Table 3.

Fully adjusted combined associations^a between air pollution during pregnancy^b and autistic traits within the borderline/clinical range.

Pollutant	Autistic traits within the borderline/clinical range				Autistic traits within the clinical range
	nc	OR (95% CI)	p-Heter	I2	nc	OR (95% CI)	p‑Heter	I2
NO2 (per Δ10 μg/m3)	6	0.95 (0.81, 1.10)	0.431	0.00%	4	0.87 (0.67, 1.14)	0.955	0.00%
NOX (per Δ20 μg/m3)	6	0.98 (0.88, 1.09)	0.438	0.00%	4	0.93 (0.78, 1.11)	0.640	0.00%
PM10 (per Δ10 μg/m3)	4	0.90 (0.68, 1.19)	0.419	0.00%	3	0.92 (0.55, 1.54)	0.368	0.00%
PM2.5 (per Δ5 μg/m3)	4	0.71 (0.37, 1.37)	0.052	61.24%	3	1.01 (0.63, 1.63)	0.472	0.00%
PMcoarse (per Δ5 μg/m3)	4	0.96 (0.72, 1.28)	0.300	18.16%	3	0.87 (0.55, 1.38)	0.320	12.33%
PM2.5absorbance (per Δ10–5m–1)	4	0.82 (0.57, 1.18)	0.244	27.95%	3	0.70 (0.44, 1.12)	0.899	0.00%
Traffic intensity on the nearest road (per Δ5,000 mv/day)	3	1.00 (0.92, 1.09)	0.721	0.00%	3	0.98 (0.85, 1.14)	0.508	0.00%
Total traffic load on all major roads within 100-m buffer (per Δ4,000,000 mv/day × m)	5	1.02 (0.89, 1.16)	0.752	0.00%	3	0.90 (0.70, 1.16)	0.691	0.00%
Abbreviations: I2, percentage of the total variability due to between-areas heterogeneity; mv, motor vehicles; p-Heter, p-value of heterogeneity using the Cochran’s Q test. aOdds ratios and 95% confidence intervals were estimated by random-effects meta-analysis by area. Models were adjusted for maternal characteristics (education, country of birth, age at delivery, prepregnancy body mass index, height, prenatal smoking, and parity), child’s sex, season at child’s birth, urbanicity at child’s birth address, and child’s age at autistic traits assessment, and evaluator of the autistic traits. Models of traffic variables were additionally adjusted for non–back-extrapolated background levels of NO2. bAir pollution levels were temporally adjusted to the exact pregnancy period except for traffic variables. cNumber of cohorts included in the meta-analysis.