“…After age (continuous)-adjusted HRs and 95% CIs were calculated (Model 1), we further adjusted the models with baseline smoking and drinking status (current, past, never), BMI (kg/m 2 ) (< 18.5, 18.5-25, ≥ 25), education level (attended school up to 18 years old, yes or no), family history of any kind of cancer (yes or no), fish consumption ('less than 1-2 times/week', '3-4 times/ week', 'almost daily'), menopausal status (yes or no), female sex-hormone used (ever, never), gynecological cancer screening (yes or no), gravid, and parity ('never', '1', '2', '3', ' ≥ 4') (Model 2). These variables were selected as covariates because they were known or suspected to confound the association (Cady et al, 1979;Sokic et al, 1994;Bosetti et al, 2001;Mack et al, 2002Mack et al, , 2003Pham et al, 2009;Santin and Furlanetto, 2011;Schonfeld et al, 2011;Kabat et al, 2012aKabat et al, , 2012bMarcello et al, 2012;Przybylik-Mazurek et al, 2012;Cash et al, 2013;Dong et al, 2013;Jung et al, 2013;Schmid et al, 2013;Brown et al, 2014;Zamora-Ros et al, 2014). Initially, we included history of blood transfusion and/or liver diseases in the final model because they were associated with a higher incidence of thyroid cancer in this population (Fujino et al, 2004).…”