Effects of Group Psychotherapy, Individual Counseling, Methylphenidate, and Placebo in the Treatment of Adult Attention-Deficit/Hyperactivity Disorder: A Randomized Clinical Trial

Importance  Attention-deficit/hyperactivity disorder (ADHD) is a neurodevelopmental disorder with high prevalence in adulthood. There is a recognized need to assess the efficacy of psychotherapy in adult ADHD.

Objective  To evaluate the efficacy of cognitive behavioral group psychotherapy (GPT) compared with individual clinical management (CM) and that of methylphenidate hydrochloride compared with placebo.

Design, Setting, and Participants  Prospective, multicenter, randomized clinical trial of 18- to 58-year-old outpatients with ADHD from 7 German study centers. Patients were recruited between January 2007 and August 2010, treatment was finalized in August 2011, and final follow-up assessments occurred in March 2013.

Interventions  Sessions of GPT and CM were held weekly for the first 12 weeks and monthly thereafter (9 months). Patients received either methylphenidate or placebo for 1 year.

Main Outcomes and Measures  The primary outcome was the change in the ADHD Index of the Conners Adult ADHD Rating Scale from baseline to the end of the 3-month intensive treatment (blinded observer ratings). Secondary outcomes included ADHD ratings after 1 year, blinded observer ratings using the Clinical Global Impression Scale, and self-ratings of depression.

Results  Among 1480 prescreened patients, 518 were assessed for eligibility, 433 were centrally randomized, and 419 were analyzed as randomized. After 3 months, the ADHD Index all-group baseline mean of 20.6 improved to adjusted means of 17.6 for GPT and 16.5 for CM, with no significant difference between groups. Methylphenidate (adjusted mean, 16.2) was superior to placebo (adjusted mean, 17.9) (difference, −1.7; 97.5% CI, −3.0 to −0.4; P = .003). After 1 year, treatment effects remained essentially stable. Descriptive analyses showed that methylphenidate was superior to placebo in patients assigned to GPT (difference, −1.7; 95% CI, −3.2 to −0.1; P = .04) or CM (difference, −1.7; 95% CI, −3.3 to −0.2; P = .03). Regarding depression, no significant differences were found. In contrast, GPT was superior to CM for all visits in the Clinical Global Impression global assessment of effectiveness.

Conclusion and Relevance  Highly structured group intervention did not outperform individual CM with regard to the primary outcome. Psychological interventions resulted in better outcomes during a 1-year period when combined with methylphenidate as compared with placebo.

Trial Registration  isrctn.org Identifier: ISRCTN54096201

Adult attention-deficit/hyperactivity disorder (ADHD) is a mental disorder affecting an estimated 2.5% of the adult population.^1-3 It is associated with numerous comorbid disorders and negative psychosocial consequences.^4-7 Most guidelines recommend a multimodal treatment approach.^8-10

Quiz Ref IDThe National Institute for Health and Care Excellence proposes methylphenidate hydrochloride as the first-line treatment for adult ADHD. Meta-analyses have shown robust moderate effect sizes for methylphenidate vs placebo in reducing ADHD symptoms.^11-14 However, up to 50% of individuals show less than a 30% decrease in symptoms.^12,15

Combined treatment with medication and individual or group cognitive behavioral therapy has demonstrated significant benefits over medication alone.^16-20 Previous pilot clinical trials have evaluated a specific cognitive group psychotherapy (GPT) program for adult ADHD.^21-24 Moderate effect sizes^23,24 of these and other psychotherapy concepts have been demonstrated.^23,25,26 However, these studies have not systematically controlled for medication and include either medicated or mixed (ie, with and without medication) patient samples. To our knowledge, the only available data come from a pilot trial that revealed a nonsignificant benefit for patients treated with stimulants vs placebo.²⁷ Thus, the effect of medication on the outcomes of psychological therapy is still largely unknown.

Specific cognitive behavioral programs have been proven more effective than unspecific control conditions (eg, relaxation, supportive therapy, and discussion groups) for adult ADHD.^23,25,26 However, the effectiveness of a highly structured group program vs a less specific treatment, eg, clinical management (CM), which simulates practice care in an optimal way, is unknown.

Given the recognized need for research,²⁸ the primary aim of our study was to demonstrate the efficacy of highly structured behavioral GPT compared with less specific treatment (eg, CM) as well as that of methylphenidate compared with placebo after 3 months. Secondary analyses included a comparison of the same effects after 1 year of treatment, the 4 treatment conditions (GPT with methylphenidate; GPT with placebo; CM with methylphenidate; and CM with placebo), measures of depression, and Clinical Global Impression Scale (CGI) score.²⁹

All methodological issues have been described in detail.^21,30 The Comparison of Methylphenidate and Psychotherapy in Adult ADHD Study (COMPAS) was a factorial, multicenter, randomized clinical trial comparing GPT with CM and methylphenidate with placebo. The full trial protocol appears in Supplement 1. Inclusion and exclusion criteria are listed in eAppendix 1 in Supplement 2. The diagnosis of ADHD, according to DSM-IV and other psychiatric symptoms, was established by psychiatric expert assessment and validated using observer rating scales and self-rating scales, including the Wender Utah Rating Scale (WURS-k; in German),^31,32 the ADHD diagnostic checklist (ADHD-DC; in German),³³ and the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I) and Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II) (in German).^34,35 Concurrent use of psychopharmacologic or psychotherapeutic treatments was not allowed outside the trial. The study received ethics committee approval from University of Freiburg. Written informed consent was obtained from all participants.

Eligible patients were randomized in batches of 14 or 15 at a time (1 × 12, 1 × 16). The randomization allowed for GPT in groups of 6 to 9 patients. Either GPT or CM plus a medication number (used to allocate either methylphenidate or placebo) was centrally assigned.^21,30 Treatments were allocated in a 1:1:1:1 ratio, stratified in blocks of 4 within the center (block size was kept confidential to help ensure concealment; the protocol stated that it was variable). Blocks were sequentially combined for application to patient batches. For patients and therapists, the study was blinded for medication and open for assignment to GPT or CM. Observers rating ADHD symptoms (ADHD-DC, Conners Adult ADHD Rating Scale [CAARS] long German version,^36-39 and CGI) were blinded to treatment allocation.

Following randomization and baseline assessment, participants received methylphenidate hydrochloride (sustained release; initial dosage of 10 mg/d; titration with 10 mg/wk over 6 weeks up to 60 mg/d; individual dosage to a maximum daily dosage of 1.3 mg/kg of body weight) or placebo. Medication adherence was assessed by pill count.

The GPT sessions followed a validated manual (eAppendix 2 in Supplement 2).^40,41 Individual CM was the active, nonpharmacological control condition chosen to simulate general practice. The CM participants received nonspecific counseling in individual sessions (15-20 minutes) (eAppendix 2 in Supplement 2).^21,30 Twelve weekly sessions of GPT and CM were followed by 10 monthly sessions over 52 weeks.^21,30 The CM sessions were audio recorded and the GPT sessions were video recorded to assess treatment fidelity by 3 blinded, independent, expert raters.^21,30

Visits for assessing primary and secondary end points took place after randomization (baseline, or time 1 [T1]), after 12 weeks of intensive treatment (T2), during maintenance after 24 weeks (T3), at the end of treatment (week 52; T4), and at 2.5 years after T1 (follow-up; T5). The primary outcome was the change in the observer-rated CAARS ADHD Index from T1 to T2. Secondary outcome measures included other CAARS subscales, the ADHD-DC, the Beck Depression Inventory,^42,43 and CGI subscales; we report results for these end points at T2 and T4.

The sample size was derived assuming an effect size (ES) of 0.33 for the 2 primary 2 × 2 comparisons (GPT or methylphenidate vs control conditions^21,30). To achieve a power of 80% for a t test at a 2-sided α of 2.5% to adjust for multiplicity, 350 patients were needed. A target of 448 randomizations was planned to compensate for dropouts. After a recruitment delay, a power of 78% (430 patients) was deemed acceptable to the study team, and no more patients were enrolled.

Changes in the CAARS ADHD Index from T1 to T2 were analyzed by randomized treatment in the full analysis set. To model a stable postdropout response, missing postbaseline data were replaced using multiple imputation through last mean carried forward (LMCF)⁴⁴ in an analysis of covariance linear model, using time, treatments, center, and baseline measurements as fixed covariates (eAppendix 3 in Supplement 2). Adjusted means per treatment were calculated from this. For the primary treatment comparisons at T2, an additional interaction term (GPT × MPH; kept if significant at 10%) was pretested.

The 2 primary comparisons were reported with confirmatory 97.5% confidence intervals (corrected for multiple testing of GPT vs CM and methylphenidate vs placebo) and descriptive 95% confidence intervals. These were statistically significant if P < .025 (P < .05 for other comparisons). If a primary treatment comparison was statistically significant, a confirmatory closed-test procedure sequentially compared GPT with methylphenidate vs CM with methylphenidate; GPT with methylphenidate vs GPT with placebo; CM with methylphenidate vs CM with placebo; and GPT with methylphenidate vs CM with placebo (descriptive reporting with nominal P values after the first nonsignificant result^21,30).

Responses were defined as decreases in the observer rating variant of the CAARS ADHD Index of 30% or more. To calculate response rates, we analyzed only complete cases and used logistic regression. Other rating scales (other CAARS subscales, ADHD-DC, and Beck Depression Inventory) were evaluated using LMCF. Complete cases of CGI subscales (ordinal data) were analyzed in a proportional odds model.

Adverse events (AEs) were evaluated according to received treatment in safety set 1 (patients who attended ≥1 GPT or CM session) and safety set 2 (patients who received ≥1 dose of methylphenidate or placebo).

All analyses were prespecified and performed using SAS version 9.2 statistical software (SAS Institute, Inc).

Descriptive pre-post ESs were calculated using LMCF means and square roots of the residual variance, averaged over LMCF imputations as standard deviation.

An interim report on recruitment, compliance, and safety (no efficacy data) was presented to the independent data monitoring committee in April 2010, based on data up to T3 for the 231 patients randomized as of May 2009. The independent data monitoring committee recommended continuing the trial without modifications.

Figure 1 illustrates patient flow.³⁰ Patients were recruited between January 2007 and August 2010, with randomization between April 10, 2007, and August 18, 2010. Treatment was finalized in August 2011, and final follow-up assessments occurred in March 2013 (data not reported here). In sum, 1480 patients were contacted. In 962 prescreened patients (65.0%), no standardized assessment for eligibility was carried out owing to lack of interest or inability to meet time requirements (391 of 962 [40.6%]) or contraindications against methylphenidate (194 of 962 [20.2%]). The remaining 518 patients were assessed for eligibility; 85 were excluded, primarily for refusing further participation (47 of 85 [55.3%]).²⁹

Of the 433 randomized individuals, 107 were randomized to GPT with methylphenidate, 109 to GPT with placebo, 110 to CM with methylphenidate, and 107 to CM with placebo. Table 1 summarizes the sociodemographic and clinical characteristics at baseline.

We obtained baseline ratings in 419 of the 433 randomized participants; these constituted the full analysis set for the primary LMCF efficacy analysis. Only one-third of the patients (138 of 419 [32.9%]) had never undergone psychiatric or psychotherapeutic outpatient treatment. The majority (277 of 419 [66.1%]) fulfilled criteria for at least 1 current or remitted Axis I disorder; 75 of 419 (17.9%) fulfilled the diagnostic criteria for at least 1 Axis II disorder. Primary outcome data (ADHD Index at T2) were available before LMCF for 91 of 103 participants (88.3%) receiving GPT with methylphenidate; 86 of 106 (81.1%) receiving GPT with placebo; 95 of 107 (88.8%) receiving CM with methylphenidate; and 80 of 103 (77.7%) receiving CM with placebo.

Treatment fidelity (GPT, CM) is shown in eAppendix 4 in Supplement 2. Interrater reliability is shown in eTable 1 in Supplement 2.

The mean (SD) daily medication dosage prescribed at T2 was 53.3 (20.4) mg total and 0.71 (0.27) mg/kg of body weight. The mean (SD) daily medication dosage was 48.8 (20.2) mg in 179 patients receiving methylphenidate and 58.5 (19.3) mg among 158 receiving placebo. The mean (SD) daily medication dosage was 53.6 (20.4) mg for 166 patients assigned to GPT and 53.1 (20.3) mg for 171 assigned to CM.

Quiz Ref IDThe test for GPT × methylphenidate interaction was nonsignificant (P = .95). Therefore, a 2 × 2 approach (vs a 4-arm approach) was applied evaluating GPT vs CM and methylphenidate vs placebo.

The study detected no advantage for the decline in ADHD symptoms for GPT vs CM (Table 2, Table 3, and Figure 2). The ADHD Index scores improved from the all-group baseline mean of 20.6 to an adjusted mean of 17.6 at T2 for GPT (n = 209; ES = −0.55) and 16.5 for CM (n = 210; ES = −0.75) (Table 2 and Figure 2). The difference between GPT and CM was nonsignificant (ADHD Index score difference for GPT vs CM, 1.1; 97.5% CI, −0.2 to 2.4; 95% CI, 0.0 to 2.2; P = .06).

Quiz Ref IDSymptoms decreased considerably more in patients assigned to methylphenidate (n = 210; adjusted mean ADHD Index score, 16.2; ES = −0.81) vs placebo (n = 209; adjusted mean ADHD Index score, 17.9; ES = −0.50) (Table 2 and Figure 2). This difference proved significant (ADHD Index score difference for methylphenidate vs placebo, –1.7; 97.5% CI, −3.0 to −0.4; 95% CI, −2.8 to −0.6; P = .003).

Because the difference between methylphenidate and placebo proved significant, we applied sequential 4-arm analyses. In patients randomized to methylphenidate, GPT vs CM produced nonsignificant findings (ADHD Index score difference, 1.1; 95% CI, −0.4 to 2.7; P = .16). Thus, confirmatory statistical testing was terminated.

The following preplanned exploratory comparisons were conducted. Comparing methylphenidate with placebo in patients assigned to GPT showed that methylphenidate was superior to placebo (ADHD Index score difference, −1.7; 95% CI, −3.2 to −0.1; P = .04). This superiority was also evident in patients randomized to CM (ADHD Index score difference for methylphenidate vs placebo, −1.7; 95% CI, −3.3 to −0.2; P = .03). In contrast, comparing both interventions (GPT and methylphenidate) with the control treatments produced a nonsignificant benefit (ADHD Index score difference for GTP with methylphenidate vs CM with placebo,−0.6; 95% CI, −2.2 to 0.9; P = .43).

At T3 and T4, the treatment effects of GPT vs CM and methylphenidate vs placebo remained stable (Table 2 and Figure 2). The slight disadvantage of GPT vs CM at T2 and T3 became a slight nonsignificant benefit at T4. In contrast, methylphenidate was significantly better than placebo during the entire study period.

Quiz Ref IDAt T2, responses were highest in the CM with methylphenidate arm at 47.4%, compared with 32.5% in CM with placebo, 29.7% in GPT with methylphenidate, and 24.4% in GPT with placebo (GPT vs CM, P = .009; methylphenidate vs placebo, P = .05). At T4, response rates were similar in all 4 treatment arms, varying between 44.1% for GPT with placebo and 52.9% for CM with methylphenidate (GPT vs CM, P = .72; methylphenidate vs placebo, P = .52) (Table 3).

For the self-ratings of the ADHD Index, GPT’s nonsignificant disadvantage vs CM at T2 and T3 became a nonsignificant advantage at T4. Methylphenidate proved superior to placebo at all 3 measurement times (Table 4).

The CAARS ratings and the ADHD-DC revealed comparable results in total and subscale scores, with no difference between GPT and CM (Table 2; eTable 2 in Supplement 2).

Regarding depression, no significant differences were found between patients treated with GPT vs CM or methylphenidate vs placebo. Methylphenidate exhibited nonsignificantly better Beck Depression Inventory ratings at all 3 times. Moreover, GPT’s nonsignificant disadvantage at T2 became a nonsignificant advantage at T4 (Table 4).

Quiz Ref IDComparison of CGI severity score between GPT and CM revealed no major differences (Table 5). Concerning CGI global change, GPT performed better than CM at all times; however, it was significant only at T4 (P = .047). Methylphenidate always performed better than placebo, but this was significant only at T3 (P = .008) (Table 5). The end point CGI global assessment of effectiveness always favored GPT over CM and methylphenidate over placebo. This difference in GPT’s favor was highly significant at T4 (P < .001) (Table 5).

Frequencies of AEs and serious AEs are shown in eTable 3 in Supplement 2. No suicides occurred. Two patients receiving methylphenidate became pregnant, despite contraception. One terminated her pregnancy for psychosocial reasons; the other experienced no AEs during delivery or early development of the child. Changes in heart rate, blood pressure, and body weight from T1 to T4 are shown in eTable 4 in Supplement 2.

The COMPAS is the first multimodal, multicenter randomized clinical trial to examine the efficacy of nonpharmacological treatments (GPT vs CM) in combination with methylphenidate or placebo. We recruited a large representative sample that compares well with other trials.^45,46 In contrast to our hypothesis, GPT could not be shown to be more effective than the CM control condition, except in CGI-related secondary outcomes. Methylphenidate was superior to placebo in nearly all outcome domains. All 4 treatment arms exhibited improvements in both symptoms and CGI ratings.

This finding contrasts with previous findings in which preliminary evidence has shown the superiority of structured disorder-oriented GPT over unspecific group control conditions.^23,25,26 However, our study used an individual CM control condition to simulate practice care in an optimal way. As a consequence, the investigated group program, although found effective in earlier preliminary studies, may not have been sufficiently effective to outperform the individual control condition. Another explanation may be that our control treatment—although performed adherent to the protocol—was not an attention placebo; instead, it included face-to-face counseling activities, which potentially responded better to the individual needs of participants than the groups. We cannot extrapolate to specific psychotherapy methods beyond the one tested.

While CM appeared superior to specific GPT after 3 months, long-term effects after 1 year favored GPT slightly. In particular, CGI global assessments of effectiveness of the interventions were significantly better for GPT at all measurement times. This is remarkable because this scale represents more general measures of well-being (eg, improved acceptance, self-esteem, coping skills), which were the focus of GPT.

Confirming preliminary evidence,²⁷ our trial clearly showed that combinations of GPT or CM with methylphenidate were superior to combinations with placebo.

Our study provides no evidence that methylphenidate reduces depressive symptoms. This contrasts with some preliminary evidence,⁴⁷ but it supports more recent research.^48,49

Our results relate well to the large childhood Multimodal Treatment Study of Children With ADHD,⁵⁰ which also found significant improvements in all treatment conditions (medication management, intensive behavioral treatment, the two combined, or standard community care). As in our trial, medication proved to be superior to intensive behavioral therapy.⁵⁰

A nonsignificant interaction term does not exclude the possibility that the effect of GPT may depend on MPH and vice versa. However, the 2-arm and 4-arm comparisons of GPT vs CM and GPT with methylphenidate vs CM with methylphenidate, as well as methylphenidate vs placebo and GPT with methylphenidate vs GPT with placebo, gave identical results, implying that our data do not suggest such dependencies.

Blinding was restricted to medication and to observer ratings of ADHD and CGI. We did not systematically assess whether blinding was effective in patients. However, because patients who received methylphenidate and those who received placebo both reported high numbers of AEs, with minimal effects to vital signs and weight, we believe the blinding was effective. As in other studies⁵¹ and in line with the nature of the disorder, compliance was a challenge: more than one-third of the randomized patients dropped out. These missing data can be viewed as a study result⁵¹ rather than as a shortcoming, since most dropouts occurred in the CM with placebo condition. Because the available data (eTable 5 in Supplement 2) and the imputed LMCF analyses of the full analysis set showed similar results, we conjecture that our findings were not confounded by dropouts.

Our findings may not be generalizable to routine care settings in which comorbidities are not excluded and patients may have more psychosocial impairments or difficulties meeting the time and effort requirements for this trial.

Despite the significant superiority of methylphenidate vs placebo on most ADHD scales, the mean differences between methylphenidate vs placebo and GPT vs CM were relatively small. Our data could not show whether superior CGI ratings of GPT reflected otherwise hidden differences in patients’ daily functioning.

The COMPAS trial sheds light on issues that, to our knowledge, have not yet been addressed. First, it almost doubles the observation period of the longest randomized study conducted globally so far.⁵² Second, it systematically addresses the effect of medication on the outcome of psychotherapy. Previous studies were pilot studies,²⁷ not placebo controlled,^22-26 or included medication-treated patients with persistent ADHD symptoms.^16,17,25 Third, we compared a highly structured GPT with a less controlled CM condition, also an untested area.

To our knowledge, COMPAS is the first trial to demonstrate long-term maintenance effects of ADHD treatments under controlled conditions. We demonstrate that psychological interventions result in better outcomes when combined with methylphenidate as compared with placebo. Our data do not suggest that highly structured group intervention outperforms individual CM, which is much easier to implement in practical care than specifically tailored and highly structured GPT.

Corresponding Author: Alexandra Philipsen, MD, Medical Campus University of Oldenburg, School of Medicine and Health Sciences, Psychiatry and Psychotherapy–University Hospital, Karl-Jaspers-Klinik, D-26111 Oldenburg, Germany (alexandra.philipsen@uni-oldenburg.de).

Submitted for Publication: April 17, 2015; final revision received August 17, 2015; accepted August 18, 2015.

Published Online: November 4, 2015. doi:10.1001/jamapsychiatry.2015.2146.

Author Contributions: Drs Philipsen and Graf had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Berger and Tebartz van Elst contributed equally.

Study concept and design: Philipsen, Jans, Graf, Alm, Retz, Ihorst, Schlander, Berger, Tebartz van Elst.

Acquisition, analysis, or interpretation of data: Philipsen, Jans, Graf, Matthies, Borel, Colla, Gentschow, Langner, Jacob, Groß-Lesch, Sobanski, Schumacher-Stien, Roesler, Retz, Retz-Junginger, Kis, Abdel-Hamid, Heinrich, Huss, Kornmann, Bürger, Perlov, Schlander, Berger, Tebartz van Elst.

Drafting of the manuscript: Philipsen, Graf, Colla, Retz, Heinrich, Tebartz van Elst.

Critical revision of the manuscript for important intellectual content: Philipsen, Jans, Graf, Matthies, Borel, Colla, Gentschow, Langner, Jacob, Groß-Lesch, Sobanski, Alm, Schumacher-Stien, Roesler, Retz-Junginger, Kis, Abdel-Hamid, Huss, Kornmann, Bürger, Perlov, Ihorst, Schlander, Berger, Tebartz van Elst.

Statistical analysis: Philipsen, Graf, Tebartz van Elst.

Obtained funding: Philipsen, Jans, Tebartz van Elst.

Administrative, technical, or material support: Philipsen, Jans, Matthies, Borel, Colla, Gentschow, Langner, Jacob, Groß-Lesch, Schumacher-Stien, Roesler, Retz-Junginger, Kis, Abdel-Hamid, Huss, Kornmann, Perlov, Schlander, Berger, Tebartz van Elst.

Study supervision: Philipsen, Colla, Gentschow, Langner, Sobanski, Roesler, Retz, Retz-Junginger, Kis, Huss, Tebartz van Elst.

Conflict of Interest Disclosures: Dr Philipsen reported serving on advisory boards, giving lectures, performing phase 3 studies, or receiving travel grants within the last 3 years from Eli Lilly and Co, Janssen-Cilag, MEDICE Arzneimittel Pütter GmbH and Co KG, Novartis, and Shire; and has authored books and articles on psychotherapy published by Elsevier, Hogrefe, Schattauer, Kohlhammer, and Karger. Dr Matthies reported receiving a speaker’s fee from Janssen-Cilag; and involvement in clinical trials conducted by Janssen-Cilag and Eli Lilly and Co. Dr Colla reported serving on advisory boards, receiving speaker’s honoraria, or performing phase 3 studies within the last 3 years with Shire, Eli Lilly and Co, and Novartis. Dr Jacob reported receiving speaker’s honoraria from MEDICE; serving as an advisory board member of Eli Lilly and Co; performing phase 3 studies at MEDICE, Novartis, Janssen-Cilag, Eli Lilly and Co; and performing an investigator-initiated trial with Eli Lilly and Co. Dr Sobanski reported receiving speaker’s honoraria from MEDICE, Eli Lilly and Co, and Novartis; serving as an advisory board member of MEDICE, Shire, and Eli Lilly and Co; and performing phase 3 studies and investigator-initiated trials with MEDICE, Novartis, Janssen-Cilag, and Eli Lilly and Co. Dr Alm reported receiving speaker’s honoraria from MEDICE; serving on the advisory board of Eli Lilly and Co; performing phase 3 studies at MEDICE, Novartis, Janssen-Cilag, Eli Lilly and Co; and performing an investigator-initiated trial with Eli Lilly and Co. Dr Roesler reported serving as an advisory board member of MEDICE, Eli Lilly and Co, and Janssen-Cilag; serving as a member of the speakers bureau of MEDICE, Eli Lilly and Co, Shire, and Novartis; and performing clinical studies for MEDICE. Dr Retz reported receiving speaker’s honoraria from and serving as an advisory board member of MEDICE, Shire, and Novartis; and performing clinical trials for BMBF, Novartis, Vifor, and MEDICE. Dr Kis reported receiving speaker’s honoraria from MEDICE, Servier, and Eli Lilly and Co; and serving as an advisory board member of MEDICE, Novartis, and Servier. Dr Huss reported serving as an advisory board member of Eli Lilly and Co, Engelhardt Arzneimittel, Janssen-Cilag, MEDICE, Novartis, Shire, and Steiner Arzneimittel within the past 5 years; serving as consultant to Engelhardt Arzneimittel, MEDICE, and Steiner Arzneimittel; receiving honoraria from Eli Lilly and Co, Engelhard Arzneimittel, Janssen-Cilag, MEDICE, Novartis, and Shire; and receiving unrestricted grants for investigator-initiated trials from Eli Lilly and Co, MEDICE, Engelhard Arzneimittel, and Steiner Arzneimittel. Mr Schlander reported receiving research support from sick funds, physicians’ associations, health technology agencies, industry associations, and biopharmaceutical enterprises, including Janssen (Johnson & Johnson) and Shire, all under unrestricted educational grant policy. Dr Tebartz van Elst reported serving on advisory boards, giving lectures, or receiving travel grants within the last 3 years from Eli Lilly and Co, Janssen-Cilag, Novartis, Shire, UCB, GlaxoSmithKline, Servier, Janssen-Cilag, and Cyberonics. No other disclosures were reported.

Funding/Support: This study was part of a multicenter research network on the psychotherapy of ADHD, funded by grants 01GV0605 and 01GV0606 from the German Federal Ministry of Education and Research. MEDICE Arzneimittel Puetter GmbH and Co KG provided the trial medication (Medikinet retard licensed as Medikinet adult and matching placebo).

Role of the Funder/Sponsor: The German Federal Ministry of Education and Research and MEDICE Arzneimittel Puetter GmbH and Co KG had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Group Information: The Comparison of Methylphenidate and Psychotherapy in Adult ADHD Study (COMPAS) Consortium includes the following members: Michael Colla, MD, Laura Gentschow, Paula Kunze, and Daina Langner, PhD, Department of Psychiatry and Psychotherapy, Campus Benjamin Franklin, Charité–University Medicine Berlin, Berlin, Germany; Bernhard Kis, MD, Mona Abdel-Hamid, PhD, Viola Heinrich, Markus Krämer, MD, and Jennifer Uekermann, PhD, Department of Psychiatry and Psychotherapy, LVR-Hospital Essen, Faculty of Medicine, University of Duisburg-Essen, Duisburg and Essen, Germany; Alexandra Philipsen, MD (coordinating investigator), Erika Graf, PhD, Swantje Matthies, MD, Marc Loewer, MD, Patricia Borel, Imke Jansen, Steffi Bonfico, Manuel Jooßens, Chiharu Sadohara, Manfred Weber, Melanie Kamp, Tatja Dopatka, Evgeniy Perlov, MD, and Harald Richter, PhD, Department of Psychiatry and Psychotherapy, University Medical Center Freiburg, University of Freiburg, Freiburg, Germany; Michael Roesler, MD, Wolfgang Retz, MD, Petra Retz-Junginger, PhD, Konstanze Roemer, Birgit Leipnitz, MD, Sabine Doyran, MD, and Monika Schulte-Altedorneburg, MD, Institute for Forensic Psychology and Psychiatry, Saarland University Medical Center and Saarland University Faculty of Medicine, Homburg/Saar, Germany; Christine Carl, PhD, and Clemens Keutler, MD, Department of Child and Adolescent Psychiatry and Psychotherapy, St Elisabethen Krankenhaus, Lörrach, Germany; Michael Huss, MD, Catherine Kornmann, Arne Bürger, Galina Chervenkova, and Patricia Meinhardt, Department of Child and Adolescent Psychiatry and Psychotherapy, University Medicine Mainz, Mainz, Germany; Esther Sobanski, MD, Barbara Alm, MD, Martina Schumacher-Stien, Simon Bukow, MD, and Sotiria Argiriou-Martin, Department of Psychiatry and Psychotherapy, Central Institute of Mental Health, Clinical Faculty Mannheim, University of Heidelberg, Mannheim, Germany; Christian Jacob, MD, Thomas Jans, PhD, Silke Groß-Lesch, MD, Monika Heine, MD, Andrea Boreatti-Hümmer, MD, Julia Heupel, Susanne Reichert, Sabine Müller, Susanne Kreiker, Alexandra Gessner, Annette Conzelmann, and Christina Bähne, Department of Psychiatry, Psychosomatics and Psychotherapy and Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, University Hospital of Würzburg, Würzburg, Germany; and Rainer Bredenkamp (director), Gabriele Ihorst, PhD, and Erika Graf, PhD, Clinical Trials Unit, University Medical Center, University of Freiburg, Freiburg, Germany. The independent data monitoring committee included Helmut Remschmidt, MD, Department of Child and Adolescent Psychiatry, Philipps University, Marburg, Germany; Gernot Wassmer, PhD, Institute for Medical Statistics, Informatics, and Epidemiology, University of Cologne, Cologne, Germany; and Norbert Wodarz, MD, Department of Psychiatry, University of Regensburg, Regensburg, Germany.

Additional Information: Other projects in our network on psychotherapy research in ADHD (speaker: Dr Philipsen; 2006-2012, Andreas Warnke, MD, University Hospital of Würzburg, Würzburg, Germany) are coordinated by Dr Jans (ADHD in mothers and children), Dr Tebartz van Elst (functional and morphometric brain mapping), and Klaus-Peter Lesch, MD, University Hospital of Würzburg (molecular genetics).

Additional Contributions: We thank Harald Richter, PhD, who was a brilliant and humorous cognitive behavioral therapy and dialectical behavioral therapy teacher and who died unexpectedly in January 2015. James Carpenter, BSc, MSc, DPhil, FHEA, London School of Hygiene and Tropical Medicine, London, England, provided advice on the conception and implementation of the LMCF analyses; he received no compensation. Roland Fischer, MD, MEDICE Arzneimittel Pütter GmbH and Co KG, Iserlohn, Germany, gave advice on preparing the study protocol and assisted in managing serious adverse events; he received no compensation. Independent supervision was carried out by Ulrike Frank, PhD, Institute for Psychology, University of Freiburg, Freiburg, Germany, in cooperation with colleagues Friederike Mayer-Bruns, MD, in private practice, and Kirsten Schehr, Dipl Psych, in private practice. Teaching was carried out by the coordinating investigator, the supervisors, and Dr Richter. Mr Schlander planned and conducted the health economic evaluation.

Correction: This article was corrected to fix an omission in author affiliations on November 12, 2015.