“…Sixty-five studies were judged to be at high risk of bias (78%) [ 13 , 15 , 16 , 19 , 21 , 37 , 41 , 43 , 45 , 46 , 48 , 50 – 54 , 56 – 62 , 65 – 71 , 73 – 78 , 80 , 81 , 83 – 95 , 96 – 100 , 103 – 106 , 111 – 115 ]. Thirty-eight (46%) studies had concerns for high or serious risk of bias for confounding [ 19 , 41 , 42 , 45 , 48 , 52 , 53 , 59 , 62 , 65 , 68 , 71 , 73 – 75 , 77 , 78 , 81 , 83 , 84 , 86 – 88 , 90 , 91 , 93 , 96 , 97 , 102 – 105 , 108 , 111 – 115 ], 54 (65%) studies were judged to be at high or unclear risk for missing data [ 13 , 15 , 16 , 19 , 40 , 41 , 43 ,…”