“…Figure 5B). Finally, for risk of attrition bias, after excluding the subgroup of two studies with unclear attrition bias ( Fujimoto et al, 2017;Hilgenstock et al, 2016), the effect for the remaining subgroup with low risk (Fujimoto et al, 2014;Godde et al, 2020;Hirtz et al, 2018;Ragert et al, 2008;Saito et al, 2019;Swissa et al, 2022;Yau et al, 2014) remained significant (scenario 1: Z = 4.36; SMD = −4.75; 95 % CI, − 6.88 to -2.61; P<0.0001; scenario 2: Z = 4.15; SMD = −4.81; 95 % CI, − 7.09 to -2.54; P<0.0001; n = 124; for detailed statistics, cf. Figure 5C).…”