Table 2

Summary of pooled odds ratios. Both fixed (FEM) and random (REM) effects models are shown

Group of studiesModelPooled unadjusted odds ratios (95% CI)Test for heterogeneity Pooled adjusted odds ratios (95% CI)Test for heterogeneity
Prenatal smoking
(all studies)
REM2.74 (2.42 to 3.10) (n = 33)χ2 (df = 32) = 120.7 (p<0.001)2.13 (1.86 to 2.43) (n = 18)χ2 (df = 17) = 54.0 (p<0.001)
FEM2.74 (2.59 to 2.90)2.17 (2.04 to 2.31)
FEM
2.82 (2.38 to 3.33) (n = 15)
2.89 (2.70 to 3.09)
χ2 (df = 14) = 59.1 (p<0.001)2.18 (1.89 to 2.51) (n = 15)
2.20 (2.05 to 2.36)
χ2 (df = 14) = 40.7 (p<0.001)
Prenatal smoking (cohort studies)REM2.75 (1.97 to 3.82) (n = 7)χ2 (df = 6) = 22.2 (p = 0.001)(n = 4)1-150 χ2 (df = 3) 14.5 (p = 0.002)
FEM2.39 (2.15 to 2.65)2.04 (1.84 to 2.25)
Prenatal smoking (case-control studies)REM2.73 (2.37 to 3.16) (n = 26)χ2 (df = 25) = 78.8 (p<0.001)2.22 (1.88 to 2.61) (n = 14)χ2 (df = 13) 37.0 (p<0.001)
FEM2.89 (2.71 to 3.09)2.26 (2.09 to 2.45)
Postnatal smokingREM2.80 (2.00 to 3.93) (n = 9)χ2 (df = 8) = 35.0 (p<0.001)(n = 4)1-150 1-151 χ2 (df = 3) = 1.18 (p= 0 .76)
FEM3.10 (2.70 to 3.56)1.94 (1.55 to 2.43)
• Schoendorf study results37 were analysed separately for black and white subjects and in all these analyses are counted as two separate studies.

• 1-150 Below the minimum of five studies for estimation of random effects.

• 1-151 Excludes reference 46 which did not control for prenatal smoking.