As reported before, the large muticenter intervention trial by Engebretson et al. (2013) who reported no effect of nonsurgical treatment of periodontitis on HbA1c levels in patients with type 2 diabetes mellitus has harshly been critized by our thought leaders. Last month, JAMA published a number of letters to the editor. One letter by Chapple, Borgnakke and Genco identified important problems in Engebretson’s paper including “problems with the study design, execution, data interpretation and reporting.”
“First, the periodontal therapy provided failed to clinically manage the periodontal infection and associated inflammatory burden. Residual plaque levels of 72% and bleeding scores of 42% are far below the consenus for expected outcomes [reference to van der Weijden et al., J Clin Periodontol 2002; 29(suppl 3): 55-71, 90-91]. Therefore, no conclusions can be drawn about the effect of clinicaly effective periodontal therapy on HbA1c in patients with type 2 diabetes.
Second, control of diabetes at baseline was predominantly good (mean HbA1c levels, 7.8%), with less than 60% of patients having HbA1c levels greater than 8.0% (HbA1c level <9.0% was an inclusion criterion). With the mean HbA1c value close to the therapeutic target, we would not expect an intervention to improve HbA1c substantially.
We are concerned about the reliance on statistical significance to justify a conclusion of no effect when the clinical therapy failed to deliver the expected standard of care.”
Drs Engebretson, Hyman and Michalowicz replied, see [pdf] here,
“We disagree with Dr Chapple and colleagues that the trial was flawed based on failure to adequately treat periodontitis and enrollment of participants with predominantly good glycemic control. Nonsurgical periodontal therapy is considered the cornerstone of periodontal therapy and is known to improve clinical and microbiological measures of disease, regardless of initial severity. We found no indication that change in HbA1c was associated with the magnitude of the periodontal treatment response (eTable 1 in the article).
“Although not reported in our article, we also found no difference in the treatment effect in groups stratified by baseline HbA1c values (P= .83 for interaction between treatment effect and baseline HbA1c).”
One has to download the table separately (see below) which is somewhat hidden in the supplementary material of the original article.
Engebretson et al. (2013) are right that patients with best response to periodontal treatment (greatest probing depth reduction, clinical attachment gain, and reduction of sites bleeding on probing) after 6 months did not differ in HbA1c changes.
It has to be stressed, though, that high plaque levels after treatment which further increased after 6 months, and very moderate bleeding on probing reduction must be considered an issue of concern. The large study have had the potential to ultimately confirm consistent conclusions made in several recent systematic reviews (Engebretson and Kocher 2013, Teeuw et al. 2010, Simpson et al. 2010) that periodontal therapy (nonsurgical, surgical, adjunct antibiotics) slightly reduces HbA1c levels by 0.36-0.4%, or to nullify these observations made in meanwhile 9 rather small trials.
Clearly, the study by Engebretson et al. (2013) stands out due to its mere size. It is the only study in which the mean effect was positive (although not significant). A meta-analysis employing a fixed effects model (FE) of all ten studies still revealed a slight but significant reduction of HbA1c 3 to 6 months after periodontal therapy of -0.1689% (95% confidence interval -0.2783, -0.0595). I2 of 49% is much larger than that reported by Engebretson and Kocher for nine studies (9%) and indicates large heterogeneity after inclusion of the new study by Engebretson et al. (2013).
Engebretson et al. (2013) had reported that nonsurgical periodontal therapy was completed after an average of 160 minutes of scaling and root planing with local anesthesia in 2 or more sessions. This is quite something. Clinical improvement of periodontal parameters is a benefit in itself. A 0.17% reduction of HbA1c in the present meta-analysis is not. Our “thought leaders” may face serious problems when new and much larger intervention studies are to be approved by ethical committees or funded by third parties, just in order to challenge the current conclusions. Let alone possible volunteers who might be denied proper periodontal treatment.
So, despite the fact that Engebretson et al. (2013) did not achieve an optimal result as regards periodontal health, their study may have the potential to end the story: Small periodontal treatment effects on HbA1c in diabetics are undeniable but hardly clinically relevant.
9 June 2014 @ 7:45 pm.
Last modified June 9, 2014.