Table 2.

Risk of bias assessment using the Newcastle–Ottawa Scale for 26 studies.

Article	Selection				Comparability		Outcome
Author (year)	Representativeness of the exposed cohorta	Selection of the non-exposed cohortb	Ascertainment of exposurec	Outcome of interest was not present at start of studyd	Adjusted for most important factore	Adjusted for additional factorsf	Assessment of outcomeg	Follow-up long enough for outcomes to occurh	Adequacy of follow up of cohortsi
Umehara Y (2017)			●	●	●	●	●
Tada T (2016)	○	●	●	●			●	●
Takeuchi Y (2015)	●	●	●	●			●	●
Honda T (2015)	●	●	●	●	●	●	●	○
Oze T (2014)	●	●	●	●	●	●	●	○
Ogawa E (2013)	●	●	●	●	●	●	●	○
Arase Y (2013)	○	●	●	●	●	●	●	●	●
Osaki Y (2012)	○	●	●	●	●	●	●	○
Watanabe S (2011)	●	●	●	●	●	●	●	○
Takahashi H (2011)	○	●	●	●	●	●	●	○
Asahina Y (2010)	●	●	●	●	●	●	●	●
Kurokawa M (2009)	●	●	●	●	●	●	●	○
Ikeda K (2007)	●	●	●	●		●	●	●	●
Kobayashi S (2006)	●	●	●	●			●	●
Soga K (2005)	○	●	●	●			●	●
Kim KI (2005)		●	●				●
Moriyama M (2003)	○	●	●	●			●	●
Ohata K (2003)	○	●	●	●			●	●
Tazawa J (2002)	○	●	●	●			●	●	●
Yoneyama K (2002)	○	●	●				●	○	●
Hirashima N (2002)	○	●	●	●			●	○
Yoshida H (1999)	●	●	●	●	●	●	●	○	●
Horiike N (1998)		●	●					●
Miyajima I (1998)	○	●	●	●			●
Onodera H (1997)	○	●	●				●		●
Kuwana K (1997)	○	●	●	●			●	○

^aSingle (○) or multi-center (●) hospital-based study.

^bDrawn from the same community as the exposed cohort.

^cBased on medical records.

^dDemonstrated.

^eAdjusted for fibrosis.

^fAdjusted for age and other factors.

^gBased on medical records.

^hMean follow-up duration > 3 years (○) and > 5 years (●).

ⁱReported number lost to follow-up.