Table 5.
Exposure | Univariable | Adjusted 1 | Adjusted 2 | Adjusted 3 | Fully Adjusted | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|
OR | 95% CI | OR | 95% CI | OR | 95% CI | OR | 95% CI | OR | 95% CI | ||
Cannabis use | Ever | – | – | – | – | – | – | – | – | – | – |
n = 4434 | Never | 2.79 | 1.70, 4.57 | 2.48 | 1.49, 4.12 | 2.55 | 1.53, 4.24 | 2.47 | 1.45, 4.19 | 2.75 | 1.18, 6.39 |
Cannabis status | None | – | – | – | – | – | – | – | – | – | – |
n = 4434 | Non‐weekly | 1.98 | 0.89, 4.42 | 1.80 | 0.80, 4.06 | 1.94 | 0.86, 4.41 | 2.48 | 1.05, 5.84 | 1.96 | 0.74, 5.21 |
Weekly | 6.83 | 3.37, 13.85 | 6.24 | 3.02, 12.92 | 5.94 | 2.83, 12.45 | 5.23 | 2.33, 11.72 | 4.44 | 1.79, 11.00 | |
CAST score 4+ | No | – | – | – | – | – | – | – | – | – | – |
n = 4137 | Yes | 8.15 | 4.00, 16.60 | 7.19 | 3.43, 15.08 | 7.23 | 3.39, 15.42 | 5.41 | 2.40, 12.20 | 4.40 | 1.70, 11.38 |
Smoking status | None | – | – | – | – | – | – | – | – | – | – |
n = 3814 | Non‐daily | 0.90 | 0.31, 2.58 | 0.86 | 0.30, 2.48 | 0.86 | 0.30, 2.50 | 0.86 | 0.29, 2.55 | 0.73 | 0.20, 2.66 |
Daily | 7.55 | 4.23, 13.47 | 6.48 | 3.49, 12.04 | 6.17 | 3.29, 11.56 | 4.78 | 2.45, 9.35 | 2.23 | 1.11, 9.39 | |
Cotinine smoking status | Non‐Smoker | – | – | – | – | – | – | – | – | – | – |
n = 2869 | Smoker | 5.44 | 2.85, 10.36 | 5.03 | 2.56, 9.88 | 4.48 | 2.26, 8.90 | 3.27 | 1.59, 6.70 | 2.54 | 1.12, 5.74 |
The educational outcome here was school dropout, which was derived from not gaining any General Certificate of State Education (GCSE) grades or receiving only U grades (graded as unclassified); four covariate models were used and each builds on the previous model. These models are grouped into proximal and distal determinants. Adjustment 1 model includes binary maternal substance use behaviours (mother smokes, mother binge drinks and mother uses cannabis). Adjustment 2 model included demographics [socio‐economic status (SES), maternal education and income] and sex. Adjustment 3 model includes the individuals previous education before substance use is more likely [standard deviation (SD) change for Key Stage 2 English and Mathematics]. The fully adjusted model includes child's substance use behaviour measured at the same time as the exposure (child drinking, child weekly cannabis use for the tobacco smoking exposures and child smoking for the cannabis use exposures) and conduct disorder. Covariates were imputed to increase n. All the testparm results were P < 0.001. CAST = Cannabis Abuse Screening Test; CI = confidence interval; OR= odds ratio.