Atomoxetine for ADHD

Completely revised 09/2026

 

Atomoxetine is an ADHD medication that is not a stimulant. It primarily inhibits the reuptake of norepinephrine, thereby increasing levels of norepinephrine and dopamine in the prefrontal cortex (PFC). Dopamine levels are not increased in the reward system (striatum, nucleus accumbens). Therefore, atomoxetine increases motivation to a lesser extent than stimulants and has no potential for abuse.
The effects build up gradually. The first changes are often noticed after 1 to 4 weeks, and the full effect is frequently not achieved until 6 to 12 weeks or later. When taken regularly, the effects last throughout the day, including in the evening and in the morning.

Atomoxetine reduces ADHD symptoms in children, adolescents, and adults significantly more than a placebo. On average, it is slightly less effective than stimulants (methylphenidate, amphetamine-based medications). Some of those for whom methylphenidate (MPH) is ineffective also respond to atomoxetine. Atomoxetine is primarily considered when stimulants are not sufficiently effective or are not tolerated, as well as in cases of comorbid anxiety disorders, tics, or substance use disorders.
The dose is determined based on body weight and increased gradually. A gradual increase and dividing the dose into two daily administrations help reduce side effects.
Common side effects include loss of appetite, nausea, fatigue, or sleep problems; in men, erectile and ejaculatory dysfunction, as well as difficulty urinating. Heart rate and blood pressure rise slightly. Large studies have found no increased risk of heart attack or stroke. Liver damage, suicidal thoughts in children and adolescents, and aggressive behavior are rare.

Atomoxetine is metabolized in the liver primarily by the enzyme CYP2D6. In about 5 to 7% of people, this enzyme is genetically programmed to function only weakly. In these individuals, significantly higher blood levels result, often leading to greater efficacy but also more side effects. Some medications, such as fluoxetine, paroxetine, or bupropion, inhibit CYP2D6 and also increase atomoxetine levels.

The effect does not diminish with long-term use. Studies have not found any withdrawal symptoms. However, many people with ADHD stop taking the medication prematurely, usually due to side effects or insufficient efficacy.

Previous names for atomoxetine were: Tomoxetine, LY139603

(E 4): Brand names: Strattera®; Agakalin®; several generic versions (Drug Directory, E 4)¹; in South Africa, also NIR®; Attencit®; Attentra®; Atteze®; Atastrat®; Concitron®; Strattex® (Narrative Review, E 4)²

 

1. Mechanism of action¶

The effects of atomoxetine take time to develop, similar to those of antidepressants. They begin after 1 to 2 weeks, and it can take up to 6 months to reach full effectiveness. (Expert consensus, E 4)³ According to a systematic review, the effect begins within 4 weeks (and possibly as early as 1 week in late responders). It takes at least 12 weeks to reach full effect. (Systematic review, E 1a)⁴

 

1.1. Elevated levels of norepinephrine and dopamine in the PFC¶

In the laboratory, atomoxetine acts primarily as a norepinephrine (transporter) reuptake inhibitor. In vivo, it also increases dopamine levels in the PFC, since the norepinephrine transporter there also reabsorbs dopamine. (Textbook, E 4)⁵

 

Contrary to other assumptions, atomoxetine therefore affects not only norepinephrine levels but also dopamine levels in the PFC. (Textbook, E 4)¹³

Atomoxetine improved attention, memory, and other executive functions in young rats and aged rhesus monkeys, which is consistent with its effects in the PFC. (Animal study, E 2b)¹⁴ A single dose reduced activation of the vmPFC in 14 healthy adults upon reward receipt. (Experimental study, E 2b)¹⁵
The increase in norepinephrine in the PFC induced by atomoxetine reached a plateau in rats and is therefore more difficult to control in a gradual manner than with other medications. (Narrative review, E 4)⁸ A measurement of norepinephrine reuptake inhibition based on the reduction in systolic blood pressure following intravenous injections of tyramine showed that with ATX (25 to 80 mg/day) a dose-dependent inhibition of norepinephrine reuptake starting at the lowest dose (25 mg/day), whereas with venlafaxine, this effect was not observed until 225 mg/day. (Clinical trial for depression, E 2b)¹⁶ In contrast, the increase in norepinephrine caused by psychostimulants such as D-amphetamine rose in a dose-dependent manner without reaching a plateau. (Narrative review, E 4)⁸

1.2. No Dopaminergic Effect of Atomoxetine in the Striatum / Nucleus Accumbens¶

Therefore, no improvements in hyperactivity/impulsivity or motivation (drive) are to be expected through this mechanism of action. Atomoxetine apparently acts through other mechanisms with regard to hyperactivity and impulsivity.

1.3. Serotonin and Atomoxetine¶

A case report described acute agitation and suicidal thoughts in an 11-year-old boy following the initiation of atomoxetine treatment, clinically similar to the reactions to SSRIs. Whether this indicates a serotonergic effect remains unclear. (Case report, E 4)¹⁹

1.4. NMDA-glutamate receptor antagonist¶

In vitro, atomoxetine blocked NMDA glutamate receptors on cultured rodent neurons in a voltage-dependent manner as a channel blocker (half-maximal inhibition at approximately 3 µM), which, according to the authors, falls within the range of clinically relevant plasma concentrations. The inhibition was noncompetitive, and the binding site appeared to overlap with the Mg²⁺ site in the channel pore. (In vitro study, E 2b)²⁰

 

1.5. Atomoxetine works differently than MPH¶

The nucleus accumbens is part of the striatum, the brain’s reward/reinforcement system, which is involved in ADHD. Atomoxetine may therefore be beneficial for people with ADHD and acute addiction problems. In cases of impairment of the reward system, methylphenidate and, in some cases, nicotine (patches) are likely to be more effective.

1.6. Specific Mechanism of Action on Hyperactivity/Impulsivity¶

In 24 healthy men, a single dose of atomoxetine (60 mg) did not improve reaction time variability or response inhibition (stop-signal reaction time) any more than citalopram did, unlike MPH (30 mg). (Double-blind crossover study, single dose, healthy subjects, E 1b)²⁹

After 6 weeks of treatment, atomoxetine moderately improved response inhibition in the Go/NoGo test in children and adolescents compared with placebo (d = 0.42), primarily due to fewer commission errors; sustained attention showed only a trend toward improvement (d = 0.17; p = 0.06). The cognitive changes did not correlate with symptom improvement (crossover RCT, double-blind, 2 × 6 weeks, N = 136, 109 completers, E 1b.)³⁰

1.7. Overview of ATX and Neurotransmitters¶

1.7.1. Binding Affinity of ATX, AMP, and MPH to DAT, NET, and SERT¶

1.7.2. Effects of ATX, AMP, and MPH on Dopamine and Norepinephrine by Brain Region¶

2. Quality of Impact¶

2.1. Effect on Symptoms¶

In a randomized, double-blind study ( ), lisdexamfetamine and atomoxetine were compared in children and adolescents who had an inadequate response to MPH. Lisdexamfetamine reduced functional impairment (as assessed by parents) in 2 of 6 domains (learning and school, social activities) and in the total score (ES 0.27) significantly more than atomoxetine. (RCT, N = 267 randomized, n = 200 completed the study, 9 weeks, E 1b)³²

In children and adolescents, atomoxetine reduced the ADHD-RS-IV total score by 3.8 points after adjustment for placebo, while guanfacine reduced it by 8.9 points. (Drug information on RCT, n = 337, E 1b)³³

In an uncontrolled observational study of 41 adults, excessive mind wandering decreased significantly after 3 months of medication. Only 4 participants received atomoxetine; the rest received methylphenidate. (Observational study, n = 4, E 2b)³⁴

In a network meta-analysis of children and adolescents, atomoxetine was superior to placebo based on clinician assessment (after approximately 12 weeks) (SMD -0.56; 95% CI -0.66 to -0.45); however, in this age group, it had the lowest mean effect size of all active ingredients and was significantly inferior to amphetamine-based medications (SMD -1.02) and MPH (SMD −0.78). Based on teacher assessment, only methylphenidate and modafinil were superior to placebo. In children, atomoxetine reduced weight (SMD −0.84). (Network meta-analysis, k = 133 RCTs, N = 24,642, of which n = 18,199 were included in the efficacy analysis, E 1a.)³⁵

In adults, the effect size for atomoxetine was an SMD of -0.45 (-0.58 to -0.32), which was comparable to that of methylphenidate (-0.49); amphetamine medications reached -0.79. Atomoxetine was discontinued due to side effects more frequently than placebo in adults, but not in children (OR 2.33; 1.28 to 4.25). (Network meta-analysis, k = 133 RCTs, N = 24,642, of which n = 18,199 were included in the efficacy analysis, E 1a.)³⁵

• g = -0.48 (-0.64 to -0.33) in adults, with poorer tolerability than placebo (discontinuations due to side effects: 9.3% vs. 3.6%) (Meta-analysis, k = 18, including 16 RCTs on atomoxetine with n = 4,103, N = 4,308, E 1a.)³⁶
• SMD -0.38 (-0.56 to -0.21) for adults in self-assessment, SMD -0.51 (-0.64 to -0.37) based on clinician assessment, similar to stimulants (-0.39 and -0.61, respectively). Atomoxetine was discontinued more frequently than placebo (OR 1.43; 1.14 to 1.80), and significantly more often due to side effects. (OR 2.43; 1.89 to 3.12) (network meta-analysis, k = 113 RCTs, N = 14,887, E 1a.)³⁷
• SMD -0.33 (-0.43 to -0.23) for self-assessment in adults, SMD -0.40 (-0.48 to -0.32) based on clinician assessment, with more treatment discontinuations (nearly 50% on atomoxetine, OR 1.39; 1.17 to 1.64) and discontinuations due to side effects (13% vs. 5%, OR 2.57; 1.78 to 3.71), and a benefit-risk ratio rated as unfavorable. (Meta-analysis, k = 12 RCTs, N = 3,375, E 1a.)³⁸

In children and adolescents, atomoxetine reduced ADHD symptoms compared with placebo (ADHD-RS-IV SMD -0.64; -0.76 to -0.52). The number needed to treat was 3.43 (2.79 to 4.45) for a response and 10.30 (5.89 to 40.62) for relapse prevention. Higher baseline symptom severity was associated with greater symptom reduction, while male gender, comorbid oppositional defiant disorder (ODD), and the hyperactive-impulsive subtype were associated with less symptom reduction. In cases of comorbid ODD, both ADHD and ODD symptoms decreased (SMD −0.42; −0.70 to −0.14). Comorbid generalized anxiety disorder or depression did not alter efficacy or side effects (meta-analysis, k = 9 RCTs, N = 1,828, n = 1,150, E 1a.)³⁹
In a direct comparison, MPH and atomoxetine did not differ overall in children and adolescents (SMD 0.09; -0.08 to 0.26). OROS-MPH was superior to atomoxetine (SMD 0.32; 0.12 to 0.53), whereas immediate release MPH was not. Only parental ratings were analyzed. Excluding the open-label studies, the difference was an SMD of 0.08 with no heterogeneity (meta-analysis, k = 9 RCTs, 7 of which were manufacturer-funded, N = 2,762, E 1a.)³⁷
With prolonged administration (at least 3 weeks), atomoxetine improved reaction time, attention, and inhibition (g = 0.36 to 0.64), but not working memory. The effect on executive functions did not differ from that of MPH (meta-analysis, k = 7 for atomoxetine, n = 829, E 1a.)⁴⁰
Atomoxetine improved quality of life (g = 0.30; 0.19 to 0.40), regardless of treatment duration, in both children and adults (amphetamine medications g = 0.51, MPH g = 0.38) (meta-analysis, k = 17 RCTs, including 11 on atomoxetine with n = 3,344, N = 5,388, E 1a.)⁴¹

 In children and adolescents, atomoxetine outperformed placebo in terms of overall ADHD symptoms (ES -0.64; -0.56 to -0.71), hyperactivity/impulsivity (ES -0.67), and inattention (ES -0.59). 44.4% versus 21.4% showed at least a 40% improvement (NNT 4), while approximately 40% remained significantly symptomatic. Quality-of-life-related endpoints improved to a slightly lesser extent (ES -0.48 to -0.25). Discontinuations due to adverse effects were more frequent (RR 1.89; 1.08 to 3.31; NNH 50), while discontinuations due to lack of efficacy were less common (RR 0.51; 0.36 to 0.74). (Meta-analysis, k = 25 RCTs, N = 3,928, E 1a.)⁴²

According to parental reports, atomoxetine significantly reduced core ADHD symptoms in children and adolescents in a pre-post comparison (ES 1.37; 1.24 to 1.51) and outperformed control conditions overall, though not other ADHD medications. In the placebo-controlled studies, the pre-post effect size was 1.18 (1.07 to 1.30) for atomoxetine and 0.53 (0.43 to 0.63) for placebo. (Meta-analysis, k = 42, of which 9 had no control group, N = 8,398, E 1a.)⁴³

In children and adolescents, atomoxetine had a moderate effect compared with placebo (SMD -0.68; -0.76 to -0.59), similar to MPH (SMD -0.75; –0.98 to –0.52) and weaker than lisdexamfetamine (SMD –1.28; –1.84 to –0.71). The discontinuation rate did not differ from that of placebo (OR 0.91; 0.66 to 1.24). Unlike with the other active ingredients, the atomoxetine studies were not significantly heterogeneous among themselves. (METASTUDY, k = 28 RCTs, 2 weeks to 3 months, N = 4,699, E 1a.)⁴⁴

In children and adolescents aged 6 to 18, effect sizes ranging from 0.6 to 1.3 were observed compared to placebo after 6 to 18 weeks, along with improvements in functional status and quality of life. The effect was similar in strength to that of MPH. (Review, k = 125, manufacturer-funded, E 4.)⁴⁵

Atomoxetine was effective in children and adolescents in Asia, Europe, North America, and Russia (overall effect size 0.62), with moderate heterogeneity (I² = 54.3%) and a smaller effect in Asia (ES 0.42 compared with 0.61 to 0.70 in the other regions). A 40% reduction in symptoms was achieved with atomoxetine or placebo. (Meta-analysis, k = 15 RCTs, N = 2,569, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)⁴⁶

• in Asia, 39.6% vs. 24.0%
• in Europe, 40.2% vs. 12.1%
• in North America, 45.3% vs. 21.7%
• in Russia, 54.2% vs. 33.3%

Among Taiwanese children and adolescents aged 6 to 16, once-daily atomoxetine reduced ADHD symptoms more effectively than placebo, as assessed by investigators, parents, and teachers (effect size 0.70). Common side effects included decreased appetite (36.1%) and nausea (16.6% vs. 0%). (RCT, 6 weeks, N = 106, n = 72, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)⁴⁷

In 5- and 6-year-olds, atomoxetine reduced ADHD symptoms more than placebo, as assessed by parents and teachers (SMD 0.7 and 0.6, respectively). A clinical response according to the CGI-I was achieved by 40% compared with 22% in the placebo group (not significant); 62% remained at least moderately impaired at the end of the study.⁴⁸

Parents’ assessments (study involving 6- to 16-year-olds) and teachers’ assessments (study involving 8- to 12-year-olds) showed the effect of atomoxetine with similar sensitivity (effect size: parents 0.69, teachers 0.63). (Pooled analysis of two RCTs, up to 7 weeks, n = 318, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)⁴⁹

In an individual-level data analysis of all nine placebo-controlled atomoxetine trials in adults, the effect size was 0.41 after 10 weeks and 0.31 after six months. (Pooled individual-patient data analysis, k = 9, N = 3,374, 1,961 short-term and 1,413 over six months, manufacturer-funded studies, author(s) employed by the manufacturer, E 1a.)⁵⁰

In the two approval studies for adults, atomoxetine reduced inattention, hyperactivity, and impulsivity to a greater extent than placebo (effect sizes 0.35 and 0.40). Fewer than 10% discontinued treatment due to side effects (7.8% and 9.3%; placebo: 4.3% and 2.4%). (2 identical RCTs, 10 weeks each, N = 280 and N = 256, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)⁵¹

In adults, atomoxetine was more effective than placebo: (meta-analysis, k = 13 RCTs, N = 1,824, E 1a.)⁵²

• the overall ADHD symptom profile (SMD -0.45; -0.54 to -0.35)
• Inattention (SMD -0.42; -0.49 to -0.35)
• Hyperactivity/Impulsivity (SMD -0.36; -0.44 to -0.29)

In adults from Japan, Korea, and Taiwan, atomoxetine reduced ADHD symptoms more effectively than placebo (CAARS-Inv:SV -14.3 vs. -8.8, ES 0.55) and improved quality of life and executive functions. Due to side effects, 5.2% discontinued treatment (placebo: 1.5%) (RCT, 10 weeks, N = 391, n = 195, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)⁵³

In adults, the effect of atomoxetine (SMD 0.46; 0.36 to 0.56) did not differ significantly from that of OROS methylphenidate (SMD 0.51; 0.40 to 0.63) in studies lasting up to 12 weeks (difference −0.05; −0.18 to 0.08), nor did the discontinuation rate (OR 0.85; 0.53 to 1.35). (Network meta-analysis, k = 15 RCTs, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)⁵⁴

After 6 weeks, children and adolescents responded to atomoxetine (53.6%) at a similar rate (at least a 40% reduction in symptoms) as they did to methylphenidate (54.4%). Atomoxetine was therefore not inferior, given a non-inferiority margin of -15% (difference -0.9%; -9.2% to 7.5%) (meta-analysis with individual-level data, k = 7 RCTs, including 2 open-label studies, N = 1,368, n = 811, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)⁵⁵

Overall, there was no significant difference in the efficacy of atomoxetine and methylphenidate in children and adolescents (SMD 0.09; -0.06 to 0.25). OROS methylphenidate was superior to atomoxetine (SMD 0.31; 0.16 to 0.47), whereas immediate release methylphenidate was not (SMD -0.05; -0.20 to 0.10) (meta-analysis, k = 11, N = 2,772, E 1a.)⁵⁶

In direct comparative studies involving children and adolescents, more people with ADHD responded to methylphenidate than to atomoxetine (RR 1.14; 1.09 to 1.20), and inattention improved slightly more with methylphenidate (SMD -0.13; -0.25 to -0.01). Atomoxetine caused drowsiness more frequently (RR for MPH 0.17; 0.11 to 0.26), but caused insomnia less frequently (RR for MPH 2.27; 1.63 to 3.15) (meta-analysis, k = 11 RCTs, 9 of which were open-label, E 1a.)⁵⁷

In 6- to 16-year-olds from China, Korea, and Mexico, atomoxetine was not inferior to immediate release methylphenidate (response rate 77.4% vs. 81.5%), but caused drowsiness more frequently (26.2% vs. 3.6%) and slightly more weight loss (-1.2 kg vs. -0.4 kg) (RCT, double-blind, approximately 8 weeks, N = 330, n = 164, E 1b.)⁵⁸

Over 24 weeks, atomoxetine and OROS methylphenidate were comparable in previously untreated 7- to 16-year-olds, for example, in terms of inattention (Cohen’s d: -3.05 vs. -3.08). Only after 2 weeks was OROS methylphenidate superior in terms of inattention (d = 0.46). Vomiting occurred more frequently with atomoxetine (18.8% vs. 6.3%), drowsiness (33.8% vs. 1.3%), and dizziness (11.3% vs. 1.3%); with methylphenidate, insomnia was more common (open-label RCT, 24 weeks, N = 160, n = 80, E 1b.)⁵⁹

Under real-world conditions, previously untreated Chinese children responded less frequently to atomoxetine than to OROS methylphenidate (63.3% vs. 84.6%) and experienced side effects more frequently (56.8% vs. 47.8%) (Prospective cohort study, 26 weeks, N = 1,021, n = 488, Phase 2b.)⁶⁰

In the Continuous Performance Test, OROS methylphenidate improved reaction time, reaction time variability, and omission errors to a greater extent than atomoxetine, under which these measures did not change significantly. The changes in test performance did not correlate with symptom improvement (crossover RCT, double-blind, 4 to 6 weeks each, N = 102, E 1b.)⁶¹

Atomoxetine and OROS methylphenidate improved response selection/inhibition, flexibility, and planning/working memory in children. Methylphenidate improved response selection and inhibition to a greater extent, as evidenced by detection performance in the CPT (slope 0.15 vs. 0.06) and rapid visual information processing (3.45 vs. 2.22). (Full text not available to us; open-label RCT; 3 months; n = 78 atomoxetine, n = 79 OROS-MPH; E 1b.)⁶²

In adults, atomoxetine improved ADHD-specific quality of life (AAQoL) compared with placebo after 10 and 24 weeks, as well as self-reported executive functions (BRIEF-A) at 10 weeks, but not general impairment (SDS) or general health (SF-36). The improvements occurred gradually and correlated only moderately with symptom reduction (r = −0.58 to −0.39), with similar correlations observed in the placebo group (r = −0.56 to −0.28). The reduction in symptoms showed barely any correlation with the non-ADHD-specific measures (SDS, SF-36). (Pooled analysis, k = 7 RCTs, 10 weeks: n = 1,085 atomoxetine, n = 1,076 placebo; 24 weeks (2 RCTs): n = 399 atomoxetine, n = 325 placebo, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)⁶³

In young adults (ages 18 to 30), atomoxetine reduced ADHD symptoms more effectively than placebo (CAARS-Inv:SV -13.6 vs. -9.3; difference 95% CI -6.35 to -2.37, Effect size 0.4) and improved quality of life (AAQoL) and self-reported executive functions (BRIEF-A). No significant effect was observed on driving behavior, social adjustment, substance use, depression, or anxiety. Due to side effects, 9.5% vs. 2.7% discontinued treatment. (RCT, 12 weeks, N = 445, n = 220, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)⁶⁴

In adults, atomoxetine (up to 80 mg) reduced the observer-rated total ADHD score (CAARS-O:L) by 13.1 points, compared with 0.4 points in the waitlist group (ES 1.8). 60.1% responded (at least a 30% reduction) compared with 0% in the waitlist group. Emotional symptoms, self-esteem, and quality of life also improved. 18.5% discontinued treatment due to side effects. (RCT with waitlist control, 12 weeks, N = 64, n = 27, per-protocol analysis of 43 study completers, manufacturer-funded, E 1b.)⁶⁵

In adults, atomoxetine (71.6 mg/day on average) reduced driving errors in three out of four categories (attention, risk-related self-control, driving skills) during a standardized driving test in real-world traffic. In a traffic-related neuropsychological test battery (ART2020), 18.2% of the atomoxetine group failed to meet the criteria for driving fitness according to German guidelines, compared to 47.6% of the waiting-list group. Self-reported critical traffic situations decreased from 12.0 to 6.8 per week. (RCT with waitlist control, 12 weeks, N = 64, of whom 43 completed the study, n = 22, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)⁶⁶

In treatment-naive children and adolescents (ages 7 to 16), atomoxetine improved school functioning (pre-post effect size based on parental assessment: d = -0.62, self-report d = -0.34) and relationships with peers (parent report d = -0.33, self-report d = -0.65) without a significant difference compared to OROS methylphenidate. Behavior at home did not improve significantly with atomoxetine. (Randomized open-label comparative trial, 24 weeks, N = 168, n = 85, E 1b.)⁶⁷

Among Indian children (ages 6 to 14), 86.2% responded to atomoxetine (mean dose 0.61 mg/kg/day) and 90.7% responded to methylphenidate (0.45 mg/kg/day) after 8 weeks (Response: symptom reduction > 25%), with no significant difference. Heart rate increased more markedly with atomoxetine (7 beats/min, p = 0.021), while weight loss was greater with methylphenidate. (Open-label RCT, 8 weeks, N = 69, E 1b.)⁶⁸

Latino and Caucasian children and adolescents (ages 6 to 18) responded similarly to atomoxetine during the first 10 to 11 weeks (ADHDRS-IV-P:I -54% and -52%, respectively), with similar tolerability. Slow CYP2D6 metabolizers were more common among Caucasians. (Pooled analysis of two open-label, non-placebo-controlled studies, N = 1,198, n = 108 Latinos, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)⁶⁹

2.2. Responding¶

In patient forums, some atomoxetine users reported that atomoxetine produced excellent effects only during the first few days—up to a maximum of about 14 days—after which those effects had significantly diminished or disappeared entirely. This pattern could recur following dosage increases, even multiple times. (ADxS experience)⁷⁰ We believe we are recognizing a pattern here that occurs more frequently with medications for ADHD that primarily have a tonic noradrenergic effect, and we suspect that this suggests that, in ADHD in general and in people with ADHD in particular, the phasic norepinephrine level is more likely to be impaired than the tonic level.
A “phasic” change refers to a short-term (stress- or exertion-induced) change in the level of a neurotransmitter or hormone.
The tonic component is the steady-state level and its typical circadian fluctuations throughout the day.
Phasic is to tonic as waves are to swell.

For information on the effectiveness of individual medications and treatment options, see ? Effect sizes of various ADHD treatment methods.

3. Indications for which atomoxetine is appropriate / inappropriate¶

 

A review found that atomoxetine, when used to treat ADHD with comorbid disorders, had effect sizes ranging from 0.47 to 2.21 on ADHD symptoms. The most marked improvement in comorbid symptoms was observed in anxiety disorders (effect size 0.40 to 1.51) and oppositional defiant disorder (effect size 0.52 to 1.10). For substance use disorders, autism, dyslexia, depression, bipolar disorder, and Tourette syndrome, the findings were mixed or limited (review, k = 24, including 14 RCTs, n = 1,348 atomoxetine, n = 832 placebo, E 4.)⁷¹ Effect sizes are predominantly from pre-post comparisons (including non-randomized studies) and are therefore not comparable with placebo-controlled values.

3.1. Non-responsive to stimulants¶

According to prevailing medical opinion, atomoxetine is recommended when neither methylphenidate nor amphetamines are effective, which, according to a review of crossover studies, is the case in about 8 to 13% of patients. (Narrative review, E 4)⁷² A reference book reports that methylphenidate, like amphetamine-based medications, is ineffective on its own in about one-third (reference book, E 4)⁷³ of people with ADHD. We recommend trying guanfacine first, especially in younger children and in people with ADHD and high blood pressure, since guanfacine showed a higher mean effect size than atomoxetine in a direct comparison, albeit with a higher rate of side effects.
Atomoxetine helped some of the MPH nonresponders. (Systematic review, E 1a)⁷⁴ According to a single study, approximately 50% of MPH non-responders and approximately 75% of MPH responders are expected to respond to atomoxetine. (Narrative review, E 4)⁷⁵

Among children and adolescents who did not respond adequately to methylphenidate, 63.6% responded to atomoxetine (mean optimal dose 40.2 mg/day) after 9 weeks, compared with 81.7% who responded to lisdexamfetamine. The median time to first response was 21 days compared with 12 days. Lisdexamfetamine reduced ADHD symptoms more significantly (ADHD-RS-IV difference -6.5; -9.3 to -3.6; ES 0.56) (RCT, 9 weeks, N = 267, n = 134, funded by the LDX manufacturer, E 1b.)⁷⁶ Atomoxetine may take longer than 9 weeks to reach its full effect.

3.2. Emotional Dysregulation¶

According to a reference book, only the disinhibition of executive functions caused by basal ganglia dysfunction responds to methylphenidate, but not emotional disinhibition controlled by the hippocampus. (Textbook, E 4)⁷⁷ Therefore, in our view, the former is likely to respond better to dopaminergic treatment, while emotion regulation and affect control are likely to be better treated with noradrenergic approaches.
However, because atomoxetine lacks dopaminergic effects in the striatum, it has a lesser stimulating effect than stimulants; this is why combination therapy is often the key to successful treatment of ADHD.
(ADxS experience): While, in our experience, atomoxetine (like guanfacine) treats emotional dysregulation significantly better (and, above all, throughout the entire day) than stimulants, studies show that stimulants are more effective—but only while they are active. The difference likely lies in the all-day effect of non-stimulants. Especially in the evening, when personal social interactions take place, the active period of stimulants has usually already ended.

 

• In two multicenter, placebo-controlled studies, 32% of adults with ADHD exhibited emotional dysregulation (irritability, affective lability, emotional hyperreactivity). These individuals responded more strongly to atomoxetine (p = 0.048). Among people with ADHD, the treatment effect on emotional dysregulation (d = 0.66; p < 0.001) was at least as great as that on core ADHD symptoms (CAARS d = 0.58) (Post hoc analysis of two RCTs, 10 weeks, N = 536, n = 170 with emotional dysregulation, manufacturer-funded, E 1b.)⁷⁸
• Atomoxetine slightly improved emotional control in adults (effect size 0.19; p = 0.013) and had a stronger effect in those with severe impairment (0.32) (pooled individual data analysis of three studies, N = 2,846, n = 829 placebo-controlled, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)⁷⁹
• Stimulants and atomoxetine combined moderately reduced emotional lability in adults (SMD -0.41; -0.57 to -0.25), which was half as effective as the reduction in core ADHD symptoms (SMD -0.8). Atomoxetine alone had a small effect (SMD -0.21; -0.34 to -0.08; k = 3), while stimulants had a stronger effect (SMD −0.57; k = 6) (meta-analysis, k = 9 RCTs, N [not specified in the full text], E 1a.)⁸⁰
• A meta-analysis with separate evaluations for each active ingredient found the following effects on emotional dysregulation in adults: (meta-analysis, k = 21 RCTs, E 1a.)⁸¹
  • Atomoxetine (k = 5) SMD 0.24 (0.15 to 0.34)
  • Methylphenidate (k = 13) SMD 0.34 (0.23 to 0.45)
  • Lisdexamfetamine (k = 3) SMD 0.34 (0.12 to 0.55); after excluding the only crossover study, SMD 0.50 (0.21 to 0.80)
• A component network meta-analysis found an SMD of -0.36 (-0.67 to -0.04) for atomoxetine in adults regarding emotional dysregulation, similar to that found for stimulants (SMD -0.38) (network meta-analysis, subanalysis k = 14, N = 2,296, E 1a.)³⁷
• In children with ADHD, atomoxetine (d = -0.93) and methylphenidate (d = -1.46) reduced emotional lability, with no significant difference between the two. The effect persisted after controlling for improvement in ADHD symptoms and was associated with increased functional connectivity between the amygdala and frontal and cingulate regions (full text not available to us, RCT, titration plus 4 weeks of stable dose, N = 152, n = 76, E 1b.)⁸²

3.3. Comorbid Anxiety Disorder¶

Positive effects of atomoxetine on comorbid anxiety disorders have been reported (Expert Consensus, E 4)⁸³; one study in children found a greater reduction in anxiety symptoms with atomoxetine than with MPH starting in the 4th week of treatment. (Comparative study, E 2b)⁸⁴

Atomoxetine reduced separation anxiety or social phobia in children and adolescents (ages 8 to 17) with ADHD and comorbid generalized anxiety disorder, separation anxiety, or social phobia over a 12-week period (ADHDRS-IV-PI -10.5 vs. -1.4, d = 1.0, p < 0.001, response rate 61.8% vs. 12.1%) and anxiety symptoms (PARS -5.5 vs. -3.2, d = 0.5, p = 0.011) over 12 weeks. Only participants who did not show improvement in anxiety during the placebo run-in phase were analyzed (n = 55 vs. 58) (RCT, N = 176, n = 87, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)⁸⁵

Atomoxetine reduced anxiety symptoms significantly more in children and adolescents with ADHD and comorbid anxiety disorder (d = 0.51) than in those with ADHD without an anxiety disorder (d = 0.05). The effect on ADHD symptoms did not depend on the presence of an anxiety disorder (crossover RCT, double-blind, 2 × 6 weeks, N = 136, n = 43 with anxiety disorder, E 1b.)³⁰

3.4. Comorbid Social Anxiety Disorder¶

Positive effects of atomoxetine on comorbid social anxiety have been reported. (Expert consensus, E 4)⁸³

In adults with ADHD and social anxiety disorder, atomoxetine (40 to 100 mg) reduced both ADHD symptoms (CAARS-Inv:SV -8.7 vs. -5.6, Effect size 0.47) and social anxiety (LSAS -22.9 vs. -14.4, effect size 0.40) more effectively than placebo (p < 0.001 for both) at every measurement time point. Trait anxiety (effect size 0.27) and quality of life (effect size 0.24) improved, while social adjustment did not improve significantly (RCT, 14 weeks, N = 442, n = 224, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)⁸⁶

3.5. SCT (Sluggish Cognitive Tempo)¶

In a study of children and adolescents (ages 10 to 16) with ADHD and/or dyslexia, atomoxetine significantly reduced SCT symptoms (K-SCT) in 7 out of 9 evaluations (3 raters × 3 diagnostic groups). The improvement in SCT persisted after adjusting for changes in ADHD symptoms and was thus largely independent of ADHD symptoms. (Secondary analysis of an RCT, E 1b)⁸⁷

Higher scores on the SCT “Sluggish/Sleepy” factor (but not “Daydreamy”) were more frequently associated with MPH nonresponse in stimulant-naive children (ages 7 to 11). There was no difference in the MPH response rate between ADHD-I and ADHD-C (combined type). (RCT, E 1b)⁸⁸

3.6. Co-occurring Depression¶

Whether atomoxetine is effective in treating depression is a matter of debate.

• No (Systematic Review, E 1a)⁸⁹(Expertenkonsens, E 4)⁹⁰(RCT, E 1b)⁹¹
• Yes (Case Report, E 4)⁹²
  • In a small open-label study without a placebo arm (n = 9 receiving atomoxetine up to 80 mg/day), symptoms of depression improved to a degree similar to that observed with paroxetine and venlafaxine. (Clinical trial for depression, E 2b)⁹³
  • Among adults with depression that did not respond adequately to sertraline, those with the 5-HTTLPR s/s genotype achieved remission more frequently with the addition of atomoxetine than those with other genotypes (81.8% versus 32.7%), whereas this was not the case with sertraline plus placebo (35.7% vs. 37.7%). (Secondary analysis of an RCT, E 1b)⁹⁴
  • In children and adolescents with ADHD and symptoms of depression or anxiety, atomoxetine alone and atomoxetine plus fluoxetine significantly reduced ADHD, depression, and anxiety symptoms. The combination had only a slight additional effect on depression but increased blood pressure and heart rate more significantly (RCT, 8 weeks, atomoxetine during the last 5 weeks, N = 173, n = 127 for the combination, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)⁹⁵
  • In adults with ADHD and depression that responded only partially to SSRIs, atomoxetine and OROS methylphenidate, when added to SSRIs, improved depressive symptoms, ADHD symptoms, and functional impairment equally. With atomoxetine, the HAMD score decreased from 22.5 to 15.6, and the HAMD response rate was 32.1% (methylphenidate: 28.1%) (randomized, rater-blinded, non-placebo-controlled trial, 12 weeks, N = 60, n = 28, E 1b.)⁹⁶

When taking SSRIs concurrently, the potential interaction via CYP2D6 should be taken into account (see below).

3.7. Co-occurring Addiction: The Merits of ATX Are Debated¶

While one school of thought favors atomoxetine for patients with comorbid substance use disorder due to its lower risk of abuse, another view sees advantages in stimulants because of their faster onset of action and greater effect size. According to the updated 2018 European consensus on the diagnosis and treatment of ADHD in adults, it is now well established that stimulants do not increase the risk of addiction in ADHD but rather significantly reduce it while being taken. (Expert Consensus, E 4)⁹⁰

Atomoxetine improved ADHD symptoms in adults with ADHD and alcohol misuse or dependence who had recently achieved abstinence (effect size 0.48; p = 0.007). There was no difference in the time to relapse into heavy drinking (p = 0.93). The cumulative number of days of heavy drinking was reduced by 26% (event rate 0.74; p = 0.023). Only 32 of 72 patients on atomoxetine and 48 of 75 on placebo completed the study. (RCT, 12 weeks, N = 147, n = 72, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)⁹⁷

Among adolescents (ages 13 to 19) with ADHD and substance use disorder who also received motivational interviewing and CBT, atomoxetine did not differ from placebo in terms of ADHD symptoms (p = 0.30) nor in substance use (p = 0.11), with a high placebo response. (RCT, 12 weeks, N = 70, n = 35, E 1b.)⁹⁸

Atomoxetine showed no potential for abuse in receptor-binding, animal, and human laboratory studies (no relevant affinity for dopamine transporters, GABA-A, or µ-opioid receptors; no self-administration in monkeys). In 46 drug users with prior experience with stimulants, single doses of up to 180 mg produced neither euphoria nor stimulant-like effects. (Narrative review; author(s) employed by the manufacturer; E 4.)⁹⁹

Among adults with ADHD and alcohol use disorder who had recently been abstinent, improvement in ADHD symptoms while taking atomoxetine correlated with reduced alcohol cravings (r = 0.29; 0.04 to 0.51). Relapses into harmful drinking were associated with a worsening of 15 of the 18 ADHD symptoms in the placebo group, but not in the atomoxetine group. Only 44% of the atomoxetine group completed the study. (Post hoc analysis of an RCT, 12 weeks, N = 147, n = 72, E 1b.)¹⁰⁰

3.8. Comorbid Tics¶

Atomoxetine did not worsen tics in children and adolescents (ages 7 to 17) with ADHD and Tourette syndrome or chronic motor tic disorder. Tic severity tended to decrease (YGTSS -5.5 vs. -3.0; p = 0.063) and decreased significantly according to the CGI-Tic scale (-0.7 vs. -0.1; p = 0.002). ADHD symptoms improved (ADHD-RS -10.9 vs. -4.9; p < 0.001), with an effect size of 0.6. Increased heart rate, weight loss, decreased appetite, and nausea were more common than with placebo. (RCT, up to 18 weeks, N = 148, n = 76, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹⁰¹

3.9. Comorbid Disorder with Oppositional Defiant Disorder (ODD)¶

Atomoxetine reduced ADHD symptoms to the same extent in children and adolescents (ages 6 to 16) with and without comorbid ODD (ES 0.85 and 0.65, respectively). Atomoxetine was not significantly more effective than placebo in reducing ODD symptoms (CPRS-R:S) in either group. The reduction in ODD symptoms was closely correlated with the degree of improvement in ADHD symptoms (r = 0.78). (Post-hoc meta-analysis, k = 3 RCTs, 6 to 8 weeks, N = 512, n = 158 with ODD, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)¹⁰²

An additional afternoon dose improved parent-rated ODD symptoms at home in children (ages 6 to 12) (p < 0.05), but did not improve ADHD symptoms or functioning at school. Twice-daily dosing was associated with less frequent abdominal pain but a more persistent decrease in appetite. (Secondary analysis of an open-label study, 8 weeks, N = 55, n = 22 on twice-daily dosing, non-randomized switch, manufacturer-funded, Phase 2b.)¹⁰³

Symptoms of oppositional defiant behavior improved less markedly with atomoxetine than ADHD symptoms. (ES -0.33) (Meta-analysis, k = 25 RCTs, N = 3,928, E 1a.)⁴²

In children aged 6 to 12 with ADHD and ODD, atomoxetine (1.2 mg/kg/day) reduced ADHD symptoms and improved the overall clinical assessment. For ODD symptoms, a benefit was observed only in the repeated-measure analysis (weeks 2 and 5), but not at the end of the study (SNAP-IV-ODD -3.7 vs. -2.9, p = 0.252). Taking the improvement in ADHD into account, the overall effect on ODD symptoms was also not significant. (RCT, 8 weeks, N = 226, n = 156, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹⁰⁴

In children and adolescents with ADHD and ODD who did not respond to a 6-week parent training program, atomoxetine (up to 1.2 mg/kg/day) reduced ADHD symptoms (SNAP-IV –8.1 vs. –2.0) and ODD symptoms (–2.7 vs. –0.3) more effectively than placebo and improved aspects of health status. Only 2 of 156 participants had previously responded to the parent training. (RCT, 8 weeks, N = 139, n = 107 atomoxetine, n = 32 placebo, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹⁰⁵

In Michelson’s 2004 ADHD relapse prevention study, atomoxetine protected against relapse regardless of comorbid ODD (overall RR 0.59; 0.43 to 0.80). Among those taking atomoxetine, 17% with ODD and 26% without ODD experienced a relapse (RR 0.67; 0.42 to 1.06); among those taking placebo, 43% with ODD and 34% without ODD experienced a relapse. (Post-hoc secondary analysis of an RCT with a tapering design following 10 weeks of open-label treatment, 9 months, N = 416, n = 292 on atomoxetine, 43% with ODD, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹⁰⁶

In Indian children with ADHD and ODD, atomoxetine and methylphenidate improved ADHD and ODD symptoms to a comparable degree. After 8 weeks, 64.3% of those on atomoxetine and 80% of those on methylphenidate no longer met the criteria for ODD. (Open-label randomized substudy, 8 weeks, N = 37, n = 22 atomoxetine, E 1b.)¹⁰⁷

3.10. Co-occurring autism spectrum disorder (ASD) = AuDHS¶

In children and adolescents with ASD and ADHD (AuDHS), atomoxetine improved parent-rated hyperactivity (SMD -0.73; -1.15 to -0.34) and inattention (SMD -0.53; –0.93 to –0.12) (2 RCTs each, n = 96), but not ADHD symptoms as assessed by teachers or clinicians. Nausea and vomiting (RR 1.91; 1.24 to 2.94), reduced sleep (RR 1.79; 1.19 to 2.70), and reduced appetite (RR 1.79; 1.17 to 2.73) occurred more frequently. The certainty of the evidence was low to very low. (Meta-analysis, k = 3 RCTs, N = 241, E 1a.)¹⁰⁸

In children and adolescents with ADHD and ASD (AuDHS), atomoxetine (1.2 mg/kg/day) reduced ADHD symptoms to a moderately greater extent than placebo (ADHD-RS difference -6.7; -10.0 to -3.4). The proportion of participants showing marked improvement (CGI-I) did not differ significantly (20.9% vs. 8.7%). Side effects were consistent with those seen in ADHD without autism. (RCT, 8 weeks, N = 97, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹⁰⁹

In the same study, atomoxetine improved, as reported by parents, hyperactivity (d = 0.6), stereotyped behavior (d = 0.5), inappropriate speech (d = 0.4), and fear of change (d = 0.4), but not social behavior. (RCT, 8 weeks, N = 97, n = 48, manufacturer-funded, E 1b.)¹¹⁰

In children with ASD and ADHD symptoms (AuDHS), atomoxetine (up to 1.8 mg/kg/day), atomoxetine plus parent training, and parent training alone each reduced ADHD symptoms more effectively than placebo (ES 0.57 to 0.98). Atomoxetine, with or without parent training, also reduced noncompliant behavior (ES 0.47 to 0.64). The response rate was 46.9% with atomoxetine compared to 19.4% with placebo. Irritability occurred more frequently with placebo than with atomoxetine. (RCT, 2×2 factorial, 10 weeks, N = 128, E 1b.)¹¹¹ Titration was slow, starting at 0.3 mg/kg, with divided doses.

1.5 years after the start of the study by Handen et al., ADHD symptoms and noncompliant behavior had worsened slightly by the end of the study, but were still significantly better than at the start. 61% continued to show at least a 30% improvement in ADHD symptoms, even though only 34% were still taking atomoxetine. (Follow-up of an RCT, N = 94, E 1b.)¹¹²

In children with developmental disorders (primarily ASD), atomoxetine improved ADHD symptoms in nearly all studies; effects on comorbid symptoms and cognition were inconsistent or undetectable. Irritability occurred more frequently than in children without developmental disorders. In one included open-label study, 77% of children with an IQ > 85 responded to treatment, but only 21% of children with a lower IQ did so. (Review, k = 11, including 2 RCTs, E 4.)¹¹³

Among children aged 3 to 6 with ADHD and autism, 37.6% experienced side effects while taking atomoxetine (target dose 1.2 to 1.8 mg/kg/day, off-label), primarily gastrointestinal side effects (24.1%), aggression (12.8%), and increased hyperactivity (9.0%). As a result, 17.3% discontinued treatment, and the side effects resolved. At the final visit, 62.4% were considered to have improved “a great deal” and 20.3% “a lot.” (Retrospective cohort study without a control group, N = 133, E 2b.)¹¹⁴

3.11. Comorbid Reading and Spelling Disorder¶

Atomoxetine improved reading skills in children and adolescents (ages 10 to 16) with dyslexia without ADHD (WJIII Word Attack ES = 0.72, Basic Reading Skills ES = 0.48, Reading Vocabulary ES = 0.73), and in children and adolescents with ADHD and dyslexia, it improved phonological processing (CTOPP Elision ES = 0.50). The improvement in reading did not correlate with the improvement in ADHD symptoms. The secondary endpoints were not corrected for multiple testing. (RCT, Phase 4 study, N = 209, n = 58 with dyslexia, of whom 29 received atomoxetine; n = 124 with ADHD and dyslexia, of whom 64 received atomoxetine; 16 weeks; author(s) employed by the manufacturer; E 1b.)¹¹⁵ Discontinuations due to adverse events: 7.5% vs. 2.2%.

3.12. Comorbid Epilepsy¶

Among children and adolescents with epilepsy and ADHD, 31.4% responded to atomoxetine after 12 weeks (mean dose 0.79 mg/kg/day). In 7.6%, seizure activity worsened; one of these patients discontinued treatment as a result. No life-threatening side effects occurred. (Retrospective cohort study, N = 105, E 2b.)¹¹⁶ The low mean dose (typically 1.2 mg/kg) could partly explain the low response rate; no control group.

3.13. ATX During Pregnancy¶

A study found no increase in severe congenital malformations following exposure to atomoxetine during the first trimester of pregnancy (n = 990 pregnancies) (adjusted prevalence ratio 0.99). Cardiac malformations and limb malformations were not statistically significantly increased; however, the estimate was uncertain due to the small number of cases. (Cohort study, E 2b)¹¹⁷
The risk of neurodevelopmental disorders in offspring was not increased after taking atomoxetine during pregnancy compared with discontinuing it before pregnancy (HR 1.03), based on a small number of exposed individuals (Swedish cohort study, n = 861,650 children of n = 572,731 mothers from 2008–2017, E 2b)¹¹⁸
The risk of birth defects associated with methylphenidate or atomoxetine (analyzed together) did not show a statistically significant change. (Meta-analysis, k = 10, n = 16,621,481, E 2b)¹¹⁹

In breastfeeding mothers taking 40 to 100 mg of atomoxetine daily, the mean concentration in breast milk was 12 ng/mL (normalized to 80 mg/day), corresponding to a relative infant dose of 0.19% (worst-case scenario: 0.65%), well below the 5% threshold. No adverse effects were reported in the breastfed infants. (Full text not available to us; pharmacokinetic study, N = 10, Phase 2b.)¹²⁰

4. Dosage¶

The dosage of atomoxetine is initially determined based on body weight. The starting dose is 0.5 mg/kg/day, and after at least 3 days, it is increased to the target dose of 1.2 mg/kg/day.
However, determining the appropriate dose is just as individualized as it is for MPH or AMP. Reports of the ability to tailor the dose based on blood levels have not been confirmed with the degree of precision required for practical therapeutic use. (Narrative review, E 4)¹²¹

To minimize side effects, a slower titration and a waiting period of at least 1 week between dose increases are recommended. (Narrative review, E 4)¹²²
(E 1a): The medication can be taken as a single daily dose in the morning or in two equal doses in the morning and in the late afternoon/early evening. A single evening dose resulted in fewer side effects but also reduced efficacy. (Systematic review, E 1a)⁸⁹(Narrativer Review, E 4)¹²²

Adults taking 40 mg twice daily experienced nausea less frequently than those taking 80 mg once daily (16.4% vs. 32.4%; p = 0.007) and had a greater reduction in symptoms. (CAARS-Inv: SV -17 vs. -13 points, p < 0.001) (RCT without a placebo arm, 6 weeks, N = 218, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹²³

In children, side effects occurred earlier with once-daily dosing than with twice-daily dosing (p = 0.007), and loss of appetite and nausea persisted longer. Rapid titration shortened the time to onset of abdominal pain, loss of appetite, and drowsiness. Children who had not previously taken stimulants experienced abdominal pain, loss of appetite, and fatigue more frequently. In adults, insomnia occurred earlier and lasted longer with twice-daily dosing and rapid titration. (Post-hoc analysis, k = 25 studies, including 22 pediatric studies; manufacturer-funded; author(s) employed by the manufacturer; E 1a.)¹²⁴

The total daily dose for children and adolescents should not exceed 1.8 mg/kg, but should not exceed 100 mg per day. (Systematic review, E 1a)⁸⁹ The U.S. prescribing information for Strattera (Lilly) specifies a maximum of 1.4 mg/kg/day or 100 mg/day for children weighing less than 70 kg, whichever is lower. (Narrative review, E 4)¹²²
A meta-analysis reports an increase in the effect of ATX up to a dose of 1.4 mg/kg, after which the effect levels off. (Meta-analysis, E 1a)¹²⁵

In Japanese children and adolescents (ages 6 to 17), atomoxetine at three doses (0.5, 1.2, and 1.8 mg/kg/day) only the 1.8 mg/kg/day dose of atomoxetine was significantly superior to placebo (ADHD-RS -11.6 vs. -8.1; p = 0.01, one-sided), with a linear dose-response relationship. The response rate among those receiving placebo was high at 55.7% (66.7% for 1.8 mg/kg/day). Loss of appetite and vomiting were more common with atomoxetine (RCT, 8 weeks, N = 245, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹²⁶

Among Korean children and adolescents, ADHD symptoms decreased in a dose-dependent manner (ADHDRS-IV scores of -9.6, -12.3, and -14.5 at doses of 0.2, 0.5, and 1.2 mg/kg/day, p = 0.024). Adverse events increased with dose (58.3% at 1.2 vs. 29.4% at 0.2 mg/kg/day) and were mostly mild to moderate in severity. (Open-label randomized trial, 6 weeks, N = 153, Phase 1b.)¹²⁷

Atomoxetine 1 time per day (60 to 100 mg) reduced ADHD symptoms in adults more effectively than placebo after 24 weeks (effect size 0.57, response rate 68% vs. 42%). Up-dosing from 40 to 80 mg after 3 days led to more discontinuations due to adverse events than after 7 days, with a comparable rate of side effects (Full text not available to us, RCT, 24 weeks, n = 268 atomoxetine, n = 234 placebo, E 1b.)¹²⁸

A single dose taken in the morning or evening reduced ADHD symptoms in children compared to placebo, with measurable effects lasting up to 24 hours after administration. When administered in the morning, atomoxetine was more effective on some scales (ADHD-RS: -17.76 vs. -13.45 vs. placebo: -9.51, response rate 61% vs. 57% vs. 35%), while when administered in the evening, it was better tolerated (at least one adverse event: 48.9% vs. 74.0%, placebo 43.5%). (RCT, 6 weeks, N = 288, manufacturer-funded, E 1b.)¹²⁹

A rapid (over 2 weeks) and a slow (over 10 weeks) overlapping switch from stimulants to atomoxetine (target dose 1.2 mg/kg/day) did not differ in children and adolescents either in terms of symptom reduction (ADHD-RS after 10 weeks: -15.0 vs. -14.3, p = 0.692) or in terms of adverse events. (Open-label RCT, 14 weeks, N = 111, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹³⁰

In a U.S. insurance database, 36.6% of adults newly prescribed atomoxetine consistently received less than the recommended target dose of 80 mg/day. The mean dose was 68.5 mg/day; women were more likely to be in the underdosed group (54% vs. 44%). (Retrospective cohort study, N = 12,412, manufacturer-funded, E 2b.)¹³¹

Among children and adolescents from non-Western countries, an initial dose of less than 0.5 mg/kg was associated with a shorter duration of atomoxetine treatment (p < 0.01). Older age, female gender, and the absence of specific phobias correlated with higher remission rates (CGI-S = 2 or lower). (Post-hoc analysis of a prospective observational study, 12 months, n = 234 ATX, of whom n = 134 had dose data, n = 221 MPH, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)¹³²

In children and adolescents (ages 6 to 16) with a robust response, a maintenance dose reduced to 0.5 mg/kg/day maintained the response for up to 8 months just as effectively as the continued dose of 1.2 to 1.8 mg/kg/day (relapse rate 2.7% vs. 2.6%). At the lower dose, heart rate increased less and weight increased more; affective lability was reported more frequently (4.5% vs. 0%). (RCT, double-blind, up to 8 months, N = 229, n = 113 low dose, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹³³

The following factors that influence the pharmacokinetics or response to atomoxetine are discussed: (Narrative review, E 4)¹²¹

• Food
• Drug interactions
• CYP2D6 gene variants
• CYP2C19 gene variants
• NET gene variants
• Dopamine-?-hydroxylase gene variants

One review describes ATX as safe and effective, even when used in combination with stimulants. (Narrative review, E 4)¹³⁴

Stimulants, most commonly MPH, were primarily combined with atomoxetine in cases of inadequate response to previous treatments. The combination improved symptom control in some, but not all, people with ADHD, without causing serious adverse events. The evidence was weak due to small, heterogeneous studies. The only RCT found no additional benefit from adding OROS methylphenidate to atomoxetine. (Systematic review, k = 16 publications, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)¹³⁵

A case report describes a 30-year-old man who suffered seizures and refractory cardiogenic shock following an atomoxetine overdose and recovered while on veno-arterial ECMO. (Case Report, E 4)¹³⁶

5. Time Until Effect Takes Effect¶

The full effect of ATX becomes apparent only after 6 to 8 weeks of treatment or later. Responders typically show some improvement as early as 4 weeks. Once full clinical efficacy is achieved, it appears to remain at a relatively constant level throughout the day. (Systematic review, E 1a)⁸⁹

The median time to response, defined as a 25% improvement in ADHD symptoms, was 3.7 weeks in pooled studies. (Further) symptom improvement may occur up to 52 weeks after the start of treatment. (Narrative review, E 4)¹³⁴

Onset of action was observed within 1 to 2 weeks, and the response continued to increase for up to 24 weeks or longer. A single dose remained effective for 24 hours despite a plasma half-life of approximately 5 hours. (Narrative review; author(s) employed by the manufacturer; E 4.)¹³⁸

The first effects of atomoxetine were observed in some cases within one week. In a 6-week study, the median was 23 days (n = 72). The effect usually developed gradually and was sometimes not fully established until after 3 months. (Review, k = 125, manufacturer-funded, E 4.)⁴⁵

In adults, atomoxetine was superior to placebo starting in the first week (effect size 0.28); the effect size increased from 0.45 at 4 weeks to 0.52 at 26 weeks, and to 0.82 at the target dose of 80 mg. The response rate (50% reduction in CAARS) continued to increase through week 26. (Pooled analysis of two RCTs, n = 518 atomoxetine, n = 485 placebo, manufacturer-funded, E 1a.)¹³⁹

In previously untreated children with a combined subtype, the clinical response to atomoxetine occurred later than with OROS methylphenidate (13 vs. 7 weeks, p < 0.001), with comparable efficacy (76.19% vs. 77.27%) and rates of adverse events (57.14% vs. 54.54%). The mean atomoxetine dose was 1.31 mg/kg/day. (Randomized, open-label comparative study, up to 20 weeks, N = 43, n = 21 atomoxetine, E 1b.)¹⁴⁰

Atomoxetine, administered once daily in the morning, remained effective in adults into the evening hours and was superior to placebo after 10 weeks and 6 months (AISRS from 38.2 to 21.4 vs. 38.6 to 25.8, p = 0.035). Discontinuations due to adverse events were more common (17.2% vs. 5.6%), and only 37.6% and 44.6%, respectively, completed the study. (RCT, 6 months, N = 501, n = 250, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹⁴¹

With a single morning dose (target dose 0.5 to 1.2 mg/kg/day), the evening and morning behavior of children and adolescents improved significantly within 2 weeks (WREMB-R evening score from 13.7 to 8.0, morning score from 4.3 to 2.4) and remained stable through week 24. (Two open-label studies, pooled, 24 weeks, n = 421, author(s) employed by the manufacturer, E 2b.)¹⁴²

Atomoxetine (target dose 1.2 mg/kg/day) improved objectively measured attention, impulse control, and motor restlessness in children on the computer-based sustained attention test with motion tracking (effect sizes 0.38 to 1.08) as well as ADHD symptoms as assessed by the examiner (ADHD-RS, ES = 1.30) and behavior in the evening (ES = 1.02) and in the morning (ES = 0.59). (RCT, 8 weeks, ages 6 to 12, N = 128, n = 63, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹⁴³

(ADxS experience): In our experience, many people with ADHD begin to see results within the first week.
(ADxS experience): People with ADHD even took ATX only on certain days and reported that its effects varied from day to day. After some time, they developed symptoms of depression. It is unknown whether there is a connection between the two.

6. Addressing / Responding¶

Response refers to whether the treatment has a noticeable effect on ADHD symptoms. People with ADHD who do not respond adequately to a medication are called nonresponders. “Nonresponding” does not mean that the medication has no effect, but simply that the effect falls short of the level of symptom improvement defined in the respective study.

6.1. Response Rates¶

• In a placebo-controlled study of children and adolescents (ages 6 to 16), atomoxetine resulted in a response (a decrease in the ADHD-RS of at least 40%) in 45% of participants, compared with 56% on OROS-MPH (Concerta) and 24% on placebo. (RCT, E 1b)¹⁴⁴
• One review reports a limited response in up to 40% of people with ADHD. (Narrative review, E 4)¹²²
• Among MPH nonresponders, the following are expected to respond to atomoxetine:
  • approx. 50% (single study in a narrative review, E 4)⁷⁵
  • 43% (RCT, E 1b)¹⁴⁴ 42% of atomoxetine nonresponders had previously responded to OROS-MPH
• Adults with ADHD
  • 28.2% responders at Week 24. 65.2% of nonresponders had discontinued ATX (post-hoc analysis of pooled studies, E 1b)¹⁴⁵ Responding: A reduction of at least 30% in the total score on the Conners’ Adult ADHD Rating Scales—Investigator Rated: Screening Version (CAARS-Inv:SV) and a score of < 3 on the Clinical Global Impressions of ADHD Severity Scale (CGI-S) at the endpoint
  • 32.9% responders at Week 10. 33.2% of nonresponders had discontinued ATX (post-hoc analysis of pooled studies, E 1b)¹⁴⁵ Responding: A reduction of at least 30% in the total score on the Conners’ Adult ADHD Rating Scales—Investigator Rated: Screening Version (CAARS-Inv:SV) and a score of < 3 on the Clinical Global Impressions of ADHD Severity Scale (CGI-S) at the endpoint

Approximately 75% of MPH responders are also expected to respond to atomoxetine. (Narrative review, E 4)⁷⁵

A combined response (CAARS reduction of 30% or less and CGI-ADHD-S score of 3 or higher) was achieved by 34.8% of adults treated with atomoxetine in the short term (placebo: 22.3%) and by 43.4% after six months (placebo 28.0%). A symptom reduction of at least 40% was observed in 41.3% (placebo 25.3%) and 44.0% (placebo 31.4%), respectively. (Pooled individual data analysis, k = 9, N = 3,374, manufacturer-funded studies, author(s) employed by the manufacturer, E 1a.)⁵⁰

Neither age, gender, prior use of stimulants, ADHD subtype, nor the baseline scores on the CAARS-Inv:SV and CGI-S were statistically significant predictors of response to ATX. Reductions in the CAARS-Inv:SV total score, the CAARS-Inv:SV subscores, and the CGI-S at week 4 in the short-term analysis and at week 4 or 10 in the long-term analysis were statistically significant predictors of response; however, they were not reliable enough for clinical practice. (Sensitivity 28.6 to 85.9%, specificity 23.8 to 86.7%). (Post-hoc analysis of pooled studies, E 1b.)¹⁴⁵

 

6.2. Therapeutic Reference Range, Reference Blood Levels¶

The therapeutic reference range (Cmax ranges for therapeutically effective doses) for atomoxetine was specified as follows: (Expert Consensus, E 4)¹⁴⁹

• 200 to 1,000 ng/mL, 60 to 90 minutes after taking 1.2 (mg/kg) per day

Time to steady state: 2 to 5 hours (expert consensus, E 4)¹⁴⁹, 3.11 hours (single dose of 10 mg, children with AuDHS). (Pharmacokinetic study, E 2b)¹⁵⁰
Laboratory value warning threshold: 2000 ng/ml (expert consensus, E 4)¹⁴⁹

The 2026 AGNP Consensus Guideline stated the following regarding atomoxetine for ADHD: (Consensus guideline, systematic literature review, 50 experts, E 4.)¹⁵¹

• Therapeutic reference range: 40 to 400 ng/mL when blood is drawn 1 hour after the last dose (pharmacokinetically expected range following a dose of 1.2 mg/kg/day with normal CYP2D6 activity)
• Expected range in slow CYP2D6 metabolizers after an equal dose: 660 to 1,600 ng/mL - Laboratory warning threshold: 660 ng/mL
• Time to steady state: 10 to 50 hours - TDM recommendation level: Level 3 (“useful”; reference range derived from concentrations below approved doses)
• No dose-related reference range due to a nonlinear dose-concentration relationship and high variability between normal and poor CYP2D6 metabolizers

Similar pharmacokinetics were observed in children and adults after adjustment for body weight. In children and adolescents, a preliminary therapeutic reference range was 100 to 400 ng/mL, measured 1 to 3 hours after administration, and was thus narrower than in adults. (Retrospective naturalistic study, N = 74, E 3.)¹⁵²

Maximum serum concentration within two hours (single 10-mg oral dose) (Pharmacokinetic Study, E 2b)¹⁵⁰

The specified therapeutic reference range is a population-based statistical value that cannot be applied directly to individual patients. Optimal neuropsychopharmacotherapy would therefore require determining the individual’s optimal therapeutic concentration range on a case-by-case basis. To do this, for example, blood levels can be measured after improvement has occurred. (Expert Consensus, E 4)¹⁴⁹

Among Chinese children, CYP2D6 intermediate metabolizers had 1.4- to 2.2-fold higher dose-adjusted plasma levels than extensive metabolizers and responded more frequently (93.55% vs. 85.71%). With a single morning dose, efficacy was better at peak concentrations of 268 ng/ml or higher (AUC = 0.710); and central nervous system and gastrointestinal side effects occurred more frequently in intermediate metabolizers at levels of 465 and 509 ng/ml, respectively. (Retrospective study, N = 385, 515 plasma levels, E 3.)¹⁵³

In children, responders had higher trough levels (approximately 12 hours after the last dose) than nonresponders (83.3 vs. 29.5 ng/ml, p < 0.01); the threshold for a 50% or greater reduction in symptoms was 64.60 ng/mL (p = 0.14). Six of the eight responders did not respond until the dose was increased from 0.72 to 1.52 mg/kg. (Nonrandomized prospective intervention study, N = 43, E 2a.)¹⁵⁴

7. Side Effects of Atomoxetine¶

7.1. Atomoxetine and Cardiovascular Problems¶

When starting atomoxetine for the first time, patients should be monitored for cardiac arrhythmias during the first week. In a self-controlled case series involving 15,472 individuals starting atomoxetine and 12,059 starting methylphenidate (Japanese health insurance data, predominantly working-age adults), an increased risk of cardiac arrhythmias was observed in the first 7 days after the first dose of atomoxetine (adjusted incidence rate ratio 6.22; 95% CI 1.90 to 20.35) and upon re-exposure (3.23; 95% CI 1.58 to 6.64). No association was found between MPH and cardiac arrhythmias, nor between either active ingredient and heart failure. (Self-controlled case series, n = 15,472 ATX, n = 12,059 new MPH users, E 2b)¹⁵⁵ The confidence interval of 1.90 to 20.35 is very wide.

After 6 weeks, atomoxetine increased the heart rate by an average of 6.4 beats per minute in 6- to 16-year-olds, Concerta (MPH) by 3.0, and the placebo by 0.4 beats per minute; the increase was significantly greater with atomoxetine than with MPH or the placebo. (RCT, E 1b.)¹⁵⁶

Systolic blood pressure increased on average (in mmHg; no significant difference from placebo): (RCT, E 1b.)¹⁵⁶

• for atomoxetine, by 3.7
• for Concerta (MPH), by 2.4
• 1.3 for the placebo group

Diastolic blood pressure increased on average (in mmHg; the increase was significantly greater with atomoxetine and MPH than with placebo): (RCT, E 1b.)¹⁵⁶

• for atomoxetine, by 3.8
• for Concerta (MPH), by 3.1
• by 0.4 for the placebo.

A large study of 2,566,995 children and adolescents with ADHD or autism found no increased risk of serious cardiovascular events—such as stroke, heart attack, or severe cardiac arrhythmias (OR 1.08; 0.78 to 1.49). (Cohort study, E 2b)¹⁵⁷
One study found an increased risk of cardiac arrhythmias during the first 7 days following initial exposure to atomoxetine (aIRR 6.22) and during subsequent exposure (aIRR 3.23). In contrast, no increased risk of cardiac arrhythmias was found with methylphenidate. (Self-controlled case series, E 2b)¹⁵⁸

A study with up to 14 years of follow-up found a 4% increased risk of cardiovascular disease for each year of ADHD medication use, particularly for high blood pressure and arterial disease. (Brief report on a case-control study, E 3.)¹⁵⁹ The risk was elevated with long-term use of methylphenidate or lisdexamfetamine, but with atomoxetine, it was elevated only during the first year of use.

Each year of ADHD medication use was associated with a 4% increased risk of cardiovascular disease (adjusted OR 1.04, 95% CI 1.03 to 1.05), particularly for high blood pressure and arterial diseases. For atomoxetine, the risk was elevated only during the first year of use (adjusted OR 1.07, 95% CI 1.01 to 1.13). (Nested case-control study, n = 10,388 cases, n = 51,672 controls from N = 278,027, E 2b.)¹⁶⁰

Compared with placebo, atomoxetine increased the heart rate in children and adolescents by 5.58 beats/min (4.67 to 6.49), the highest value among the active ingredients studied, and systolic blood pressure by 1.07 mmHg (0.36 to 1.79). Stimulants did not increase cardiovascular parameters to a greater extent than atomoxetine. (Full text not available to us; network meta-analysis; k = 102 RCTs; N = 22,702; n = 13,315 children and adolescents; n = 9,387 adults; E 1a.)¹⁶¹

In children and adolescents, atomoxetine increased systolic (SMD 0.16; 0.04 to 0.27) and diastolic blood pressure (SMD 0.22; 0.10 to 0.34) and heart rate by a small to moderate amount (SMD 0.43; 0.26 to 0.60), with no significant difference compared to methylphenidate or amphetamine. Cardiovascular side effects occurred in 13.1% of people with ADHD taking atomoxetine; only 1 out of 126 people with ADHD discontinued treatment for this reason (meta-analysis, k = 18, of which 7 were on atomoxetine, N = 5,837, n = 961 atomoxetine, E 1a.)¹⁶²

Children and adolescents taking atomoxetine showed a greater increase in heart rate (SMD 0.86; 0.11 to 1.62; k = 4) and systolic blood pressure (SMD 0.366; 0.23 to 0.51; k = 3) than those on MPH, with no difference in cardiac adverse events (OR 0.88; 0.51 to 1.51; k = 5) (meta-analysis, k = 22, N = 46,107, E 1a.)¹⁶³

Atomoxetine slightly increased systolic blood pressure in adults and diastolic blood pressure in children and adolescents, and increased heart rate in all age groups (by 5 to 9 beats per minute). These increases occurred early on, stabilized, and resolved after discontinuation; the QTc interval did not change. Palpitations were more common in adults than in the placebo group (3.7% vs. 0.8%). (Pooled analysis of 5 RCTs, up to 10 weeks, n = 612 atomoxetine, n = 474 placebo, long-term n = 169, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)¹⁶⁴

In the Italian ADHD registry, diastolic blood pressure (+1.60 mmHg) and heart rate (+2.93/min) increased significantly after 6 months of atomoxetine treatment, and after 12 months, heart rate (+3.26/min) increased significantly; with MPH, only heart rate increased after 6 months (+2.01/min). ECG abnormalities occurred predominantly with methylphenidate (10.0% vs. 4.1% at 12 months, RR 2.41; 1.04 to 5.60) and were mostly clinically insignificant (observational study, up to 24 months, N = 1,758, E 2b.)¹⁶⁵

In adults, starting atomoxetine treatment was not associated with an increased risk of sudden death or ventricular arrhythmia (HR 0.41; 0.10 to 1.75), stroke (HR 1.30; 0.52 to 3.29), or heart attack (HR 0.56; 0.16 to 2.00). The wide confidence intervals did not rule out a moderately increased risk (cohort study, n = 20,995 atomoxetine users, N = 238,183 matched non-users, E 2b.)¹⁶⁶

Adults taking atomoxetine did not have a higher risk of stroke (RR 1.38; 0.42 to 4.54) or transient ischemic attacks (TIA) (RR 0.31; 0.04 to 2.63) than those taking stimulants. Compared with the general population, the risk of a TIA was increased after starting ADHD medication (HR 3.44; 1.13 to 10.60), but the risk of a stroke was not (HR 0.71; 0.34 to 1.47) (Cohort study, mean follow-up 1.5 years, n = 21,606 atomoxetine, n = 21,606 stimulants, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)¹⁶⁷

A study found no prolongation of the QTcF or QTcB to more than 500 ms or an increase of more than 60 ms. (RCT, E 1b)¹⁶⁸

7.2. Atomoxetine increases histamine levels¶

(E 2b): Atomoxetine increased extracellular histamine in the PFC in rats, as did other MPHs and AMPs. (Animal study, rat, E 2b)¹⁶⁹(Tierstudie, Ratte, E 2b)¹⁷⁰

(ADxS experience): People with histamine intolerance often report problems when taking ADHD medications. It is unclear whether this is related to central histamine release, even though histamine intolerance is primarily controlled by peripheral histamine.A person with ADHD and histamine intolerance reported that she could not tolerate AMP or sustained-release MPH at all, but was able to tolerate immediate-release MPH in low doses.

7.3. Other Side Effects of Atomoxetine¶

Among 6- to 16-year-olds, weight decreased by 0.6 kg after 6 weeks on atomoxetine, decreased by 0.9 kg on Concerta (MPH), and increased by 1.1 kg on placebo. The weight loss was significantly greater with both active ingredients than with placebo, and greater with MPH than with atomoxetine. (RCT, E 1b)¹⁵⁶
Atomoxetine has been associated with a low incidence of elevated serum aminotransferase levels and with rare cases of acute, clinically evident liver damage. (Drug Database, E 4)¹⁷¹
Taking atomoxetine for more than 3 years was associated with a 44% increased risk of type 2 diabetes in adults (OR 1.44; 1.01 to 2.04). (Registry study, E 2b)¹⁷² MPH was associated with a reduced risk of type 2 diabetes during the first 3 years of use, while lisdexamfetamine showed no association.

Studies found the following side effects of atomoxetine (as a percentage of people with ADHD);

• Loss of appetite 14.9% (pooled analysis of RCTs, E 1a)¹⁷³
• Insomnia: 11.3% (pooled analysis of RCTs, E 1a)¹⁷³
• erectile dysfunction in adult men: 8.0% vs. 1.9% with placebo (pooled analysis of RCTs, E 1a)¹⁷⁴
• Delayed onset of urination (the start of urination) in adult men (in male adolescents and women, the profile was similar to that of placebo): 6.9% vs. 2.4% for placebo (pooled analysis of RCTs, E 1a)¹⁷⁴
• Drowsiness: 6.0% (not significantly more frequent than with placebo in the short-term studies, 6.0% vs. 4.2%; significantly more frequent in the long-term studies, 6.4% vs. 3.4%). (Pooled analysis of RCTs, E 1a)¹⁷³
• Decreased libido in adult men (in male adolescents and women, the profile was similar to that of placebo): 4.6% vs. 3.0% for placebo. (Pooled analysis of RCTs, E 1a)¹⁷⁴
• Dysuria in adult men (in male adolescents and women, the profile was similar to that of placebo): 3.7% vs. 1.5% for placebo. (Pooled analysis of RCTs, E 1a)¹⁷⁴
• Ejaculation disorders in adult men (in male adolescents and women, the profile was similar to that of placebo): 2.8% vs. 1.1% for placebo. (Pooled analysis of RCTs, E 1a)¹⁷⁴
• Weakened urine stream in adult men (in adolescent males and women, the profile was similar to that of the placebo): 2.5% vs. 0.6% for the placebo. (Pooled analysis of RCTs, E 1a)¹⁷⁴

Atomoxetine administered twice daily reduced sleep latency in children more than MPH administered three times daily (12.1 vs. 39.2 minutes; p < 0.001). MPH was more effective at reducing nighttime awakenings. There was no difference in ADHD efficacy. (Double-blind crossover study, approximately 7 weeks each, N = 85, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹⁷⁵

In the Italian ADHD registry, serious adverse events occurred more frequently with atomoxetine than with MPH (12.18 versus 3.51 per 100 person-years, IRR 3.47; 2.35 to 5.13), after adjusting for comorbidities, particularly: (registry study, 5 years, n = 753 atomoxetine, n = 1,350 methylphenidate, E 2b.)¹⁷⁶

• gastrointestinal (IRR 4.56; 2.00 to 10.43)
• neuropsychiatric (IRR 2.54; 1.36 to 4.74) and
• cardiovascular (IRR 3.43; 1.21 to 9.76)
• Suicidal thoughts (All five cases occurred only in the atomoxetine group)

Even mild side effects were more common with atomoxetine than with MPH, including: (Registry study, 5 years, n = 753 atomoxetine, n = 1,350 methylphenidate, E 2b.)¹⁷⁶

• Loss of appetite
• Weight loss
• Stomach pain
• Irritability
• Dizziness

Suicidal behavior, aggression, psychosis, seizures, liver damage, and QT prolongation were rare in children and adolescents treated with atomoxetine. A more pronounced increase in heart rate or blood pressure (20 beats/min or more, from 15 to 20 mmHg) affected 8 to 12%; growth retardation appeared to resolve over 2 to 5 years (systematic review, k = 70, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)¹⁷⁷

Over a period of 3 to 4 years, no new side effects occurred with atomoxetine compared to the start of treatment. Less than 6% exhibited aggressive behavior, less than 1.6% had suicidal thoughts or suicidal behavior, and no more than 2% had clinically significant changes in liver function tests; growth, vital signs, and ECG showed no clinically significant changes. 1.7% discontinued treatment due to side effects (pooled analysis of open-label long-term follow-up studies from 13 RCTs and 3 open-label studies; only patients treated with atomoxetine for at least 3 years were included, mean duration 4.8 years, n = 714, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)¹⁷⁸

After 2 years of atomoxetine treatment, weight (2.7 percentiles, corresponding to 0.87 kg) and height (2.2 percentiles, corresponding to 0.44 cm) were only slightly below expected values. The children who were smallest at the start rose in percentile rank, while the largest fell. (Pooled analysis of 13 studies, 2 years or longer, N = 412, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)¹⁷⁹

After 5 years of atomoxetine treatment, weight was initially below the expected value (with the greatest deviation occurring at 15 months) and returned to the expected level after 36 months; height returned to the expected level after 24 months. Some children who were taller or heavier than average at the start remained below expected values after 5 years. (Open-label study, up to 5 years, N = 1,312, n = 61 over 5 years, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)¹⁸⁰

Atomoxetine did not delay sexual maturation. The time to the first change in Tanner stage (464.3 vs. 433.1 days, p = 0.33) and the proportion of participants with at least one stage advance (27.1% vs. 31.9%) did not differ from placebo. The age at the onset of puberty was consistent with U.S. normative data. The Tanner stages were only a secondary endpoint; about half of the participants had been pretreated with stimulants. (RCT, relapse prevention after 10 weeks of open-label treatment, 18 months, N = 416, n = 292, analyzed n = 394, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)¹⁸¹

41 of 7,961 patients treated with atomoxetine in clinical trials showed liver abnormalities (primarily mild elevations in transaminases without liver failure). Among 351 spontaneous reports in the first 4 years after market launch, atomoxetine was considered the probable cause of liver damage in 3 cases. All cases of liver damage resolved after discontinuation of the drug. (Analysis of study and spontaneous reporting data, k = 45 studies, n = 7,961, approximately 4.3 million exposed individuals after market launch, manufacturer-funded, author(s) employed by the manufacturer, E 3.)¹⁸²

Hepatobiliary adverse events were reported more frequently with atomoxetine than with MPH (RR 2.41; 2.11 to 3.11), particularly drug-induced liver injury (RR 6.42; 4.45 to 9.06), hepatitis (RR 6.48; 2.95 to 14.24), and jaundice (RR 2.19; 1.12 to 4.27) (Pharmacovigilance study, EudraVigilance, N = 421 reports, n = 232 atomoxetine, E 3.)¹⁸³

Seizures associated with atomoxetine occurred in 0.1 to 0.2% of patients in clinical trials, no more frequently than with placebo or MPH. The postmarketing reporting rate (8 per 100,000 treated patients) was within the range of the incidence in the general population. (Analysis of study and spontaneous reporting data; k = 31 studies with n = 5,831 patients treated with atomoxetine; approximately 2.2 million exposed individuals in the spontaneous reporting database; manufacturer-funded; author(s) employed by the manufacturer; E 3.)¹⁸⁴

The risk of seizures while taking atomoxetine was lower in previously untreated children and adolescents than while taking stimulants (RR 0.72; 0.37 to 1.38), even after adjusting for initial cohort assignment (RR 0.90; 0.54 to 1.49). (Retrospective cohort study with propensity score matching, ages 6 to 17 years, 6 months, n = 13,398 atomoxetine, n = 13,322 stimulants, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)¹⁸⁵

Sexual and urological disorders are associated with a possible nonselective peripheral effect on the adrenergic nerve endings in the smooth muscle cells of the sphincter and the urethral arteries. (Network meta-analysis, E 1a)¹⁸⁶ Atomoxetine improved nocturnal enuresis in children. (RCT, E 1b)¹⁸⁷

A warning is issued regarding the (rare) possibility (official information, E 4)¹⁸⁸

• serotonin syndrome, particularly in cases of overdose or when taking concomitant serotonergic medications
• severe aggression in children and adolescents, including thoughts of harming others

In controlled studies, aggression or hostility occurred in less than 2% of treated patients. In children, these events were not significantly more common with atomoxetine than with placebo (21 vs. 9 people with ADHD, 1.6% vs. 1.1%, risk ratio 1.33; 0.67 to 2.64), and were equally common compared to MPH (7 vs. 4 people with ADHD). In adults, no such events occurred with ATX (meta-analysis, k = 14 pediatric and 3 adult studies, N = 2,114 in the pediatric placebo database, n = 1,308, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)¹⁸⁹

A 21-year-old man developed serotonin syndrome—characterized by profuse sweating, marked agitation, drowsiness, a mild fever, palpitations, prolonged QT interval, muscle twitching, tremors, and heightened reflexes. After administration of activated charcoal and a generous infusion, the symptoms resolved within three days. (Case report, E 4)¹⁹⁰

Another study found that, within one year of starting ADHD medication, adults had an 8-fold increased risk of hospitalization due to first-episode psychosis or mania (0.38% vs. 0.048%). (Registry study, E 2b)¹⁹¹ Since this study compares people with ADHD who are taking medication—rather than people with ADHD who are not taking medication—to the general population, no reliable conclusions can be drawn regarding the contribution of ADHD medications. In Iceland, atomoxetine was primarily prescribed for people with ADHD who have a recent history of active substance use; 10 out of 15 people taking atomoxetine had a diagnosis of substance use disorder. The absolute risk of a first-time episode of psychosis or mania upon initiation of medication was dependent on the active ingredient:

• Atomoxetine: 0.60% (highest)
• Amphetamine-based medications: 0.33%
• Methylphenidate: 0.19%

In the WHO VigiBase database, psychotic symptoms were not reported more frequently among 13- to 25-year-olds who were treated with atomoxetine compared to people who were treated with methylphenidate (RR 0.89; 0.70 to 1.13), unlike with amphetamines (RR 1.61; 1.26 to 2.06) (pharmacovigilance study, N = 13,863 reports, n = 4,014 atomoxetine, E 3.)¹⁹² Major depression was reported more frequently with atomoxetine in the negative control group (RR 1.74), which could be an indication of indication bias.

In most cases, the use of atomoxetine for off-label purposes was harmless; the most common consequences were drowsiness (5.7%), tachycardia (4.5%), and nausea (3.1%). Severe poisoning occurred in 21 cases, including 9 with seizures. 8.7% of the ingestions were suicide attempts (retrospective analysis of U.S. poison control center data from 2002 to 2010, n = 12,370 single-substance exposures, E 3.)¹⁹³

7.4. Contraindications for Atomoxetine¶

Contraindications include

• taking MAO inhibitors within the last 14 days
• Narrow-angle glaucoma
• severe heart and vascular diseases that may worsen with even a slight increase in blood pressure or heart rate

Atomoxetine carries a warning about an increased risk of suicidal thoughts in children and adolescents.
A large meta-analysis of children and adolescents found no significantly higher rates of aggression (7.5% vs. 6.0%) or suicidal thoughts (1.3% vs. 0.9%) with atomoxetine than with placebo. (Meta-analysis, k = 25 RCTs, N = 3,928, E 1a)⁴²

Suicidal thoughts occurred rarely in children and adolescents treated with atomoxetine, but more frequently than in those receiving a placebo (0.37% versus 0%, incidence difference 0.46; 0.09 to 0.83). The relative risk was not significantly increased (2.92; 0.63 to 13.57). There were no suicides; one suicide attempt occurred among those taking atomoxetine. Compared with MPH, the frequency of suicide-related events did not differ. The number needed to harm for an additional suicide-related event was 227 (meta-analysis, k = 14 RCTs, 6 to 18 weeks, n = 1,357 atomoxetine, n = 851 placebo, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)¹⁹⁵

Suicidal thoughts or suicidal behavior occurred in 0.37% of children and adolescents treated with atomoxetine (placebo 0.07%, MHRR 1.57, p = 0.42) and in adults at a rate of 0.11% (placebo 0.12%, MHRR 0.96), with no significant difference. There were no suicides and one suicide attempt. (Meta-analysis, k = 32, N = 3,883 children and 3,365 adults, manufacturer-funded, author(s) employed by the manufacturer, E 1a.)¹⁹⁶

In a direct comparison, the incidence of suicide-related events did not differ between atomoxetine (3 out of 559) and MPH (2 out of 465) (MHRR 0.52; 0.06 to 4.54). There were no cases of suicide attempts or suicides. (Meta-analysis, k = 5 RCTs, 6 to 9 weeks, N = 1,024, manufacturer-funded, E 1a.)¹⁹⁷

Children and adolescents treated with atomoxetine did not have a higher risk of suicide or suicide attempts in the first year than those treated with stimulants, neither as first-line therapy (HR 0.95; 0.47 to 1.92) nor as second-line therapy (HR 0.71; 0.30 to 1.67). (Cohort study, N = 279,315 for first-line therapy, N = 220,215 for second-line therapy, E 2b.)¹⁹⁸

In English primary care practices, suicidal thoughts among 2- to 18-year-olds did not occur more frequently in the 12 months following the start of atomoxetine than before (RR 0.71; 0.48 to 1.07); they occurred less frequently among 12- to 18-year-olds (RR 0.51; 0.30 to 0.87) and were not significantly more frequent among 2- to 11-year-olds (RR 1.27; 0.64 to 2.49). The case-by-case assessment classified 13 of 60 cases as possibly or probably caused by atomoxetine. (Cohort study, prescription-event monitoring, N = 4,509, research institution receives donations from the manufacturer, E 2b.)¹⁹⁹

Suicidal behavior was significantly higher in the 2 months prior to the start of atomoxetine treatment (OR 2.63; 2.28 to 3.04) and was only slightly higher during treatment compared with periods without treatment (OR 1.17; 1.00 to 1.38). Compared with the period immediately before the start of treatment, the risk was lower while on atomoxetine (OR 0.45; 0.37 to 0.54). (Cohort study with a within-individual design, N = 830,352, of whom 6% were on atomoxetine, E 2b.)²⁰⁰

8. Metabolism of Atomoxetine¶

8.1. Elimination via normal CYP2D6 metabolism¶

8.2. Elimination with Reduced CYP2D6 Metabolism¶

About 5 to 7% of people have a genetic polymorphism that makes them poor metabolizers. In slow metabolizers, the half-life increases to about 21.6 hours. (Narrative review, E 4)²⁰⁵

For more information, see CYP2D6 Metabolizing Enzyme and Effects and Duration of Action of ADHD Medications.

8.3. Clearance with Increased CYP2D6 Metabolism¶

A small number of people are ultra-rapid metabolizers. The half-life is not affected by this and is approximately 5 hours for them, just as it is for normal metabolizers. (Narrative Review, E 4)²⁰⁵

9. Interactions¶

9.1. Atomoxetine and CYP2D6¶

(E 3): The metabolism of atomoxetine by CYP2D6 depends on genetic parameters, which complicates dosing. A study of 33 Japanese children and adolescents described the prediction of atomoxetine plasma concentrations using simple physiology-based pharmacokinetic models and one-compartment models. (Modeling study, E 3)²¹¹(Übersicht, E 4)²¹²

Among children and adolescents, normal or ultra-rapid CYP2D6 metabolizers were more likely to discontinue atomoxetine due to lack of efficacy (aHR = 1.9, p = 0.04), while slow or intermediate metabolizers were more likely to discontinue due to side effects. When the genotype was known prior to prescribing, slow and intermediate metabolizers received lower initial doses (median 0.5 vs. 0.8 mg/kg/day). (Full text not available to us; retrospective study, N = 108, E 3.)²¹³

The CYP2D6 phenotype and dopamine transporter genotype tended to influence the course of the response to atomoxetine. After 4 weeks of titration, symptoms and doses were similar across the CYP2D6 groups, which is why genotyping appeared to be of little use in guiding treatment. Patients with the 9/9 dopamine transporter genotype responded less well to low/moderate doses but may still have shown a response at higher doses (54/72 mg) or later in the course of treatment. (Double-blind crossover study, 4 weeks, N = 199, E 1b.)²¹⁴

9.1.1. In people with ADHD and genetically reduced CYP2D6 metabolism¶

9.1.1.1. Atomoxetine without CYP2D6 inhibitors¶

In patients with poor CYP2D6 metabolism, the average blood level of atomoxetine was 10 times higher than in patients with extensive CYP2D6 metabolism.
High response rate to atomoxetine (80% according to the manufacturer) accompanied by increased side effects, which, however, usually do not lead to discontinuation of the medication. Titration starting with low doses (adults: 40 mg/day; children: 0.5 mg/kg/day) is recommended. (Narrative review, E 4)²¹⁵
(ADxS experience): We are aware of cases of side effects associated with dose escalation that were significantly reduced by a gradual dose escalation (8 mg/day increments, increased every 4 days).

Children and adolescents with poor CYP2D6 metabolism showed a greater reduction in symptoms at doses up to 1.8 mg/kg/day (effect size 0.5; 0.1 to 0.9), a greater increase in heart rate and diastolic blood pressure (p < 0.001), less weight gain, and more frequent loss of appetite and tremor than extensive metabolizers (pooled analysis; efficacy from 4 RCTs, 6 to 8 weeks, n = 576 extensive metabolizers and n = 33 poor metabolizers; safety from all studies, N = 3,254, n = 237 slow metabolizers, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)²¹⁶

In adults, there were no significant differences in common side effects or vital signs between CYP2D6 intermediate metabolizers and extensive/ultra-rapid metabolizers. Slow metabolizers experienced dry mouth, erectile dysfunction, hyperhidrosis, insomnia, and urinary retention more frequently, as well as greater increases in diastolic blood pressure and heart rate, and a greater decrease in BMI (open-label phase of an RCT, 12 weeks, 80 to 100 mg/day, n = 1,039 extensive/ultra-rapid, n = 780 intermediate, n = 117 poor metabolizers, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)²¹⁷

9.1.1.2. Atomoxetine when taken concomitantly with CYP2D6 inhibitors¶

In patients with poor CYP2D6 metabolism, additional administration of CYP2D6 inhibitors did not further increase the atomoxetine blood levels—which were already approximately 10 times higher—because, in the absence of CYP2D6 activity, no enzyme can be inhibited. (Narrative review, E 4)²¹⁵

9.1.2. In people with ADHD and genetically enhanced CYP2D6 metabolism¶

9.1.2.1. Atomoxetine without CYP2D6 inhibitors¶

In cases of extensive CYP2D6 metabolism without concomitant use of CYP2D6 inhibitors, average atomoxetine blood levels were lower compared to cases of limited CYP2D6 metabolism. Some of these patients had peak plasma concentrations below 200 ng/mL 1 to 2 hours after dosing. Low response to atomoxetine (60% in manufacturer studies). A dose increase may be necessary, possibly to over 100 mg/day in adults. (Narrative review, E 4)²¹⁵

9.1.2.2. Atomoxetine when taken concomitantly with CYP2D6 inhibitors¶

In cases of extensive CYP2D6 metabolism and concomitant use of CYP2D6 inhibitors, the peak plasma concentration of atomoxetine should be monitored after 1–2 hours. CYP2D6 inhibitors can increase atomoxetine blood levels, which increases both the likelihood of a response and the risk of side effects. When CYP2D6 inhibitors are administered alongside atomoxetine in people with ADHD and genetically enhanced CYP2D6 metabolism, atomoxetine blood levels should be monitored regularly. (Narrative review, E 4)²¹⁵

9.1.3. CYP2D6 inhibitors¶

CYP2D6 metabolizes, among other things: (Prescribing Information, E 4)²¹⁸

• Class I antiarrhythmic drugs
• Beta-blockers
• HT3 receptor antagonists
• Amphetamine and its derivatives
• Opioids

CYP2D6 inhibitors include, among others:

• (E 4): Fluoxetine (strong). (Narrative review, E 4)²¹⁵(Fachinformation, E 4)²¹⁸
• (E 4): Paroxetine (strong). (Narrative review, E 4)²¹⁵(Fachinformation, E 4)²¹⁸
  • Paroxetine (multiple doses of 20–40 mg/day over 6 days) increased atomoxetine exposure in healthy extensive CYP2D6 metabolizers following a single dose of atomoxetine (25 mg) (AUC₀–∞) by a factor of 5.8 and the peak concentration by a factor of 1.7. At the same time, the exposure to the active metabolite 4-hydroxyatomoxidine-O-glucuronide decreased by a factor of 1.6. (Pharmacokinetic study, open-label, non-randomized, Phase 2b)²¹⁹
• Bupropion (moderate according to the source, strong according to the FDA). (Narrative review, E 4)²¹⁵
• Duloxetine (moderate). (Narrative review, E 4)²¹⁵
• Sertraline (low). (Prescribing Information, E 4)²¹⁸. Sertraline itself was not significantly metabolized by CYP2D6. (Pharmacokinetic Study, E 2b)²²⁰

9.1.4. CYP2D6 and DAT1 gene variants influence the effects of atomoxetine and MPH¶

10. Long-term effects: No tolerance effects with atomoxetine¶

A meta-analysis of 87 randomized, placebo-controlled, double-blind studies (duration: 3 to 28 weeks) found no evidence of a decline in the efficacy of methylphenidate, amphetamine-based medications, atomoxetine, or alpha-2 agonists as treatment duration increased. (Meta-analysis, k = 87 RCTs, E 1a)²²²

Among 6- to 15-year-olds who had responded to 12 weeks of open-label atomoxetine treatment, continued atomoxetine administration over 9 months was more effective than placebo at preventing relapses (relapse rate 22.3% vs. 37.9%, p = 0.002, NNT 6). A 50% worsening of symptoms occurred less frequently (28.4% vs. 47.6%, p < 0.001, NNT 5) (RCT, withdrawal design, 9 months, N = 416, n = 292, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)²²³

In adolescents (ages 12 to 18), symptom improvement with atomoxetine (ADHD-RS score reduced by 20.2 points from 34.7, mean final dose 1.41 mg/kg/day) was maintained for up to 24 months without a dose increase, with no evidence of tolerance development. 16.5% discontinued treatment due to lack of efficacy, and 5.2% due to side effects. Height, weight, blood pressure, pulse, laboratory tests, and ECG showed no clinically significant abnormalities. (Pooled analysis of 13 studies, 6 of which were double-blind with an open-label long-term phase; n = 601; 219 with at least 2 years of treatment; manufacturer-funded; author(s) employed by the manufacturer; E 2b.)²²⁴

In adults, atomoxetine reduced ADHD symptoms (CAARS-Inv:SV) by 30.2% and functional impairment (Sheehan Disability Scale) by 25.3% (both p < 0.001). Only 18.0% completed the study; 26.3% discontinued due to lack of efficacy and 12.2% due to side effects. The side effects were predominantly consistent with the expected noradrenergic effects. (Open-label long-term study, up to 4 years, n = 384, of whom 69 remained in the study until the end, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)²²⁵

In a 48-week open-label extension study of Japanese adults, 39.5% discontinued treatment; 15.9% of all participants discontinued due to side effects, primarily nausea (4.3%). Nausea occurred in 56.2% of participants, predominantly during the first 3 months (49.4%). ADHD symptoms, quality of life, and executive functions improved significantly (p < 0.05). (Open-label extension study, 48 weeks, n = 233, manufacturer-funded, author(s) employed by the manufacturer, E 2b.)²²⁶

11. Discontinuation of Atomoxetine¶

In 4 studies, there was no evidence of withdrawal symptoms after 9 to 10 weeks of treatment. A tapering regimen was not deemed necessary. In adults, allergic reactions occurred more frequently with a gradual tapering regimen (6.5% vs. 1.1% with abrupt discontinuation); according to the authors, this was likely coincidental. (RCT, E 1b)²²⁷
(ADxS experience): In our experience, gradually tapering off atomoxetine reduces the risk of withdrawal symptoms.

After discontinuing atomoxetine, relapses occurred less frequently and more slowly than after discontinuing stimulants or extended-release guanfacine. The authors considered methodological differences among the discontinuation studies to be a possible cause (narrative review of randomized discontinuation studies, manufacturer-funded, E 4.)²²⁸

In adults, after 6 months of treatment, the response rate remained at 50% for at least 6 months after discontinuation. (Narrative review, author(s) employed by the manufacturer, E 4.)¹³⁸

In children and adolescents (ages 6 to 15), after 1 year of atomoxetine treatment, continuing treatment for an additional 6 months was more effective at preventing relapses than switching to placebo (2.5% vs. 12.2%, RR 5.6; 1.2 to 25.6). After discontinuation, symptoms generally did not return to the severity seen at the start of the study, though there was considerable individual variability. The authors concluded that some patients might benefit from a drug-free trial after one year of successful treatment. (RCT, tapering design, 6 months after 1 year of treatment, N = 163, n = 81, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)²²⁹

In adults who had responded to atomoxetine, the improvements in executive functions (BRIEF-A, self- and raters’ assessments) were maintained with continued atomoxetine treatment over 25 weeks. Under placebo, these improvements deteriorated but did not return to baseline levels. (RCT, withdrawal design, 25 weeks, N = 524, n = 266, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)²³⁰

Adults who showed greater improvement in executive functions while taking atomoxetine were more likely to experience a decline in executive functions and relapse after discontinuing the medication. Overall, the improvement in executive function was maintained for up to 6 months after discontinuation, and this maintenance was associated with reduced ADHD symptoms. (RCT, discontinuation design, 6 months after 24 weeks of atomoxetine, N = 524, n = 258 placebo, author(s) employed by the manufacturer, E 1b.)²³¹

After discontinuing long-term treatment, the symptoms “difficulty waiting” and “careless errors” predicted an increased risk of relapse in adults prior to randomization (p < 0.05). In children, “not listening” was associated with a lower likelihood of maintaining a response. Changes in quality of life correlated moderately (r = -0.61) with changes in symptoms in children and weakly (r = -0.38) in adults. (Post-hoc analysis of two RCTs with a withdrawal design, placebo groups: n = 82 children and n = 258 adults; manufacturer-funded; author(s) employed by the manufacturer; E 1b.)²³²

The common side effects of atomoxetine usually occurred during the first week of treatment and resolved after a median of 3 to 53 days. Sexual side effects in men (12.6%) subsided most slowly (median 53 days). Increases in blood pressure and heart rate returned to baseline levels after discontinuation. (Open-label preliminary phase n = 2,017, followed by an RCT with a withdrawal design, 25 weeks, N = 524, n = 258 placebo, manufacturer-funded, author(s) employed by the manufacturer, E 1b.)²³³

12. Adherence to Atomoxetine Treatment¶

In some studies, adherence to atomoxetine decreased from 64% after 3 months to 44% after 1 year. The 1-year persistence rate was 62%, which was similar to that of short-acting stimulants (60%) but lower than that of long-acting stimulants (81%). Between 2.3% and 6.7% discontinued treatment due to side effects, and between 8.4% and 26% due to lack of efficacy. Factors influencing treatment outcomes included age, gender, duration of treatment, initial dose, comorbidities, and cost coverage. (Systematic review, k = 31, author(s) employed by the manufacturer, E 1a.)²³⁴

Among adults who switched from methylphenidate to a second-line medication, adherence after 12 months was lower with atomoxetine than with lisdexamfetamine (proportion of days covered: 0.30 vs. 0.48). 85% vs. 63% had discontinued treatment (adjusted HR for discontinuation with LDX 0.53; 0.48 to 0.59), with 44.2% of atomoxetine patients discontinuing as early as the first month. (Retrospective cohort study, U.S. insurance data, N = 3,392, n = 674 atomoxetine, author(s) employed by the manufacturer of the comparator drug, E 2b.)²³⁵

Among Chinese children and adolescents, OROS methylphenidate and atomoxetine showed comparable efficacy after 8 weeks (remission rates of 35.3% vs. 37.1%). After 1 year, only 3.4% of the atomoxetine group were still taking the medication regularly, compared with 21.8% in the OROS-MPH group (p < 0.05). The most common reasons for non-adherence were side effects and lack of efficacy. (Open-label RCT, 8 weeks plus 1 year of follow-up, 6- to 16-year-olds, n = 118 atomoxetine, n = 119 OROS-MPH, partially funded by the manufacturer of the comparator drug, E 1b.)²³⁶