic reviews?
Faggion’s questionable average repeats what one of our main professional organizations, the European Federation of Periodontology, has claimed in their now outdated Manifesto, approximately 0.4% HbA1c reduction after periodontal therapy, when the true effect seems to be rather below 0.3%.
Randomised clinical trials consistently demonstrate that mechanical periodontal therapy associates with approximately a 0.4% reduction in HbA1C at 3-months, a clinical impact equivalent to adding a second drug to a pharmacological regime for diabetes.
Thus, the overview of systematic reviews about the effect of non-surgical periodontal therapy on diabetic control must be seen as an undesired attempt to reset old claims which have been put into perspective after the update of a 2015 Cochrane review by Simpson et al. who conclude,
There is low quality evidence that the treatment of periodontal disease by SRP does improve glycaemic control in people with diabetes, with a mean percentage reduction in HbA1c of 0.29% at 3-4 months; however, there is insufficient evidence to demonstrate that this is maintained after 4 months.There was no evidence to support that one periodontal therapy was more effective than another in improving glycaemic control in people with diabetes mellitus. In clinical practice, ongoing professional periodontal treatment will be required to maintain clinical improvements beyond 6 months. Further research is required to determine whether adjunctive drug therapies should be used with periodontal treatment. Future RCTs should evaluate this, provide longer follow-up periods, and consider the inclusion of a third ‘no treatment’ control arm. Larger, well conducted and clearly reported studies are needed in order to understand the potential of periodontal treatment to improve glycaemic control among people with diabetes mellitus. In addition, it will be important in future studies that the intervention is effective in reducing periodontal inflammation and maintaining it at lowered levels throughout the period of observation.
3 August 2016 @ 10:59 am.
Last modified August 3, 2016.