Amphetamines for attention deficit hyperactivity disorder (ADHD) in adults

Xavier Castells; Lídia Blanco‐Silvente; Ruth Cunill

doi:10.1002/14651858.CD007813.pub3

Amphetamines for attention deficit hyperactivity disorder (ADHD) in adults

Authors' declarations of interest

Version published: 09 August 2018 Version history

https://doi.org/10.1002/14651858.CD007813.pub3

Collapse all Expand all

Abstract

available in

Background

Attention deficit hyperactivity disorder (ADHD) is a childhood‐onset disorder characterised by inattention, hyperactivity, and impulsivity. ADHD can persist into adulthood and can affects individuals' social and occupational functioning, as well as their quality of life and health. ADHD is frequently associated with other mental disorders such as substance use disorders and anxiety and affective disorders. Amphetamines are used to treat adults with ADHD, but uncertainties about their efficacy and safety remain.

Objectives

To examine the efficacy and safety of amphetamines for adults with ADHD.

Search methods

In August 2017, we searched CENTRAL, MEDLINE, Embase, PsycINFO, 10 other databases, and two trials registers, and we ran citation searches for included studies. We also contacted the corresponding authors of all included studies, other experts in the field, and the pharmaceutical company, Shire, and we searched the reference lists of retrieved studies and reviews for other published, unpublished, or ongoing studies. For each included study, we performed a citation search in Web of Science to identify any later studies that may have cited it.

Selection criteria

We searched for randomised controlled trials comparing the efficacy of amphetamines (at any dose) for ADHD in adults aged 18 years and over against placebo or an active intervention.

Data collection and analysis

Two review authors extracted data from each included study. We used the standardised mean difference (SMD) and the risk ratio (RR) to assess continuous and dichotomous outcomes, respectively. We conducted a stratified analysis to determine the influence of moderating variables. We assessed trials for risk of bias and drew a funnel plot to investigate the possibility of publication bias. We rated the quality of the evidence using the GRADE approach, which yielded high, moderate, low, or very low quality ratings based on evaluation of within‐trial risk of bias, directness of evidence, heterogeneity of data; precision of effect estimates, and risk of publication bias.

Main results

We included 19 studies that investigated three types of amphetamines: dexamphetamine (10.2 mg/d to 21.8 mg/d), lisdexamfetamine (30 mg/d to 70 mg/d), and mixed amphetamine salts (MAS; 12.5 mg/d to 80 mg/d). These studies enrolled 2521 participants; most were middle‐aged (35.3 years), Caucasian males (57.2%), with a combined type of ADHD (78.8%). Eighteen studies were conducted in the USA, and one study was conducted in both Canada and the USA. Ten were multi‐site studies. All studies were placebo‐controlled, and three also included an active comparator: guanfacine, modafinil, or paroxetine. Most studies had short‐term follow‐up and a mean study length of 5.3 weeks.

We found no studies that had low risk of bias in all domains of the Cochrane 'Risk of bias’ tool, mainly because amphetamines have powerful subjective effects that may reveal the assigned treatment, but also because we noted attrition bias, and because we could not rule out the possibility of a carry‐over effect in studies that used a cross‐over design.

Sixteen studies were funded by the pharmaceutical industry, one study was publicly funded, and two studies did not report their funding sources.

Amphetamines versus placebo

Severity of ADHD symptoms: we found low‐ to very low‐quality evidence suggesting that amphetamines reduced the severity of ADHD symptoms as rated by clinicians (SMD −0.90, 95% confidence interval (CI) −1.04 to −0.75; 13 studies, 2028 participants) and patients (SMD −0.51, 95% CI −0.75 to −0.28; six studies, 120 participants).

Retention: overall, we found low‐quality evidence suggesting that amphetamines did not improve retention in treatment (risk ratio (RR) 1.06, 95% CI 0.99 to 1.13; 17 studies, 2323 participants).

Adverse events: we found that amphetamines were associated with an increased proportion of patients who withdrew because of adverse events (RR 2.69, 95% CI 1.63 to 4.45; 17 studies, 2409 participants).

Type of amphetamine: we found differences between amphetamines for the severity of ADHD symptoms as rated by clinicians. Both lisdexamfetamine (SMD −1.06, 95% CI −1.26 to −0.85; seven studies, 896 participants; low‐quality evidence) and MAS (SMD −0.80, 95% CI −0.93 to −0.66; five studies, 1083 participants; low‐quality evidence) reduced the severity of ADHD symptoms. In contrast, we found no evidence to suggest that dexamphetamine reduced the severity of ADHD symptoms (SMD −0.24, 95% CI −0.80 to 0.32; one study, 49 participants; very low‐quality evidence). In addition, all amphetamines were efficacious in reducing the severity of ADHD symptoms as rated by patients (dexamphetamine: SMD −0.77, 95% CI −1.14 to −0.40; two studies, 35 participants; low‐quality evidence; lisdexamfetamine: SMD −0.33, 95% CI −0.65 to −0.01; three studies, 67 participants; low‐quality evidence; MAS: SMD −0.45, 95% CI −1.02 to 0.12; one study, 18 participants; very low‐quality evidence).

Dose at study completion: different doses of amphetamines did not appear to be associated with differences in efficacy.

Type of drug‐release formulation: we investigated immediate‐ and sustained‐release formulations but found no differences between them for any outcome.

Amphetamines versus other drugs

We found no evidence that amphetamines improved ADHD symptom severity compared to other drug interventions.

Authors' conclusions

Amphetamines improved the severity of ADHD symptoms, as assessed by clinicians or patients, in the short term but did not improve retention to treatment. Amphetamines were associated with higher attrition due to adverse events. The short duration of studies coupled with their restrictive inclusion criteria limits the external validity of these findings. Furthermore, none of the included studies had an overall low risk of bias. Overall, the evidence generated by this review is of low or very low quality.

PICOs

Population

Intervention

Comparison

Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Plain language summary

available in

Amphetamines for attention deficit hyperactivity disorder in adults

Background

Attention deficit hyperactivity disorder (ADHD) is a childhood‐onset psychiatric disorder that can persist into adulthood in up to 50% of patients. From a clinical point of view, ADHD is characterised by hyperactivity, mood instability, irritability, difficulties in maintaining attention, lack of organisation, and impulsive behaviours. Occurrence of other disorders at the the same time is common, especially mood disorders and substance abuse. Amphetamines (a type of stimulant) are thought to improve ADHD symptoms, but there are concerns about how safe they are for regular use by patients with ADHD.

Review question

We examined whether treatment with amphetamines improves the symptoms of ADHD in adults.

Study characteristics

Reviewers found 19 studies, which enrolled 2521 patients. Most patients were male (57.2%), middle‐aged (mean age 35.3 years) Caucasians (84.5%). These studies compared amphetamines to placebo (something that looks like an amphetamine but with no active ingredient), and three studies also compared amphetamines with other drugs such as guanfacine, modafinil, and paroxetine. In this review, we assessed the effects of three different kinds of amphetamines: dexamphetamine (from 10.2 to 21.8 mg/d), lisdexamfetamine (from 30 to 70 mg/d), and mixed amphetamine salts (MAS) (from 12.5 to 80 mg/d). Treatment length ranged from one to 20 weeks. Eighteen studies were conducted in the USA and one study in Canada and the USA. Ten studies were conducted at multiple sites. Study funding was reported in all but two studies. Sixteen studies were funded by the manufacturer, and one was funded by government agencies.

All amphetamines reduced the severity of ADHD symptoms as rated by patients. Lisdexamfetamine and MAS also reduced the severity of ADHD symptoms as rated by clinicians, but dexamphetamine did not. Overall, amphetamines did not make people more likely to stay in treatment and were associated with higher risk of treatment ending early as the result of adverse events. We found no evidence suggesting that higher doses worked better than lower ones. We did not find any difference in effectiveness between amphetamines that act for longer periods of time versus those that act for shorter periods of time. Therefore, it appears that short‐term treatment with amphetamines reduces the severity of ADHD symptoms, but studies assessing the effects of amphetamines for longer periods of time are needed. We found no differences in effectiveness between amphetamines and guanfacine, modafinil, or paroxetine.

Quality of the evidence

The quality of the evidence was low to very low for all outcomes for several reasons, namely, it was possible for patients to know the treatment they were taking; the number of studies and included patients was low, leading to imprecise results for many outcomes; the studies had problems in their design; and, for some outcomes, results varied across trials.

Authors' conclusions

available in

Implications for practice

Amphetamines appear to improve the severity of attention deficit hyperactivity disorder (ADHD) symptoms in adults in the short term. Nevertheless, compared to placebo, amphetamines do not increase retention in treatment overall and are associated with higher risk of dropping out as the result of adverse events. Furthermore, blinding failure could occur in the included studies, leading to an overestimation of amphetamine efficacy. For this reason, we considered evidence on the efficacy of amphetamines for ADHD in adults, generated by this review, to be of low or very low quality.

Evidence from this review does not provide a sound basis on which to support the use of higher doses of amphetamines or sustained‐release formulations to achieve greater efficacy. However, we did find differences between the types of amphetamines used: lisdexamfetamine was efficacious for reducing the severity of ADHD symptoms independently of the rater, but no evidence showed an effect of dexamphetamine or MAS on the severity of ADHD symptoms, respectively, as rated by clinicians or participants. This could provide indirect, low‐quality evidence in favour of lisdexamfetamine over dexamphetamine and MAS.

Implications for research

The external validity of studies that have investigated the efficacy of amphetamines for ADHD in adults could be greater. This could be achieved by including patients with comorbid disorders such as substance use disorder or major depressive disorder. Studies with longer follow‐up periods are also needed to demonstrate the long‐term efficacy of amphetamines.

Use of objective outcomes that cannot be influenced by blinding failure, such as the number of accidents or problems at work or at home, would improve the reliability of findings. Nevertheless, it must be acknowledged that using these types of outcomes will make studies less feasible because large samples will be needed to demonstrate differences between the interventions studied.

Given that other drugs, such as atomoxetine or methylphenidate, have been shown to reduce ADHD symptoms in adults, it would be of great interest to compare the efficacy of amphetamines versus the efficacy of these interventions.

In addition, changes in comorbidity profiles with current DSM‐5 criteria mean that much of the work reviewed will need to be revalidated.

Summary of findings

Open in table viewer

Summary of findings for the main comparison. Amphetamines versus placebo for attention deficit hyperactivity disorder (ADHD) in adults

Amphetamines versus placebo for attention deficit hyperactivity disorder (ADHD) in adults
Patient or population: adult patients with attention deficit hyperactivity disorder (ADHD) Settings: outpatients Intervention: amphetamines Comparison: placebo
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	Number of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Placebo	Amphetamines
*Dexamphetamine*
ADHD symptom severity: clinician rated Assessed with ADHD‐RS‐IV Follow‐up: post intervention (mean 20 weeks)	‐	Mean clinician‐rated ADHD symptom severity score in the intervention groups was 0.24 standard deviations lower (0.80 lower to 0.32 higher)	‐	49 (1 study)	⊕⊝⊝⊝ Very low^a,b,c	An SMD of 0.24 can be considered a small effect size.
ADHD symptom severity: patient rated Assessed with DSM‐IV ADHD Behavior Checklist for Adults Follow‐up: post intervention (mean 2 weeks)	‐	Mean patient‐rated ADHD symptom severity score in the intervention groups was 0.77 standard deviations lower (1.14 lower to 0.4 lower)	‐	35 (2 studies)	⊕⊕⊝⊝ Low^a,c,d	An SMD of 0.77 can be considered a medium effect size.
*Lisdexamfetamine*
ADHD symptom severity: clinician rated Assessed with ADHD‐RS‐IV and CAARS Follow‐up: post intervention (1‐10 weeks)	‐	Mean clinician‐rated ADHD symptom severity score in the intervention groups was 1.06 standard deviations lower (1.26 lower to 0.85 lower)	‐	896 (7 studies)	⊕⊕⊝⊝ Low^c,e,f,g	An SMD of 1.06 can be considered a large effect size.
ADHD symptom severity: patient rated Assessed with CAARS Follow‐up: post intervention (1‐4 weeks)	‐	Mean patient‐rated ADHD symptom severity score in the intervention groups was 0.33 standard deviations lower (0.65 lower to 0.01 lower)	‐	67 (3 studies)	⊕⊕⊝⊝ Low^c,d,h	An SMD of 0.33 can be considered a medium effect size.
*Mixed amphetamine salts*
ADHD symptom severity: clinician rated Assessed with ADHD‐RS‐IV and AISRS Follow‐up: post intervention (3‐13 weeks)	‐	Mean clinician‐rated ADHD symptom severity score in the intervention groups was 0.80 standard deviations lower (0.93 lower to 0.66 lower)	‐	1083 (5 studies)	⊕⊕⊝⊝ Low^c,e	An SMD of 0.8 can be considered a small effect size.
ADHD symptom severity: patient rated Assessed with CAARS Follow‐up: post intervention (mean 1 week)	‐	Mean patient‐rated ADHD symptom severity score in the intervention groups was 0.45 standard deviations lower (1.02 lower to 0.12 higher)	‐	18 (1 study)	⊕⊝⊝⊝ Very low^b,c,h	An SMD of 0.45 can be considered a medium effect size.
*All amphetamines*
Retention in treatment Assessed with the proportion of randomised participants that completed the study Follow‐up: post intervention (1‐20 weeks)	Study population		RR 1.06 (0.99 to 1.13)	2323 (17 studies)	⊕⊕⊝⊝ Low^a,i	‐
	708 per 1000	750 per 1000 (701 to 800)
	Moderate
	800 per 1000	848 per 1000 (792 to 904)
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ADHD: attention deficit hyperactivity disorder; ADHD‐RS‐IV: Attention Deficit Hyperactivity Disorder Rating Scale, Fourth Version; AISRS: Adult Attention Deficity Hyperactivity Disorder Investigator Rating Scale; CAARS: Conners' Adult Attention Deficit Hyperactivity Disorder Rating Scales;CI: confidence interval; DSM‐IV:Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition;SMD: standardised mean difference.
GRADE Working Group grades of evidence. High quality: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate quality: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low quality: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low quality: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aThe certainty of the evidence was downgraded by one level owing to unclear risk of detection and performance bias because it is unclear whether blinding can be achieved in placebo‐controlled studies given the powerful behavioural effects of amphetamines. ^bThe certainty of the evidence was downgraded by two levels owing to imprecision because the 95% CI is wide, indicating that the intervention effect for this outcome can range from a small, worsening effect to a large benefit. ^cThe statistical power to detect publication bias for this comparison in this review is low. ^dThe certainty of the evidence was downgraded by one level owing to imprecision because the 95% CI is rather wide, indicating that the intervention effect for this outcome can range from a moderate to a large benefit. ^eThe certainty of the evidence was downgraded by two levels owing to unclear risk of detection and performance bias (it is unclear whether blinding can be achieved in placebo‐controlled studies given the powerful behavioural effects of amphetamines), high risk of attrition bias (large proportion of participants discontinued treatment or differences between study groups in discontinuation rates), and high risk of other bias (such as the possibility of carry‐over effect in cross‐over studies without a washout phase). ^fThe certainty of the evidence was downgraded by one level owing to moderate statistical heterogeneity. ^gThe certainty of the evidence was upgraded by one level because a large and precise effect size was observed. ^hThe certainty of the evidence was downgraded by one level owing to unclear risk of detection and performance bias (it is unclear whether blinding can be achieved in placebo‐controlled studies given the powerful behavioural effects of amphetamines) and high risk of other bias (such as the possibility of carry‐over effect in cross‐over studies without a washout phase). ⁱThe certainty of the evidence was downgraded by one level owing to inconsistency (this comparison includes three different types of amphetamines at a wide range of doses, and the analysis showed moderate heterogeneity).

Background

available in

Description of the condition

Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder affecting 7% of children and adolescents worldwide (Thomas 2015). ADHD is reported to continue into adulthood in 15% to 50% of children given this diagnosis (Faraone 2006; Lara 2009). Factors associated with persistence of the disorder into adulthood are the presence of comorbidity, ADHD severity, and ADHD treatment (Caye 2016). The prevalence of ADHD in adults has been estimated at 2.5% to 5% (Simon 2009; Willcutt 2012).

Childhood ADHD is characterised by inattention, hyperactivity, and impulsivity. Inattention is often presented as distractibility, difficulty in sustaining attention on tasks or activities, trouble in organising tasks or activities, and forgetfulness. Hyperactivity and impulsivity are usually manifested as an inability to be still or to undertake quiet activities, being fidgety, talking excessively, or having trouble awaiting turns (Thapar 2016). The clinical characteristics of ADHD in adults are more subtle than in children, with hyperactivity and impulsivity often manifesting as restlessness and talkativeness (Kessler 2010; Kooij 2009). In addition, symptoms of emotional dysregulation, such as irritability, emotional lability, and emotional reactivity, are usually described in adults with ADHD (Corbisiero 2013; Retz 2012). Symptoms of ADHD thus vary across the lifespan, with improvements in hyperactivity and impulsivity usually observed (Kessler 2010). However, inattention is thought to remain unchanged, and executive functions are significantly altered (Riccio 2005). This leads to an inability to perform complex activities as a consequence of lack of activity planning, inadequate time management, high distractibility, and lack of attention (Riccio 2005). The cluster of ADHD symptoms includes the clinical expression of neuropsychological dysfunction in several executive functions, such as working memory and impulse inhibition (Schoechlin 2005), and in reward and motivation (Castellanos 2006; Sonuga‐Barke 2008).

Adults with ADHD are at higher risk of developing comorbid psychiatric disorders such as anxiety and mood and substance use disorders (Kessler 2006). In addition, a high prevalence of antisocial personality disorder has been observed in this population (Biederman 2006; Young 2005). Particularly worrying is the prevalence of substance misuse amongst adults with ADHD, which has been reported to be twice as high as that of the general population (Biederman 2006; Levin 1998). Similarly, an inverse association has been found in patients with substance use disorders, in whom the estimated prevalence of ADHD is 23% (Van Emmerik‐van Oortmerssen 2012). Adults with ADHD tend to have more social problems that affect their work and family life (Biederman 1993). Furthermore, they have poorer driving performance and are more frequently involved in car accidents (Barkley 2002). Recently, ADHD has been associated with increased mortality (Dalsgaard 2015).

ADHD is usually diagnosed using the criteria of the Diagnostic and Statistical Manual for Mental Disorders (DSM) or theInternational Classification of Diseases (ICD). Diagnostic criteria differ between the DSM and the ICD, and these criteria have varied across different versions of the DSM. For example, to qualify for ADHD, both DSM‐IV and DSM‐IV‐TR require that patients have six out of nine symptoms of inattention or hyperactivity/impulsivity, that these symptoms have begun before the age of seven years, and that they clearly impair social, academic, or occupational function in two or more settings. In contrast, the latest version of the DSM ‐ DSM‐5 ‐ requires only five out of nine symptoms of inattention or hyperactivity/impulsivity that have begun before the age of 12 years and that interfere with social, academic, or occupational functioning in two or more settings, for adults to qualify for ADHD. In addition, and for the first time, DSM‐5 allows a diagnosis of ADHD to be made in patients with autism spectrum disorder (ASD). For a diagnosis of ADHD based on ICD‐10 criteria, six symptoms of inattention, three symptoms of hyperactivity, and one symptom of impulsivity that are present before the age of six years, and that impair social, academic, or occupational function in two or more settings, are needed. The presence of a comorbid ASD is incompatible with a diagnosis of ADHD according to ICD‐10 diagnostic criteria.

Description of the intervention

Amphetamines are drugs, structurally related to catecholamines, that increase dopamine (DA) and norepinephrine (NE) concentrations at the synapse. In healthy individuals, these catecholaminergic actions result in psychostimulant effects (Hardman 2001). Because of their stimulant activity within the central nervous system, amphetamines have been studied for the treatment of several disorders, including narcolepsy (Nishino 2007), obesity (Ioannides‐Demos 2005), amphetamine dependence (Shearer 2002), cocaine dependence (Castells 2016), and ADHD (Wilens 2003).

Use of amphetamines for the treatment of adults with ADHD has been increasing during the past decade and recently surpassed use of methylphenidate in the USA (Safer 2016). Different types of amphetamines are available for the treatment of ADHD, such as lisdexamfetamine, dexamphetamine (or dextroamphetamine), and mixed amphetamine salts (MAS), which contain d‐amphetamine and l‐amphetamine at a ratio of 3:1. Amphetamines are metabolised in the liver, and their half‐lives are 10 to 15 hours for MAS (10 to 12 hours for d‐amphetamine and 12 to 15 hours for l‐amphetamine) and around 12 hours for dexamphetamine (Markowitz 2017). Lisdexamfetamine is a prodrug with a half‐life of around 0.6 hours that is metabolised to dexamphetamine (Markowitz 2017). All amphetamine derivatives are administered orally. Lisdexamfetamine is administered once a day, and MAS and dextroamphetamine may be administered once or twice a day depending on the formulation (immediate‐release versus extended‐release) (Markowitz 2017). Recommended dosages range from 5 mg/d to 40 mg/d for MAS (FDA 2015a), from 30 mg/d to 70 mg/d for lisdexamfetamine (FDA 2015b), and from 5 mg/d to 40 mg/d for dexamphetamine (FDA 2007).

How the intervention might work

From a neurobiological perspective, ADHD is characterised by a hypofunction in frontal‐striatal, cerebellar circuits that results in executive function impairment, including decreased attention and reduced ability to plan activities and inhibit inappropriate actions. A dysfunction in dopaminergic neurotransmission has been observed in these circuits. Amphetamines increase dopamine (DA) and norepinephrine (NE) concentrations at the synapse. Although the precise mechanism of action is not well understood, it seems that these drugs act on the dopamine transporter (DAT) and presumably cause inversion of the transport direction of DAT, resulting in an efflux of dopamine from the presynaptic neuron towards the synapse. Some have proposed that amphetamines get into the presynaptic neuron through the DAT and cause exocytosis of vesicles containing DA. In addition to increased DA release, amphetamines inhibit catecholamine metabolism through catechol‐O‐methyltransferasse (COMT) (for a review of the mechanism of action of amphetamines, see Carboni 2004; Fleckenstein 2007; Heal 2013; and Sulzer 2005). Thus, by promoting DA release from the presynaptic neuron and inhibiting COMT, amphetamines increase dopamine at the synapse, which yields improvement in executive function and ADHD symptoms (for a review of the neurobiological basis of ADHD and the mechanism of action of psychostimulants, see Arnsten 2006; Grace 2002; Heal 2013; and Swanson 2007).

Why it is important to do this review

The number of medicines containing amphetamines and the number of clinical trials assessing the efficacy of these medicines for adults with ADHD have been increasing over past decades (Cunill 2016; Heal 2013). Furthermore, prescription of amphetamines for adults with ADHD has also increased (Safer 2016). In addition, after publication of the first version of this review in 2011 (Castells 2011a), lisdexamfetamine was approved for the treatment of adults with ADHD in several European countries (Ermer 2016; MHRA 2015). Despite this increase in the number of clinical trials and prescriptions of amphetamines, no new systematic review has focused on the efficacy of amphetamines in adults.

A number of factors appear to modify the efficacy of drugs used to treat ADHD. For instance, the efficacy of other stimulants seems to be lower in patients with ADHD and comorbid substance use disorders (Cunill 2015; Koesters 2008), implying that stimulants may be less useful in these patients and thus stressing the importance of adapting ADHD treatment to patient characteristics. Furthermore, the efficacy of methylphenidate is lower with lower doses (Castells 2011b; Faraone 2004), as well as with long‐acting drug‐release formulations (Peterson 2007). For this reason, we plan to carry out subgroup analyses of these factors. In addition, because pharmaceutical industry funding has been associated with positive trial results (Bekelman 2003; Riera 2017), the type of funding (i.e. with and without pharmaceutical industry funding) also merits a subgroup analysis.

Amphetamines have been blamed for causing 20 deaths among patients in Canada receiving these drugs for the treatment of ADHD, and these drugs were temporarily pulled from the market in that country (Kondro 2005). Amphetamines moreover can cause withdrawal effects (Phillips 2014), and they can be misused (Weyandt 2016). Therefore, we also aim to review the adverse effects of amphetamines, with a special emphasis on cardiovascular and psychiatric outcomes.

Finally, the change in diagnostic criteria with the introduction of DSM‐5, which permits a diagnosis of ADHD in individuals with autism spectrum disorders, will allow investigation of the efficacy and safety of amphetamines in patients with this comorbidity.

Objectives

available in

To examine the efficacy and safety of amphetamines for adults with ADHD.

Methods

available in

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs).

Types of participants

Adults (over 18 years of age) with ADHD, diagnosed by any standardised criteria (e.g. DSM‐III, DSM‐III‐R, DSM‐IV, DSM‐IV‐TR, DSM‐5, ICD‐10). The presence of comorbid disorders was not an exclusion criterion.

Types of interventions

Any amphetamine (including amphetamine, dextroamphetamine, lisdexamfetamine, or mixed amphetamine salts (MAS; Adderall)), given at any dose, compared with placebo or an active intervention(s).

We did not exclude studies with additional interventions if these were provided to both study groups.

Types of outcome measures

Primary outcomes

Severity of ADHD symptoms, assessed by clinicians and patients using a standardised instrument (e.g. the ADHD Rating Scale‐IV (ADHD‐RS‐IV; DuPaul 1998), Conners' Adult ADHD Rating Scale (CAARS; Conners 1999))

Secondary outcomes

Efficacy outcomes
1. Clinical impression of severity at study end, measured by the Clinical Global Impression (CGI) ‐ Severity (CGI‐S) scale (Guy 1976)
2. Clinical impression of improvement at study end, assessed by the CGI‐Improvement (CGI‐I) scale (Guy 1976)
3. Proportion of participants achieving a reduction of at least 30% in severity of ADHD symptoms
4. Proportion of participants achieving a CGI‐I score of 1 or 2
5. Proportion of participants achieving a reduction of at least 30% in severity of ADHD symptoms and a CGI‐I score of 1 or 2
6. Global functioning: social, occupational, and psychological functioning of adults with ADHD at study end, assessed by a standardised instrument
7. Depressive symptoms: severity of depressive symptoms at study completion, assessed by a standardised instrument
8. Anxiety: severity of anxiety symptoms at study completion, assessed by a standardised instrument
9. Retention: proportion of randomised participants that completed the study
Adverse events
1. Proportion of participants withdrawn owing to any cardiovascular adverse event
2. Proportion of participants withdrawn owing to medication abuse
3. Proportion of participants withdrawn owing to any psychiatric adverse event
4. Proportion of participants withdrawn owing to any adverse event

Search methods for identification of studies

Electronic searches

We searched the following databases and trial registers in July 2016 and again in August 2017.

Cochrane Central Register of Controlled Trials (CENTRAL; 2017, Issue 7) in the Cochrane Library, which contains the Cochrane Developmental, Psychosocial and Learning Problems Specialised Register (searched 21 August 2017).
MEDLINE Ovid (1946 to August week 2 2017).
MEDLINE In‐Process and Other Non‐indexed Citations Ovid (searched 21 August 2017).
MEDLINE Epub Ahead of Print Ovid (searched 21 August 2017).
Embase Ovid (1974 to 2017 week 34).
PsycINFO Ovid (1967 to August week 2 2017).
CINAHL Plus EBSCOhost (Cumulative Index to Nursing and Allied Health Literature; 1937 to 23 August 2017).
Science Citation Index Web of Science (SCI; 1970 to 22 August 2017).
Social Science Citation Index Web of Science (SSCI; 1970 to 22 August 2017).
Conference Proceedings Citation Index ‐ Science Web of Science (CPCI‐S; 1990 to 22 August 2017).
Conference Proceedings Citation Index ‐ Social Science and Humanities Web of Science (CPCI‐SS&H; 1990 to 22 August 2017).
Cochrane Database of Systematic Reviews (CDSR; 2017, Issue 8), part of the Cochrane Library (searched 21 August 2017).
Database of Abstracts of Reviews of Effects (DARE; 2015, Issue 2), part of the Cochrane Library (final issue searched 29 July 2016).
Worldcat (www.worldcat.org; searched 23 August 2017).
Clinicaltrials.gov (clinicaltrials.gov; searched 23 August 2017).
World Health Organization International Clinical Trials Registry Platform (WHO ICTRP; apps.who.int/trialsearch; searched 23 August 2017).

We did not apply any language or date restrictions.

We have listed the search strategies for this update in Appendix 1, along with previous search strategies in Appendix 2.

Searching other resources

We contacted the corresponding authors of all included studies, experts in the field, and the pharmaceutical company, Shire, and we inspected the reference lists of retrieved studies and relevant reviews to identify any other published, unpublished, or ongoing studies. In addition, for each included study, we performed a citation search in Web of Science to identify any later studies that may have cited it.

Data collection and analysis

Selection of studies

Having removed duplicates, two review authors (XC and RC) independently assessed the titles and abstracts of all remaining records yielded by the search strategy for eligibility, discarding those that were clearly irrelevant. Next, we acquired the full‐text reports of those records deemed potentially eligible and assessed them against our inclusion criteria (see Criteria for considering studies for this review). When we identified unpublished trials, we contacted the study co‐ordinators to request the data. We resolved disagreements by discussion, until reaching a consensus, or in consultation with a third review author (LB). We recorded our selection process in a PRISMA diagram (Moher 2009).

Data extraction and management

Two review authors (XC and RC) independently inspected the full‐text reports of included studies and extracted data onto a piloted data extraction sheet (Appendix 3). We resolved disagreements by discussion, until reaching a consensus, or in consultation with a third review author (LB).

Regarding our primary outcomes (severity of ADHD symptoms), we collected both change scores (the difference between ADHD symptom severity score at study end compared to baseline) and endpoint scores (ADHD symptom severity score at study end) but gave preference to change scores over endpoint scores. For all secondary outcomes (efficacy outcomes and adverse events), we collected endpoint scores.

We emailed study authors to request any missing data or information, when necessary. We also contacted the authors of all cross‐over trials to obtain first period data on ADHD symptoms. We made a second approach if no answer was obtained by one month after the first email (see Dealing with missing data).

Two review authors (XC and RC) entered data into Review Manager 5 (RevMan 5) (Review Manager 2014).

Assessment of risk of bias in included studies

In accordance with the Cochrane 'Risk of bias' tool (Higgins 2017a), as well as the criteria set out in Appendix 4, two review authors (XC and RC) independently assessed the risk of bias in each included study as high, low, or unclear, for each of the following domains: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other bias. Review authors sought to resolve any differences by discussion, with MC adjudicating in cases for which this was not possible.

Measures of treatment effect

Continuous outcome data

We calculated the standardised mean difference (SMD) and 95% confidence intervals (CIs) because included studies used different scales to assess the severity of ADHD symptoms. We used Hedges’ method for calculating SMD with individual study weights calculated as the inverse of the variance.

For both clinician‐ and patient‐rated severity of ADHD symptoms, we entered data into RevMan using the generic inverse variance to combine data from parallel and cross‐over studies in the manner recommended by Elbourne 2002 (see Unit of analysis issues section for additional details).

Dichotomous outcome data

We calculated the risk ratio (RR) and presented it with 95% CIs.

Unit of analysis issues

Cross‐over trials

To combine parallel‐group studies with cross‐over studies, we calculated the correlation coefficient between active and control periods and used it to calculate effect sizes (Elbourne 2002). We used data from the first study period, when available, when we could not apply these recommendations.

We could calculate the correlation coefficient of the outcome score between active and control periods from only two studies (Taylor 2000; Taylor 2001). We applied the correlation coefficient to the other cross‐over studies using the most conservative correlation coefficient (r = 0.44) in the main analysis (Taylor 2001); we used the least conservative one (r = 0.61) in a sensitivity analysis (Taylor 2000).

Multiple treatment groups

When several independent treatment groups were available (e.g. amphetamine + psychotherapy versus placebo + psychotherapy; amphetamine + fake psychotherapy versus placebo + fake psychotherapy), we included them as independent studies. In studies with multiple and correlated interventions (e.g. amphetamine 10 mg versus placebo; amphetamine 20 mg versus placebo), we combined the intervention groups into a single group and included them in the meta‐analysis as a single comparison. For binary data, we summed sample sizes and numbers of participants with the event across groups. We combined continuous data using the formulae described in Section 7.7.3.8, "Combining groups", of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a).

Dealing with missing data

We used the number of randomised participants as the denominator for dichotomous variables. For continuous data, we entered into Review Manager 2014 the sample size used in calculations of the mean and the standard deviation. We did not use any imputations to deal with missing data.

We emailed study authors to request missing data or information, when necessary. We also contacted Shire after the corresponding authors directed us to this pharmaceutical company to obtain the information requested (Castells 2009b [pers comm]). See Characteristics of included studies for data requested and subsequently provided.

Assessment of heterogeneity

We assessed clinical heterogeneity by comparing differences in study populations, interventions, and outcomes, and we evaluated methodological heterogeneity by comparing study designs.

We investigated statistical heterogeneity using tau², Chi² test, and the I² statistic, which determines the proportion of variability due to heterogeneity (I² from 0 to 40% = not important statistical heterogeneity; I² from 30% to 60% = moderate heterogeneity; I² from 50% to 90% = substantial heterogeneity; and I² from 75% to 100% = considerable heterogeneity (Deeks 2017)).

Assessment of reporting biases

We drew funnel plots to investigate any relationship between effect size and study precision (closely related to sample size) when we identified a sufficient number of studies (al least 10 studies) (Egger 1997).

Data synthesis

We used RevMan 5 to perform the analyses (Review Manager 2014).

We calculated weighted averages and 95% CIs using the inverse variance method for continuous outcomes and the Mantel‐Haenszel method for dichotomous outcomes. We pooled data using a random‐effects model because we noted marked between‐study heterogeneity as regards study design (studies with cross‐over and parallel designs were included) and length of follow‐up (from two to 20 weeks).

We examined the efficacy of amphetamines for reducing the severity of ADHD symptoms in adults by means of continuous outcome variables (using change scores, e.g. change in ADHD symptom severity score from baseline to study completion; and endpoint scores, e.g. ADHD symptom severity score at study completion) and binary ones (e.g. proportion of patients achieving a reduction of at least 30% in the severity of ADHD symptoms). The primary efficacy outcome (severity of ADHD symptoms) combined change scores and endpoint scores, but we prioritised change scores when both types of scores were available in the same study. We preferred change scores because they are more precise than endpoint scores, as long as they were adjusted for baseline severity. We analysed studies reporting response rates separately.

Summarising the quality of the evidence

Using GRADEpro (GRADEPro GDT 2015), we constructed a 'Summary of findings’ table for the comparison of amphetamines versus placebo for ADHD in adults, for the following outcomes assessed post intervention: 'severity of ADHD symptoms' (primary outcome) assessed by clinicians and patients for each amphetamine; and retention in treatment (secondary outcome). Two review authors (XC and RC) assessed the quality of the evidence for each of these outcomes using the GRADE approach, resolving disagreements by discussion until reaching a consensus. GRADE offers a structured process for appraising the quality of evidence and developing recommendations based on the extent to which one can be confident that the estimates of effect are correct (Guyatt 2011a). The assessment may result in high‐, moderate‐, low‐ or very low‐quality ratings based on evaluation of five categories: within‐trial risk of bias, directness of evidence, heterogeneity of data, precision of effect estimates, and risk of publication bias (Balshem 2011; Guyatt 2011b; Guyatt 2011c; Guyatt 2011d; Guyatt 2011e; Guyatt 2011f).

Subgroup analysis and investigation of heterogeneity

Irrespective of whether we found statistical heterogeneity, we conducted the following subgroup analyses when we had sufficient studies (i.e. at least one study in each subgroup).

Comorbidities: the presence of a comorbidity (drug use disorder, major depressive disorder) versus no comorbidity.
Type of amphetamine (amphetamine, dextroamphetamine, lisdexamfetamine, or MAS).
Dose at study completion (equal to and above the median dose versus below the median dose). We performed this subgroup analysis separately for each type of amphetamine because no pharmacological equivalence was available for the three types of amphetamine that have been studied in adults with ADHD (dextroamphetamine, lisdexamfetamine, and MAS (which consists of a fixed‐dose mixture of racaemic amphetamine aspartate monohydrate, racaemic amphetamine sulphate, dextroamphetamine saccharide, and dextroamphetamine sulphate)).
Type of drug‐release formulation (immediate‐release versus long‐acting release).

We conducted subgroup analyses for the following outcomes only: severity of ADHD symptoms rated by clinicians and patients; retention in treatment; and proportion of participants withdrawn owing to any adverse events (because the number of studies measuring these outcomes was large enough to carry out these analyses). We calculated the pooled effect size (RR or SMD) for each subgroup. We investigated whether there were between‐subgroup differences by means of the Chi² test, using a random‐effects model.

We did not conduct our planned subgroup analysis for study funding (with versus without pharmaceutical industry funding) (Castells 2009a), as only was study was not funded by the pharmaceutical industry. See Differences between protocol and review.

Sensitivity analysis

We performed sensitivity analyses in which we restricted the meta‐analysis of each outcome to those studies that had low risk of bias on that outcome. We had intended to restrict the analysis to studies that had low risk of bias in all domains (Castells 2009a), but this was not possible as no studies fulfilled this criterion. Instead, we used our assessments of incomplete outcome data and other potential sources of bias, whose scores showed between‐study variability, and conducted sensitivity analyses that included only studies scoring low risk of bias on these specific domains.

We conducted three post hoc sensitivity analyses. First, we borrowed the correlation coefficient from Taylor 2000 to calculate the effect size of six cross‐over studies (Dupaul 2012; Kay 2009; Martin 2014a/Martin 2014b; Spencer 2001; Wigal 2010) (see Unit of analysis issues). Second, we calculated the pooled risk difference for the outcomes of 'proportion of participants withdrawn owing to cardiovascular adverse events' and 'proportion of participants withdrawn owing to any adverse event', to include studies that had no events for these outcomes. Third, we excluded from the analysis one cross‐over study (Spencer 2001), which had a carry‐over effect, to determine wether the carry‐over effect may have biased the results of this review.

Results

Description of studies

See Characteristics of included studies, Characteristics of excluded studies, and Characteristics of ongoing studies.

Results of the search

Our searches for this update yielded 3414 records, from which we identified and discarded 1391 duplicates. We screened titles and abstracts of the remaining 2023 records and retrieved 39 full‐text reports for further examination. Of these, we excluded eight reports that did not meet our inclusion criteria (Criteria for considering studies for this review), and we identified eight secondary publications of previously included studies. We included 12 new studies (from 18 reports) and identified five ongoing studies, which, when combined with studies included in the previous version of the review (Castells 2011a), gives a total of 19 included studies (from 33 reports) and seven ongoing studies (from seven reports). See Figure 1.

Figure 1

Flow diagram.

Included studies

This review includes 19 studies (Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a/Martin 2014b; Spencer 2001; Spencer 2008; Taylor 2000; Taylor 2001; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010).

Study design

Of the 19 studies included in this review, six used a cross‐over design. All studies compared amphetamine versus placebo, and three studies also compared amphetamine versus an active intervention: guanfacine (Taylor 2001), modafinil (Taylor 2000), or paroxetine (Weiss 2006). One study investigated two types of amphetamines (lisdexamfetamine and MAS); thus we have included two drug versus placebo comparisons in the review (Martin 2014a/Martin 2014b).

Setting

Eighteen studies were conducted in the USA (Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a/Martin 2014b; Spencer 2001; Spencer 2008; Taylor 2000; Taylor 2001; Waxmonsky 2014; Weisler 2006; Weisler 2017; Wigal 2010). One study was conducted in both Canada and the USA (Weiss 2006).

Ten studies were multi‐centre; that is, participants were enrolled and were followed up at more than one study site (Adler 2008; Adler 2013; Brams 2012; Frick 2017; Levin 2015; Spencer 2008; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010).

Participants

The included studies randomised 2521 participants, mostly males (57.2%). Most participants were middle‐aged Caucasians (mean age, 35.3 years) with a combined type of ADHD (78.8%). (For a detailed description of participant characteristics, see Table 1.) Sample sizes ranged from 17 participants in Taylor 2001 to 420 participants in Adler 2008.

Open in table viewer

Table 1. Participants' baseline characteristics

Characteristic	Descriptive statistics	N studies (N patients)
Gender: male	N = 1435 (57.2%)	19 (2507)
Age	Mean = 35.3 (range = 20.2 to 41.2) years	19 (2507)
Race: Caucasian	N = 2006 (84.5%)	15 (2373)
Combined ADHD	N = 1341 (78.8%)	11 (1701)
Predominantly inattentive ADHD	N = 344 (20.2%)
Predominantly hyperactive/impulsive ADHD	N = 28 (1.6%)
Comorbid SUD as inclusion criterion	N = 158 (6.3%)	19 (2507)
Comorbid depressive disorders as inclusion criteria	N = 0	19 (2507)
Comorbid anxiety disorders as inclusion criteria	N = 0	19 (2507)
Treated previously with stimulants	N = 306 (41.1%)	8 (744)

ADHD: attention deficit hyperactivity disorder.
N: number.
SUD: substance use disorder.

Interventions

These studies investigated three types of amphetamines: dextroamphetamine in three studies (Taylor 2000; Taylor 2001; Weiss 2006); lisdexamfetamine in nine studies (Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Dupaul 2012; Kollins 2014; Martin 2014a; Waxmonsky 2014; Wigal 2010); and MAS in eight studies (Frick 2017; Kay 2009; Levin 2015; Martin 2014b; Spencer 2001; Spencer 2008; Weisler 2006; Weisler 2017).

Doses studied ranged from 10.2 mg/d in Taylor 2001 to 21.8 mg/d in Taylor 2000 for dextroamphetamine; from 30 mg/d in Adler 2008 and Dupaul 2012 to 70 mg/d in Adler 2008 and Dupaul 2012 for lisdexamfetamine; and from 12.5 mg/d in Weisler 2017 to 80 mg/d in Levin 2015 for MAS.

Duration

Duration of study interventions ranged from one week in Dupaul 2012 to 20 weeks in Weiss 2006, with a mean of 5.3 weeks (37.2 days). Only three studies were longer than eight weeks in duration (Adler 2013; Levin 2015; Weiss 2006).

Sponsorship

All but two studies reported their funding sources (Taylor 2000; Taylor 2001). With the exception of one study ‐ Levin 2015 ‐ all studies were funded by the pharmaceutical industry.

Excluded studies

In total, we excluded 17 studies (eight studies in this update and nine studies in the previous review) for the following reasons: 12 (70.6%) studies were not RCTs; three (17.6%) studies were not conducted in adults with ADHD (two studies were performed in children and one in individuals who had ADHD symptoms who did not qualify for the ADHD disorder), one (5.9%) study was not controlled with placebo or an active control, and one (5.9%) study did not investigate amphetamines. See Characteristics of excluded studies tables and Figure 1.

Ongoing studies

Seven clinical trials were still ongoing when we completed this update (NCT00202605; NCT00514202; NCT00928148; NCT01863459; NCT02635035; NCT02803229; NCT03153488); two of these ‐ NCT00514202 and NCT00202605 ‐ were already identified in the previous version (Castells 2011a).

Four of these studies were completed (NCT00202605; NCT00514202; NCT00928148; NCT01863459), two are recruiting participants (NCT02635035; NCT02803229), and one is not yet recruiting (NCT03153488). Five studies investigated MAS (NCT00202605; NCT00514202; NCT00928148; NCT02803229; NCT03153488), and two studies investigated lisdexamfetamine (NCT01863459; NCT02635035). Four studies have a cross‐over design (NCT00202605; NCT00928148; NCT01863459; NCT02635035), and three studies have a parallel design (NCT00514202; NCT02803229; NCT03153488). Three studies included patients with ADHD and comorbid disorders (NCT00514202; NCT01863459; NCT02803229). See Characteristics of ongoing studies.

Risk of bias in included studies

We have provided a comprehensive description of the risk of bias for each study in the 'Risk of bias' tables beneath the Characteristics of included studies tables. We have summarised this information in Figure 2.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Note: scores for blinding of participants, personnel, and outcome assessors refer to amphetamines vs placebo only comparisons; scores on all remaining domains refer to amphetamines vs placebo, guanfacine, modafinil, or paroxetine.

Allocation

Sequence generation

Four studies reported on how the random sequence was generated, and so we considered their risk of bias to be low (Adler 2013; Kay 2009; Levin 2015; Weisler 2017). The 16 remaining studies did not report on how the random sequence was generated, and so we judged them to be at unclear risk of bias (Adler 2008; Biederman 2012; Brams 2012; Dupaul 2012; Frick 2017; Kollins 2014; Martin 2014a; Martin 2014b; Spencer 2001; Spencer 2008; Taylor 2000; Taylor 2001; Waxmonsky 2014; Weisler 2006; Weiss 2006; Wigal 2010).

Allocation concealment

Three studies reported on their method of allocation concealment, and so we considered their risk of bias to be low (Adler 2013; Dupaul 2012; Weisler 2017). The 17 remaining studies did not report on their method of allocation concealment, and so we judged them to be at unclear risk of bias (Adler 2008; Biederman 2012; Brams 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a; Martin 2014b; Spencer 2001; Spencer 2008; Taylor 2000; Taylor 2001; Waxmonsky 2014; Weisler 2006; Weiss 2006; Wigal 2010).

Blinding

Blinding of participants and personnel (performance bias)

Blinding of participants and personnel was intended in all included studies (Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a/Martin 2014b; Spencer 2001; Spencer 2008; Taylor 2000; Taylor 2001; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010). Nevertheless, we deemed all studies to be unclear risk of performance bias, given that amphetamines have powerful subjective effects that may reveal the assigned treatment (Childs 2009; Johanson 1980; Makris 2004; Makris 2007; Wachtel 1992).

Blinding of outcome assessment (detection bias)

Blinding of outcome assessment was intended in all included studies (Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a/Martin 2014b; Spencer 2001; Spencer 2008; Taylor 2000; Taylor 2001; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010). Nevertheless, we deemed all studies to be unclear risk of detection bias, given that amphetamines have powerful subjective effects that may reveal the assigned treatment (Childs 2009; Johanson 1980; Makris 2004; Makris 2007; Wachtel 1992).

Incomplete outcome data

Study outcomes can be influenced by attrition because reasons for dropping out from the study may differ between active intervention and placebo groups. This selective attrition makes intervention groups that were similar at baseline different at the end of the study. This appears to be the case in studies investigating the efficacy of amphetamines for adults with ADHD. As discussed later, the proportion of participants dropping out owing to AEs was higher amongst those receiving amphetamines than placebo, suggesting that attrition was somehow related to the experimental intervention. This selective attrition can lead to bias. This is particularly true for studies with a higher dropout rate (Adler 2013), and for those with statistically significant differences in the number of dropouts between study groups (Brams 2012; Frick 2017; Spencer 2008); we rated these studies at high risk of attrition bias. In such an instance, no statistical method of dealing with missing data appears to guarantee unbiased results. Conversely, this type of bias seems unlikely amongst those studies for which attrition was low (Adler 2008; Taylor 2000; Taylor 2001); we considered these studies to be at low risk of attrition bias. For the remaining 12 studies, we judged the risk of attrition bias to be unclear because attrition was moderate but imputation methods for missing data were applied, or attrition was low but missing data were not imputed (Biederman 2012; Dupaul 2012; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a/Martin 2014b; Spencer 2001; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010).

Selective reporting

For 13 studies, the protocol was available and outcomes stated in the protocol were reported in the article (Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a/Martin 2014b; Spencer 2008; Waxmonsky 2014; Weisler 2017; Wigal 2010). We considered these studies to be at low risk of reporting bias. For the six remaining studies, the study protocol was not available and thus we considered them to be at unclear risk of reporting bias (Dupaul 2012; Spencer 2001; Taylor 2000; Taylor 2001; Weisler 2006; Weiss 2006).

Other potential sources of bias

We considered seven studies to be at high risk of other bias (Dupaul 2012; Kay 2009; Martin 2014a; Martin 2014b; Spencer 2001; Waxmonsky 2014; Wigal 2010), mainly because they had a cross‐over design with no washout phase, and thus the possibility of a carry‐over effect could not be ruled out. Indeed, in one of these studies the carry‐over effect was evident (Spencer 2001). The carry‐over effect can yield an underestimation of the effect of the intervention and can bias the result towards the null for both effectiveness and AE outcomes. For one study ‐ Frick 2017 ‐ we judged the risk of other bias to be unclear because there was a long period of time between performance of the study and publication of the main results and, in addition, secondary results were published before the main results were published. For the 12 remaining studies, groups were balanced at baseline and no other potential source of bias was found; thus we considered them to be at low risk of other bias (Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Kollins 2014; Levin 2015; Spencer 2008; Taylor 2000; Taylor 2001; Weisler 2006; Weisler 2017; Weiss 2006).

Summary

We did not rate any study as having low risk of bias overall because we considered all of them to be at unclear or high risk of bias in at least one domain of the Cochrane 'Risk of bias' tool. For all studies, we considered the risk of performance and detection bias to be unclear because it is likely that participants or clinicians would have detected the medication, given the powerful behavioural effects of amphetamines. This bias is unlikely to occur if amphetamines are compared with other psychostimulants such as modafinil or methylphenidate. Furthermore, attrition bias is likely in several studies, and the possibility of a carry‐over effect could not be ruled out in studies using a cross‐over design.

Effects of interventions

See: Summary of findings for the main comparison Amphetamines versus placebo for attention deficit hyperactivity disorder (ADHD) in adults

Amphetamines versus placebo

We were able to perform meta‐analyses for most outcomes, given the high availability of data: 'ADHD symptom severity: clinician' (68.4%), 'Retention in treatment' (84.2%), and 'Proportion of participants withdrawn owing to any adverse event' (84.2%). We present the results for each outcome below, along with results for the outcomes of 'severity of ADHD symptoms' assessed by clinicians and patients for each amphetamine, and we present results for 'retention to treatment' in summary of findings Table for the main comparison.

Primary outcomes: severity of ADHD symptoms

We found evidence to suggest that amphetamines are more efficacious than placebo in reducing the severity of ADHD symptoms whether assessed by clinicians (standardised mean difference (SMD) −0.90, 95% confidence interval (CI) −1.04 to −0.75; 13 studies, 2028 participants; Analysis 1.1; Figure 3; Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Frick 2017; Kollins 2014; Levin 2015; Spencer 2001; Spencer 2008; Waxmonsky 2014; Weisler 2017; Weiss 2006; Wigal 2010) or by patients (SMD −0.51, 95% CI −0.75 to −0.28; six studies, 120 participants; Analysis 1.2; Dupaul 2012; Kollins 2014; Martin 2014a; Martin 2014b; Taylor 2000; Taylor 2001). Statistical heterogeneity for severity of ADHD symptoms was moderate when rated by clinicians (I² = 47%) and was low (I² = 13%) when rated by patients. We drew a funnel plot for clinician‐rated efficacy and detected no asymmetry (Figure 4).

Figure 3

Forest plot of comparison: 1 Amphetamines vs placebo for ADHD in adults, outcome: 1.1 Severity of ADHD symptoms: clinician rated.

Figure 4

Funnel plot of comparison: 1 Amphetamines vs placebo for ADHD in adults, outcome: 1.1 Severity of ADHD symptoms: clinician rated.

Secondary outcomes

Efficacy outcomes

We found evidence that amphetamines are more efficacious than placebo in reducing the severity of ADHD symptoms, irrespective of the efficacy definition used.

Clinical impression of severity at study end (SMD −1.09, 95% CI −1.57 to −0.61; two studies, 78 participants; Analysis 1.3; Spencer 2001; Waxmonsky 2014).
Clinical impression of improvement at study end (one study, 263 participants; Analysis 1.4; Weisler 2017).
Proportion of participants achieving a reduction of at least 30% in the severity of ADHD symptoms (risk ratio (RR) 1.52, 95% CI 1.19 to 1.95; two studies, 381 participants; Analysis 1.5; Levin 2015; Weisler 2006).
Proportion of participants achieving a CGI‐I score of 1 or 2 (RR 2.47, 95% CI 2.10 to 2.90; eight studies, 1707 participants; Analysis 1.6; Adler 2008; Adler 2013; Frick 2017; Levin 2015; Spencer 2008; Waxmonsky 2014; Weisler 2006; Weiss 2006).
Proportion of participants achieving a reduction of at least 30% in the severity of ADHD symptoms and a CGI‐I score of 1 or 2 in another study (one study, 61 participants; Analysis 1.7; Biederman 2012).

We found no statistical heterogeneity for any of these outcomes.

We conducted a meta‐analysis of two studies with 110 participants (Biederman 2012; Weiss 2006), which revealed no differences between the groups given amphetamines and those given placebo in global functioning (SMD 0.54, 95% CI −0.34 to 1.42; Analysis 1.8), depressive symptoms (SMD 0.16, 95% CI −0.22 to 0.53; Analysis 1.9), or anxiety symptoms (SMD 0.13, 95% CI −0.24 to 0.51; Analysis 1.10). Nevertheless, few studies provided data on these outcomes in a way that was suitable for meta‐analysis.

In another meta‐analysis of 17 studies with 2323 participants (Adler 2008; Adler 2013; Biederman 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a; Martin 2014b; Spencer 2001; Spencer 2008; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010), we found no evidence to suggest that amphetamines improve retention in treatment (RR 1.06, 95% CI 0.99 to 1.13; low‐quality evidence; Analysis 1.11; Figure 5). This latter analysis showed moderate statistical heterogeneity (I² = 40%), but we detected no asymmetry in the funnel plot (not shown).

Figure 5

Forest plot of comparison: 1 Amphetamines vs placebo for ADHD in adults, outcome: 1.11 Retention in treatment.

Adverse events

A meta‐analysis of three studies with 699 participants showed that a higher proportion of participants in the amphetamine group than in the placebo group dropped out owing to cardiovascular adverse events. However, this difference was not statistically significant (RR 2.18, 95% CI 0.39 to 12.04; Analysis 1.12; Adler 2008; Dupaul 2012; Weisler 2006).

We conducted a meta‐analysis of 17 studies with 2409 participants and found that the proportion of participants who dropped out owing to any adverse event was higher in the amphetamine group than in the placebo group (RR 2.69, 95% CI 1.63 to 4.42; Analysis 1.13; Figure 6) (Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a; Martin 2014b; Spencer 2008; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010). However, it must be noted that the proportion of participants who were withdrawn owing to any adverse event was low, even in the amphetamines arm (7.6%).

Figure 6

Forest plot of comparison: 1 Amphetamines vs placebo for ADHD in adults, outcome: 1.13 Proportion of participants withdrawn owing to any adverse event.

We found no statistical heterogeneity for any adverse events.

No study reported data on the remaining two outcomes: 'proportion of participants withdrawn owing to medication abuse' and 'proportion of participants withdrawn owing to any psychiatric adverse event'.

Subgroup analyses

Comorbidity

We found no evidence to suggest that comorbidity influenced the effects of amphetamines on:

severity of ADHD symptoms assessed by clinicians (SMD −0.90, 95% CI −1.04 to −0.75; 13 studies, 2028 participants; Analysis 2.1; Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Frick 2017; Kollins 2014; Levin 2015; Spencer 2001; Spencer 2008; Waxmonsky 2014; Weisler 2017; Weiss 2006; Wigal 2010);
severity of ADHD symptoms assessed by participants (SMD −0.51, 95% CI −0.75 to −0.28; six studies, 120 participants; Analysis 2.2; Dupaul 2012; Kollins 2014; Martin 2014a; Martin 2014b; Taylor 2000; Taylor 2001);
retention in treatment (RR 1.06, 95% CI 0.99 to 1.13; 17 studies, 2323 participants; Analysis 2.3; Adler 2008; Adler 2013; Biederman 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a; Martin 2014b; Spencer 2001; Spencer 2008; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010); or
proportion of participants withdrawn owing to any adverse event (RR 2.69, 95% CI 1.63 to 4.42; 17 studies, 2409 participants; Analysis 2.4; Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a; Martin 2014b; Spencer 2008; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010).

Types of amphetamines

Included studies assessed the effects of three amphetamines: dexamphetamine, lisdexamfetamine, and MAS. We found differences between these three types of amphetamines in the reduction in severity of ADHD symptoms assessed by clinicians (Analysis 3.1): both lisdexamfetamine and MAS were more efficacious than placebo in reducing the severity of ADHD symptoms (lisdexamfetamine: SMD −1.06, 95% CI −1.26 to −0.85; seven studies, 896 participants; Figure 3; Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Kollins 2014; Waxmonsky 2014; Wigal 2010; MAS: SMD −0.80, 95% CI −0.93 to −0.66; five studies, 1083 participants; Figure 3; Frick 2017; Levin 2015; Spencer 2001; Spencer 2008; Weisler 2006), but not dexamphetamine (SMD −0.24, 95% CI −0.80 to 0.32; one study, 49 participants; Figure 3; Weiss 2006).

We also found evidence to suggest that both dexamphetamine and lisdexamfetamine are more efficacious than placebo in reducing the severity of ADHD symptoms as assessed by participants (dexamphetamine: SMD −0.77, 95% CI −1.14 to −0.40; two studies, 35 participants; Taylor 2000; Taylor 2001; lisdexamfetamine: SMD −0.33, 95% CI −0.65 to −0.01; three studies, 67 participants; Dupaul 2012; Kollins 2014; Martin 2014a), but not MAS (SMD −0.45, 95% CI −1.02 to 0.12; one study, 18 participants; Martin 2014b). See Analysis 3.2.

We found no between‐group differences in retention in treatment (RR 1.06, 95% CI 0.99 to 1.13; 17 studies, 2323 participants; Analysis 3.3; Adler 2008; Adler 2013; Biederman 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a; Martin 2014b; Spencer 2001; Spencer 2008; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010), or in the proportion of participants withdrawn owing to adverse events (RR 2.69, 95% CI 1.63 to 4.42; 17 studies, 2409 participants; Analysis 3.4; Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a; Martin 2014b; Spencer 2008; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010).

Dose at study completion

The amphetamine maintenance dose was available for all but two studies (Biederman 2012; Weiss 2006). We studied the influence of dose by splitting available comparisons into two groups (lower versus higher doses). The cutoff for delimiting low and high doses was 16 mg/d for dexamphetamine, 53.4 mg/d for lisdexamfetamine, and 50 mg/d for MAS. Four studies compared three doses of amphetamines versus placebo (Adler 2008; Dupaul 2012; Frick 2017; Weisler 2006), thus providing three amphetamine versus placebo comparisons. We combined two of these three comparisons into the same subgroup, thereby leaving two amphetamine (higher and lower doses) versus placebo comparisons (see Unit of analysis issues for the explanation on methods used to combine multiple and correlated interventions). Two studies ‐ Levin 2015 and Weisler 2017 ‐ compared two amphetamine doses versus placebo, which we combined into the same subgroup because both were above or below the median dose (Levin 2015; Weisler 2017).

We found no evidence that dose influenced the effects of:

dexamphetamine on severity of ADHD symptoms as assessed by participants (SMD −0.77, 95% CI −1.14 to −0.40; two studies, 35 participants; Analysis 4.1; Taylor 2000; Taylor 2001);
lisdexamfetamine on severity of ADHD symptoms as assessed by clinicians (SMD −1.02, 95% CI −1.22 to −0.82; six studies, 885 participants; Analysis 5.1; Adler 2008; Adler 2013; Brams 2012; Kollins 2014; Waxmonsky 2014; Wigal 2010), or as assessed by participants (SMD −0.35, 95% CI −0.61 to −0.10; three studies, 67 participants; Analysis 5.2; Dupaul 2012; Kollins 2014; Martin 2014a); retention in treatment (RR 1.00, 95% CI 0.93 to 1.08; five studies, 712 participants; Analysis 5.3; Adler 2008; Adler 2013; Dupaul 2012; Kollins 2014; Martin 2014a); or the proportion of participants withdrawn owing to any adverse event (RR 2.72, 95% CI 1.09 to 6.75; six studies, 828 participants; Analysis 5.4; Adler 2008; Adler 2013; Brams 2012; Dupaul 2012; Kollins 2014; Martin 2014a); and
MAS on severity of ADHD symptoms as assessed by clinicians (SMD −0.81, 95% CI −0.94 to −0.69; five studies, 1083 participants; Analysis 6.1; Frick 2017; Levin 2015; Spencer 2001; Spencer 2008; Weisler 2017); retention in treatment (RR 1.16, 95% CI 1.05 to 1.28; eight studies, 1569 participants; Analysis 6.2; Frick 2017; Kay 2009; Levin 2015; Martin 2014b; Spencer 2001; Spencer 2008; Weisler 2006; Weisler 2017); or the proportion of participants withdrawn owing to any adverse event (RR 3.73, 95% CI 2.16 to 6.44; seven studies, 1539 participants; Analysis 6.3; Frick 2017; Kay 2009; Levin 2015; Martin 2014b; Spencer 2008; Weisler 2006; Weisler 2017).

Type of drug‐release formulation

We found no evidence to suggest that the type of drug‐release formulation influences the effects of amphetamines on:

severity of ADHD symptoms as assessed by clinicians (SMD −0.90, 95% CI −1.04 to −0.75; 13 studies, 2028 participants; Analysis 7.1; Adler 2008; Adler 2013; Biederman 2012; Brams 2012; Frick 2017; Kollins 2014; Levin 2015; Spencer 2001; Spencer 2008; Waxmonsky 2014; Weisler 2006; Weiss 2006; Wigal 2010);
severity of ADHD symptoms as assessed by participants (SMD −0.51, 95% CI −0.75 to −0.27; six studies, 120 participants; Analysis 7.2; Dupaul 2012; Kollins 2014; Martin 2014a; Martin 2014b; Taylor 2000; Taylor 2001); or
retention in treatment (RR 1.06, 95% CI 0.99 to 1.13; 17 studies, 2323 participants; Analysis 7.3; Adler 2008; Adler 2013; Biederman 2012; Dupaul 2012; Frick 2017; Kay 2009; Kollins 2014; Levin 2015; Martin 2014a; Martin 2014b; Spencer 2001; Spencer 2008; Waxmonsky 2014; Weisler 2006; Weisler 2017; Weiss 2006; Wigal 2010).

Study funding

Twelve studies were funded by the pharmaceutical industry; only one was a non‐commercial study. We decided post hoc not to conduct a subgroup analysis of study funding given the difference in the number of studies included within each subgroup, which could compromise the validity of these analyses.

Sensitivity analyses

We performed sensitivity analyses by limiting analyses to those studies scoring low risk of bias on two specific domains of the Cochrane 'Risk of bias' tool, namely, incomplete outcome data and other potential sources of bias. Findings from these analyses were similar to those of the primary analyses (incomplete outcome data: Analysis 8.1; Analysis 8.2; other potential sources of bias: Analysis 9.1; Analysis 9.2; Analysis 9.3).

We ran another sensitivity analysis by changing the statistical model from a random‐effects model, which we used to pool data in the main analysis, to a fixed‐effect model. We observed similar results for efficacy outcomes (Analysis 10.1; Analysis 10.2; Analysis 10.3; Analysis 1.4; Analysis 1.5; Analysis 10.6; Analysis 10.7; Analysis 10.8; Analysis 10.9; Analysis 10.10), with the exception of retention in treatment, which was higher for amphetamines than for placebo (Analysis 10.11: RR 1.10, 95% CI 1.04 to 1.16; 17 studies, 2323 participants). We observed similar results for adverse events when using the fixed‐effect model (Analysis 10.12; Analysis 10.13).

We also conducted three post hoc sensitivity analyses.

We repeated the analysis of the severity of ADHD symptoms as rated by clinicians and participants after calculating the effect size of four studies (Dupaul 2012; Martin 2014a/Martin 2014b; Spencer 2001; Wigal 2010) using the least conservative correlation coefficient (see Unit of analysis issues). Results of these analyses (clinician rated: SMD −0.90, 95% CI −1.05 to −0.76; 13 studies, 2028 participants; Analysis 11.1; participant rated: SMD −0.47, 95% CI −0.69 to −0.25; six studies, 120 participants; Analysis 11.2) were comparable with results of the original analyses (clinician rated: SMD −0.90, 95% CI −1.04 to −0.75; 13 studies, 2028 participants; Analysis 1.1; patient rated: SMD −0.51, 95% CI −0.75 to −0.28; six studies, 120 participants; Analysis 1.2).
In the second analysis, we re‐analysed the outcomes of 'proportion of participants withdrawn owing to any cardiovascular adverse event' and 'proportion of participants withdrawn owing to any adverse event', calculating the risk difference (RD) (Analysis 12.1: RD 0.02, 95% CI −0.00 to 0.04; three studies, 699 participants; Analysis 12.2: RD 0.04, 95% CI 0.01 to 0.06; 17 studies, 2409 participants, respectively). This yielded similar findings to the previous analyses (Analysis 1.12: RR 2.18, 95% CI 0.39 to 12.04; three studies, 699 participants; Analysis 1.13: RR 2.69, 95% CI 1.64 to 4.42; 17 studies, 2409 participants, respectively).
In the third analysis, we removed one study (Spencer 2001), which was showing a carry‐over effect from the analysis on severity of ADHD symptoms (Analysis 13.1: SMD −0.90, 95% CI −1.05 to −0.74; 12 studies, 1998 participants), and obtained similar results to the primary analysis (Analysis 1.1: SMD −0.90, 95% CI −1.04 to −0.75; 13 studies, 2028 participants), suggesting that inclusion of this study did not bias the results of this review.

Amphetamines versus guanfacine

Only one study (17 participants) compared the efficacy of amphetamines versus guanfacine (Taylor 2001).

Primary outcomes: severity of ADHD symptoms

Taylor 2001 found no evidence to suggest that amphetamines are superior to guanfacine in reducing the severity of ADHD symptoms as rated by participants (Analysis 14.1).

Secondary outcomes

Taylor 2001 did not provide data on any of our secondary outcomes.

Amphetamines versus modafinil

Only one study (22 participants) compared the efficacy of amphetamines versus modafinil (Taylor 2000).

Primary outcomes: severity of ADHD symptoms

Taylor 2000 found no evidence to suggest that amphetamines are superior to modafinil in reducing the severity of ADHD symptoms as rated by participants (Analysis 15.1).

Secondary outcomes

Taylor 2000 did not provide data on any of our secondary outcomes.

Amphetamines versus paroxetine

Only one study (98 participants) compared the efficacy of amphetamines versus paroxetine (Weiss 2006).

Primary outcomes: severity of ADHD symptoms

Weiss 2006 found no evidence to suggest that amphetamines are superior to paroxetine in reducing the severity of ADHD symptoms as rated by clinicians (Analysis 16.1).

Secondary outcomes

Efficacy outcomes

Weiss 2006 found evidence indicating that amphetamines are more efficacious than paroxetine in increasing the proportion of participants achieving a CGI‐I score of 1 or 2 (Analysis 16.2), but are not more efficacious than paroxetine in improving global functioning (Analysis 16.3), reducing symptoms of depression (Analysis 16.4) or anxiety (Analysis 16.5), or improving retention in treatment (Analysis 16.6).

Adverse events

Weiss 2006 found no evidence that amphetamines are more efficacious than paroxetine in reducing the proportion of participants withdrawn owing to any adverse event (Analysis 16.7).

Discussion

available in

Summary of main results

Amphetamines showed mixed results in the treatment of adults with attention deficit hyperactivity disorder (ADHD). We found low‐ to very low‐quality evidence suggesting that amphetamines were more efficacious than placebo in reducing the severity of ADHD symptoms, irrespective of the rater, and low‐quality evidence that they did not improve retention in treatment. Furthermore, amphetamines were associated with increased risk of dropping out owing to adverse events. Amphetamines were not effective in improving depressive and anxiety symptoms nor global functioning.

This review found that amphetamines reduced the severity of ADHD symptoms in adults in the short term. This finding was consistent across all analyses that were conducted using different efficacy definitions and statistical models. Furthermore, in most analyses, effect sizes of amphetamines for improving ADHD symptoms were moderate to high according to conventional cutoffs (Cohen 1988).

The included studies were of short duration, lasting an average of only 5.3 weeks. This is notable for three reasons. First, ADHD is a chronic disorder, and pharmacological treatment is usually administered over long periods of time. Second, because severity tends to lessen with age (Biederman 2000; Faraone 2006; Hill 1996), the possibility that the efficacy of amphetamines in adult ADHD is less after long‐term amphetamine treatment cannot be ruled out and should be studied through clinical trials with a longer follow‐up period. Third, some reports suggest that the efficacy of drugs used to treat ADHD tends to decrease progressively over time (Cunill 2016; MTA 2004; Riera 2017). Therefore, given that most included studies were of short duration, it is possible that effect sizes of amphetamines are smaller over the long term.

As a group, amphetamines did not improve retention in treatment. Retention can be interpreted as a risk‐benefit outcome because it reflects the combined evaluation of efficacy and safety (Castells 2013; Cunill 2013; Schhneider 2006; Stroup 2003). This result cannot be considered a positive one, as it is always desirable for any intervention to show a lower discontinuation rate than placebo, suggesting that the efficacy of the medication outweighs its side effects.

We found between‐study variability in relation to the severity of ADHD symptoms as assessed by clinicians. This resulted in moderate statistical heterogeneity. We investigated the source of this heterogeneity via four subgroup analyses (comorbidities, types of amphetamines, dose at study completion, and type of drug‐release formulation). Even though we found an effect for the type of amphetamine on the severity of ADHD symptoms, with lisdexamfetamine and MAS showing larger effect sizes than dexamphetamine, this factor did not entirely explain the between‐study variability, as within‐subgroup statistical heterogeneity remained evident. We also found moderate statistical heterogeneity for 'retention to treatment', but no subgroup analyses could control for such heterogeneity, which is likely to be explained by other co‐variates or a combination of them.

As stated above, we found that the type of amphetamine influenced clinician‐rated ADHD efficacy: although both lisdexamfetamine and MAS reduced the severity of ADHD symptoms compared to placebo, dexamphetamine did not. The type of amphetamine did not influence participant‐rated ADHD efficacy, retention to treatment, or adverse events. This result, along with the fact that dextroamphetamine has been infrequently studied, provides indirect and low‐quality evidence preferring lisdexamfetamine and MAS over dextroamphetamine.

Studies have investigated a wide range of doses, and higher and lower doses of amphetamine have shown similar results. This finding is consistent with that of clinical trials that have investigated the efficacy of multiple amphetamine doses and found no differences between treatment arms (Adler 2008; Weisler 2006). Methylphenidate given at a wide range of doses has also been investigated, and findings regarding its dose‐response effects have been contradictory: some studies suggested a positive relationship (Castells 2011b; Faraone 2004; Medori 2008), and others found no association with dose (Koesters 2008; Spencer 2007).

Amphetamines have a short half‐life and must be administered two or three times a day. To facilitate treatment compliance, sustained‐release formulations have been developed. This systematic review showed that amphetamines delivered via immediate‐release and slow‐release formulations had similar efficacy and tolerability. This is relevant, as long‐acting formulations have been found to be less efficacious than short‐acting ones (Peterson 2007).

Few studies included participants with comorbid disorders, which contrasts with the high prevalence of other psychiatric conditions diagnosed in patients with ADHD (Kessler 2006). The presence of comorbid disorders did not modify efficacy, retention to treatment, nor adverse events. This finding is consistent with those of a recent study that did not find comorbidity to modify the effects of pharmacological treatment in adults with ADHD (Cunill 2016).

We did not assess the effects of study sponsorship on efficacy, retention to treatment, or adverse events, because with the exception of one study, all were sponsored by the pharmaceutical industry. Other studies have shown that studies with a commercial sponsor report more favourable outcomes than are reported by independent studies (Lundh 2017; Riera 2017).

Failure to identify an impact of amphetamines on depressive and anxiety symptoms could be a consequence of the strict inclusion criteria adopted by most included studies, which excluded patients with major depressive or bipolar disorders. Baseline depression and anxiety scores therefore were low, leaving little room for improvement. Another possible interpretation is that the effects of amphetamines on ADHD symptoms are independent of effects on mood and anxiety. The number of studies included in these analyses was low, limiting our ability to draw conclusions.

Amphetamines have been compared with only three drugs (guanfacine, modafinil, and paroxetine) in three small clinical trials. Therefore, it is not surprising that no differences have been found for most outcomes.

Overall completeness and applicability of evidence

The overall completeness and applicability of evidence related to the efficacy and safety of amphetamines are limited by two factors. First, the dearth of data on patients with ADHD with comorbid disorders such as substance abuse or major depressive disorder. This is particularly notable given the high prevalence of comorbid psychiatric disorders in patients with ADHD (Biederman 2006; Levin 1998; Van Emmerik‐van Oortmerssen 2012; Young 2005), which is expected to increase further with the use of DSM‐5, as it permits a diagnosis of ADHD in patients with autism spectrum disorders. Second, the short duration of studies, which contrasts with the chronic course and long‐term treatment of the disorder. However, strengths must also be acknowledged. This review includes a systematic and exhaustive search that permitted us to identify all amphetamine trials performed in adults with ADHD. In addition, we were able to obtain a large quantity of missing data of interest from the trial authors of a number of studies included in this review.

Quality of the evidence

We did not find any study that was free of bias. Most articles reported neither on how the random sequence was generated nor how it was concealed. Therefore, we were not able to differentiate between reporting problems and study bias. However, even if these processes had been performed correctly, no study would have been rated as free of bias because amphetamines have intense behavioural effects, and participants and raters may have detected the administered study medication. This detection may have caused a blinding failure, which might have exaggerated the efficacy of the intervention (Schultz 1995); this type of bias is less likely to occur when amphetamines are compared to other psychostimulants such as modafinil (Taylor 2000). However, no study assessed whether blinding had failed, and the fact that all studies were scored at unclear risk of bias on this domain was based on the review authors' opinion, which, in turn, was based on ample evidence that amphetamines have intense behavioural and haemodynamic effects that can unmask the intervention being studied (Childs 2009; Johanson 1980; Makris 2004; Makris 2007; Wachtel 1992). Use of a nocebo (i.e. an active placebo that produces noticeable side effects that may convince the person that he/she is being treated with the active drug) has been proposed as a means of reducing the possibility of unblinding (Storebø 2015); however, this type of comparator has ethical problems, as it conflicts with the principle of non‐maleficence. A better alternative to nocebos would be the use of objective outcomes (e.g. accidents, legal or work problems), which have a lower risk of performance and detection bias than subjective outcomes (e.g. ADHD symptom severity). Use of objective, clinically meaningful outcomes, such as accidents or legal or work problems, would also improve the external validity of the findings of clinical trials including patients with ADHD. The validity of the outcome variables used to determine the efficacy of amphetamines for ADHD symptoms is an important question. The clinical interpretation of a reduction of 30% in the severity of ADHD symptoms or a change in the number of units on the ADHD Rating Scale is not straightforward. Thus, it would be helpful to use outcomes with greater clinical interpretability to improve our understanding of the effect of an intervention for this disorder; by way of example, one could monitor the proportion of patients achieving 'symptomatic remission' (i.e. the proportion of patients who fail to meet the full ADHD diagnostic criteria) (Biederman 2000; Keck 1998).

Indirectness moreover may have jeopardised the quality of evidence in this review. Indirectness can arise from combining different medications (e.g. different amphetamines), different doses of the same medication, or studies with important follow‐up differences, thereby hindering the possibility of making precise recommendations. Uncertainties regarding the indirectness of some estimations, the imprecision of some calculations, the existence of statistical heterogeneity, and the possibility of blinding failure mean that no result can be deemed to provide high‐quality or moderate‐quality evidence. Thus, it is likely that new research may change the main findings of this review.

We ran two post hoc sensitivity analyses excluding studies at unclear or high risk of bias on two specific domains of the Cochrane 'Risk of bias' tool: 'incomplete outcome data' and 'other biases'. These analyses yielded results similar to those of the primary ones, which suggests that our findings are robust against the two potential sources of bias.

Potential biases in the review process

We conducted a comprehensive search across several bibliographic databases and trial registers, without language restrictions. We also contacted the pharmaceutical industry and corresponding authors of included publications to enquire about additional studies that we may have missed. We did not, however, inspect FDA and EMA websites, and thus we cannot rule out the possibility that the review process is biased. However, we found no evidence of reporting bias, as suggested by a symmetrical funnel plot, but it must be highlighted that the sensitivity and precision of these tests are low.

We were able to obtain relevant data from almost all studies. We were able to obtain endpoint or change scores or response rates of clinician‐ or patient‐rated scales assessing the severity of ADHD symptoms in ways suitable for meta‐analysis, either directly from the study report or from the study authors. In addition, we were able to obtain data on all‐cause treatment discontinuation from 17 out of 19 studies.

With regards to the methods used, some studies applied a modified intention‐to‐treat (ITT) approach, where only participants who provided at least one post‐randomisation outcome were included in the efficacy analysis (Adler 2008; Adler 2013; Brams 2012; Spencer 2008; Weisler 2006; Weisler 2017). Not including all randomised participants may cause attrition bias. To minimise this source of bias, we used an ITT approach to calculate the risk ratio (RR) of these studies. Proceeding in this way yields more conservative efficacy results because it assumes that all individuals who left the study did not have the outcome. Provided that most studies provided short‐term follow‐up, and given that ADHD is a chronic disorder whose severity does not change after short periods of time, it seems reasonable to assume that participants who left the study were not treatment responders. Even if this is not the case, we expect this will have minimal influence on the results because the proportion of participants excluded from the efficacy analysis of those studies that used a modified ITT approach was low (consistently below 3% of the randomised sample).

We advise caution when interpreting the results of between‐subgroup comparisons. Given that these comparisons are indirect ones, head‐to‐head comparisons are needed to confirm their findings.

Agreements and disagreements with other studies or reviews

We rated the quality of the evidence in this review as low to very low (Otasowie 2014; Punja 2016; Storebø 2015), which is comparable with the quality of evidence reported by other Cochrane Reviews on this topic. Factors that limit the validity and quality of the evidence of systematic reviews of pharmacological treatment for ADHD are recurrent and include attrition bias, the possibility of blinding failure, imprecise results, and statistical heterogeneity (Castells 2011b; Castells 2013; Cunill 2016; Otasowie 2014; Peterson 2007; Punja 2016). Improving the quality of studies investigating the efficacy and safety of pharmacological treatment for ADHD has become a priority to increase the reliability of study findings.

Figure 1

Flow diagram.

Navigate to figure in ReviewOpen in new tab

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Navigate to figure in ReviewOpen in new tab

Figure 3

Forest plot of comparison: 1 Amphetamines vs placebo for ADHD in adults, outcome: 1.1 Severity of ADHD symptoms: clinician rated.

Navigate to figure in ReviewOpen in new tab

Figure 4

Funnel plot of comparison: 1 Amphetamines vs placebo for ADHD in adults, outcome: 1.1 Severity of ADHD symptoms: clinician rated.

Navigate to figure in ReviewOpen in new tab

Figure 5

Forest plot of comparison: 1 Amphetamines vs placebo for ADHD in adults, outcome: 1.11 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Figure 6

Forest plot of comparison: 1 Amphetamines vs placebo for ADHD in adults, outcome: 1.13 Proportion of participants withdrawn owing to any adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 1.1

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 1 ADHD symptom severity: clinician‐rated.

Navigate to figure in ReviewOpen in new tab

Analysis 1.2

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 2 ADHD symptom severity: patient‐rated.

Navigate to figure in ReviewOpen in new tab

Analysis 1.3

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 3 Clinical impression of severity at study end.

Navigate to figure in ReviewOpen in new tab

Analysis 1.4

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 4 Clinical impression of improvement at study end.

Navigate to figure in ReviewOpen in new tab

Analysis 1.5

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 5 Proportion of participants achieving a reduction ≥ 30% in severity of ADHD symptoms.

Navigate to figure in ReviewOpen in new tab

Analysis 1.6

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 6 Proportion of participants achieving a CGI‐Improvement score of 1 or 2.

Navigate to figure in ReviewOpen in new tab

Analysis 1.7

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 7 Proportion of participants achieving a reduction ≥ 30% in severity of ADHD symptoms and a CGI‐Improvement score of 1 or 2.

Navigate to figure in ReviewOpen in new tab

Analysis 1.8

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 8 Global functioning.

Navigate to figure in ReviewOpen in new tab

Analysis 1.9

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 9 Depressive symptoms.

Navigate to figure in ReviewOpen in new tab

Analysis 1.10

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 10 Anxiety symptoms.

Navigate to figure in ReviewOpen in new tab

Analysis 1.11

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 11 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Analysis 1.12

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 12 Proportion of participants withdrawn owing to any cardiovascular adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 1.13

Comparison 1 Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 13 Proportion of participants withdrawn owing to any adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 2.1

Comparison 2 Subgroup analysis 1: comorbidity, Outcome 1 ADHD symptom severity: clinician‐rated.

Navigate to figure in ReviewOpen in new tab

Analysis 2.2

Comparison 2 Subgroup analysis 1: comorbidity, Outcome 2 ADHD symptom severity: patient‐rated.

Navigate to figure in ReviewOpen in new tab

Analysis 2.3

Comparison 2 Subgroup analysis 1: comorbidity, Outcome 3 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Analysis 2.4

Comparison 2 Subgroup analysis 1: comorbidity, Outcome 4 Proportion of patients withdrawn owing to any adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 3.1

Comparison 3 Subgroup analysis 2: type of amphetamine, Outcome 1 ADHD symptom severity: clinician‐rated.

Navigate to figure in ReviewOpen in new tab

Analysis 3.2

Comparison 3 Subgroup analysis 2: type of amphetamine, Outcome 2 ADHD symptom severity: patient‐rated.

Navigate to figure in ReviewOpen in new tab

Analysis 3.3

Comparison 3 Subgroup analysis 2: type of amphetamine, Outcome 3 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Analysis 3.4

Comparison 3 Subgroup analysis 2: type of amphetamine, Outcome 4 Proportion of participants withdrawn owing to any adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 4.1

Comparison 4 Subgroup analysis 3: dose of dexamphetamine, Outcome 1 ADHD symptom severity: patient rated.

Navigate to figure in ReviewOpen in new tab

Analysis 5.1

Comparison 5 Subgroup analysis 3: dose of lisdexamfetamine, Outcome 1 ADHD symptom severity: clinician rated.

Navigate to figure in ReviewOpen in new tab

Analysis 5.2

Comparison 5 Subgroup analysis 3: dose of lisdexamfetamine, Outcome 2 ADHD symptom severity: patient rated.

Navigate to figure in ReviewOpen in new tab

Analysis 5.3

Comparison 5 Subgroup analysis 3: dose of lisdexamfetamine, Outcome 3 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Analysis 5.4

Comparison 5 Subgroup analysis 3: dose of lisdexamfetamine, Outcome 4 Proportion of participants withdrawn owing to any adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 6.1

Comparison 6 Subgroup analysis 3: dose of mixed amphetamine salts, Outcome 1 ADHD symptom severity: clinician rated.

Navigate to figure in ReviewOpen in new tab

Analysis 6.2

Comparison 6 Subgroup analysis 3: dose of mixed amphetamine salts, Outcome 2 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Analysis 6.3

Comparison 6 Subgroup analysis 3: dose of mixed amphetamine salts, Outcome 3 Proportion of participants withdrawn owing to any adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 7.1

Comparison 7 Subgroup analysis 4: type of drug‐release formulation, Outcome 1 ADHD symptom severity: clinician rated.

Navigate to figure in ReviewOpen in new tab

Analysis 7.2

Comparison 7 Subgroup analysis 4: type of drug‐release formulation, Outcome 2 ADHD symptom severity: patient rated.

Navigate to figure in ReviewOpen in new tab

Analysis 7.3

Comparison 7 Subgroup analysis 4: type of drug‐release formulation, Outcome 3 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Analysis 8.1

Comparison 8 Sensitivity analysis: incomplete subjective outcome data, Outcome 1 ADHD symptom severity: clinician rated.

Navigate to figure in ReviewOpen in new tab

Analysis 8.2

Comparison 8 Sensitivity analysis: incomplete subjective outcome data, Outcome 2 ADHD symptom severity: patient rated.

Navigate to figure in ReviewOpen in new tab

Analysis 9.1

Comparison 9 Sensitivity analysis: other potential sources of bias, Outcome 1 ADHD symptom severity: clinician rated.

Navigate to figure in ReviewOpen in new tab

Analysis 9.2

Comparison 9 Sensitivity analysis: other potential sources of bias, Outcome 2 ADHD symptom severity: patient rated.

Navigate to figure in ReviewOpen in new tab

Analysis 9.3

Comparison 9 Sensitivity analysis: other potential sources of bias, Outcome 3 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Analysis 10.1

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 1 ADHD symptom severity: clinician‐rated.

Navigate to figure in ReviewOpen in new tab

Analysis 10.2

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 2 ADHD symptom severity: patient‐rated.

Navigate to figure in ReviewOpen in new tab

Analysis 10.3

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 3 Clinical impression of severity at study end.

Navigate to figure in ReviewOpen in new tab

Analysis 10.4

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 4 Clinical impression of improvement at study end.

Navigate to figure in ReviewOpen in new tab

Analysis 10.5

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 5 Proportion of participants achieving a reduction ≥ 30% in severity of ADHD symptoms.

Navigate to figure in ReviewOpen in new tab

Analysis 10.6

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 6 Proportion of participants achieving a CGI‐Improvement score of 1 or 2.

Navigate to figure in ReviewOpen in new tab

Analysis 10.7

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 7 Proportion of participants achieving a reduction ≥ 30% in severity of ADHD symptoms and a CGI‐Improvement score of 1 or 2.

Navigate to figure in ReviewOpen in new tab

Analysis 10.8

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 8 Global functioning.

Navigate to figure in ReviewOpen in new tab

Analysis 10.9

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 9 Depressive symptoms.

Navigate to figure in ReviewOpen in new tab

Analysis 10.10

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 10 Anxiety symptoms.

Navigate to figure in ReviewOpen in new tab

Analysis 10.11

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 11 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Analysis 10.12

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 12 Proportion of participants withdrawn owing to any cardiovascular adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 10.13

Comparison 10 Sensitivity analysis: fixed‐effect model, Outcome 13 Proportion of participants withdrawn owing to any adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 11.1

Comparison 11 Post hoc sensitivity analysis 1: calculation of effect sizes using correlation coefficient from Taylor 2000, Outcome 1 ADHD symptom severity: clinician rated.

Navigate to figure in ReviewOpen in new tab

Analysis 11.2

Comparison 11 Post hoc sensitivity analysis 1: calculation of effect sizes using correlation coefficient from Taylor 2000, Outcome 2 ADHD symptom severity: patient rated.

Navigate to figure in ReviewOpen in new tab

Analysis 12.1

Comparison 12 Post hoc sensitivity analysis 2: pooled risk difference for proportion of participants withdrawn owing to cardiovascular adverse events and any adverse event, Outcome 1 Proportion of participants withdrawn owing to any cardiovascular adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 12.2

Comparison 12 Post hoc sensitivity analysis 2: pooled risk difference for proportion of participants withdrawn owing to cardiovascular adverse events and any adverse event, Outcome 2 Proportion of participants withdrawn owing to any adverse event.

Navigate to figure in ReviewOpen in new tab

Analysis 13.1

Comparison 13 Post hoc sensitivity analysis 3: exclusion of cross‐over study, Outcome 1 ADHD symptom severity: clinician rated.

Navigate to figure in ReviewOpen in new tab

Analysis 14.1

Comparison 14 Amphetamines vs guanfacine for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 1 ADHD symptom severity: patient rated.

Navigate to figure in ReviewOpen in new tab

Analysis 15.1

Comparison 15 Amphetamines vs modafinil for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 1 ADHD symptom severity: patient rated.

Navigate to figure in ReviewOpen in new tab

Analysis 16.1

Comparison 16 Amphetamines vs paroxetine for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 1 ADHD symptom severity: clinician rated.

Navigate to figure in ReviewOpen in new tab

Analysis 16.2

Comparison 16 Amphetamines vs paroxetine for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 2 Proportion of participants achieving a CGI‐Improvement score of 1 or 2.

Navigate to figure in ReviewOpen in new tab

Analysis 16.3

Comparison 16 Amphetamines vs paroxetine for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 3 Global functioning.

Navigate to figure in ReviewOpen in new tab

Analysis 16.4

Comparison 16 Amphetamines vs paroxetine for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 4 Depressive symptoms.

Navigate to figure in ReviewOpen in new tab

Analysis 16.5

Comparison 16 Amphetamines vs paroxetine for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 5 Anxiety symptoms.

Navigate to figure in ReviewOpen in new tab

Analysis 16.6

Comparison 16 Amphetamines vs paroxetine for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 6 Retention in treatment.

Navigate to figure in ReviewOpen in new tab

Analysis 16.7

Comparison 16 Amphetamines vs paroxetine for adult attention deficit hyperactivity disorder (ADHD) in adults, Outcome 7 Proportion of participants withdrawn owing to any adverse event.

Navigate to figure in ReviewOpen in new tab

Summary of findings for the main comparison. Amphetamines versus placebo for attention deficit hyperactivity disorder (ADHD) in adults

Amphetamines versus placebo for attention deficit hyperactivity disorder (ADHD) in adults
Patient or population: adult patients with attention deficit hyperactivity disorder (ADHD) Settings: outpatients Intervention: amphetamines Comparison: placebo
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	Number of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Placebo	Amphetamines
*Dexamphetamine*
ADHD symptom severity: clinician rated Assessed with ADHD‐RS‐IV Follow‐up: post intervention (mean 20 weeks)	‐	Mean clinician‐rated ADHD symptom severity score in the intervention groups was 0.24 standard deviations lower (0.80 lower to 0.32 higher)	‐	49 (1 study)	⊕⊝⊝⊝ Very low^a,b,c	An SMD of 0.24 can be considered a small effect size.
ADHD symptom severity: patient rated Assessed with DSM‐IV ADHD Behavior Checklist for Adults Follow‐up: post intervention (mean 2 weeks)	‐	Mean patient‐rated ADHD symptom severity score in the intervention groups was 0.77 standard deviations lower (1.14 lower to 0.4 lower)	‐	35 (2 studies)	⊕⊕⊝⊝ Low^a,c,d	An SMD of 0.77 can be considered a medium effect size.
*Lisdexamfetamine*
ADHD symptom severity: clinician rated Assessed with ADHD‐RS‐IV and CAARS Follow‐up: post intervention (1‐10 weeks)	‐	Mean clinician‐rated ADHD symptom severity score in the intervention groups was 1.06 standard deviations lower (1.26 lower to 0.85 lower)	‐	896 (7 studies)	⊕⊕⊝⊝ Low^c,e,f,g	An SMD of 1.06 can be considered a large effect size.
ADHD symptom severity: patient rated Assessed with CAARS Follow‐up: post intervention (1‐4 weeks)	‐	Mean patient‐rated ADHD symptom severity score in the intervention groups was 0.33 standard deviations lower (0.65 lower to 0.01 lower)	‐	67 (3 studies)	⊕⊕⊝⊝ Low^c,d,h	An SMD of 0.33 can be considered a medium effect size.
*Mixed amphetamine salts*
ADHD symptom severity: clinician rated Assessed with ADHD‐RS‐IV and AISRS Follow‐up: post intervention (3‐13 weeks)	‐	Mean clinician‐rated ADHD symptom severity score in the intervention groups was 0.80 standard deviations lower (0.93 lower to 0.66 lower)	‐	1083 (5 studies)	⊕⊕⊝⊝ Low^c,e	An SMD of 0.8 can be considered a small effect size.
ADHD symptom severity: patient rated Assessed with CAARS Follow‐up: post intervention (mean 1 week)	‐	Mean patient‐rated ADHD symptom severity score in the intervention groups was 0.45 standard deviations lower (1.02 lower to 0.12 higher)	‐	18 (1 study)	⊕⊝⊝⊝ Very low^b,c,h	An SMD of 0.45 can be considered a medium effect size.
*All amphetamines*
Retention in treatment Assessed with the proportion of randomised participants that completed the study Follow‐up: post intervention (1‐20 weeks)	Study population		RR 1.06 (0.99 to 1.13)	2323 (17 studies)	⊕⊕⊝⊝ Low^a,i	‐
	708 per 1000	750 per 1000 (701 to 800)
	Moderate
	800 per 1000	848 per 1000 (792 to 904)
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ADHD: attention deficit hyperactivity disorder; ADHD‐RS‐IV: Attention Deficit Hyperactivity Disorder Rating Scale, Fourth Version; AISRS: Adult Attention Deficity Hyperactivity Disorder Investigator Rating Scale; CAARS: Conners' Adult Attention Deficit Hyperactivity Disorder Rating Scales;CI: confidence interval; DSM‐IV:Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition;SMD: standardised mean difference.
GRADE Working Group grades of evidence. High quality: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate quality: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low quality: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low quality: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aThe certainty of the evidence was downgraded by one level owing to unclear risk of detection and performance bias because it is unclear whether blinding can be achieved in placebo‐controlled studies given the powerful behavioural effects of amphetamines. ^bThe certainty of the evidence was downgraded by two levels owing to imprecision because the 95% CI is wide, indicating that the intervention effect for this outcome can range from a small, worsening effect to a large benefit. ^cThe statistical power to detect publication bias for this comparison in this review is low. ^dThe certainty of the evidence was downgraded by one level owing to imprecision because the 95% CI is rather wide, indicating that the intervention effect for this outcome can range from a moderate to a large benefit. ^eThe certainty of the evidence was downgraded by two levels owing to unclear risk of detection and performance bias (it is unclear whether blinding can be achieved in placebo‐controlled studies given the powerful behavioural effects of amphetamines), high risk of attrition bias (large proportion of participants discontinued treatment or differences between study groups in discontinuation rates), and high risk of other bias (such as the possibility of carry‐over effect in cross‐over studies without a washout phase). ^fThe certainty of the evidence was downgraded by one level owing to moderate statistical heterogeneity. ^gThe certainty of the evidence was upgraded by one level because a large and precise effect size was observed. ^hThe certainty of the evidence was downgraded by one level owing to unclear risk of detection and performance bias (it is unclear whether blinding can be achieved in placebo‐controlled studies given the powerful behavioural effects of amphetamines) and high risk of other bias (such as the possibility of carry‐over effect in cross‐over studies without a washout phase). ⁱThe certainty of the evidence was downgraded by one level owing to inconsistency (this comparison includes three different types of amphetamines at a wide range of doses, and the analysis showed moderate heterogeneity).

Summary of findings for the main comparison. Amphetamines versus placebo for attention deficit hyperactivity disorder (ADHD) in adults

Navigate to table in Review

Table 1. Participants' baseline characteristics

Characteristic	Descriptive statistics	N studies (N patients)
Gender: male	N = 1435 (57.2%)	19 (2507)
Age	Mean = 35.3 (range = 20.2 to 41.2) years	19 (2507)
Race: Caucasian	N = 2006 (84.5%)	15 (2373)
Combined ADHD	N = 1341 (78.8%)	11 (1701)
Predominantly inattentive ADHD	N = 344 (20.2%)
Predominantly hyperactive/impulsive ADHD	N = 28 (1.6%)
Comorbid SUD as inclusion criterion	N = 158 (6.3%)	19 (2507)
Comorbid depressive disorders as inclusion criteria	N = 0	19 (2507)
Comorbid anxiety disorders as inclusion criteria	N = 0	19 (2507)
Treated previously with stimulants	N = 306 (41.1%)	8 (744)
ADHD: attention deficit hyperactivity disorder. N: number. SUD: substance use disorder.

Table 1. Participants' baseline characteristics

Navigate to table in Review

Comparison 1. Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician‐rated Show forest plot	13		Std. Mean Difference (Random, 95% CI)	‐0.90 [‐1.04, ‐0.75]

2 ADHD symptom severity: patient‐rated Show forest plot	6		Std. Mean Difference (Random, 95% CI)	‐0.51 [‐0.75, ‐0.28]

3 Clinical impression of severity at study end Show forest plot	2	78	Std. Mean Difference (IV, Random, 95% CI)	‐1.09 [‐1.57, ‐0.61]

4 Clinical impression of improvement at study end Show forest plot	1	263	Std. Mean Difference (IV, Random, 95% CI)	‐0.75 [‐1.01, ‐0.48]

5 Proportion of participants achieving a reduction ≥ 30% in severity of ADHD symptoms Show forest plot	2	381	Risk Ratio (M‐H, Random, 95% CI)	1.52 [1.19, 1.95]

6 Proportion of participants achieving a CGI‐Improvement score of 1 or 2 Show forest plot	8	1707	Risk Ratio (M‐H, Random, 95% CI)	2.47 [2.10, 2.90]

7 Proportion of participants achieving a reduction ≥ 30% in severity of ADHD symptoms and a CGI‐Improvement score of 1 or 2 Show forest plot	1	61	Risk Ratio (M‐H, Random, 95% CI)	2.54 [1.34, 4.82]

8 Global functioning Show forest plot	2	110	Std. Mean Difference (IV, Random, 95% CI)	0.54 [‐0.34, 1.42]

9 Depressive symptoms Show forest plot	2	110	Std. Mean Difference (IV, Random, 95% CI)	0.16 [‐0.22, 0.53]

10 Anxiety symptoms Show forest plot	2	110	Std. Mean Difference (IV, Random, 95% CI)	0.13 [‐0.24, 0.51]

11 Retention in treatment Show forest plot	17	2323	Risk Ratio (M‐H, Random, 95% CI)	1.06 [0.99, 1.13]

11.1 Dexamphetamine	1	49	Risk Ratio (M‐H, Random, 95% CI)	0.79 [0.54, 1.17]
11.2 Lisdexamfetamine	8	873	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.94, 1.08]
11.3 Mixed amphetamine salts	8	1401	Risk Ratio (M‐H, Random, 95% CI)	1.14 [1.02, 1.28]
12 Proportion of participants withdrawn owing to any cardiovascular adverse event Show forest plot	3	699	Risk Ratio (M‐H, Random, 95% CI)	2.18 [0.39, 12.04]

13 Proportion of participants withdrawn owing to any adverse event Show forest plot	17	2409	Risk Ratio (M‐H, Random, 95% CI)	2.69 [1.64, 4.42]

13.1 Dexamphetamine	1	49	Risk Ratio (M‐H, Random, 95% CI)	1.70 [0.31, 9.27]
13.2 Lisdexamfetamine	9	989	Risk Ratio (M‐H, Random, 95% CI)	1.79 [0.72, 4.42]
13.3 Mixed amphetamine salts	7	1371	Risk Ratio (M‐H, Random, 95% CI)	3.50 [1.86, 6.59]

Comparison 1. Amphetamines vs placebo for adult attention deficit hyperactivity disorder (ADHD) in adults

Navigate to table in Review

Comparison 2. Subgroup analysis 1: comorbidity

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician‐rated Show forest plot	13		Std. Mean Difference (Random, 95% CI)	‐0.90 [‐1.04, ‐0.75]

1.1 With comorbidity	2		Std. Mean Difference (Random, 95% CI)	‐0.76 [‐1.11, ‐0.41]
1.2 Without comorbidity	11		Std. Mean Difference (Random, 95% CI)	‐0.91 [‐1.07, ‐0.76]
2 ADHD symptom severity: patient‐rated Show forest plot	6		Std. Mean Difference (Random, 95% CI)	‐0.51 [‐0.75, ‐0.28]

2.1 With comorbidity	1		Std. Mean Difference (Random, 95% CI)	‐0.66 [‐1.44, 0.12]
2.2 Without comorbidity	5		Std. Mean Difference (Random, 95% CI)	‐0.50 [‐0.77, ‐0.23]
3 Retention in treatment Show forest plot	17	2323	Risk Ratio (M‐H, Random, 95% CI)	1.06 [0.99, 1.13]

3.1 With comorbidity	2	158	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.77, 1.33]
3.2 Without comorbidity	15	2165	Risk Ratio (M‐H, Random, 95% CI)	1.07 [0.99, 1.15]
4 Proportion of patients withdrawn owing to any adverse event Show forest plot	17	2409	Risk Ratio (M‐H, Random, 95% CI)	2.69 [1.64, 4.42]

4.1 With comorbidity	2	158	Risk Ratio (M‐H, Random, 95% CI)	2.67 [0.12, 60.93]
4.2 Without comorbidity	15	2251	Risk Ratio (M‐H, Random, 95% CI)	2.69 [1.63, 4.45]

Comparison 2. Subgroup analysis 1: comorbidity

Navigate to table in Review

Comparison 3. Subgroup analysis 2: type of amphetamine

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician‐rated Show forest plot	13		Std. Mean Difference (Random, 95% CI)	‐0.90 [‐1.04, ‐0.75]

1.1 Dexamphetamine	1		Std. Mean Difference (Random, 95% CI)	‐0.24 [‐0.80, 0.32]
1.2 Lisdexamfetamine	7		Std. Mean Difference (Random, 95% CI)	‐1.06 [‐1.26, ‐0.85]
1.3 Mixed amphetamine salts	5		Std. Mean Difference (Random, 95% CI)	‐0.80 [‐0.93, ‐0.66]
2 ADHD symptom severity: patient‐rated Show forest plot	6		Std. Mean Difference (Random, 95% CI)	‐0.51 [‐0.75, ‐0.28]

2.1 Dexamphetamine	2		Std. Mean Difference (Random, 95% CI)	‐0.77 [‐1.14, ‐0.40]
2.2 Lisdexamfetamine	3		Std. Mean Difference (Random, 95% CI)	‐0.33 [‐0.65, ‐0.01]
2.3 Mixed amphetamine salts	1		Std. Mean Difference (Random, 95% CI)	‐0.45 [‐1.02, 0.12]
3 Retention in treatment Show forest plot	17	2323	Risk Ratio (M‐H, Random, 95% CI)	1.06 [0.99, 1.13]

3.1 Dexamphetamine	1	49	Risk Ratio (M‐H, Random, 95% CI)	0.79 [0.54, 1.17]
3.2 Lisdexamfetamine	8	873	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.94, 1.08]
3.3 Mixed amphetamine salts	8	1401	Risk Ratio (M‐H, Random, 95% CI)	1.14 [1.02, 1.28]
4 Proportion of participants withdrawn owing to any adverse event Show forest plot	17	2409	Risk Ratio (M‐H, Random, 95% CI)	2.69 [1.64, 4.42]

4.1 Dexamphetamine	1	49	Risk Ratio (M‐H, Random, 95% CI)	1.70 [0.31, 9.27]
4.2 Lisdexamfetamine	9	989	Risk Ratio (M‐H, Random, 95% CI)	1.79 [0.72, 4.42]
4.3 Mixed amphetamine salts	7	1371	Risk Ratio (M‐H, Random, 95% CI)	3.50 [1.86, 6.59]

Comparison 3. Subgroup analysis 2: type of amphetamine

Navigate to table in Review

Comparison 4. Subgroup analysis 3: dose of dexamphetamine

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: patient rated Show forest plot	2		Std. Mean Difference (Random, 95% CI)	‐0.77 [‐1.14, ‐0.40]

1.1 Lower dose	1		Std. Mean Difference (Random, 95% CI)	‐0.55 [‐1.10, ‐0.00]
1.2 Higher dose	1		Std. Mean Difference (Random, 95% CI)	‐0.93 [‐1.40, ‐0.46]

Comparison 4. Subgroup analysis 3: dose of dexamphetamine

Navigate to table in Review

Comparison 5. Subgroup analysis 3: dose of lisdexamfetamine

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician rated Show forest plot	6		Std. Mean Difference (Random, 95% CI)	‐1.02 [‐1.22, ‐0.82]

1.1 Lower dose	2		Std. Mean Difference (Random, 95% CI)	‐0.98 [‐1.41, ‐0.55]
1.2 Higher dose	5		Std. Mean Difference (Random, 95% CI)	‐1.04 [‐1.31, ‐0.78]
2 ADHD symptom severity: patient rated Show forest plot	3		Std. Mean Difference (Random, 95% CI)	‐0.35 [‐0.61, ‐0.10]

2.1 Lower dose	1		Std. Mean Difference (Random, 95% CI)	‐0.33 [‐0.78, 0.12]
2.2 Higher dose	3		Std. Mean Difference (Random, 95% CI)	‐0.36 [‐0.67, ‐0.05]
3 Retention in treatment Show forest plot	5	712	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.93, 1.08]

3.1 Lower dose	2	322	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.89, 1.14]
3.2 Higher dose	4	390	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.89, 1.14]
4 Proportion of participants withdrawn owing to any adverse event Show forest plot	6	828	Risk Ratio (M‐H, Random, 95% CI)	2.72 [1.09, 6.75]

4.1 Lower dose	3	335	Risk Ratio (M‐H, Random, 95% CI)	2.98 [0.56, 15.72]
4.2 Higher dose	4	493	Risk Ratio (M‐H, Random, 95% CI)	2.61 [0.88, 7.75]

Comparison 5. Subgroup analysis 3: dose of lisdexamfetamine

Navigate to table in Review

Comparison 6. Subgroup analysis 3: dose of mixed amphetamine salts

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician rated Show forest plot	5		Std. Mean Difference (Random, 95% CI)	‐0.81 [‐0.94, ‐0.69]

1.1 Lower dose	3		Std. Mean Difference (Random, 95% CI)	‐0.78 [‐0.94, ‐0.63]
1.2 Higher dose	3		Std. Mean Difference (Random, 95% CI)	‐0.86 [‐1.06, ‐0.66]
2 Retention in treatment Show forest plot	8	1569	Risk Ratio (M‐H, Random, 95% CI)	1.16 [1.05, 1.28]

2.1 Lower dose (50 mg/d)	5	962	Risk Ratio (M‐H, Random, 95% CI)	1.13 [0.96, 1.32]
2.2 Higher dose (50 mg/d)	5	607	Risk Ratio (M‐H, Random, 95% CI)	1.21 [1.09, 1.35]
3 Proportion of participants withdrawn owing to any adverse event Show forest plot	7	1539	Risk Ratio (M‐H, Random, 95% CI)	3.73 [2.16, 6.44]

3.1 Lower dose	5	962	Risk Ratio (M‐H, Random, 95% CI)	3.59 [1.84, 7.00]
3.2 Higher dose	4	577	Risk Ratio (M‐H, Random, 95% CI)	4.03 [1.56, 10.42]

Comparison 6. Subgroup analysis 3: dose of mixed amphetamine salts

Navigate to table in Review

Comparison 7. Subgroup analysis 4: type of drug‐release formulation

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician rated Show forest plot	13		Std. Mean Difference (Random, 95% CI)	‐0.90 [‐1.04, ‐0.75]

1.1 Immediate‐release formulations	1		Std. Mean Difference (Random, 95% CI)	‐0.91 [‐1.38, ‐0.44]
1.2 Sustained‐release formulations	12		Std. Mean Difference (Random, 95% CI)	‐0.90 [‐1.05, ‐0.74]
2 ADHD symptom severity: patient rated Show forest plot	6		Std. Mean Difference (Random, 95% CI)	‐0.51 [‐0.75, ‐0.27]

2.1 Immediate‐release formulations	3		Std. Mean Difference (Random, 95% CI)	‐0.67 [‐0.98, ‐0.37]
2.2 Sustained‐release formulations	3		Std. Mean Difference (Random, 95% CI)	‐0.33 [‐0.65, ‐0.01]
3 Retention in treatment Show forest plot	17	2323	Risk Ratio (M‐H, Random, 95% CI)	1.06 [0.99, 1.13]

3.1 Immediate‐release formulations	2	41	Risk Ratio (M‐H, Random, 95% CI)	1.13 [0.91, 1.40]
3.2 Sustained‐release formulations	15	2282	Risk Ratio (M‐H, Random, 95% CI)	1.06 [0.98, 1.13]

Comparison 7. Subgroup analysis 4: type of drug‐release formulation

Navigate to table in Review

Comparison 8. Sensitivity analysis: incomplete subjective outcome data

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician rated Show forest plot	1		Std. Mean Difference (Random, 95% CI)	Totals not selected

2 ADHD symptom severity: patient rated Show forest plot	2		Std. Mean Difference (Random, 95% CI)	‐0.77 [‐1.14, ‐0.40]

Comparison 8. Sensitivity analysis: incomplete subjective outcome data

Navigate to table in Review

Comparison 9. Sensitivity analysis: other potential sources of bias

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician rated Show forest plot	9		Std. Mean Difference (Random, 95% CI)	‐0.84 [‐1.02, ‐0.66]

2 ADHD symptom severity: patient rated Show forest plot	3		Std. Mean Difference (Random, 95% CI)	‐0.75 [‐1.07, ‐0.43]

3 Retention in treatment Show forest plot	9	1661	Risk Ratio (M‐H, Random, 95% CI)	1.04 [0.96, 1.13]

Comparison 9. Sensitivity analysis: other potential sources of bias

Navigate to table in Review

Comparison 10. Sensitivity analysis: fixed‐effect model

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician‐rated Show forest plot	13		Std. Mean Difference (Fixed, 95% CI)	‐0.89 [‐0.98, ‐0.79]

1.1 Dexamphetamine	1		Std. Mean Difference (Fixed, 95% CI)	‐0.24 [‐0.80, 0.32]
1.2 Lisdexamfetamine	7		Std. Mean Difference (Fixed, 95% CI)	‐1.04 [‐1.19, ‐0.90]
1.3 Mixed amphetamine salts	5		Std. Mean Difference (Fixed, 95% CI)	‐0.80 [‐0.93, ‐0.66]
2 ADHD symptom severity: patient‐rated Show forest plot	6		Std. Mean Difference (Fixed, 95% CI)	‐0.51 [‐0.73, ‐0.29]

2.1 Dexamphetamine	2		Std. Mean Difference (Fixed, 95% CI)	‐0.77 [‐1.13, ‐0.41]
2.2 Lisdexamfetamine	3		Std. Mean Difference (Fixed, 95% CI)	‐0.33 [‐0.65, ‐0.01]
2.3 Mixed amphetamine salts	1		Std. Mean Difference (Fixed, 95% CI)	‐0.45 [‐1.02, 0.12]
3 Clinical impression of severity at study end Show forest plot	2	78	Std. Mean Difference (IV, Fixed, 95% CI)	‐1.09 [‐1.57, ‐0.61]

4 Clinical impression of improvement at study end Show forest plot	1	263	Std. Mean Difference (IV, Fixed, 95% CI)	‐0.75 [‐1.01, ‐0.48]

5 Proportion of participants achieving a reduction ≥ 30% in severity of ADHD symptoms Show forest plot	2	381	Risk Ratio (M‐H, Fixed, 95% CI)	1.52 [1.18, 1.95]

6 Proportion of participants achieving a CGI‐Improvement score of 1 or 2 Show forest plot	8	1707	Risk Ratio (M‐H, Fixed, 95% CI)	2.52 [2.14, 2.97]

7 Proportion of participants achieving a reduction ≥ 30% in severity of ADHD symptoms and a CGI‐Improvement score of 1 or 2 Show forest plot	1	61	Risk Ratio (M‐H, Fixed, 95% CI)	2.54 [1.34, 4.82]

8 Global functioning Show forest plot	2	110	Std. Mean Difference (IV, Fixed, 95% CI)	0.56 [0.17, 0.95]

9 Depressive symptoms Show forest plot	2	110	Std. Mean Difference (IV, Fixed, 95% CI)	0.16 [‐0.22, 0.53]

10 Anxiety symptoms Show forest plot	2	110	Std. Mean Difference (IV, Fixed, 95% CI)	0.13 [‐0.24, 0.51]

11 Retention in treatment Show forest plot	17	2323	Risk Ratio (M‐H, Fixed, 95% CI)	1.10 [1.04, 1.16]

11.1 Dexamphetamine	1	49	Risk Ratio (M‐H, Fixed, 95% CI)	0.79 [0.54, 1.17]
11.2 Lisdexamfetamine	8	873	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.96, 1.11]
11.3 Mixed amphetamine salts	8	1401	Risk Ratio (M‐H, Fixed, 95% CI)	1.15 [1.07, 1.24]
12 Proportion of participants withdrawn owing to any cardiovascular adverse event Show forest plot	2	675	Risk Ratio (M‐H, Fixed, 95% CI)	2.51 [0.32, 19.54]

13 Proportion of participants withdrawn owing to any adverse event Show forest plot	17	2409	Risk Ratio (M‐H, Fixed, 95% CI)	2.99 [1.86, 4.83]

13.1 Dexamphetamine	1	49	Risk Ratio (M‐H, Fixed, 95% CI)	1.70 [0.31, 9.27]
13.2 Lisdexamfetamine	9	989	Risk Ratio (M‐H, Fixed, 95% CI)	1.77 [0.78, 4.02]
13.3 Mixed amphetamine salts	7	1371	Risk Ratio (M‐H, Fixed, 95% CI)	4.05 [2.14, 7.67]

Comparison 10. Sensitivity analysis: fixed‐effect model

Navigate to table in Review

Comparison 11. Post hoc sensitivity analysis 1: calculation of effect sizes using correlation coefficient from Taylor 2000

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician rated Show forest plot	13		Std. Mean Difference (Random, 95% CI)	‐0.90 [‐1.05, ‐0.76]

2 ADHD symptom severity: patient rated Show forest plot	6		Std. Mean Difference (Random, 95% CI)	‐0.47 [‐0.69, ‐0.25]

Comparison 11. Post hoc sensitivity analysis 1: calculation of effect sizes using correlation coefficient from Taylor 2000

Navigate to table in Review

Comparison 12. Post hoc sensitivity analysis 2: pooled risk difference for proportion of participants withdrawn owing to cardiovascular adverse events and any adverse event

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Proportion of participants withdrawn owing to any cardiovascular adverse event Show forest plot	3	699	Risk Difference (M‐H, Random, 95% CI)	0.02 [‐0.00, 0.04]

2 Proportion of participants withdrawn owing to any adverse event Show forest plot	17	2409	Risk Difference (M‐H, Random, 95% CI)	0.04 [0.01, 0.06]

2.1 Dexamphetamine	1	49	Risk Difference (M‐H, Random, 95% CI)	0.05 [‐0.12, 0.23]
2.2 Lisdexamfetamine	9	989	Risk Difference (M‐H, Random, 95% CI)	0.01 [‐0.02, 0.04]
2.3 Mixed amphetamine salts	7	1371	Risk Difference (M‐H, Random, 95% CI)	0.06 [0.02, 0.10]

Comparison 12. Post hoc sensitivity analysis 2: pooled risk difference for proportion of participants withdrawn owing to cardiovascular adverse events and any adverse event

Navigate to table in Review

Comparison 13. Post hoc sensitivity analysis 3: exclusion of cross‐over study

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician rated Show forest plot	12		Std. Mean Difference (Random, 95% CI)	‐0.90 [‐1.05, ‐0.74]

Comparison 13. Post hoc sensitivity analysis 3: exclusion of cross‐over study

Navigate to table in Review

Comparison 14. Amphetamines vs guanfacine for adult attention deficit hyperactivity disorder (ADHD) in adults

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: patient rated Show forest plot	1		Std. Mean Difference (Random, 95% CI)	Totals not selected

Comparison 14. Amphetamines vs guanfacine for adult attention deficit hyperactivity disorder (ADHD) in adults

Navigate to table in Review

Comparison 15. Amphetamines vs modafinil for adult attention deficit hyperactivity disorder (ADHD) in adults

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: patient rated Show forest plot	1		Std. Mean Difference (Random, 95% CI)	Totals not selected

Comparison 15. Amphetamines vs modafinil for adult attention deficit hyperactivity disorder (ADHD) in adults

Navigate to table in Review

Comparison 16. Amphetamines vs paroxetine for adult attention deficit hyperactivity disorder (ADHD) in adults

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ADHD symptom severity: clinician rated Show forest plot	1		Std. Mean Difference (IV, Random, 95% CI)	Totals not selected

2 Proportion of participants achieving a CGI‐Improvement score of 1 or 2 Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

3 Global functioning Show forest plot	1		Std. Mean Difference (IV, Random, 95% CI)	Totals not selected

4 Depressive symptoms Show forest plot	1		Std. Mean Difference (IV, Random, 95% CI)	Totals not selected

5 Anxiety symptoms Show forest plot	1		Std. Mean Difference (IV, Random, 95% CI)	Totals not selected

6 Retention in treatment Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

7 Proportion of participants withdrawn owing to any adverse event Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

Comparison 16. Amphetamines vs paroxetine for adult attention deficit hyperactivity disorder (ADHD) in adults

Navigate to table in Review

Cochrane Review language

Website language

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

PICOs

PICOs

Population

Intervention

Comparison

Outcome

Plain language summary

Amphetamines for attention deficit hyperactivity disorder in adults

Visual summary

Authors' conclusions

Implications for practice

Implications for research

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Continuous outcome data

Dichotomous outcome data

Unit of analysis issues

Cross‐over trials

Multiple treatment groups

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Summarising the quality of the evidence

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Results

Description of studies

Results of the search

Included studies

Study design

Setting

Participants

Interventions

Duration

Sponsorship

Excluded studies

Ongoing studies

Risk of bias in included studies

Allocation

Sequence generation

Allocation concealment

Blinding

Blinding of participants and personnel (performance bias)

Blinding of outcome assessment (detection bias)

Incomplete outcome data

Selective reporting

Other potential sources of bias

Summary