Table 2.

Risk of bias summary.

Study, year	Sequence generation	Allocation concealment	Participants and assessor blinding	Treatment provider blinding	Incomplete outcome data addressed	Free of selective reporting	others
Chen and Li [30] (2012)	Low	Unclear[b]	High[c]	High	Low	Low	Low
Inoue et al. [31] (2006)	Low	Low	Low	High	Low	Low	Low
Inoue et al. [32] (2009)	Low	Unclear[b]	High[c]	High	Low	Low	Low
Liu et al. [33] (2014)	Low	Low	Low	High	Low	Low	Low
Lu et al. [34] (2010)	Unclear[a]	Unclear[b]	Low	High	Low	Low	Low
Mejuto-Vázquez et al. [35] (2014)	Low	Low	High[d]	High	Low	Low	Low
Maeda et al. [36] (2013)	Unclear[a]	Unclear[b]	Unclear[e]	High	Low	Low	High[f]
Nabeta and Kawakita [37] (2002)	Low	Unclear[b]	Low	High	Low	Low	Low
Shin et al. [38] (2013)	Low	Low	High[c]	High	Low	Low	Low
Stival et al. [39] (2014)	Low	Unclear[b]	Low	High	Low	Low	Low
Su et al. [40] (2010)	Low	Low	Low	High	Low	Low	Low
Yang et al. [41] (2012)	Low	Low	Low	High	Low	Low	Low
Zhang et al. [42] (2015)	Low	Low	Low	High	Low	Low	Low

^[a]Lu et al. 2010 and Maeda et al. 2013 RCT claimed to have randomly assigned participants but did not describe the methods in detail; ^[b]Chen and Li 2012, Inoue et al. 2009, Maeda et al. 2013, Lu et al. 2010, Nabeta and Kawakita 2002, and Stival et al. 2014 did not mention allocation concealment; ^[c]Chen and Li 2012, Inoue et al. 2009, and Shin et al. 2013 compared acupuncture versus analgesia injection, and the participants, who were also the outcome assessors, could not be blinded; ^[d] Mejuto-Vázquez et al. 2014 compared acupuncture versus no treatment, and the participants, who were also the outcome assessors, could not be blinded; ^[e]Maeda et al. 2013 RCT used nonpenetrating sham acupuncture as control but did not evaluate the credibility of the sham; ^[f]for Maeda et al. 2013 RCT, the baseline was not comparable in the two groups.