“…Within groups according to symptom status, heterogeneity was mostly low (Table 2 ). The OR for preterm birth was 1.44 (95% CI 0.92–2.26, I 2 0%) in two studies with ≥ 50% participants with symptoms [ 24 , 29 ], 0.84 (95% CI 0.45–1.58, I 2 88%, prediction interval 0.10–6.81) in seven studies with < 50% participants with symptoms [ 43 , 45 – 47 , 65 , 78 , 79 ], 1.12 (95% CI 0.94–1.35, I 2 0%, prediction interval 0.75–1.68) in four studies only with participants without self-reported or observed symptoms [ 55 , 56 , 70 , 82 , 83 ], and 1.05 (95% CI 0.87–1.26, I 2 2%, prediction interval 0.70–1.57) in 17 studies in which symptom status was not reported [ 27 , 28 , 32 , 37 , 40 , 44 , 50 , 53 , 61 , 63 , 64 , 66 , 68 , 72 , 74 – 76 ]. The prediction intervals for all groups with sufficient data included the null value.…”