Abstract
PURPOSE: Paan (betel leaf and betel nut quid) used with or without tobacco has been positively associated with oral cancer. Oral submucous fibrosis (OSMF), a precancerous condition caused by paan, lies on the causal pathway between paan use and oral cancer. The purpose of this analysis was to estimate the effect of paan consumption on oral cancer risk when it is mediated by OSMF. METHODS: We used mediation methods proposed by VanderWeele, which are based on causal inference principles, to characterize the total, direct, and indirect effects of paan, consumed with and without tobacco, on oral cancer mediated by OSMF. We reanalyzed case-control data collected from three hospitals in Karachi, Pakistan, between July 1996 and March 1998. RESULTS: For paan without tobacco, the total effect on oral cancer was OR 7.39, 95 % CI 1.01, 38.11, the natural indirect effect (due to OSMF among paan users) was OR 2.48, 95 % CI 0.99, 10.44, and the natural direct effect (due to paan with OSMF absent) was OR 3.32, 95 % CI 0.68, 10.07. For paan with tobacco, the total direct effect was OR 15.68, 95 % CI 3.00, 54.90, the natural indirect effect was OR 2.18, 95 % CI 0.82, 5.52, and the natural direct effect was OR 7.27, 95 % CI 2.15, 20.43. CONCLUSIONS: Paan, whether or not it contained tobacco, raised oral cancer risk irrespective of OSMF. Oral cancer risk was higher among those who used paan with tobacco.