Environmental exposures in early-life and general health in childhood

Amine, Ines; Guillien, Alicia; Philippat, Claire; Anguita-Ruiz, Augusto; Casas, Maribel; de Castro, Montserrat; Dedele, Audrius; Garcia-Aymerich, Judith; Granum, Berit; Grazuleviciene, Regina; Heude, Barbara; Haug, Line Småstuen; Julvez, Jordi; López-Vicente, Mónica; Maitre, Léa; McEachan, Rosemary; Nieuwenhuijsen, Mark; Stratakis, Nikos; Vafeiadi, Marina; Wright, John; Yang, Tiffany; Yuan, Wen Lun; Basagaña, Xavier; Slama, Rémy; Vrijheid, Martine; Siroux, Valérie

doi:10.1186/s12940-023-01001-x

Environmental Health

Table 4 Results from LASSO Exposures selected in the final model

From: Environmental exposures in early-life and general health in childhood

Exposures	Transformation^a	LASSO	Multivariable model from LASSO results
Exposures	Transformation^a	% of selection among the 20 imputed datasets	Estimate change by IQR [95% CI]	p-value^b
During pregnancy
Presence of a bluespace	-^c	60%	0.13 [-0.02; 0.28]	0.08
Passive smoking (yes)	-	90%	-0.11 [-0.21; -0.01]	0.02
Cotinine levels	-	65%
< 18.5 µg/L			reference
18.5–50 µg/L (passive smoking)			-0.05 [-0.18; 0.08]	0.43
> 50 µg/L (active smoking)			-0.1 [-0.22; 0.01]	0.09
During childhood
Sodas	-	100%
Low (< 1 time/month)			reference
Medium (1 time/month – 1 time/week)			0.09 [0.00; 0.18]	0.06
High (> 1 time/week)			-0.01 [-0.12; 0.10]	0.85
Vegetables	-	100%
Low (< 6 times/week)			reference
Medium (6–9 times/week)			0.00 [-0.09; 0.1]	0.94
High (> 9 times/week)			0.10 [0.02; 0.19]	0.02
Bakery products	-	100%
Low (< 2 times/week)			reference
Medium (2–6 times/week)			0.10 [0.00; 0.19]	0.04
High (> 6 times/week)			0.00 [-0.11; 0.10]	0.93
Caffeinated drink	-	100%
Low (Never)			reference
Medium (1 time/month)			0.00 [-0.11; 0.10]	0.93
High (> 1 time/month)			-0.10 [-0.21; 0.00]	0.05
Pet (yes)	-	100%	0.14 [0.06; 0.22]	< 0.01
Contact with family and friends	-	100%
Daily			reference
Once a week			-0.08 [-0.16; 0.00]	0.05
Less than once a week			0.06 [-0.10; 0.23]	0.44
Physical activity		95%	0.06 [0.00; 0.12]	0.06
Benzene indoor	Log	100%	-0.08 [-0.14; -0.03]	< 0.01
PM_2.5 indoor	Log	100%	-0.05 [-0.10; -0.01]	0.01
MEPA	Log	100%	-0.05 [-0.09; -0.01]	0.01
Co	Log	100%	0.04 [0.01; 0.08]	0.02
Cu	Log	100%	-0.10 [-0.15; -0.05]	< 0.01
PFOA	Log	100%	0.08 [0.03; 0.13]	< 0.01
DDE	Log	100%	0.08 [0.01; 0.14]	0.02
HCB	Log	100%	0.07 [0.00; 0.13]	0.04

Population: n = 870 children from the HELIX subcohort. Methods: first, penalized linear regression model was applied, with regularization parameter (lambda) optimized with tenfold cross validation. Then, all exposures selected for at least 10 imputed datasets (50% of selection) were included in a multivariable linear regression adjusted for the covariates. All exposures with a p-value > 10% were removed one by one from the final model. Covariates: cohort, child age, maternal education and age, parental country of birth, season of birth, pre-pregnancy BMI, plus the breastfeeding duration for postnatal exposures only
Co cobalt, Cu Copper, DDE Dichlorodiphenyldichloroethylene, HCB Hexachlorobenzene, MEPA Methyl-paraben, PFOA: Perfluorooctanoate, PM: Particulate matter
^aThe transformation was applied before the standardisation on the interquartile range (IQR). This last transformation consisted in removing the mean and dividing by the IQR
^bt-test on the estimated coefficients, based on the Rubin’s rule
^c “- “ means that no transformation was applied, mainly because the variable was categorical

Back to article page

ISSN: 1476-069X

Contact us

General enquiries: journalsubmissions@springernature.com