Table 3

Predictive values of secondhand smoke exposure measures during pregnancy along with sociodemographic, pregnancy characteristics for predicting SGA birth (N=545) Guangdong, China, 2009–2011

Variables	AUROC			H-L test p Value
Variables	Estimate (95% CI)	Differences	p Value	H-L test p Value
Model 1 (Sociodemographics†)	0.581 (0.525 to 0.637)	−0.103	0.009	0.571
Model 2 (Sociodemographics+Pregnancy Characteristics‡)	0.665 (0.612 to 0.719)	−0.019	0.619	0.673
Model 3 (Sociodemographics+Pregnancy Characteristics+SHS exposure measures)
Self-report, min/day (as a continuous variable)	0.684 (0.631 to 0.737)	Reference		0.406
Serum cotinine, ng/mL (as a continuous variable)	0.669 (0.616 to 0.722)	−0.015	0.695	1.000
CYP2A64*	0.669 (0.616 to 0.722)	−0.015	0.695	0.577
Self-report+Serum cotinine	0.684 (0.631 to 0.737)	0.000	1.000	1.000
Self-report+CYP2A64*	0.686 (0.634 to 0.738)	0.002	0.958	0.612
Serum cotinine+CYP2A64*	0.671 (0.619 to 0.724)	−0.013	0.732	0.997
Serum cotinine+Self-report+CYP2A64*	0.687 (0.635 to 0.739)	0.003	0.937	1.000
Model 4 (Sociodemographics+Pregnancy Characteristics+SHS exposure measures)
Self-report (as a binary variable§)	0.678 (0.626 to 0.731)	−0.006	0.874	0.164
Serum cotinine (as a binary variable¶)	0.672 (0.619 to 0.724)	−0.012	0.753	0.414
CYP2A64*	0.669 (0.616 to 0.722)	−0.015	0.695	0.577
Self-report+Serum cotinine	0.681 (0.629 to 0.733)	−0.004	0.937	0.063
Self-report+CYP2A64*	0.682 (0.630 to 0.734)	−0.002	0.958	0.338
Serum cotinine+CYP2A64*	0.673 (0.621 to 0.725)	−0.011	0.772	0.540
Serum cotinine+Self-report+CYP2A64*	0.684 (0.632 to 0.735)	−0.001	1.000	0.264

†Maternal age and educational levels.
‡Maternal prepregnancy body mass index, parity, gestational diabetes, gestational hypertension and gestational weight gain velocity during the second and third trimesters (kg/week).
§Conventional cut-off point (0 vs >0 min/day).
¶Conventional cut-off point (<2 vs ≥2 ng/mL).
AUROC, area under the receiver operating curve; SGA, small-for-gestational-age; SHS, secondhand smoke.