Skip to main content
. 2021 Mar 11;22:101351. doi: 10.1016/j.pmedr.2021.101351

Table 3.

GEE model results for longitudinal analysis of cannabis ever-use, current use, and regular use among cases of ever use for three cohorts of high school students taking part in COMPASS (Canada) at four consecutive time points starting in a given baseline year. Students in two cohorts (C1, C2) graduated prior to legalization, students in the third (C3) attended grade 12 in the year following legalization.

Longitudinal analysis Model 4 Ever use vs never-use (ref) n = 5 400
Model 5 Current use vs ever- and never-use (ref) n = 5 400
Model 6 Regular use vs occasional use (ref) n = 2 446
AOR (95% CI) p AOR (95% CI) p AOR (95% CI) p
Cohort
C1 (2013–14) 1.02 (0.79, 1.32) 0.8855 1.03 (0.68, 1.55) 0.8919 0.93 (0.52, 1.65) 0.8048
C2 (2014–15) (ref)
C3 (2015–16) 0.85 (0.64, 1.12) 0.2426 0.72 (0.46, 1.14) 0.1632 0.64 (0.32, 1.28) 0.2058
Gender
Female (ref)
Male 1.07 (0.97, 1.19) 0.1849 1.46 (1.29, 1.66) <0.0001 1.50 (1.30, 1.71) <0.0001
Grade
9 (ref)
10 2.60 (2.21, 3.07) <0.0001 2.63 (1.97, 3.51) <0.0001 1.19 (0.76, 1.85) 0.4430
11 6.44 (5.39, 7.70) <0.0001 6.22 (4.65, 8.31) <0.0001 1.70 (1.13, 2.55) 0.0108
12 11.19 (9.29, 13.48) <0.0001 9.78 (7.27, 13.17) <0.0001 1.96 (1.30, 2.96) 0.0014
Cohort 1 * Grade
9 (ref)
10 1.10 (0.87, 1.39) 0.4321 1.18 (0.80, 1.76) 0.4018 1.38 (0.75, 2.55) 0.2998
11 0.86 (0.67, 1.11) 0.2499 0.87 (0.58, 1.31) 0.5199 1.03 (0.57, 1.88) 0.9127
12 0.88 (0.68, 1.14) 0.3317 0.87 (0.58, 1.32) 0.5196 1.02 (0.56, 1.86) 0.9401
Cohort 3 * Grade
9 (ref)
10 1.18 (0.93, 1.51) 0.1699 1.51 (0.97, 2.37) 0.0682 1.68 (0.81, 3.50) 0.1655
11 1.12 (0.86, 1.46) 0.3926 1.34 (0.85, 2.11) 0.2039 1.51 (0.74, 3.09) 0.2541
12 1.20 (0.91, 1.58) 0.1874 1.44 (0.91, 2.28) 0.1188 1.60 (0.79, 3.26) 0.1946

indicates school year at baseline when participating students were attending grade 9. Samples comprise only students participating in COMPASS in all four grades from grade 9 to 12. C2 was chosen as the reference cohort to facilitate comparison between students unexposed to the cannabis legalization process and those who were directly exposed. All models controlled for province, ethnicity, spending money and student-level clustering. AORs for main effects can be interpreted as the increased risk when all other variables held at their reference levels. AORs for interaction terms represent the multiplicative increase in risk among the corresponding non-reference categories.