... The remaining 25 studies controlled for smoking either by direct adjustment, stratification or by including socioeconomic status as a proxy variable for tobacco smoking (Chen et al., 2010;D'Ippoliti et al., 2015;Chen et al., 2004;Chiou et al., 1995;Yang et al., 2013;Chung et al., 2013;Ferreccio et al., 2000;Steinmaus et al., 2013;Steinmaus et al., 2014a;Mendez Jr. et al., 2017;Chen et al., 1986;Mostafa et al., 2008;Heck et al., 2009;Steinmaus et al., 2010;Dauphine et al., 2013;Ferreccio et al., 2013;Melak et al., 2014;Steinmaus et al., 2014b;Baastrup et al., 2008;Garcia-Esquinas et al., 2013;Hsu et al., 2013;Argos et al., 2014;Wu et al., 2016;Hsu et al., 2017;Nuvolone et al., 2023). Among the 15 studies included in the meta-analysis (Hopenhayn-Rich et al., 1998;Tsuda et al., 1995;Chen et al., 2004;Chen et al., 1988a;Chiou et al., 1995;Yang et al., 2013;Chung et al., 2013;Morales et al., 2000;Su et al., 2011;Tsai et al., 1999;Ferreccio et al., 2000;Steinmaus et al., 2013;Steinmaus et al., 2014a;Roh et al., 2018;Ferdosi et al., 2016), eight (six ecological and two cohort) studies did not account for tobacco smoking (Hopenhayn-Rich et al., 1998;Tsuda et al., 1995;Chen et al., 1988a;Morales et al., 2000;Su et al., 2011;Tsai et al., 1999;Roh et al., 2018;Ferdosi et al., 2016). Although the remaining studies accounted for tobacco smoking, more than half of these did not control for other important covariates, including socioeconomic status or other diseaserelated characteristics (e.g., body mass index, family history of cancer, or comorbidities) (Chiou et al., 1995;Yang et al., 2013;Chung et al., 2013;Steinmaus et al., 2014a). ...