Table 2.
Program Effects on Substance Use Outcomes at One-year Follow-up
| Any TND versus SCC a | TND+MI versus TND-only a | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| At One-year Follow-up |
Net Effect | At One-year Follow-up |
Net Effect | ||||||
| Substance use status b | Any TND (%) |
SCC (%) |
OR (95% CI) b | p c | TND +MI (%) |
TND- only (%) |
OR (95% CI) b | p c | |
| Cigarette Use | 35.0 | 40.0 | 0.80 (0.61–1.04) | 0.079 | 36.1 | 35.9 | 1.01 (0.73–1.41) | 0.481 | |
| Alcohol Use | 43.8 | 53.3 | 0.68 (0.51–0.91) | 0.011 | 42.4 | 46.0 | 0.86 (0.61–1.22) | 0.243 | |
| Drunk on alcohol | 27.5 | 36.07 | 0.67 (0.46–1.98) | 0.039 | 27.1 | 30.1 | 0.86 (0.54–1.36) | 0.289 | |
| Marijuana Use | 37.0 | 41.1 | 0.84 (0.60–1.18) | 0.190 | 32.6 | 41.5 | 0.68 (0.45–1.02) | 0.058 | |
| Hard Drug Use | 15.4 | 21.1 | 0.68 (0.48–0.96) | 0.035 | 13.8 | 18.3 | 0.71 (0.46–1.09) | 0.103 | |
| Times of substance used | Any TND |
SCC | IRR (95%CI) d | p c | TND +MI |
TND- only |
IRR (95%CI) d | p c | |
| Cigarette Use | 2.63 | 2.97 | 0.87 (0.77–0.99) | 0.035 | 2.47 | 2.79 | 0.88 (0.77–1.01) | 0.065 | |
| Alcohol Use | 1.72 | 1.91 | 0.91 (0.83–1.00) | 0.046 | 1.70 | 1.74 | 1.00 (0.89–1.12) | 0.481 | |
| Drunk on alcohol | 1.44 | 1.58 | 0.91 (0.82–1.01) | 0.068 | 1.37 | 1.51 | 0.94 (0.83–1.06) | 0.193 | |
| Marijuana Use | 2.36 | 2.52 | 0.95 (0.84–1.07) | 0.247 | 2.22 | 2.51 | 0.92 (0.83–1.02) | 0.073 | |
| Hard Drug Use | 1.48 | 1.70 | 0.87 (0.77–0.98) | 0.026 | 1.41 | 1.55 | 0.92 (0.80–1.05) | 0.147 | |
| Index indicators of substance use | Any TND |
SCC | β±se (std) | p c | TND +MI |
TND- only |
β±se (std) | p c | |
| Substance use index e | −0.09 | 0.19 | −0.28±0.12 | 0.014 | −0.07 | −0.12 | 0.05±0.15 | 0.38 | |
| Hard drug use index f | −0.05 | 0.09 | −0.14±0.06 | 0.023 | −0.08 | −0.03 | −0.05±0.08 | 0.27 | |
Notes:
Program Condition: Standard Care Control (SCC), TND Classroom Program Only (TND-only), or TND plus Motivational Interviewing (TND+MI). Any TND: TND-only combined with TND+MI (i.e., received either program condition). All models were adjusted for the corresponding substance use at baseline, age, gender, ethnicity, propensity-to-attrition score, and survey method at follow-up. In addition, perceived quality of classroom program was added as a covariate for the evaluation regarding “TND+MI versus TND-only.
OR: Odds Ratio. The outcomes were coded as binary indicators of substance use. Evaluation was conducted with two-level random coefficients modeling; program effects were assumed to be fixed and school effects were assumed to be random. Binary outcomes for each type of substance use were linked to the linear combinations of predictors with a logit link function. Exp(beta) is thus the OR for the substance use variable at 1-year follow-up. Analysis was adjusted for the corresponding substance use at baseline, age, gender, ethnicity, propensity-to-attrition score, and survey method at follow-up.
p for one-tailed hypothesis test generated from t test for evaluations on both types of outcomes.
IRR: incident rate ratio. The outcomes were coded as 12-level count (times) of substance use. Evaluation was conducted with two-level random coefficients modeling; program effects were assumed to be fixed and school effects were assumed to be random. A zero-inflated negative binomial (ZINB) distribution was applied to model the number of times of substance use. The mean for the ZINB distribution was linked to the linear combinations of predictors with a log link function. The IRR is thus the relative increase of the average number of times of substance use. Analysis was adjusted for the corresponding substance use at baseline, age, gender, ethnicity, propensity-to-attrition score, and survey method at follow-up.
Continuous index of overall substance use, coded as the log of an equally weighted sum of the log-transformed 12-level indicators of cigarette use, alcohol use, marijuana use, and hard drug use. Evaluation was conducted with two-level random coefficients modeling; program effects were assumed to be fixed and school effects were assumed to be random. The outcome was standardized to std=1; thus the metric for beta is “std”, or can be interpreted as the “effect size” for means comparison.
An index that is the log of the average number of times of different types of hard drug use in the last 30 days. Hard drugs measured included cocaine, hallucinogens, stimulants, inhalants, ecstasy, pain killers, tranquilizers, and “other” hard drugs.