Companion reviews: ADHD — stimulants and the evidence, Adult ADHD and the evidence, Diagnostic Thresholds and the Evidence, Placebo, unblinding and the evidence, Depression — antidepressants and the evidence, Antipsychotics and the evidence, Mood stabilizers and the evidence, and Benzodiazepines and the Evidence.

---

Executive summary

Two folk models dominate discussion of ADHD medications beyond stimulants: that non-stimulants are broadly equivalent first-line options to stimulants (the substitutability claim), and its inversion, that only stimulants work and everything else is ineffective (the stimulant-exceptionalism claim). The opened evidence supports neither cleanly.

Fourteen things the evidence actually establishes:

1. Child guanfacine and clonidine show clinician-rated symptom reductions versus placebo, though effect sizes are smaller than stimulants and parent ratings for guanfacine are not significant. Cortese et al. (2018) network meta-analysis of children and adolescents found guanfacine clinician-rated SMD −0.67 (95% CI −0.85 to −0.50) versus placebo at closest to 12 weeks. Clonidine clinician-rated SMD −0.71 (95% CI −1.17 to −0.24). Guanfacine parent-rated SMD −0.23 (95% CI −0.90 to 0.45), not significant. Guanfacine dropout for adverse events OR 2.64 (95% CI 1.20 to 5.81). Child clonidine CGI-I OR 2.78 (95% CI 0.91 to 8.53), not significant. The authors' search closed 7 April 2017 and identified no adult clonidine or guanfacine trials.

2. Official FDA boxed warnings apply to atomoxetine and viloxazine for suicidal ideation, and to lisdexamfetamine for abuse and addiction potential, but not to guanfacine or clonidine. The FDA Strattera (atomoxetine) label carries a boxed warning: "WARNING: SUICIDAL THOUGHTS AND BEHAVIORS IN PEDIATRIC PATIENTS 6 YEARS OF AGE AND OLDER." Short-term pediatric trials found suicidal ideation in 0.4% (5 of 1,357) including one attempt versus 0% (0 of 851) placebo; no suicides; all in ages 6–12 in month 1. The adult ADHD analysis "did not reveal an increased risk of suicidal ideation or behavior." The FDA Qelbree (viloxazine ER) label carries a boxed warning: "WARNING: SUICIDAL THOUGHTS AND BEHAVIORS." Clinical studies found higher rates of suicidal thoughts and behavior in patients with ADHD treated with Qelbree than in patients treated with placebo: pediatric 9 of 1,019 (0.9%) versus 2 of 463 (0.4%); adult 3 of 189 (1.6%) versus 0 of 183; no completed suicides. The FDA Vyvanse label carries a boxed warning: "WARNING: ABUSE, MISUSE, AND ADDICTION" stating "VYVANSE has a high potential for abuse and misuse, which can lead to the development of a substance use disorder, including addiction." The opened 2018 Intuniv (guanfacine XR) label and the 2010 Kapvay (clonidine XR) label have no boxed warning.

3. Child guanfacine XR shows labeled ADHD-RS reductions; the opened Biederman 2008 trial prints the LS-means the label rounds. The FDA Intuniv 2018 label reports Study 1 ADHD-RS-IV placebo-subtracted LS-mean 2 mg −7.4 (95% CI −11.3 to −3.5); 3 mg −7.5 (95% CI −11.4 to −3.6); 4 mg −10.0 (95% CI −13.9 to −6.1). Study 2: 1–4 mg −6.8 to −7.9 versus placebo. The indication states "INTUNIV® is indicated for the treatment of Attention Deficit Hyperactivity Disorder (ADHD) as monotherapy and as adjunctive therapy to stimulant medications" in ages 6–17. No US adult ADHD claim on this label. Biederman et al. (2008) report the same Study 1 population (NCT00152009): N = 345 ages 6–17; ADHD-RS-IV LS-mean −16.18 / −16.43 / −18.87 (2/3/4 mg) versus −8.48 placebo. Placebo-adjusted −7.70 (95% CI −12.25 to −3.15), −7.95 (95% CI −12.50 to −3.40), −10.39 (95% CI −14.97 to −5.82). Post-hoc effect sizes 0.64 / 0.66 / 0.86. Adverse-event discontinuation 16.2% versus 1.2% placebo. The label rounds the LS-means to −15.9 / −16.0 / −18.5; the paper prints −16.18 / −16.43 / −18.87. Both report the same trial.

4. Child clonidine XR shows labeled ADHD-RS improvement in two trials, but the label prints no LS-mean or SMD; the clinician SMD comes from Cortese, not the label. The FDA Kapvay 2010 label states the indication: "KAPVAY™ is a centrally acting alpha2-adrenergic agonist indicated for the treatment of attention deficit hyperactivity disorder (ADHD) as monotherapy or as adjunctive therapy to stimulant medications." Study 1 monotherapy N = 236; Study 2 adjunct N = 198; both "statistically significantly superior" on ADHD-RS-IV at week 5. This 2010 label prints no LS-mean, SMD, or response rate. Treatment-period maximum placebo-subtracted at 0.2 / 0.4 mg/day: systolic blood pressure −4.0 / −8.8 mm Hg; diastolic blood pressure −4.0 / −7.3; heart rate −4.0 / −7.7 bpm. No adult ADHD trial. No boxed warning. The child clonidine clinician SMD −0.71 (95% CI −1.17 to −0.24) is from Cortese et al. (2018), not the Kapvay label.

5. Viloxazine is outside Cortese 2018 and has labeled pediatric and adult effects plus a suicidality box. The Cortese et al. (2018) search closed 7 April 2017; viloxazine is outside that network meta-analysis. The FDA Qelbree label reports pediatric ADHD-RS-5 placebo-subtracted LS-mean: Study 1 ages 6–11, 100 mg −5.8 (95% CI −8.9 to −2.6), 200 mg −6.9 (95% CI −10.0 to −3.8); Study 2 ages 6–11, 200 mg −6.0 (95% CI −10.0 to −1.9), 400 mg −5.8 (95% CI −9.9 to −1.7); Study 3 ages 12–17, 200 mg −4.5 (95% CI −8.4 to −0.6), 400 mg −5.1 (95% CI −8.9 to −1.3). Adult AISRS label Study 4: −3.7 (95% CI −6.2 to −1.2). Adult heart rate ≥20 bpm any time 52 of 178 (29%) versus 23 of 181 (13%). The indication is ADHD in adults and pediatric patients 6 years and older. Nasser et al. (2022) report Study 4 (NCT04016779): adults 18–65, week-6 AISRS LS-mean change from baseline −15.5 versus −11.7; difference −3.7 (95% CI −6.3 to −2.2); p = 0.0040. AISRS responders ≥30% improvement 60.0% (78 of 130) versus 47.6% (68 of 143). Responders ≥50% improvement not significant. Adverse-event discontinuation 9.0% versus 4.9%. [contested] The paper prints 95% CI −6.3 to −2.2; the label prints 95% CI −6.2 to −1.2. Both report the same trial. No SMD is provided in the opened viloxazine sources. The boxed-warning counts are as in headline 2.

6. Adult named-compound leftover: lisdexamfetamine shows labeled ADHD-RS reductions and a dedicated Cochrane review; bupropion shows a dedicated Cochrane SMD. The FDA Vyvanse label reports adults 18–55, Study 7 (Adler et al. 2008 trial identity on the label): investigator ADHD-RS LS-mean 30 mg −16.2, 50 mg −17.4, 70 mg −18.6 versus placebo −8.2. Placebo-subtracted −8.0 (95% CI −11.5 to −4.6), −9.2 (95% CI −12.6 to −5.7), −10.4 (95% CI −13.9 to −6.9). Study 9 adult withdrawal: treatment failure 9% versus 75% placebo. The indication includes ADHD in adults and pediatric patients ≥6 years, and adult binge eating disorder. Castells et al. (2018) Cochrane review found adult clinician-rated ADHD severity versus placebo: all amphetamines SMD −0.90 (95% CI −1.04 to −0.75); 13 studies, 2,028 participants; low- to very-low quality. Named compound lisdexamfetamine SMD −1.06 (95% CI −1.26 to −0.85); 7 studies, 896 participants. Mixed amphetamine salts SMD −0.80 (95% CI −0.93 to −0.66). Dexamphetamine clinician SMD −0.24 (95% CI −0.80 to 0.32), not significant (1 study, 49 participants). Adverse-event withdrawal relative risk 2.69 (95% CI 1.63 to 4.45). Verbeeck et al. (2017) Cochrane review of bupropion found six RCTs, 438 adults. Severity SMD −0.50 (95% CI −0.86 to −0.15); 129 participants, 3 studies; GRADE low. Response ≥30% improvement RR 1.50 (95% CI 1.13 to 1.99). CGI-I 1 or 2 RR 1.78 (95% CI 1.27 to 2.50). Adverse-event withdrawal RR 1.20 (95% CI 0.35 to 4.10), not significant. Clinician-rated-only sensitivity SMD −0.36 (95% CI −0.79 to 0.07), 87 participants, not significant. The Cortese adult clinician atomoxetine SMD −0.45 (95% CI −0.58 to −0.32), bupropion SMD −0.46 (95% CI −0.85 to −0.07), and methylphenidate SMD −0.49 (95% CI −0.64 to −0.35) remain re-bindable from the opened 2018 network meta-analysis as supporting text when rater is named.

7. Adult extended-release methylphenidate shows dedicated Cochrane evidence with investigator and self-rated SMDs. Boesen et al. (2022) Cochrane review of adult ER-MPH versus placebo found investigator-rated SMD −0.42 (95% CI −0.49 to −0.36; 18 trials, 4,183 participants); self-rated SMD −0.37 (95% CI −0.43 to −0.30; 16 trials, 3,799 participants on summary-of-findings table); GRADE very low. Adverse-event discontinuation RR 2.36 (95% CI 1.62 to 3.43; 15 trials, 3,963 participants). Any adverse event RR 1.27 (95% CI 1.19 to 1.37). Serious adverse event RR 1.43 (95% CI 0.85 to 2.43), not significant. Median duration 8 weeks (24 trials, 5,066 participants). This is not pooled with Cortese 2018 adult methylphenidate clinician SMD −0.49.

8. Adult immediate-release methylphenidate shows a thinner Cochrane body with investigator mean difference and participant SMD not significant. Cândido et al. (2021) Cochrane review of adult IR-MPH versus placebo found no pooled clinician SMD. Investigator-rated AISRS mean difference −20.70 (95% CI −23.97 to −17.43; 1 trial, 146 participants); participant-rated SMD −0.59 (95% CI −1.25 to 0.06; 2 trials, 138 participants), not significant; GRADE very low. Adverse-event withdrawal not pooled. Gastrointestinal RR 1.96 (95% CI 1.13 to 2.95); appetite RR 1.77 (95% CI 1.06 to 2.96). Ten trials, 497 adults; 6–18 weeks duration.

9. Adult guanfacine extended-release shows a Japan RCT with ADHD-RS difference −4.28; US Intuniv label remains pediatric ages 6–17. Iwanami et al. (2020) report a Japanese adult RCT: N = 201 randomized (full analysis set 100 guanfacine / 100 placebo), mean ages 31.1 versus 33.8 years; week-10 ADHD-RS-IV LS-mean −11.55 ± 1.10 versus −7.27 ± 1.07; difference −4.28 (95% CI −6.67 to −1.88); effect size 0.52; adverse-event discontinuation 19.8% versus 3.0%. The paper reports guanfacine approved for ADHD in adults in Japan. US Intuniv 2018 label remains ages 6–17; no adult ADHD claim. Cortese et al. (2018) "no adult guanfacine trials" reflects the search closed 7 April 2017.

10. A later clonidine extended-release label prints ADHD-RS LS-means the 2010 Kapvay label lacked. Actavis/Teva clonidine hydrochloride extended-release DailyMed revised October 2023 reports pediatric ages 6–17 ADHD-RS-IV placebo-subtracted LS-mean: Study 1 0.2 mg −8.5 (95% CI −12.2 to −4.8), 0.4 mg −9.1 (95% CI −12.8 to −5.5); Study 2 adjunct −4.5 (95% CI −7.8 to −1.1). No boxed warning. No adult trial. The 2010 Kapvay label blood pressure and heart rate effects and indication remain as shipped; Cortese 2018 child clonidine clinician SMD −0.71 remains as shipped.

11. Intuniv Study 2 manuscript prints ADHD-RS LS-means the label summarizes as a range; the 2 mg dose is outside that range. Sallee et al. (2009) report intent-to-treat N = 306 ages 6–17; placebo-adjusted ADHD-RS-IV LS-mean −6.75 / −5.41 / −7.34 / −7.88 at 1 / 2 / 3 / 4 mg (p = 0.0041 / 0.0176 / 0.0016 / 0.0006). Post-hoc effect sizes 0.53 / 0.43 / 0.58 / 0.62. Adverse-event discontinuation 7.4% versus 7.6%. Combined mean change from baseline −19.6 (SD 13.9) versus −12.2 (SD 13.0). Ages 13–17 subgroup (n = 80) not significant at any dose. No 95% CI on the four primary contrasts. The opened Intuniv 2018 label Study 2 range "−6.8 to −7.9" does not include the paper's 2 mg −5.41; both are printed.

12. A third child guanfacine extended-release RCT shows ADHD-RS effect sizes 0.75 and 0.78; not Study 1 or 2. Newcorn et al. (2013) report children ages 6–12, N = 333 (AM dosing 107 / PM dosing 114 / placebo 112). ADHD-RS-IV mean change from baseline −19.8 (AM) / −20.1 (PM) versus −11.0 placebo; authors' effect sizes 0.75 / 0.78; adverse-event discontinuation 16 of 221 guanfacine versus 0 placebo. Somnolence 44.3% versus 12.5%.

13. Pediatric viloxazine 812P304 is a sequential-testing negative trial the Qelbree label does not print. Nasser et al. (2021) Psychopharmacology Bulletin report adolescents ages 12–17, intent-to-treat 292. ADHD-RS-5 LS-mean change from baseline −18.3 ± 1.36 (400 mg) / −16.7 ± 1.39 (600 mg) versus −13.2 ± 1.38 placebo. Placebo-adjusted −5.1 ± 1.93 (p = 0.0082) / −3.5 ± 1.93 (p = 0.0712). Sequential testing: 600 mg first and not significant, so neither dose considered superior. Responders ≥50% improvement 48.2% / 46.0% versus 32.9%. Adverse-event discontinuation 4.0% / 5.1% versus 1.0% (combined 4.5%). This trial is not Qelbree label Studies 1–3.

14. Labeled pediatric viloxazine RCTs lack original manuscripts; BJCP 2022 reprints treatment effect estimates. Nasser et al. (2022) British Journal of Clinical Pharmacology report end-of-study treatment effect (positive value = larger improvement versus placebo): 812P301 100 mg 5.46 (95% CI 2.21 to 8.70), 200 mg 6.32 (95% CI 3.19 to 9.45); 812P303 200 mg 6.39 (95% CI 2.26 to 10.52), 400 mg 5.99 (95% CI 1.75 to 10.24); 812P302 200 mg 5.18 (95% CI 1.18 to 9.17), 400 mg 5.83 (95% CI 1.95 to 9.70); 812P304 400 mg 4.48 (95% CI 0.62 to 8.35). Qelbree label placebo-subtracted LS-mean (already on this page): Study 1 100 mg −5.8, 200 mg −6.9; Study 2 200 mg −6.0, 400 mg −5.8; Study 3 200 mg −4.5, 400 mg −5.1. Both are printed; not pooled.

What the evidence does not establish:

• That non-stimulants are equivalent to stimulants (guanfacine and clonidine clinician SMDs −0.67 and −0.71 versus Cortese child clinician amphetamines −1.02 and methylphenidate −0.78)
• That atomoxetine or viloxazine are free from suicidality risk (both carry boxed warnings for suicidal ideation in pediatric patients)
• That guanfacine or clonidine work in adults (one Japanese adult guanfacine RCT Iwanami 2020 difference −4.28; US labels have no adult ADHD indication)
• That bupropion is an effective ADHD treatment when rated by clinicians only (Verbeeck clinician-only sensitivity SMD −0.36, not significant)
• That adult immediate-release methylphenidate shows significant participant-rated effects (Cândido 2021 participant SMD −0.59, not significant)
• Long-term functional outcomes for non-stimulant ADHD medications

The structural problem that conditions all of this: Cortese et al. (2018) closed its search in April 2017; viloxazine is outside that network. Guanfacine and clonidine parent and teacher ratings are less consistent than clinician ratings. Atomoxetine and viloxazine both carry boxed warnings for suicidal ideation in pediatric patients (though adult atomoxetine analysis did not find increased risk). Lisdexamfetamine is a prodrug amphetamine with a dedicated Cochrane review (Castells 2018) finding adult clinician SMD −1.06, but it is not pooled with Cortese's adult amphetamine SMD −0.79. Bupropion's clinician-only sensitivity analysis (Verbeeck 2017) is not significant. Adult methylphenidate now has dedicated Cochrane reviews: Boesen 2022 extended-release investigator SMD −0.42, Cândido 2021 immediate-release investigator mean difference −20.70 but participant SMD not significant. Adult guanfacine has a Japanese RCT (Iwanami 2020) difference −4.28, but no US adult indication. Child guanfacine has three opened RCTs (Biederman 2008, Sallee 2009, Newcorn 2013), not all matching label summaries. Clonidine extended-release has a later label printing LS-means the 2010 Kapvay label omitted. Viloxazine has a negative sequential-testing trial (812P304) the label does not print. The folk models—that non-stimulants are substitutable first-line options or that only stimulants work—both overclaim what opened sources support.

---

How to read this review

The single most important heuristic

A medication effect without naming the source, rater, and population is not a clean estimate. Guanfacine clinician-rated SMD −0.67 is Cortese 2018; guanfacine parent-rated SMD −0.23 is not significant. Viloxazine adult AISRS placebo-subtracted −3.7 with two different confidence intervals (Nasser paper versus Qelbree label) reporting the same trial. Lisdexamfetamine adult clinician SMD −1.06 is Castells 2018, not Cortese 2018. Any claim that "this medication works" must specify which outcome, in which population, rated by whom, versus what comparator, from which opened source.

Four structural problems

Boxed warnings for suicidality. Atomoxetine and viloxazine both carry FDA boxed warnings for suicidal thoughts and behaviors in pediatric patients. Atomoxetine pediatric suicidal ideation 0.4% (5 of 1,357) versus 0% (0 of 851) placebo; viloxazine pediatric 0.9% (9 of 1,019) versus 0.4% (2 of 463); adult 1.6% (3 of 189) versus 0% (0 of 183). No completed suicides. Adult atomoxetine analysis did not reveal increased risk.

Rater variance. Guanfacine parent-rated SMD −0.23 is not significant, but clinician-rated SMD −0.67 is significant. Clonidine clinician-rated SMD −0.71 is significant, but clonidine CGI-I OR 2.78 is not significant. Bupropion pooled severity SMD −0.50 is significant, but clinician-rated-only sensitivity SMD −0.36 is not significant. Adult extended-release methylphenidate investigator-rated SMD −0.42 and self-rated SMD −0.37 are similar (Boesen 2022), but immediate-release methylphenidate investigator-rated mean difference −20.70 is significant while participant-rated SMD −0.59 is not significant (Cândido 2021). The magnitude and significance of effects depend on who rates the outcome.

Network meta-analysis coverage. Cortese et al. (2018) closed its search 7 April 2017. Viloxazine received initial U.S. approval in 2021; it is outside the Cortese network. Adult guanfacine Iwanami 2020 Japan RCT is outside the Cortese network. Boesen et al. (2022) and Cândido et al. (2021) are dedicated adult methylphenidate Cochrane reviews; their estimates are not pooled with Cortese's adult methylphenidate clinician SMD −0.49. Castells et al. (2018) is a dedicated adult amphetamine Cochrane review; its lisdexamfetamine SMD −1.06 is not pooled with Cortese's adult amphetamine clinician SMD −0.79. Verbeeck et al. (2017) is a dedicated bupropion Cochrane review; its pooled SMD −0.50 includes non-clinician raters, and its clinician-only sensitivity SMD −0.36 is not significant. Each opened source has its own inclusion criteria, rater restrictions, and named-compound subgroups. Pooled estimates from different meta-analyses are not directly comparable.

Label-manuscript discrepancies. Sallee et al. (2009) report Intuniv Study 2 with 2 mg placebo-adjusted LS-mean −5.41, but the FDA Intuniv 2018 label reports Study 2 range "−6.8 to −7.9"; the 2 mg dose is outside that range. Nasser et al. (2022) adult viloxazine paper prints 95% CI −6.3 to −2.2, but the Qelbree label prints 95% CI −6.2 to −1.2 for the same trial. The 2010 Kapvay label prints no ADHD-RS LS-mean; the 2023 revised clonidine label prints Study 1 LS-means −8.5 and −9.1. Viloxazine adolescent trial 812P304 is negative by sequential-testing protocol but is not printed on the Qelbree label.

Conventions

Claims are marked [contested] where the literature genuinely disagrees on a number (Nasser paper 95% CI −6.3 to −2.2 versus Qelbree label 95% CI −6.2 to −1.2 for the same trial).

Confidence intervals are omitted where they could not be verified against the primary text, rather than reconstructed.

---

Child Alpha-2 Agonists (Guanfacine and Clonidine)

Cortese et al. (2018) conducted a network meta-analysis of comparative efficacy and tolerability of medications for ADHD in children, adolescents, and adults. Published in The Lancet Psychiatry 5:727–738, DOI 10.1016/S2215-0366(18)30269-4, PMC6109107. The study included 133 double-blind RCTs (81 in children and adolescents, 51 in adults, 1 in a mixed population).

Search dates: Databases searched from inception to 7 April 2017.

Efficacy analysis: Symptom severity rated by clinicians, teachers, or parents at time point closest to 12 weeks. Efficacy analysis included 10,068 children and adolescents and 8,131 adults.

Child and adolescent guanfacine results:

• Clinician-rated SMD versus placebo: −0.67 (95% CI −0.85 to −0.50)
• Parent-rated SMD versus placebo: −0.23 (95% CI −0.90 to 0.45), not significant
• Dropout for adverse events OR versus placebo: 2.64 (95% CI 1.20 to 5.81)

Child and adolescent clonidine results:

• Clinician-rated SMD versus placebo: −0.71 (95% CI −1.17 to −0.24)
• CGI-I response OR versus placebo: 2.78 (95% CI 0.91 to 8.53), not significant

The paper states: "We did not identify any trials of clonidine or guanfacine in adults." This reflects the search closure date of 7 April 2017.

Figure note: Effect sizes restated from Cortese et al. (2018) network meta-analysis. Guanfacine parent-rated SMD was not significant. Both medications show smaller clinician-rated effects than amphetamines (SMD −1.02) or methylphenidate (SMD −0.78) in the same analysis.

Interpretation: Guanfacine shows a significant clinician-rated effect (SMD −0.67) but no significant parent-rated effect (SMD −0.23). Clonidine shows a significant clinician-rated effect (SMD −0.71) but no significant CGI-I response. Both medications show smaller effect sizes on clinician ratings than amphetamines (child clinician SMD −1.02) or methylphenidate (child clinician SMD −0.78) in the same network meta-analysis.

1.2 Guanfacine XR label and the Biederman 2008 trial: ADHD-RS LS-means

The FDA Intuniv (guanfacine extended-release) label revised 2018 reports two pivotal trials in pediatric ADHD.

Indication (exact): "INTUNIV® is indicated for the treatment of Attention Deficit Hyperactivity Disorder (ADHD) as monotherapy and as adjunctive therapy to stimulant medications."

Population: Efficacy tables report ages 6–17 years. No US adult ADHD claim on this label.

Study 1 results (label Table 16):

ADHD-RS-IV placebo-subtracted LS-mean:

• 2 mg: −7.4 (95% CI −11.3 to −3.5)
• 3 mg: −7.5 (95% CI −11.4 to −3.6)
• 4 mg: −10.0 (95% CI −13.9 to −6.1)

Study 2 results (label text): 1–4 mg range −6.8 to −7.9 versus placebo.

Biederman et al. (2008) report the same Study 1 population. Published in Pediatrics 121:e73–e84, DOI 10.1542/peds.2006-3695. ClinicalTrials.gov identifier NCT00152009. Study labeled as Intuniv Study 1 on FDA label.

Population: N = 345, ages 6–17 years.

Duration: 8 weeks.

ADHD-RS-IV LS-mean change from baseline:

• 2 mg: −16.18
• 3 mg: −16.43
• 4 mg: −18.87
• Placebo: −8.48

Placebo-adjusted LS-mean (95% CI):

• 2 mg: −7.70 (95% CI −12.25 to −3.15)
• 3 mg: −7.95 (95% CI −12.50 to −3.40)
• 4 mg: −10.39 (95% CI −14.97 to −5.82)

Post-hoc effect sizes: 0.64 / 0.66 / 0.86 for 2 / 3 / 4 mg.

Adverse-event discontinuation: 16.2% versus 1.2% placebo.

Interpretation: The label and the Biederman paper report the same trial. The label rounds the LS-means to −15.9 / −16.0 / −18.5 versus the paper's −16.18 / −16.43 / −18.87, but the placebo-subtracted confidence intervals are similar. Effect sizes are moderate (0.64–0.86 by post-hoc calculation), smaller than methylphenidate teacher ratings (Cortese child teacher SMD −0.82) and substantially smaller than amphetamine clinician ratings (Cortese child clinician SMD −1.02). Adverse-event discontinuation is substantial (16.2% versus 1.2%).

1.3 Clonidine XR label: indication, blood pressure, and what the label does not print

The FDA Kapvay (clonidine extended-release) label approved 2010 reports two pivotal trials in pediatric ADHD.

Indication (exact): "KAPVAY™ is a centrally acting alpha2-adrenergic agonist indicated for the treatment of attention deficit hyperactivity disorder (ADHD) as monotherapy or as adjunctive therapy to stimulant medications."

Population: Efficacy described for pediatric patients. No adult ADHD trial. No boxed warning.

Study 1: Monotherapy, N = 236.

Study 2: Adjunctive therapy to stimulant medications, N = 198.

Efficacy statement (label text): Both studies found Kapvay "statistically significantly superior" to placebo on ADHD-RS-IV total score at week 5 (primary endpoint).

What the label does not print: The 2010 Kapvay label prints no LS-mean, no SMD, and no response rate for ADHD-RS-IV outcomes.

Blood pressure and heart rate effects (treatment-period maximum placebo-subtracted):

At 0.2 mg/day:

• Systolic blood pressure: −4.0 mm Hg
• Diastolic blood pressure: −4.0 mm Hg
• Heart rate: −4.0 bpm

At 0.4 mg/day:

• Systolic blood pressure: −8.8 mm Hg
• Diastolic blood pressure: −7.3 mm Hg
• Heart rate: −7.7 bpm

Interpretation: The label states superiority on ADHD-RS-IV but does not quantify the effect size. The child clonidine clinician-rated SMD −0.71 (95% CI −1.17 to −0.24) cited in this review comes from Cortese et al. (2018), not the Kapvay label. The label reports blood pressure and heart rate reductions as expected for an alpha2-adrenergic agonist. No adult ADHD trial is reported. The Jain et al. (2011) publication body is unopened; no Jain LS-means are cited here.

---

Part II — Boxed warnings: atomoxetine, viloxazine, and lisdexamfetamine

2.1 Atomoxetine: boxed warning for suicidal ideation in pediatric patients, no adult signal

The FDA Strattera (atomoxetine) label carries a boxed warning (exact): "WARNING: SUICIDAL THOUGHTS AND BEHAVIORS IN PEDIATRIC PATIENTS 6 YEARS OF AGE AND OLDER."

Pediatric suicidality:

Short-term placebo-controlled trials found suicidal ideation in:

• 0.4% (5 of 1,357) of atomoxetine-treated pediatric patients
• 0% (0 of 851) of placebo-treated patients

Details:

• All five cases were ages 6–12
• All occurred within the first month of treatment
• One case was a suicide attempt
• No completed suicides

Adult suicidality:

The label states: "Adult ADHD analysis did not reveal an increased risk of suicidal ideation or behavior."

Indication (exact): Indicated for ADHD in adults and pediatric patients 6 years and older.

Adult Studies 6 and 7 (label description):

• Michelson et al. 2003 trial identity (N = 280 and N = 256, 10 weeks, investigator CAARS 18-item)
• Label states outcomes were "statistically significantly improved" versus placebo
• The label prints no CAARS LS-mean or SMD

Adult adverse-event discontinuation: 11.3% (61 of 541) versus 3% (12 of 405) placebo.

Adult cardiovascular effects (Table 7):

Maximum heart rate increase ≥20 bpm:

• Atomoxetine: 22%
• Placebo: 8%

Interpretation: The boxed warning states "WARNING: SUICIDAL THOUGHTS AND BEHAVIORS IN PEDIATRIC PATIENTS 6 YEARS OF AGE AND OLDER." Five cases of suicidal ideation including one attempt in 1,357 pediatric patients (0.4%) versus zero in 851 placebo (0%) is a small absolute number, but the relative risk is undefined (zero denominator). All cases were ages 6–12 in month 1. Adult analysis did not find increased risk. Adult adverse-event discontinuation is substantial (11.3% versus 3%). The label identifies the adult trials as Michelson et al. 2003 but prints no LS-mean or SMD for CAARS outcomes. The Michelson 2003 body is unopened; no CAARS LS-mean or effect size is cited here.

The Cortese et al. (2018) adult clinician atomoxetine SMD −0.45 (95% CI −0.58 to −0.32) remains available as a supporting line when the rater is named.

2.2 Viloxazine: boxed warning for suicidality, pediatric and adult ADHD-RS / AISRS labeled effects

The FDA Qelbree (viloxazine extended-release) label carries a boxed warning (exact): "WARNING: SUICIDAL THOUGHTS AND BEHAVIORS."

The warning states: "In clinical studies, higher rates of suicidal thoughts and behavior were reported in patients with ADHD treated with Qelbree than in patients treated with placebo."

Pediatric suicidality:

• 9 of 1,019 (0.9%) versus 2 of 463 (0.4%) placebo

Adult suicidality:

• 3 of 189 (1.6%) versus 0 of 183 (0%) placebo

No completed suicides.

Indication (exact): ADHD in adults and pediatric patients 6 years and older.

Pediatric ADHD-RS-5 placebo-subtracted LS-mean (label):

Study 1 (ages 6–11):

• 100 mg: −5.8 (95% CI −8.9 to −2.6)
• 200 mg: −6.9 (95% CI −10.0 to −3.8)

Study 2 (ages 6–11):

• 200 mg: −6.0 (95% CI −10.0 to −1.9)
• 400 mg: −5.8 (95% CI −9.9 to −1.7)

Study 3 (ages 12–17):

• 200 mg: −4.5 (95% CI −8.4 to −0.6)
• 400 mg: −5.1 (95% CI −8.9 to −1.3)

Adult AISRS (Adult ADHD Investigator Symptom Rating Scale) placebo-subtracted LS-mean (label Study 4):

• −3.7 (95% CI −6.2 to −1.2)

Adult heart rate effect:

Heart rate increase ≥20 bpm any time:

• 52 of 178 (29%) viloxazine
• 23 of 181 (13%) placebo

Nasser et al. (2022) report of Study 4:

Published in CNS Drugs 36:897–915, DOI 10.1007/s40263-022-00938-w, PMC9328182. ClinicalTrials.gov identifier NCT04016779.

Population: Adults 18–65 years, ADHD.

Duration: 6 weeks.

AISRS LS-mean change from baseline:

• Viloxazine: −15.5
• Placebo: −11.7
• Difference: −3.7 (95% CI −6.3 to −2.2); p = 0.0040

AISRS responders ≥30% improvement:

• 60.0% (78 of 130) viloxazine
• 47.6% (68 of 143) placebo

AISRS responders ≥50% improvement: Not significant.

Adverse-event discontinuation: 9.0% versus 4.9%.

[contested] Confidence interval discrepancy:

• Nasser paper: 95% CI −6.3 to −2.2
• Qelbree label: 95% CI −6.2 to −1.2

Both report the same trial (Study 4, NCT04016779). The point estimate (−3.7) matches. The confidence intervals do not. The paper and the label are cited separately here.

Interpretation: Viloxazine is outside the Cortese et al. (2018) network meta-analysis (search closed 7 April 2017; viloxazine initial U.S. approval 2021). The boxed warning for suicidality applies to both pediatric and adult patients. Pediatric suicidal ideation rates are 0.9% versus 0.4% placebo; adult rates are 1.6% versus 0% placebo; no completed suicides. Pediatric ADHD-RS-5 placebo-subtracted effects range from about −4.5 to −6.9 across studies and doses. Adult AISRS placebo-subtracted effect is −3.7. No SMD is provided in the opened viloxazine sources. Adult heart rate increase ≥20 bpm occurs in 29% versus 13% placebo. The confidence interval discrepancy between the paper and the label for the same trial is noted as [contested].

Figure note: Two printed 95% CIs for the same trial (NCT04016779). Point estimate −3.7 matches; confidence intervals do not. Qelbree label: −6.2 to −1.2. Nasser 2022 CNS Drugs: −6.3 to −2.2. Both are cited separately here.

2.3 Lisdexamfetamine: boxed warning for abuse, misuse, and addiction

The FDA Vyvanse (lisdexamfetamine dimesylate) label carries a boxed warning (exact): "WARNING: ABUSE, MISUSE, AND ADDICTION."

The warning states: "VYVANSE has a high potential for abuse and misuse, which can lead to the development of a substance use disorder, including addiction."

Indication (exact): ADHD in adults and pediatric patients ≥6 years, and moderate to severe binge eating disorder in adults.

Adult Study 7 (label, ages 18–55):

The label identifies this as the Adler et al. 2008 trial. The Adler 2008 manuscript body is unopened; numbers cited here are from the Vyvanse label only.

Investigator ADHD-RS LS-mean:

• 30 mg: −16.2
• 50 mg: −17.4
• 70 mg: −18.6
• Placebo: −8.2

Placebo-subtracted (95% CI):

• 30 mg: −8.0 (95% CI −11.5 to −4.6)
• 50 mg: −9.2 (95% CI −12.6 to −5.7)
• 70 mg: −10.4 (95% CI −13.9 to −6.9)

Adult Study 9 (withdrawal design):

Treatment failure:

• 9% lisdexamfetamine
• 75% placebo

Interpretation: Lisdexamfetamine is a prodrug amphetamine. The boxed warning for abuse, misuse, and addiction applies. Adult ADHD-RS placebo-subtracted effects range from −8.0 to −10.4 across doses. The withdrawal-design Study 9 shows treatment failure in 9% versus 75% placebo, supporting efficacy during active treatment. The Adler 2008 manuscript body is unopened; the label provides the LS-means cited here.

Figure note: Suicidal ideation rates from short-term placebo-controlled trials. No completed suicides in any trial. Atomoxetine pediatric: 0.4% vs 0%; adult analysis did not reveal increased risk. Viloxazine pediatric: 0.9% vs 0.4%; adult: 1.6% vs 0%. Guanfacine XR and clonidine XR have no boxed warnings.

---

Part III — Adult named-compound trials: lisdexamfetamine Cochrane, bupropion Cochrane, and leftover Cortese rows

3.1 Lisdexamfetamine: Castells 2018 Cochrane review of adult amphetamines

Castells et al. (2018) conducted a Cochrane systematic review of amphetamines for ADHD in adults. Published in Cochrane Database of Systematic Reviews 2018;8:CD007813, DOI 10.1002/14651858.CD007813.pub3, PMC6513464.

Population: Adults with ADHD.

Included studies: 19 studies, 2,521 adults (six used a cross-over design).

Primary outcome: Clinician-rated ADHD severity versus placebo.

All amphetamines pooled (clinician-rated):

SMD −0.90 (95% CI −1.04 to −0.75); 13 studies, 2,028 participants; low- to very-low quality.

Named compound lisdexamfetamine (clinician-rated):

SMD −1.06 (95% CI −1.26 to −0.85); 7 studies, 896 participants.

Named compound mixed amphetamine salts (clinician-rated):

SMD −0.80 (95% CI −0.93 to −0.66).

Named compound dexamphetamine (clinician-rated):

SMD −0.24 (95% CI −0.80 to 0.32), not significant; 1 study, 49 participants.

Adverse-event withdrawal:

Relative risk 2.69 (95% CI 1.63 to 4.45).

Interpretation: This is a dedicated Cochrane review of adult amphetamines, distinct from the Cortese et al. (2018) network meta-analysis. Castells finds adult clinician-rated lisdexamfetamine SMD −1.06, larger than the Cortese adult amphetamine clinician SMD −0.79 (95% CI −0.99 to −0.58). The two estimates are not pooled; they come from different meta-analyses with different inclusion criteria. Lisdexamfetamine is a prodrug amphetamine with a dedicated evidence base. Adverse-event withdrawal RR 2.69 is substantial. Quality of evidence is low to very low.

Figure note: Two separate meta-analyses (Cortese 2018 and Castells 2018 Cochrane). Do not pool these estimates. Castells includes named-compound breakouts: lisdexamfetamine SMD −1.06, mixed amphetamine salts SMD −0.80, dexamphetamine not significant.

3.2 Bupropion: Verbeeck 2017 Cochrane review, pooled and clinician-only sensitivity

Verbeeck et al. (2017) conducted a Cochrane systematic review of bupropion for ADHD in adults. Published in Cochrane Database of Systematic Reviews 2017;(10):CD009504, DOI 10.1002/14651858.CD009504.pub2, PMC6485546.

Search dates: Search current to February 2017.

Population: Adults with ADHD.

Included studies: Six RCTs, 438 adults.

Primary outcome: ADHD symptom severity (pooled clinician and self-report).

Pooled severity SMD:

SMD −0.50 (95% CI −0.86 to −0.15); 129 participants, 3 studies; GRADE low.

Response ≥30% improvement:

RR 1.50 (95% CI 1.13 to 1.99).

CGI-I 1 or 2:

RR 1.78 (95% CI 1.27 to 2.50).

Adverse-event withdrawal:

RR 1.20 (95% CI 0.35 to 4.10), not significant.

Clinician-rated-only sensitivity analysis:

SMD −0.36 (95% CI −0.79 to 0.07), 87 participants, not significant.

Interpretation: The pooled severity SMD −0.50 includes both clinician and self-report ratings. When restricted to clinician-rated outcomes only, the effect is smaller (SMD −0.36) and not statistically significant. Response rates (≥30% improvement RR 1.50; CGI-I 1 or 2 RR 1.78) are significant. This is a dedicated Cochrane review of bupropion, distinct from the Cortese et al. (2018) network meta-analysis. Cortese reports adult bupropion clinician SMD −0.46 (95% CI −0.85 to −0.07). The two estimates are not pooled; they come from different meta-analyses. Verbeeck's clinician-only sensitivity suggests the pooled effect may be driven by self-report ratings.

3.3 Re-bindable Cortese 2018 adult clinician rows: atomoxetine, bupropion, methylphenidate

Cortese et al. (2018) adult clinician-rated SMDs versus placebo (search closed 7 April 2017):

• Amphetamines: SMD −0.79 (95% CI −0.99 to −0.58)
• Methylphenidate: SMD −0.49 (95% CI −0.64 to −0.35)
• Bupropion: SMD −0.46 (95% CI −0.85 to −0.07)
• Atomoxetine: SMD −0.45 (95% CI −0.58 to −0.32)
• Modafinil: SMD 0.16 (95% CI −0.28 to 0.59), not significant

Adult tolerability (dropout for adverse events OR versus placebo):

• Atomoxetine: OR 2.33 (95% CI 1.28 to 4.25)
• Methylphenidate: OR 2.39 (95% CI 1.40 to 4.08)
• Modafinil: OR 4.01 (95% CI 1.42 to 11.33)

Interpretation: These rows are re-bindable as supporting text when the rater (clinician) is named. They are not new leftover headlines; they were printed as a single paragraph (§2.2) in the live stimulant essay. Atomoxetine SMD −0.45 and bupropion SMD −0.46 are clinician-rated and similar in magnitude. Methylphenidate SMD −0.49 is also clinician-rated. Modafinil is not significant. Dropout for adverse events is substantial for atomoxetine (OR 2.33), methylphenidate (OR 2.39), and modafinil (OR 4.01).

---

Part IV — Adult methylphenidate Cochrane reviews: extended-release and immediate-release

4.1 Extended-release methylphenidate: Boesen 2022 Cochrane review

Boesen et al. (2022) conducted a Cochrane systematic review of extended-release methylphenidate for ADHD in adults. Published in Cochrane Database of Systematic Reviews 2022;2:CD012857, DOI 10.1002/14651858.CD012857.pub2, PMC8869321.

Population: Adults with ADHD.

Included studies: 24 trials, 5,066 adults.

Median duration: 8 weeks.

Investigator-rated ADHD severity versus placebo:

SMD −0.42 (95% CI −0.49 to −0.36); 18 trials, 4,183 participants; GRADE very low.

Self-rated ADHD severity versus placebo:

SMD −0.37 (95% CI −0.43 to −0.30); 16 trials, 3,799 participants on summary-of-findings table; GRADE very low.

The paper's Analysis 1.4 header prints 17 studies for self-rated outcomes, but the same participant count and SMD as the summary-of-findings table reporting 16 studies.

Adverse-event discontinuation:

RR 2.36 (95% CI 1.62 to 3.43); 15 trials, 3,963 participants.

Any adverse event:

RR 1.27 (95% CI 1.19 to 1.37).

Serious adverse event:

RR 1.43 (95% CI 0.85 to 2.43), not significant.

Interpretation: This is a dedicated Cochrane review of adult extended-release methylphenidate. Investigator-rated SMD −0.42 and self-rated SMD −0.37 are similar in magnitude. Both rater types show significant effects with very low quality evidence. This is not pooled with Cortese et al. (2018) adult methylphenidate clinician SMD −0.49 (95% CI −0.64 to −0.35). The two estimates come from different meta-analyses with different inclusion criteria. Adverse-event discontinuation RR 2.36 is substantial.

4.2 Immediate-release methylphenidate: Cândido 2021 Cochrane review

Cândido et al. (2021) conducted a Cochrane systematic review of immediate-release methylphenidate for ADHD in adults. Published in Cochrane Database of Systematic Reviews 2021;1:CD013011, DOI 10.1002/14651858.CD013011.pub2, PMC8092481.

Population: Adults with ADHD.

Included studies: 10 trials, 497 adults.

Duration range: 6–18 weeks.

Investigator-rated AISRS mean difference versus placebo:

MD −20.70 (95% CI −23.97 to −17.43); 1 trial, 146 participants; GRADE very low.

Participant-rated SMD versus placebo:

SMD −0.59 (95% CI −1.25 to 0.06); 2 trials, 138 participants; not significant; GRADE very low.

No pooled clinician SMD.

Adverse-event withdrawal: Not pooled.

Gastrointestinal adverse events:

RR 1.96 (95% CI 1.13 to 2.95).

Appetite loss:

RR 1.77 (95% CI 1.06 to 2.96).

Interpretation: This is a thinner Cochrane body than the extended-release review. The investigator-rated mean difference −20.70 on AISRS is significant, but participant-rated SMD −0.59 is not significant. The mean difference is not converted to SMD here. No pooled clinician SMD is available. Adverse-event withdrawal rates are not pooled. This is not pooled with Boesen 2022 extended-release or Cortese 2018 adult methylphenidate clinician SMD −0.49. Gastrointestinal effects and appetite loss are significant.

Figure note: Three separate meta-analyses with different inclusion criteria and rater restrictions. Do not pool these estimates. Cândido 2021 reports one investigator-rated mean difference (not SMD) and a participant SMD that was not significant. Boesen 2022 and Cortese 2018 are separate networks.

---

Part V — Adult guanfacine extended-release: Japan RCT, no US adult indication

5.1 Iwanami 2020: Japanese adult guanfacine XR RCT

Iwanami et al. (2020) conducted a randomized placebo-controlled trial of guanfacine extended-release in Japanese adults with ADHD. Published in Journal of Clinical Psychiatry 81(3):19m12979, DOI 10.4088/JCP.19m12979, PMID 32297719. ClinicalTrials.gov identifier JapicCTI-163231.

Population: Japanese adults ≥18 years with ADHD.

Randomized: N = 201 (full analysis set 100 guanfacine / 100 placebo).

Mean ages: 31.1 years (guanfacine) versus 33.8 years (placebo).

Duration: 10 weeks.

ADHD-RS-IV LS-mean change from baseline (week 10):

• Guanfacine: −11.55 ± 1.10
• Placebo: −7.27 ± 1.07
• Difference: −4.28 (95% CI −6.67 to −1.88)

Effect size: 0.52.

Adverse-event discontinuation: 19.8% versus 3.0%.

Regulatory status:

The paper states guanfacine is approved for ADHD in adults in Japan.

US regulatory status:

The US FDA Intuniv 2018 label indication remains ages 6–17 years. No adult ADHD claim on the US label.

Interpretation: This is an adult guanfacine extended-release trial in a Japanese population. The ADHD-RS-IV LS-mean difference −4.28 with effect size 0.52 is significant. Adverse-event discontinuation is substantial (19.8% versus 3.0%). Cortese et al. (2018) statement "we did not identify any trials of clonidine or guanfacine in adults" is accurate for that review's search closure date of 7 April 2017. This trial was published in 2020. The US Intuniv label has no adult ADHD indication.

---

Part VI — Clonidine extended-release: later label with LS-means

6.1 Actavis/Teva 2023 label versus 2010 Kapvay label

The Actavis/Teva clonidine hydrochloride extended-release DailyMed label revised October 2023 (setid 0100c70d-7fde-46a1-8374-940550a27e43) prints ADHD-RS-IV LS-means the 2010 Kapvay label did not report.

Pediatric ages 6–17 ADHD-RS-IV placebo-subtracted LS-mean:

Study 1 (monotherapy):

• 0.2 mg: −8.5 (95% CI −12.2 to −4.8)
• 0.4 mg: −9.1 (95% CI −12.8 to −5.5)

Study 2 (adjunctive):

• −4.5 (95% CI −7.8 to −1.1)

No boxed warning. No adult trial.

What the 2010 Kapvay label reported:

The 2010 Kapvay label (as cited in Part I section 1.3 of this review) stated both studies found Kapvay "statistically significantly superior" to placebo on ADHD-RS-IV but printed no LS-mean, no SMD, and no response rate. The 2010 label did report blood pressure and heart rate effects.

Interpretation: The 2023 revised label quantifies the ADHD-RS-IV effect sizes the 2010 label omitted. Study 1 monotherapy placebo-subtracted LS-means are −8.5 and −9.1 for 0.2 and 0.4 mg. Study 2 adjunctive therapy placebo-subtracted LS-mean is −4.5. These LS-means are consistent with the Cortese et al. (2018) child clonidine clinician-rated SMD −0.71 (95% CI −1.17 to −0.24) from the network meta-analysis. The 2010 Kapvay label blood pressure and heart rate effects (Part I section 1.3) remain as shipped. The Jain et al. (2011) manuscript body remains unopened.

---

Part VII — Child guanfacine extended-release: three opened RCTs

7.1 Intuniv Study 2 manuscript: Sallee 2009

Sallee et al. (2009) report a randomized placebo-controlled trial of guanfacine extended-release in children and adolescents with ADHD. Published in Journal of the American Academy of Child and Adolescent Psychiatry 48:155–165, DOI 10.1097/CHI.0b013e318191769e. ClinicalTrials.gov identifier NCT00150618.

Population: N = 306 (intent-to-treat), ages 6–17 years.

Duration: 9 weeks (3-week escalation, 3-week maintenance, 3-week taper).

ADHD-RS-IV placebo-adjusted LS-mean:

• 1 mg: −6.75 (p = 0.0041)
• 2 mg: −5.41 (p = 0.0176)
• 3 mg: −7.34 (p = 0.0016)
• 4 mg: −7.88 (p = 0.0006)

Post-hoc effect sizes: 0.53 / 0.43 / 0.58 / 0.62 for 1 / 2 / 3 / 4 mg.

Adverse-event discontinuation: 7.4% (combined guanfacine) versus 7.6% placebo.

Combined mean change from baseline: −19.6 (SD 13.9) versus −12.2 (SD 13.0) placebo.

Ages 13–17 subgroup (n = 80): Not significant at any dose.

No 95% CI on the four primary contrasts.

FDA Intuniv 2018 label Study 2 summary:

The label reports Study 2: 1–4 mg range −6.8 to −7.9 versus placebo.

Discrepancy:

The Sallee 2009 paper's 2 mg placebo-adjusted LS-mean −5.41 is outside the label's reported range "−6.8 to −7.9." Both report the same trial (NCT00150618 identified as Intuniv Study 2 on the label).

Interpretation: The paper prints the four dose-specific LS-means the label summarizes as a range. The 2 mg dose LS-mean −5.41 is outside the label range. Both are printed here; no reconciliation is attempted. Post-hoc effect sizes range from 0.43 to 0.62. Adverse-event discontinuation rates are similar (7.4% versus 7.6%). The ages 13–17 subgroup did not show significant effects at any dose. This trial is distinct from Biederman et al. (2008) Study 1 (Part I section 1.2).

7.2 Newcorn 2013: AM versus PM dosing

Newcorn et al. (2013) report a randomized placebo-controlled trial of guanfacine extended-release comparing morning and evening dosing in children with ADHD. Published in Journal of the American Academy of Child and Adolescent Psychiatry 52:921–930, DOI 10.1016/j.jaac.2013.06.006. ClinicalTrials.gov identifier NCT00997984.

Population: Children ages 6–12, N = 333 (AM dosing 107 / PM dosing 114 / placebo 112).

Duration: Week 8 (visit 10 LOCF).

ADHD-RS-IV mean change from baseline:

• AM dosing: −19.8
• PM dosing: −20.1
• Placebo: −11.0

Authors' effect sizes: 0.75 (AM) / 0.78 (PM).

Adverse-event discontinuation: 16 of 221 guanfacine versus 0 of 112 placebo.

Somnolence: 44.3% (guanfacine) versus 12.5% (placebo).

Interpretation: This trial is not Intuniv Study 1 (Biederman 2008) or Study 2 (Sallee 2009). The effect sizes 0.75 and 0.78 are larger than Sallee's post-hoc range (0.43–0.62) and similar to Biederman's higher doses (0.64–0.86). Adverse-event discontinuation is 16 of 221 (7.2%) versus 0 placebo. Somnolence is substantial (44.3% versus 12.5%). Both AM and PM dosing show similar efficacy.

Figure note: Three separate guanfacine XR trials with different durations, populations, and dosing schedules. Do not pool these estimates. Intuniv label Study 1 matches Biederman 2008. Sallee 2009 2 mg dose (−5.41) is outside the label's stated Study 2 range. Newcorn 2013 is not Study 1 or 2.

---

Part VIII — Pediatric viloxazine: negative sequential-testing trial and treatment effect reprints

8.1 Viloxazine 812P304: sequential-testing negative trial

Nasser et al. (2021) report a randomized placebo-controlled trial of viloxazine extended-release in adolescents with ADHD. Published in Psychopharmacology Bulletin 51(2):43–64, PMC8146561. ClinicalTrials.gov identifier NCT03247556.

Population: Adolescents ages 12–17 years, intent-to-treat N = 292.

Duration: 7 weeks (2-week titration + 5-week maintenance).

ADHD-RS-5 LS-mean change from baseline:

• 400 mg: −18.3 ± 1.36
• 600 mg: −16.7 ± 1.39
• Placebo: −13.2 ± 1.38

Placebo-adjusted:

• 400 mg: −5.1 ± 1.93 (p = 0.0082)
• 600 mg: −3.5 ± 1.93 (p = 0.0712)

Sequential testing procedure:

The study tested 600 mg first. Because 600 mg was not significant (p = 0.0712), sequential testing protocol determined neither dose was considered superior to placebo.

Responders ≥50% improvement:

• 400 mg: 48.2%
• 600 mg: 46.0%
• Placebo: 32.9%

Adverse-event discontinuation:

• 400 mg: 4.0%
• 600 mg: 5.1%
• Placebo: 1.0%
• Combined viloxazine: 4.5%

Regulatory status:

This trial (812P304) is not the Qelbree label Studies 1–3 (Parts II section 2.2).

Interpretation: This is a negative trial by sequential-testing protocol. The 600 mg dose tested first was not significant, so neither dose was considered superior despite the 400 mg dose p-value of 0.0082. Responder rates ≥50% improvement favor both doses versus placebo. Adverse-event discontinuation is low (4.5% combined versus 1.0%). The Qelbree label does not print this trial's outcomes. No pairwise 95% CI is available in this body.

Figure note: Sequential-testing negative trial (NCT03247556). 600 mg tested first; not significant (p=0.0712), so neither dose considered superior despite 400 mg p=0.0082. No 95% CI provided; only point estimates and standard errors shown. This trial is not Qelbree label Studies 1–3.

8.2 Nasser 2022 BJCP: treatment effect reprints from labeled trials

Nasser et al. (2022) report treatment effect estimates from pediatric viloxazine trials. Published in British Journal of Clinical Pharmacology 88:4828–4838, DOI 10.1111/bcp.15412, PMC9796605.

End-of-study treatment effect (positive value = larger improvement versus placebo):

812P301 (ages 6–11):

• 100 mg: 5.46 (95% CI 2.21 to 8.70)
• 200 mg: 6.32 (95% CI 3.19 to 9.45)

812P303 (ages 6–11):

• 200 mg: 6.39 (95% CI 2.26 to 10.52)
• 400 mg: 5.99 (95% CI 1.75 to 10.24)

812P302 (ages 12–17):

• 200 mg: 5.18 (95% CI 1.18 to 9.17)
• 400 mg: 5.83 (95% CI 1.95 to 9.70)

812P304 (ages 12–17):

• 400 mg: 4.48 (95% CI 0.62 to 8.35)

The paper's 600 mg confidence interval is omitted here (lower bound inconsistent with treatment effect ± standard error).

Qelbree label placebo-subtracted LS-means (already cited in Part II section 2.2):

Study 1 (ages 6–11): 100 mg −5.8, 200 mg −6.9 Study 2 (ages 6–11): 200 mg −6.0, 400 mg −5.8 Study 3 (ages 12–17): 200 mg −4.5, 400 mg −5.1

Interpretation: The BJCP 2022 treatment effects are similar in magnitude to the Qelbree label placebo-subtracted LS-means but not identical. Both are printed here; neither is converted to SMD. The estimates are not pooled. The original Clinical Therapeutics and Journal of Clinical Psychiatry manuscripts for Studies 1–3 remain unopened.

Figure note: Qelbree label and BJCP 2022 use opposite sign conventions. Label: placebo-subtracted LS-mean (negative = improvement). BJCP: treatment effect (positive = improvement). Do not pool or average these estimates. The magnitudes are similar but not identical. Both are printed separately here.

Figure note: Adverse-event discontinuation rates from selected trials. Guanfacine XR: 16.2% (Biederman 2008), 7.4% (Sallee 2009), 19.8% (Iwanami 2020 Japan adult). Atomoxetine adult: 11.3%. Viloxazine adult: 9.0%. Viloxazine 812P304: 4.5%.

---

Part IX — What the evidence does not establish, and gaps in the opened sources

9.1 What the evidence does not establish

Non-stimulants are equivalent to stimulants. Guanfacine clinician SMD −0.67 and clonidine clinician SMD −0.71 (Cortese child) are smaller than Cortese child clinician amphetamines SMD −1.02 and methylphenidate SMD −0.78. Atomoxetine adult clinician SMD −0.45 and bupropion adult clinician SMD −0.46 (Cortese) are smaller than amphetamines adult clinician SMD −0.79 and methylphenidate adult clinician SMD −0.49. Adult extended-release methylphenidate investigator SMD −0.42 (Boesen 2022) is smaller than child methylphenidate. Viloxazine adult AISRS placebo-subtracted −3.7 is not directly comparable to SMDs from other meta-analyses.

Atomoxetine or viloxazine are free from suicidality risk. Both carry FDA boxed warnings for suicidal ideation in pediatric patients. Atomoxetine pediatric suicidal ideation 0.4% (5 of 1,357) versus 0% (0 of 851) placebo; viloxazine pediatric 0.9% (9 of 1,019) versus 0.4% (2 of 463); viloxazine adult 1.6% (3 of 189) versus 0% (0 of 183). Adult atomoxetine analysis did not find increased risk. No completed suicides.

Guanfacine or clonidine work in US adults. One Japanese adult guanfacine RCT (Iwanami 2020) shows ADHD-RS-IV difference −4.28 (95% CI −6.67 to −1.88) with effect size 0.52, but the US Intuniv label has no adult ADHD indication. The US Kapvay label has no adult ADHD trial. Cortese et al. (2018) found no adult guanfacine or clonidine trials (search closed 7 April 2017).

Bupropion is an effective ADHD treatment when rated by clinicians only. Verbeeck et al. (2017) clinician-rated-only sensitivity SMD −0.36 (95% CI −0.79 to 0.07), not significant. The pooled severity SMD −0.50 (95% CI −0.86 to −0.15) includes self-report ratings. Response rates (≥30% improvement RR 1.50; CGI-I 1 or 2 RR 1.78) are significant, but these outcomes may include self-report components.

Immediate-release methylphenidate shows significant adult participant-rated effects. Cândido et al. (2021) investigator-rated AISRS mean difference −20.70 is significant, but participant-rated SMD −0.59 (95% CI −1.25 to 0.06) is not significant.

Long-term functional outcomes for non-stimulant ADHD medications. No opened source in this review reports long-term functional outcomes (academic achievement, occupational outcomes, substance use, arrests) for guanfacine, clonidine, atomoxetine, viloxazine, bupropion, or methylphenidate beyond acute symptom reduction.

9.2 Gaps in the opened sources: located-only, no numbers cited

The following studies are located but not opened. No numbers from these sources are cited in this review:

• Michelson et al. (2001) Pediatrics 108:E83 and Michelson et al. (2002) American Journal of Psychiatry 159:1896–1901. Child atomoxetine. Bodies unopened.
• Michelson et al. (2003) Biological Psychiatry 53:112–120. Adult atomoxetine. The Strattera label identifies Studies 6 and 7 as Michelson et al. 2003. Body unopened.
• Jain et al. (2011) Journal of the American Academy of Child and Adolescent Psychiatry 50:171–179. Child clonidine XR monotherapy and adjunct. Body unopened.
• Adler et al. (2008) Journal of Clinical Psychiatry 69:1364–1373. Adult lisdexamfetamine Study 7. Numbers cited in Part II section 2.3 are from the Vyvanse label, not the manuscript. Body unopened.
• Wilens et al. (2005) Biological Psychiatry 57:793–801. Body unopened. Not cited in this review.
• Kollins et al. (2011) Kapvay Study 2 adjunct manuscript. Body unopened.
• Wilens et al. (2012) guanfacine adjunct. Body unopened.
• Castells et al. (2011) methylphenidate meta-regression. Body unopened.
• Nasser et al. (2020, 2021) Clinical Therapeutics and Journal of Clinical Psychiatry 200 mg and 400 mg pediatric viloxazine original manuscripts. Pediatric numbers cited in Parts II and VIII come from the Qelbree label and the Nasser 2022 BJCP treatment effect reprint, not the individual trial manuscripts. Bodies unopened.

These gaps may be named in a future review if the bodies are opened. For this review, numbers from opened sources listed in the fourteen headlines are cited.

---

Conclusion

The folk models—that non-stimulants are broadly equivalent first-line options to stimulants, or that only stimulants work and everything else is ineffective—both overclaim what opened sources support.

Guanfacine and clonidine show clinician-rated symptom reductions (Cortese child clinician SMD −0.67 and −0.71) with three opened child guanfacine RCTs (Biederman 2008, Sallee 2009, Newcorn 2013) and a later clonidine label printing LS-means. Guanfacine parent ratings are not significant. Adult guanfacine has one Japanese RCT (Iwanami 2020) with ADHD-RS-IV difference −4.28, but no US adult indication. Atomoxetine and viloxazine carry boxed warnings for suicidal ideation in pediatric patients. Adult methylphenidate has dedicated Cochrane reviews: Boesen 2022 extended-release investigator SMD −0.42, Cândido 2021 immediate-release investigator mean difference −20.70 but participant SMD not significant. Adult lisdexamfetamine shows labeled ADHD-RS placebo-subtracted −8.0 to −10.4 and Castells 2018 Cochrane clinician SMD −1.06, not pooled with Cortese's adult amphetamine estimate. Bupropion shows pooled severity SMD −0.50, but clinician-only sensitivity is not significant. Viloxazine has a negative sequential-testing adolescent trial (812P304) the Qelbree label does not print.

The structural problem is that each opened source has its own inclusion criteria, rater restrictions, and named-compound subgroups. Cortese et al. (2018) closed its search in April 2017; viloxazine and adult guanfacine are outside that network. Boesen 2022 and Cândido 2021 are dedicated adult methylphenidate Cochrane reviews; Castells 2018 is a dedicated adult amphetamine Cochrane review; Verbeeck 2017 is a dedicated bupropion Cochrane review. Pooled estimates from different meta-analyses are not directly comparable. Child guanfacine labels and manuscripts do not always match (Sallee 2009 2 mg LS-mean −5.41 is outside the Intuniv label Study 2 range "−6.8 to −7.9"). Boxed warnings for suicidality (atomoxetine, viloxazine) and abuse potential (lisdexamfetamine) are present. Rater variance is substantial: guanfacine parent ratings are not significant, bupropion clinician-only sensitivity is not significant, immediate-release methylphenidate participant ratings are not significant, and effect sizes depend on who rates the outcome.

The evidence supports acute symptom reduction for several non-stimulant and methylphenidate medications, with effect sizes smaller than stimulants and substantial variability by rater, population, named compound, and formulation. Long-term functional outcomes are not established. The folk models of substitutability and stimulant-exceptionalism both exceed what the opened sources show.