As shown in , 1334 adolescents were pre-screened, 450 were consented and screened, and 303 were randomized to OROS-MPH or placebo and included in ITT analyses. There were no statistically significant differences between groups in study completion, compliance with weekly research visits or CBT session attendance, or medication compliance. The most conservative measure of medication compliance, determined by weekly pill counts of the number returned divided by number of pills prescribed and assuming that non-returned pills/bottles were not taken, was 79%. Medication compliance, based on adolescent self-reports, was 82.3% overall.
DEMOGRAPHIC AND BASELINE CHARACTERISTICS
There were no statistically significant differences between groups at baseline on demographics, primary outcomes or other relevant clinical characteristics (). The mean (standard deviation [SD]) age was 16.5 (1.3) years; and 78.9% were male. Race and ethnicity were self-classified. Race: Caucasian, 61.7%; African American, 23.2%; other, 15.1%. Ethnicity: Hispanic, 15.2%. At baseline, participants had moderately severe ADHD (mean ADHD-RS=38.7 (8.9)) and reported using non-tobacco substances about half of the days in the past month (mean=14.6/28 days).
Baseline Clinical Characteristics by Treatment Group
Prior to conducting longitudinal analyses, we confirmed that there were no statistically significant differences between groups with regard to the pattern the percentage of primary outcome data that was missing (ADHD-RS: OROS-MPH=21.2%; placebo=24.5%, P>0.05; days of substance use assessed: OROS-MPH=11.9%; placebo=16.5%, P>0.05). Additionally, completers (N=227) were not different from non-completers (N=76) on baseline demographic or clinical characteristics (ADHD-RS scores; days of past-28 day substance use; number of substance use disorders- all P>0.05) and the proportion of completers did not differ by treatment assignment (P>0.05).
ADHD Primary ADHD Outcome
illustrates the longitudinal course of DSM-IV ADHD-RS scores over time by treatment group. There was a clinically and statistically significant decrease in ADHD score in both treatment groups but no between-group difference: estimated decrease from baseline to study end for the OROS-MPH + CBT group was −19.2 (95% CI, −17.1 to −21.2; P<0.001) and for the placebo + CBT group was −21.2 (95% CI, −19.1 to −23.2, P<0.001). Likelihood-based criterion determined that the longitudinal course of adolescent DSM-IV ADHD-RS score in the ITT sample followed a cubic curve with random subject and linear time effects and serially correlated residual errors.
Change in DSM-IV Attention-Deficit/Hyperactivity Disorder Rating Scale (ADHD-RS) Scores by Treatment Group
Secondary ADHD Outcomes ADHD-RS Parent Informant
Parent ADHD-RS scores were significantly lower in participants treated with OROS-MPH + CBT compared to placebo + CBT at 8 weeks (OROS-MPH, N=85, mean score=26.0 (10.5); placebo, N=79, mean score=30.4(12.5); mean difference between groups=4.4; 95% CI, 0.8-7.9; P=0.0163 t-test) and at 16 weeks (OROS-MPH, N=84, mean score=24.0 (11.8); placebo, N=68, mean score=30.9 (13.0); mean difference between groups=6.7; 95% CI, 2.9-10.9; P<0.001 t-test).
When the logit model with a linear time effect and random intercept that adjusted for a significant site effect was implemented via the general linear mixed model, the proportion of ADHD treatment responders (CGI-I score of 1 or 2) was surprisingly low in OROS-MPH + CBT (25.5%) and not significantly different (P=0.715) from placebo + CBT (23.2%). Rates of treatment response were even lower with multiple imputation for missing data and not significantly different (P=0.418) between OROS-MPH + CBT (23.4%) and placebo + CBT (19.1%). Similarly, there was not a statistically significant difference (P=0.64) in treatment responders based on participants’ last CGI-I rating (OROS-MPH + CBT=23.6 %, N=144; placebo + CBT=21.3%, N=141). Of completers, 26.1% were treatment responders in the OROS-MPH + CBT group (N=119) and 22.2% were treatment responders in the placebo + CBT group (N=108) which was also not statistically different between groups (P=0.50). As a validity check of the CGI-I measure, CGI-I responders were compared to non-responders with regard to change in ADHD-RS scores in completers. Responders (CGI-I=1, 2) had significantly greater decline in ADHD-RS scores (mean= −26.4 [11.6 SD]) compared to non-responders (CGI-I>2; mean= −18.8 [11.7], P<0.001; mean difference= −7.6, 95% CI, −11.2 to −4.0; Cohen’s d=0.66).
ARCQ Change in “Problem-Solving Ability” and “Focused Coping Skills”
Compared to placebo, adolescents treated with OROS-MPH + CBT reported significantly greater improvement in their “problem solving ability” (median difference between groups=7, favoring OROS-MPH; 95% CI, 3-8; P=0.002, Cohen’s d=0.35) and acquisition of “focused coping skills” (median difference between groups=4, favoring OROS-MPH; 95% CI, 1-7; P=0.02, Cohen’s d=0.25).
MGH Liking Scale
In response to the MGH Liking Scale question “Do you feel that your medication has been effective in treating your ADD/ADHD symptoms?”, the median affirmative response was significantly greater in participants treated with OROS-MPH (median=6, (5,8) 25th, 5th percentile, N=144) than in those treated with placebo (median=5 (2,6), N=141; P<0.001, Cohen’s d=0.56).
Substance Use Primary Substance Outcome
shows the longitudinal course of days of past 28-day nontobacco drug use (TLFB) over time by treatment group, indicating that there was a clinically and statistically significant decrease in both the OROS-MPH + CBT (−5.7 days, 95% CI, −7.4 to −4.0; P<0.001) and placebo + CBT groups (−5.2 days; 95% CI, −7.0 to −3.5; P<0.001) but no between-group difference (Chi-square=3.5, 3df, P=0.321). Likelihood-based criteria determined that the longitudinal course of adolescent-reported days of non-tobacco drug use in the past 28 days in the ITT sample followed a quadratic curve with random subject and linear time effects. shows the response of the primary ADHD and substance outcome variables based on estimates from the longitudinal models (also see Figures and ).
Change in Past 28 Day Non-Tobacco Drug Use by Treatment Group
Estimated Measures of Attention-Deficit/Hyperactivity Disorder (ADHD) and Substance Use Disorder (SUD) by Treatment Group in the Intent-To-Treat Sample (N=303)
Secondary Substance Outcomes
Adolescents treated with OROS-MPH + CBT had significantly more negative UDS (3.8, 95% CI, 3.0-4.6) compared to participants treated with placebo + CBT (2.8, 95% CI, 2.1-3.5; P=0.05, Cohen’s d=0.22) and there was no difference between groups in the number of UDS collected (11.7, 11.3, respectively).
MEDICATION TOLERABILITY, SAFETY, ADVERSE EVENTS
Ninety-six percent of participants treated with OROS-MPH achieved the maximum daily dose of 72mg, and 86.6% sustained this dose throughout the trial. Participants treated with OROS-MPH had more treatment-emergent study-related AEs per subject (mean=2.4, 95% CI, 1.9-2.9) compared to placebo (mean=1.6, 95% CI, 1.3-1.9; P=0.02). More participants treated with OROS-MPH had permanent reductions in their medication dose compared to placebo (10.7% vs. 6.8%, respectively). Specific AEs treated with placebo (panic/anxiety (1); auditory hallucinations/vomiting (1); nausea/upper abdominal pain (2); influenza (1); rash (1); nervousness (1); depressed mood (1); mass (1)). Eleven SAEs occurred throughout the trial. Four SAEs occurred in participants treated with OROS-MPH, only one of which was judged as study-related, in which a participant was briefly hospitalized for evaluation of psychosis after attending a “rave” at which he ingested unknown substances. Seven SAEs occurred in participants treated with placebo, four of which were judged as possibly study-related and resulted in brief hospitalizations for syncope, drug toxicity, asthma, and increased aggression, respectively.
Seventy percent of participants (OROS-MPH, 68%; placebo, 72%) reported using drugs or alcohol on days they took study medication, but only 4 (2.8%) taking OROS-MPH and 3 (2.1%) taking placebo, reported possible adverse interactions between medication and drugs/alcohol taken on the same day.
Medication Abuse and Diversion
There were no statistically significant differences between OROS-MPH and placebo, respectively, on self-reported medication abuse (“taking more medication than prescribed”, 4.8% vs. 2.8%, P>0.05) or diversion (“selling medication to others”, 2.1% vs. 1.4%, respectively, P>0.05; “letting others take your medication”, 3.5% vs. 1.4%, respectively, P>0.05) as assessed by the MGH Abuse and Diversion Questionnaire.
Assessment of Study Blind
At study exit, adolescents and clinicians were asked to guess whether the adolescent had been on active medication or placebo during the trial. Of participants taking OROS-MPH, 59.8% guessed correctly; clinician guesses were 59.7% correct. Of participants taking placebo, 66.7% guessed correctly; clinician guesses were 67.3% correct. The percent of correct guesses was significantly greater than chance based on chi-square tests of association (p<0.001) for both the adolescents and clinicians. However, given that there were no significant group differences on measures relying on adolescent or clinician ratings, it is unlikely that the correct guesses biased the trial toward finding a medication effect.