Table 4.

Risk of Bias Classification

Risk of Bias	High Risk	Low Risk	Uncertain Risk
Random sequence generation	Dongmei et al., 2011 [19] Jin et al., 2013 [20] Tsai et el., 2017 [21]	None	None
Allocation concealment	Dongmei et al., 2011 [19] Jin et al., 2013 [20] Tsai et el., 2017 [21]	None	None
Blinding of participants and professionals	Dongmei et al., 2011 [19] Jin et al., 2013 [20] Tsai et el., 2017 [21]	None	None
Blinding of outcome evaluators	None	Dongmei et al., 2011 [19] Jin et al., 2013 [20]	Tsai et el., 2017 [21]
Incomplete outcomes data	None	Dongmei et al., 2011 [19] Jin et al., 2013 [20] Tsai et el., 2017 [21]	None
Selective outcome reporting	None	Dongmei et al., 2011 [19] Jin et al., 2013 [20] Tsai et el., 2017 [21]	None

The risk of bias was graded for each domain as high, low, or unclear. For incomplete outcome data in individual studies, we stipulated a low risk of bias for a loss to follow-up of less than 10% and a difference of less than 5% in missing data between the intervention/exposure and control groups