S Petti1, M Masood, G A Messano, C Scully. 1. Department of Public Health and Infectious Diseases, Sapienza University, Rome, Italy. stefano.petti@uniroma1.it
Abstract
BACKGROUND: Alcohol drinking is associated with high oral cancer (OC) risk. This association is particularly evident in tobacco smoking/betel quid (BQ) chewing subjects. In a previous stratified meta-analysis (Petti S et al., Cancer Epidemiol 2012) we reported that drinking was inversely associated with OC in non-smoking BQ non-chewing individuals, while this association was reversed in smoking individuals. However, the previous study could be excessively influenced by a large primary study, which yielded more than 50% of the weight of all the primary studies. Therefore, we updated this analysis using only recent studies. METHODS: Case-control studies published between 2010 and 2012 were searched. In each study, non-smoking/ BQ non-chewing exposed (ever routine drinkers) and unexposed (never drinkers) subjects with (cases)/without (controls) OC were extracted and odds ratio (OR) calculated. Between-study heterogeneity was assessed with Cochran's Q. Publication bias was formally assessed with trim and fill method. Sensitivity analysis to inclusion criteria was made. The pooled OR was assessed with the fixed- and random-effect methods and corrected for publication bias. RESULTS: Seven of these studies met the inclusion criteria: they were not heterogeneous enough. Correction for publication bias was not necessary and provided only one missing study. The OR estimates were 0.70 (95% confidence interval -95CI, 0.51-0.98), 0.70 (95CI, 0.51-0.96), 0.75 (95CI, 0.54-1.03) with the three methods. Sensitivity analysis did not change the OR estimates considerably. CONCLUSIONS: This analysis corroborated the results of the previous analysis, confirming that drinking was inversely associated with OC in non-smoking, BQ non-chewing subjects.
BACKGROUND:Alcohol drinking is associated with high oral cancer (OC) risk. This association is particularly evident in tobacco smoking/betel quid (BQ) chewing subjects. In a previous stratified meta-analysis (Petti S et al., Cancer Epidemiol 2012) we reported that drinking was inversely associated with OC in non-smoking BQ non-chewing individuals, while this association was reversed in smoking individuals. However, the previous study could be excessively influenced by a large primary study, which yielded more than 50% of the weight of all the primary studies. Therefore, we updated this analysis using only recent studies. METHODS: Case-control studies published between 2010 and 2012 were searched. In each study, non-smoking/ BQ non-chewing exposed (ever routine drinkers) and unexposed (never drinkers) subjects with (cases)/without (controls) OC were extracted and odds ratio (OR) calculated. Between-study heterogeneity was assessed with Cochran's Q. Publication bias was formally assessed with trim and fill method. Sensitivity analysis to inclusion criteria was made. The pooled OR was assessed with the fixed- and random-effect methods and corrected for publication bias. RESULTS: Seven of these studies met the inclusion criteria: they were not heterogeneous enough. Correction for publication bias was not necessary and provided only one missing study. The OR estimates were 0.70 (95% confidence interval -95CI, 0.51-0.98), 0.70 (95CI, 0.51-0.96), 0.75 (95CI, 0.54-1.03) with the three methods. Sensitivity analysis did not change the OR estimates considerably. CONCLUSIONS: This analysis corroborated the results of the previous analysis, confirming that drinking was inversely associated with OC in non-smoking, BQ non-chewing subjects.