Sir, we are responding to the letters 'Yet another cautionary tale?' by P. Hujoel (doi.org/10.1038/s41415-023-5951-1) and 'Very high risk of bias' by A. P. Pires dos Santos, P. Nadanovsky and D. Nunan (doi.org/10.1038/s41415-023-5949-8).

Both Hujoel and Pires dos Santos et al. suggested that our results may be biased due to residual confounding by smoking or health awareness. However, residual confounding can be minimised if the analyses were done correctly and if the investigators made careful adjustments for potential confounding factors.1 To further refute these criticisms, we conducted a restricted analysis among only never-smokers. The results were quite similar to what we had presented in the current paper,1 with HR = 0.52 (0.22-1.23), p = 0.14. The non-significant p-value is due to the sample size reduction arising from restriction to never-smokers as we explained in our book chapter.2 When we adjusted for the health awareness marker, regular dental check-ups, the results were also similar to the results presented in the current publication: HR = 0.50 (0.29-0.88) p = 0.02. Thus, we are confident that good oral hygiene is driving the CVD mortality risk reduction in our study.1 Clearly, oral self-care is inversely related to inflammatory markers such as systemic CRP and oral innate immune marker salivary lysozyme levels. But it is not associated with periodontitis or BMI. See Figure 1.

Fig. 1
figure 1

Oral hygiene self-care and inflammatory markers

Full size image

Pires dos Santos et al. emphasised the limitations of self-reported data. We concur with them, that self-reported data are less reliable. However, oral hygiene self-care does not have any other instrument but self-report. It is unfortunate that these scholars do not recognise the benefits of oral hygiene self-care but only see the deficiencies of self-report. Similarly, physical activities are difficult to assess and especially by self-report.3 However, many studies use the frequency of exercise as a proxy and are accepted as valid.4 Now physical activities are recognised as beneficial for health.5 The same acceptance should be given to self-reported oral hygiene performance.

We would like to point out to Hujoel and Pires dos Santos et al. that the risk reduction of 50% is relative to those who did not 'brush or floss'. It is not an absolute risk reduction. A classic example of differences between 'relative risk' and 'absolute risk' can be found in the JUPITER trial for rosuvastatin [see comment6]. The CVD event rate in the placebo group was approximately 3% and the same in the statin group was 1.6%, thus, although this trial reported a highly significant relative risk reduction of 44%, the absolute risk reduction was only 1.4%. Per our calculation: (251/8901 - 142/8901) x 100 = 1.4%.

We would like to offer a word of caution to Hujoel who wrote 'a failure to take hormone replacement therapy in post-menopausal women caused cardiovascular disease, that insufficient intake of dietary carotenoids caused cancer, and that periodontitis during pregnancy caused adverse pregnancy outcomes'. These are transposition of reported study results. Even if the relationship is causal, increased or decreased risk is not the same as 'disease' or 'non-disease' occurrence. One should not invert the reported results because 'estrogen replacement therapy decreased the risk of CVD' and 'a failure to take hormone replacement therapy caused cardiovascular disease' are two different events in inverse direction as we have explicated.7 We also would like to inform Pires dos Santos et al. that the ROBINS-E tool is not universally accepted as useful.8 Lastly, we thank BDJ and its reviewers for giving us the opportunity to discuss these issues openly and fairly.