A quantitative bias analysis to assess the impact of unmeasured confounding on associations between diabetes and periodontitis
(Journal of Clinical Periodontology; doi: 10.1111/jcpe.13386)
Talal S. Alshihayb, Elizabeth A. Kaye, Yihong Zhao, Cataldo W. Leone, Brenda Heaton
Two conditions can occur together partly because they share influences such as smoking, age or social circumstances. This study examined how those influences can change estimates of the relationship between diabetes and periodontitis.
Researchers used repeated medical and dental examinations from a long-running study of mainly white men in Boston. They analysed both directions of the relationship and used statistical simulations to explore the possible effects of an unmeasured factor.
Estimated associations changed after accounting for measured factors and remained uncertain. The simulations showed that assumptions about missing information could alter the apparent strength of the link. Both main estimates were compatible with no clear association in this particular dataset. Simulations themselves depend on assumptions and do not reveal exactly which hidden factors were present.
This is mainly a study about interpreting research, not a new treatment finding. It explains why one observational study cannot prove that gum disease causes diabetes or vice versa. Considering smoking or age statistically helps, but does not remove every source of uncertainty. The findings should encourage careful reading of headlines, while decisions about an individual’s dental and diabetes care remain based on their actual clinical needs.
The main contribution is to show how research methods affect the conclusions we draw. Results depend on the data available, the measurements chosen and the assumptions made in analysis. A more detailed method may reduce one problem without removing every possible source of bias. These findings do not measure the benefit of a new treatment. For a patient reading research, useful questions are whether disease was measured reliably, whether comparison groups were similar and whether the conclusions match what the study actually tested.
Read the full article Back to Research
Periodontitis is the 6th most prevalent condition globally
Periodontitis and diabetes are bidirectionally linked
Diabetic complications are increased if you have both diseases
Successful periodontal treatment can improve blood glucose control
Periodontitis is the 6th most prevalent condition globally
Periodontitis and diabetes are bidirectionally linked
Diabetic complications are increased if you have both diseases
Successful periodontal treatment can improve blood glucose control
