One has to praise the authors of the systematic review on “The epidemiological evidence behind (sic!) the association between periodontitis and incident atherosclerotic cardiovascular disease” (Dietrich et al. 2013) [pdf] that they had managed to get the awful print-out of their paper quickly corrected, see the Corrigendum here. First I had even entertained suspicion that results of the systematic search should be rather concealed than reported. Tables in the Corrigendum are now more informative and can be scrutinized by attentive readers who are still interested in the above question (I have to admit, though, that I am more and more bored of repeatedly published “strong evidence for minor effects”).
Now, Table 3 seems in fact to reproduce Dietrich’s and coworkers’ original findings of their cohort study of 2008 [pdf] (by and large either not significant or inconclusive associations of bone loss or cumulative probing depth and incident coronary heart disease) with an important exception: figures were rounded to only one decimal place. Authors should know that rounding can only be done once, namely by the investigators of the original paper. Systematic reviews of selected papers must not produce new results by rounding. Never.
Any associations in the selected, by Dietrich et al. (2013), cohort studies (a design which might allow for at least some speculation on causation) were either not significant, inconclusive or weak at best. Authors did not attempt to do a meta-analysis. But I would bet if they had done it, the overall risk for incident atherosclerotic cardiovascular disease would be increased, based on the reviewed cohort studies, by a factor of not more than 1.3 (possibly not even significant) in case of more advanced periodontitis. A risk ratio which had been claimed already a decade or more ago.
The association in the selected case-control studies was seemingly stronger. But should they even be considered in evidence-based medicine for anything else than formulating a hypothesis which has then to be tested in well-conducted prospective cohort studies or ultimately intervention studies to eventually answer the question about causation which seems to plague a number of periodontists for more than 25 years? Actually, in the presence of new cohort studies they should no longer be considered. It is all about Sir Bradford Hill’s criteria for causation, decades old, heavily criticized but still valid. And then, if cohort studies won’t suggest strong association, then intervention studies are not justified. Keep in mind that, in any RCT, hundreds of patients with more or less severe periodontitis must be randomized and half of them be deprived of any proper periodontal therapy for years in order to prove a so far not even calculated but probably very low attributable risk.
Since the attributable risk, which describes the proportion of all cases of the outcome (incidence of atherosclerotic cardiovascular disease) in the target population that are attributable to the exposure (periodontitis), depends on the prevalence of the risk factor, recent rather questionable attempts to exaggerate prevalence, extent and severity of periodontitis in the adult U.S. population by Eke et al. (2012) are directly related to the current AAP/EFP recover mission.
29 May 2013 @ 6:43 pm.
Last modified May 30, 2013.