“…However, there was indirect evidence that exposure measurement methods were accurate, as described by the criteria for a rating of “low” risk of bias ( Attarchi et al, 2013 , Bauer et al, 1991 , Chang and Chang, 2009 , Chang et al, 2003 , Chen et al, 2017 , Du et al, 2007 , Guo et al, 2012 , Melamed et al, 1997 , Seixas et al, 2001 , Singh et al, 2012 , Souza et al, 2001 , Starck et al, 1999 , Vihma, 1981 , Whittaker et al, 2014 , Zhao et al, 1991 ). Lastly, twenty eight studies ( Ahmed et al, 2001 , Attarchi et al, 2012 , Chang et al, 2012 , He et al, 2017 , Hong et al, 1998 , Hu et al, 2005 , Ivanovich et al, 1994 , Johnson et al, 2006 , Kock et al, 2004 , Kovacevic and Belojevic, 2006 , Landen et al, 2004 , Lee, 1999 , Lv et al, 2003 , Morata et al, 1997a , Morata et al, 1997b , Nasir and Rampal, 2012 , Noweir, 1984 , Osibogun et al, 2000 , Pawlaczyk-Luszczynska et al, 2016 , Sancini et al, 2014 , Shakhatreh et al, 2000 , Solecki, 2008 , Sriopas et al, 2017 , Toppila et al, 2001 , Wu et al, 1987 , Xie et al, 2015 , Xue et al, 2018 , Yuan et al, 2005 ) were classified as low risk of bias because, in general, the reviewers considered direct evidence that exposure was consistently assessed, using well-established methods that directly measure exposure ( Fig. 3 ).…”