“…Forty-two RCTs were rated as having a low risk of bias in random sequence generation: Of them, 26 [ 25 , 30 – 33 , 35 – 38 , 40 , 41 , 45 , 47 , 51 , 53 , 54 , 60 , 62 – 64 , 67 , 71 , 74 , 79 , 87 , 88 ] used various computerized randomization programs, 11 [ 43 , 44 , 50 , 55 , 58 , 66 , 68 , 72 , 78 , 81 , 85 ] used random number tables, 3 [ 27 – 29 ] used coin tossing, and 2 [ 24 , 39 ] used block randomization. In terms of allocation in the RCTs, 19 studies [ 24 , 26 , 30 – 32 , 38 – 41 , 45 , 58 , 62 , 64 , 67 , 71 , 79 , 84 , 87 , 88 ] described proper allocation concealment (the use of sealed envelopes or independent researchers). Because of the nature of the interventions, performance bias was high in most studies; only 2 studies [ 41 , 47 ] were assessed as having a low risk of bias in participant blinding due to the use of nonpenetrating SA.…”