Is the misinterpretation of association and causation a never-ending story?

Abstract

Dear editor, we congratulate you for the editorial “Opportunistic Dentistry is Harming Our Credibility”.1 We applaud the initiative to revive the discussion on association and causation. Even though the topic has been exhaustively explored, there seems to be a need to extrapolate results in order to get a better audience to journals or newspapers, more patients into the clinics, justify grants, more respect or admiration from other colleagues, and so on. Indeed, opportunistic dentistry is harming our credibility among our peers and patients. Frequently, we are asked by colleagues and patients to elucidate whether causation is present in the almost 100 associations described so far between periodontitis and other health conditions.1 Spurious associations can be easily found when the respect for the existence of biological plausibility and temporal relationship between exposure and outcome are neglected. We recommend reading the case of dental flossing and obesity as an example.2 For the time being, we accept that “there is still no definitive evidence that treating oral disease has any clinically meaningful effect on the prevention, treatment, or outcomes of any systemic disease.”3 It is not surprising that individuals with systemic diseases may present several (un)healthy conditions at the same time, as a consequence of the exposure to and interaction of a multitude of shared risk factors. When a study detects a positive association between periodontitis and a systemic condition, it only shows that this group of people, in addition to periodontitis, also has a higher chance of suffering from the systemic condition under investigation. It is entirely different from saying that the development of periodontitis caused or contributed to the onset and progression of a specific systemic disease. Causal inference studies are scarce in the dental field and, even when found, suffer from relevant methodological limitations in data quality (particularly insufficient sample size and inaccurate analytical approach), which preclude the drawing of causal relationships.