“…The reporting quality was classified as high in 15 studies (75.0%), while five studies (25.0%) had medium reporting quality [ 62 , 66 , 71 , 72 , 73 ]. The risk of bias was classified as very high in two studies [ 58 , 71 ], high in six studies [ 59 , 62 , 65 , 66 , 70 , 72 ], with some concerns in three studies [ 56 , 61 , 68 ], and low in nine studies [ 43 , 57 , 60 , 62 , 63 , 67 , 69 , 73 , 74 ] ( Figure 2 ). Taken together, the higher concerns were due to confounding, the measurement of the exposure, and missing data, as shown in Figure 3 .…”