Can Unmeasured Confounders explain Oral Health-Head and Neck Cancer association?
Elango P1, Nicolau B1, Farsi N2, Grant AV3, Rousseau MC4, Madathil S1
1Faculty of Dental Medicine and Oral Health Sciences, Ï㽶ÊÓƵ, Montreal, QC, Canada; 2Department of Preventive Dental sciences, King Abdul Aziz University, Jeddah, Saudi Arabia 3Department of Anaesthesia, Faculty of Medicine and Health Sciences, Alan Edwards Centre for Research on Pain, Ï㽶ÊÓƵ, Montreal, QC, Canada 4Epidemiology and Biostatistics Unit, Centre Armand-Frappier Santé Biotechnologie, Institut national de la recherche scientifique, Laval, Canada.
Background: Studies have reported strong associations between oral health indicators and head and neck cancer (HNC). However, the validity of these results has been challenged due to potential
mediators, and unmeasured risk factors that influence these links. Negative controls, a tool routinely being employed to detect bias in observational studies, are widespread in epidemiology but notably
absent in oral health research.
Objective: To estimate the extent to which the association between oral health indicators and HNC risk is due to unmeasured confounders using a negative control exposure analysis.
Methods Data were drawn from a hospital-based case-control study conducted in four main referral hospitals in Montreal, Canada. HNC incident cases (n=389) and controls (n=429) frequency matched
for sex and age were recruited from the same hospitals. In-person interviews collected information on several life course exposures. Oral rinse and brush specimens were analysed for human papillomavirus (HPV) infection. We estimated odds ratios (OR) and 95% confidence intervals (CI) for the associations between oral health indicators and HNC using logistic regression, adjusting for tobacco, alcohol and HPV. Sexually Transmitted Diseases (STDs), which has not been associated with HNC before, was the negative control exposure. The association between STDs and HNC risk was adjusted for confounders including HPV and assumed a null association. Any association observed between STDs and HNC risk would indicate bias due to unmeasured confounders in the association between oral health and HNC.
Results- Complete denture [OR= 1.33, 95%CI (0.93-1.90)] and having more than nine missing teeth [OR=1.31, 95%CI (0.93-1.83)] were associated with HNC risk. Similar results were obtained when
stratified by HNC subsite. Negative control analysis yielded a null finding, indicating no considerable bias due to unmeasured confounders.
Conclusion- Our findings support the associations between oral health indicators and HNC risk observed in previous studies.
Keywords: Head and Neck Cancers, Oropharyngeal Cancers, Oral Cancers, Epidemiology, Negative Controls, Bias Analysis