Table 4.
Associations between childhood exposures and change in behavioral problems, showing exposures selected into the multiple-exposure modelsa (out of 123 candidate exposures, N = 1287).
|
Internalizing scale |
Externalizing scale |
ADHD index |
|||||
|---|---|---|---|---|---|---|---|
| Childhood exposure (IQR or reference category) | Exposure family | aMR [95%CI]b | p-value | aMR [95%CI]b | p-value | aMR [95%CI]b | p-value |
| Road traffic load (100 m) (1138814 vehicules/day m) |
Road traffic | 1.08 [0.98; 1.18] | 0.123 | ||||
| Indoor benzene (0.99 µg/m3) | Indoor air | 1.06 [0.99; 1.13] | 0.121 | ||||
| Indoor PM2.5 (6.5 µg/m3) | Indoor air | 1.09 [1.03; 1.16] | 0.005 | ||||
| Second-hand smoking | Tobacco smoking | 1.13 [1.01; 1.26] | 0.034 | 1.08 [0.95; 1.23] | 0.217 | 1.05 [0.94; 1.18] | 0.395 |
| Cotinine | Tobacco smoking | 1.09 [0.93; 1.28] | 0.297 | ||||
| Total hours of sleep | Lifestyle | 0.89 [0.83; 0.96] | 0.001 | 0.91 [0.85; 0.99] | 0.027 | 0.94 [0.87; 1.01] | 0.083 |
| Bakery products | Lifestyle/Diet | ||||||
| Medium (vs. low) | 1.00 [0.88; 1.13] | 0.997 | 1.03 [0.90; 1.18] | 0.649 | |||
| High (vs. low) | 1.18 [1.03; 1.36] | 0.017 | 1.15 [0.98; 1.34] | 0.081 | |||
| Dairy products | Lifestyle/Diet | ||||||
| Medium (vs. low) | 0.89 [0.79; 1.01] | 0.060 | |||||
| High (vs. low) | 0.98 [0.87; 1.12] | 0.806 | |||||
| Meat intake | Lifestyle/Diet | ||||||
| Medium (vs. low) | 0.99 [0.87; 1.11] | 0.818 | |||||
| High (vs. low) | 1.11 [0.99; 1.25] | 0.067 | |||||
| Ready made food | Lifestyle/Diet | ||||||
| Medium (vs. low) | 1.13 [0.99; 1.29] | 0.073 | 1.10 [0.97; 1.25] | 0.147 | |||
| High (vs. low) | 1.25 [1.09; 1.44] | 0.001 | 1.20 [1.05; 1.37] | 0.008 | |||
| Sweets | Lifestyle/Diet | ||||||
| Medium (vs. low) | 1.03 [0.90; 1.17] | 0.690 | 1.04 [0.92; 1.17] | 0.543 | |||
| High (vs. low) | 1.14 [1.00; 1.31] | 0.058 | 1.14 [1.01; 1.30] | 0.041 | |||
| Fastfood | Lifestyle/Diet | ||||||
| Medium (vs. low) | 1.03 [0.91; 1.15] | 0.668 | |||||
| High (vs. low) | 1.12 [0.93; 1.34] | 0.247 | |||||
| Caffeinated drinks | Lifestyle/Diet | ||||||
| Medium (vs. low) | 1.00 [0.86; 1.16] | 0.974 | |||||
| High (vs. low) | 1.14 [1.01; 1.30] | 0.037 | |||||
| Fish and seafood | Lifestyle/Diet | ||||||
| Medium (vs. low) | 0.90 [0.80; 1.03] | 0.116 | |||||
| High (vs. low) | 0.99 [0.86; 1.14] | 0.910 | |||||
| Healthy diet (KIDMED score) | Lifestyle/Diet | 0.93 [0.87; 0.99] | 0.034 | ||||
| Lead (5.91 µg/L) | Metal | 1.06 [1.00; 1.13] | 0.061 | 1.12 [1.05; 1.19] | 2.4E-4 | ||
| Copper (350 µg/L) | Metal | 1.10 [1.03; 1.17] | 0.002 | ||||
| Dimethylthiophosphate (4.89 µg/g) | OP pesticides | 0.91 [0.85; 0.97] | 0.007 | 0.92 [0.86; 0.98] | 0.013 | ||
| Sum of polychlorinated biphenyls (27.6 ng/g lipids) | OC compounds | 0.89 [0.82; 0.98] | 0.016 | 0.93 [0.84; 1.02] | 0.131 | 0.93 [0.84; 1.05] | 0.259 |
| Dichlorodiphenyldichloroethylene, DDE (34.0 ng/g lipids) | OC compounds | 0.95 [0.86; 1.04] | 0.242 | ||||
| Polybrominated diphenyl ether 47 (0.21 ng/g lipids) | OC compounds | 1.04 [1.00; 1.09] | 0.050 | ||||
| Monoethyl phthalate (59.84 µg/g) | Phtalates | 1.06 [0.98; 1.15] | 0.119 | ||||
| Social participation | Social capital | ||||||
| 1 organisation (vs none) | 0.98 [0.88; 1.10] | 0.775 | |||||
| >1 organisations (vs none) | 0.83 [0.71; 0.97] | 0.016 | |||||
| Contact with family and friends | Social capital | ||||||
| Once a week (vs. less than once a week) | 0.79 [0.63; 1.00] | 0.051 | |||||
| Almost daily (vs. less than once a week) | 0.69 [0.55; 0.87] | 0.002 | |||||
| House crowding | Social capital | 0.89 [0.85; 0.94] | <0.001 | ||||
Abbreviations: CI = confidence interval; IQR, interquartile range; cat., categories.
The selection of exposures was determined by the least absolute shrinkage and selection operator (LASSO) algorithm applying log link function.
The change in behavioral problems was modelled with negative binomial regression which model the ratio of the mean score for one unit change in the exposure (the coefficients obtained needed to be exponentiated). The results are presented as adjusted mean ratios (aMR) with corresponding 95% confidence intervals (CIs). In all of these models, the exposure estimate was reported for one interquartile range (IQR) increase in exposure levels—for example, a ratio of mean score of 1.25 for an IQR increase in the exposure meant that those with an exposure at the 75th percentile had a 25% higher mean score, compared with those with an exposure at the 25th percentile. These coefficients are adjusted for the other exposures and for cohort, maternal age, maternal education level, maternal pre-pregnancy body mass index, parity, parental country of birth, child age, child sex, and child height.