Metilfenidat s produljenim otpuštanjem za poremećaj s nedostatkom pažnje i hiperaktivnošću (ADHD) u odraslih
Abstract
Background
Attention deficit hyperactivity disorder (ADHD) is a psychiatric diagnosis increasingly used in adults. The recommended first‐line pharmacological treatment is central nervous system (CNS) stimulants, such as methylphenidate, but uncertainty remains about its benefits and harms.
Objectives
To assess the beneficial and harmful effects of extended‐release formulations of methylphenidate in adults diagnosed with ADHD.
Search methods
We searched CENTRAL, MEDLINE, Embase, nine other databases and four clinical trial registries up to February 2021. We searched 12 drug regulatory databases for clinical trial data up to 13 May 2020. In addition, we cross‐referenced all available trial identifiers, handsearched reference lists, searched pharmaceutical company databases, and contacted trial authors.
Selection criteria
Randomised, double‐blind, parallel‐group trials comparing extended‐release methylphenidate formulations at any dose versus placebo and other ADHD medications in adults diagnosed with ADHD.
Data collection and analysis
Two review authors independently extracted data. We assessed dichotomous outcomes as risk ratios (RRs), and rating scales and continuous outcomes as mean differences (MDs) or standardised mean differences (SMDs). We used the Cochrane risk of bias tool to assess risks of bias, and GRADE to assess the certainty of the evidence. We meta‐analysed the data using a random‐effects model. We assessed three design characteristics that may impair the trial results' 'generalisability'; exclusion of participants with psychiatric comorbidity; responder selection based on previous experience with CNS stimulants; and risk of withdrawal effects. Our prespecified primary outcomes were functional outcomes, self‐rated ADHD symptoms, and serious adverse events. Our secondary outcomes included quality of life, ADHD symptoms rated by investigators and by peers such as family members, cardiovascular variables, severe psychiatric adverse events, and other adverse events.
Main results
We included 24 trials (5066 participants), of which 21 reported outcome data for this review. We also identified one ongoing study. We included documents from six drug regulatory agencies covering eight trials. Twenty‐one trials had an outpatient setting and three were conducted in prisons. They were primarily conducted in North America and Europe. The median participant age was 36 years. Twelve trials (76% of participants) were industry‐sponsored, four (14% of participants) were publicly funded with industry involvement, seven (10% of participants) were publicly funded, and one had unclear funding. The median trial duration was eight weeks. One trial was rated at overall unclear risk of bias and 20 trials were rated at overall high risk of bias, primarily due to unclear blinding of participants and investigators, attrition bias, and selective outcome reporting. All trials were impaired in at least one of the three design characteristics related to 'generalisability'; for example, they excluded participants with psychiatric comorbidity such as depression or anxiety, or included participants only with a previous positive response to methylphenidate, or similar drugs. This may limit the trials’ usefulness for clinical practice, as they may overestimate the benefits and underestimate the harms.
Extended‐release methylphenidate versus placebo (up to 26 weeks)
For the primary outcomes, we found very low‐certainty evidence that methylphenidate had no effect on ‘days missed at work’ at 13‐week follow‐up (mean difference (MD) −0.15 days, 95% confidence interval (CI) −2.11 to 1.81; 1 trial, 409 participants) or serious adverse events (risk ratio (RR) 1.43, CI 95% CI 0.85 to 2.43; 14 trials, 4078 participants), whereas methylphenidate improved self‐rated ADHD symptoms (small‐to‐moderate effect; SMD −0.37, 95% CI −0.43 to −0.30; 16 trials, 3799 participants). For secondary outcomes, we found very low‐certainty evidence that methylphenidate improved self‐rated quality of life (small effect; SMD −0.15, 95% CI −0.25 to −0.05; 6 trials, 1888 participants), investigator‐rated ADHD symptoms (small‐to‐moderate effect; SMD −0.42, 95% CI −0.49 to −0.36; 18 trials, 4183 participants), ADHD symptoms rated by peers such as family members (small‐to‐moderate effect; SMD −0.31, 95% CI −0.48 to −0.14; 3 trials, 1005 participants), and increased the risk of experiencing any adverse event (RR 1.27, 95% CI 1.19 to 1.37; 14 trials, 4214 participants). We rated the certainty of the evidence as ‘very low’ for all outcomes, primarily due to high risk of bias and 'indirectness of the evidence'. One trial (419 participants) had follow‐up at 52 weeks and two trials (314 participants) included active comparators, hence long‐term and comparative evidence is limited.
Authors' conclusions
We found very low‐certainty evidence that extended‐release methylphenidate compared to placebo improved ADHD symptoms (small‐to‐moderate effects) measured on rating scales reported by participants, investigators, and peers such as family members. Methylphenidate had no effect on 'days missed at work' or serious adverse events, the effect on quality of life was small, and it increased the risk of several adverse effects. We rated the certainty of the evidence as ‘very low’ for all outcomes, due to high risk of bias, short trial durations, and limitations to the generalisability of the results. The benefits and harms of extended‐release methylphenidate therefore remain uncertain.
PICO
Laički sažetak
Metilfenidat s produljenim otpuštanjem za poremećaj s nedostatkom pažnje i hiperaktivnošću (ADHD) u odraslih
Cilj sustavnog pregleda
Ovaj sustavni pregled istražio je pozitivne i negativne učinke metilfenidata i usporedio ih s placebom ili drugim lijekovima kod odraslih osoba kojima je dijagnosticiran poremećaj s nedostatkom pažnje i hiperaktivnošću (engl. attention deficit hyperactivity disorder, ADHD). Posebna pažnja bila je usmjerena na učinke na svakodnevno funkcioniranje, kao što je broj dana izostanaka s posla i učinke na simptome ADHD‐a i kvalitetu života koje procjenjuju sami pacijenti, ali i liječnik ili istraživač.
Dosadašnje spoznaje
ADHD je psihijatrijska dijagnoza koju karakteriziraju poteškoće s koncentracijom, hiperaktivnost i impulzivno ponašanje, koje često utječu na društveni život, kvalitetu rada na poslu i održavanje osobnih odnosa. Sve je više odraslih osoba dijagnosticirano s ADHD‐om i uzima lijekove. Liječenje ADHD‐a treba se sastojati od niza različitih pristupa, uključujući psihoterapiju, socijalne intervencije, druge tipove nemedicinskih oblika liječenja i farmakološke lijekove. Metilfenidat i amfetamin se preporučuju kao lijekovi izbora za ADHD. Metilfenidat je lijek koji povećava aktivnost središnjeg živčanog sustava (SŽS). Ovaj sustavni pregled fokusiran je na metilfenidat koji se daje u obliku tablete koja se treba uzimati jednom ili dvaput dnevno, tzv. „formulacija s produljenim oslobađanjem“. Metilfenidat je testiran u mnogim kliničkim istraživanjima na odraslima s ADHD‐om, no ukupni pozitivni i negativni učinci su nesigurni zbog zabrinutosti oko dizajna istraživanja i opisivanja rezultata. Metilfenidat se koristi i kod djece i adolescenata s ADHD‐om.
Značajke istraživanja
U ovaj su sustavni pregled uključeni dokazi objavljeni do veljače 2021. godine. Autori su pronašli 24 istraživanja (5,066 ispitanika) s dijagnozom ADHD‐a, a pronašli su i jedno istraživanje koje je u tijeku. U svakom istraživanju ispitanici su podijeljeni u dvije skupine; jedna je skupina primala metilfenidat s produljenim oslobađanjem, a druga placebo, poznat i kao „šećerna pilula“. Dva su istraživanja uključivala još jedan lijek za ADHD kao treću skupinu. Medijan dobi (tj. srednja dob s popisa svih dobi poredanih od najmlađih do najstarijih) sudionika bila je 36 godina. Istraživanja su uglavnom provođena u Europi i Sjevernoj Americi na ambulantnim bolesnicima, što znači da sudionici nisu bili hospitalizirani tijekom ispitivanja. Istraživanja su trajala oko osam tjedana. Polovicu istraživanja sponzorirale su tvrtke koje također prodaju lijekove, što je moglo utjecati na način na koji su ta istraživanja provedena.
Ključni rezultati
U usporedbi s placebom, metilfenidat u tableti s produljenim oslobađanjem smanjio je ozbiljnost simptoma ADHD‐a kada su ih ocijenili sudionici, istraživači, kao i članovi obitelji ili supružnici. Metilfenidat nije smanjio broj dana izostajanja s posla, a učinak na kvalitetu života koji su ocijenili sudionici bio je neznatan. Ispitivanja nisu otkrila povećani rizik od ozbiljnih štetnih događaja, ali metilfenidat je povećao ukupan rizik od bilo kakvih štetnih događaja.
Kvaliteta dokaza
Pouzdanost ovih dokaza ocijenjena je kao „vrlo niska“, što znači da još uvijek nije sigurno nadmašuju li pozitivni učinci metilfenidata negativne učinke. Sva su istraživanja bila pod visokim rizikom od pristranosti u načinu na koji su osmišljena i provedena, što je moglo utjecati na rezultate (učiniti ih pristranima). Prema tome, prijavljeni pozitivni i negativni učinci u ispitivanjima nisu pouzdani i mogli bi se bitno promijeniti ako se provedu nova ispitivanja. Također, ograničeno je znanje o dugoročnim učincima metilfenidata, koji se često uzima godinama, jer je većina ispitivanja trajala oko osam tjedana, a samo je jedno ispitivanje trajalo 52 tjedna.
Autori su identificirali nekoliko problema s ispitivanjima: na primjer, sudionicima koji uz ADHD imaju i druge psihijatrijske dijagnoze često nije bilo dopušteno sudjelovati, iako ljudi s ADHD‐om često imaju više psihijatrijskih dijagnoza, poput anksioznosti i depresije. To ograničava korisnost ispitivanja za važne skupine odraslih s ADHD‐om. Također, često su smjele sudjelovati samo osobe koje su prethodno uzimale metilfenidat ili slične lijekove s dobrim rezultatima. To znači da bi istraživanja mogla podcijeniti ozbiljnost i broj ljudi koji imaju problema s uzimanjem lijekova. Kako bi se dobile pouzdanije procjene pozitivnih i negativnih učinaka metilfenidata, potrebno je provesti nova klinička ispitivanja, lišena problema istaknutih u ovom sustavnom pregledu.
Authors' conclusions
Summary of findings
| Extended‐release methylphenidate compared to placebo for ADHD | ||||||
| Patient or population: adults diagnosed with ADHD | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with extended‐release methylphenidate | |||||
| Functional outcome:days missed at work/ school/ training
Time of assessment: 13 weeks | The mean number of days missed at work, school or training in the control group was 2.4 | The mean number of days missed at work, school or training was 0.15 days fewer | ‐ | 409 | ⊕⊝⊝⊝ | ‐ |
| ADHD symptoms (self‐rated)
Time of assessment: median 8 weeks (range 2 ‐ 26) | ‐ | The mean ADHD score in the intervention groups was 0.37 SDlower | ‐ | 3799 | ⊕⊝⊝⊝ | An SMD of 0.37 may be considered a small‐to‐moderate effect size |
| Serious adverse events
Time of assessment: median 8 weeks (range 4 ‐ 52) | Study population | RR1.43 | 4078 | ⊕⊝⊝⊝ | ‐ | |
| 12 per 1000 | 17 per 1000 | |||||
| Quality of life (self‐rated)
Time of assessment: median 6 weeks (range 4 ‐ 52) | ‐ | The mean quality of life score in the intervention groups was 0.15 SD lower (0.25 lower to 0.05 lower) | ‐ | 1888 | ⊕⊝⊝⊝ | An SMD of 0.15 may be considered a small effect size |
| ADHD symptoms (investigator‐rated)
Time of assessment: median 8 weeks (range 2 ‐ 26) | ‐ | The mean investigator‐rated ADHD score in the intervention groups was 0.42 SD lower | ‐ | 4183 | ⊕⊝⊝⊝ | An SMD of 0.42 may be considered a small‐to‐moderate effect size |
| ADHD symptoms (peer‐rated)
Time of assessment: median 7.5 weeks (range 6 ‐ 9) | ‐ | The mean peer‐rated ADHD score in the intervention groups was 0.31 SD lower | ‐ | 1005 | ⊕⊝⊝⊝ | An SMD of 0.31 may be considered a small‐to‐moderate effect size |
| Adverse events (proportion experiencing any adverse event)
Time of assessment: median 8 weeks (range 4 ‐ 52) | Study population | RR1.27 | 4214 | ⊕⊝⊝⊝ | ‐ | |
| 641 per 1000 | 814 per 1000 | |||||
| *The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aWe downgraded two levels due to risk bias; primarily due to blinding issues caused by methylphenidate's effects, high attrition (dropout) rates, and selective reporting of the included trials. | ||||||
Background
Description of the condition
The diagnosis of attention deficit hyperactivity disorder (ADHD) for adults is described in the Diagnostic and Statistical Manual of Mental Disorders (DSM‐5; APA 2013) and in the International Classification of Disorders (ICD‐11; WHO 2019a). Individuals diagnosed with ADHD are characterised by having problems with concentration, hyperactivity, and impulsivity, with substantial impact on a person’s private, social, and professional life. The diagnostic criteria for ADHD have changed over time (Lange 2010; Doernberg 2016). The DSM‐5 criteria require the presence of five or more symptoms of inattention or hyperactivity/impulsivity for at least six months. The symptoms should affect two or more domains of social, academic, and work/school functioning, and they should have been present before the age of 12 years (APA 2013). The DSM‐5 criteria operate with three diagnostic subtypes: 'combined presentation', 'predominantly inattentive presentation', and 'predominantly hyperactive/impulsive presentation' (APA 2013). The ICD‐11 criteria are similar to the DSM system but do not define an age cut‐off for symptom occurrence, although they state that most experience symptoms prior to the age of 12. The ICD‐11 criteria define five subtypes, including three subtypes identical to those in DSM‐5, and two more, entitled 'other specified presentation' and 'unspecified presentation' (WHO 2019a).
A systematic review of epidemiological studies reported a pooled prevalence of adult ADHD of 2.5% (Simon 2009), with the most recent World Health Organization (WHO) Mental Health Survey reporting a prevalence of 2.8% (range 0.6% to 7.3%) across 10 countries (Fayyad 2017). The prevalence of adult ADHD was higher in males than in females (odds ratio (OR) 1.6, 95% confidence interval (CI) 1.3 to 1.9) (Fayyad 2017). According to the US National Epidemiologic Survey on Alcohol and Related Conditions, women diagnosed with ADHD exhibit less hyperactive behaviour compared to men, but a similar level of 'risky behaviour', and higher levels of anxiety (Cortese 2018b).
The youngest children in a school class have a higher risk of getting a diagnosis of ADHD than the oldest (Elder 2010; Evans 2010; Morrow 2012; Sayal 2017; Layton 2018; Root 2019). A recent systematic review of 334 studies concluded that ADHD is overdiagnosed in children and adolescents (Kazda 2021). Interestingly, follow‐up epidemiological studies have reported that between 5% and 22% of those diagnosed in childhood fulfil the diagnostic ADHD criteria in adulthood only (Faraone 2006; Moffitt 2015; Agnew‐Blais 2016; Caye 2016). Adults with ADHD also often have psychiatric comorbidities (Kessler 2006; Jacob 2007; Sobanski 2007a; Cumyn 2009; Grosz‐Lesch 2013; Philipsen 2014), such as depression, anxiety or substance abuse disorders. For these reasons, the diagnosis of adult ADHD and its categorisation as a 'neurodevelopmental disorder', the treatment modalities, the impact of socioeconomic factors, and the natural course of the symptoms for those diagnosed are debated (Barkley 2002; Timimi 2004; Coghill 2004; Marcovitch 2004; Bailly 2005; Asherson 2010; Moncrieff 2010; Moncrieff 2011; MHRA 2010; MEB 2016).
Description of the intervention
Methylphenidate is a central nervous system (CNS) stimulant related to amphetamine. It inhibits dopamine and noradrenergic reuptake transporters, thus increasing the synaptic concentration of these neurotransmitters (Challman 2000; Iversen 2006). Methylphenidate can be administered as immediate‐release or extended‐release oral tablet formulations. The extended‐release formulations use different coatings and release mechanisms to ensure a prolonged release of methylphenidate (Coghill 2013; Maldonado 2013), and are prescribed more frequently than immediate‐release formulations (Treceño 2012; Beau‐Lejdstrom 2016). The UK National Institute for Health and Care Excellence (NICE) state in their ADHD treatment guideline that pharmacological treatment, such as CNS stimulants, should be used as part of a multimodal approach consisting also of social and psychological interventions (NICE 2018).
Several extended‐release methylphenidate formulations have been approved for ADHD in adults. The US Food and Drug Administration (FDA) approved the first drug for this indication in 2005 with the authorisation of Novartis’s dex‐methylphenidate (FDA 2007). The German Bundesinstitut für Arzneimittel und Medicinprodukte (BfArM), was the first European drug regulator to approve a formulation for the indication in adults in 2011, with the approval of Medice’s extended‐release formulation (BfArM 2011). Methylphenidate has been the most prescribed CNS stimulant for ADHD, but in some countries, e.g. in the USA, amphetamines are now more widely prescribed for use in adult ADHD (Safer 2016; Piper 2018).
In the USA, methylphenidate is classified as a 'schedule II controlled substance', meaning that it is authorised for medical use but that it has a "high potential for abuse which may lead to severe psychological or physical dependence" (US DEA 2021). The most recent systematic review on diversion and misuse of prescription stimulants, such as methylphenidate, found that it is a "significant public health concern" (Faraone 2020a).
How the intervention might work
Methylphenidate has central and peripheral stimulatory effects. The drug's central effects are primarily due to increased synaptic concentrations of dopamine and norepinephrine, which are suggested to have behavioural and cognitive effects in adults with ADHD (Foquest 2019, p 22), but the exact mode of therapeutic action is unknown (FDA 2017, p 20). The main hypothesis underlying CNS stimulant use for ADHD is that these neurotransmitters activate specific brain regions related to attention and inhibition control (Cortese 2020). A systematic review of single‐dose methylphenidate studies in healthy adults reported on improvements in several cognitive domains (Linssen 2014). Methylphenidate is also used as a cognitive enhancement drug (i.e. 'study drug') by students and adults (Sussman 2006; Wilens 2008; Franke 2013). A systematic review of methylphenidate's impact on maths and reading skills in children found only small‐to‐moderate effects (Kortekaas‐Rijlaarsdam 2019), and a systematic review of observational studies on methylphenidate's impact on academic achievements (e.g. school tests, failure to pass a class, grades) reported that the evidence on methylphenidate's benefits is uncertain (De Faria 2021).
Methylphenidate's peripheral stimulatory effects include increases in heart rate and blood pressure (Mick 2013). Observational studies of CNS stimulant use in children and in adults have reported an increased risk of cardiovascular events (Holick 2009; Schelleman 2012; Dalsgaard 2014; Shin 2016) or no increased risk (Cooper 2011; Habel 2011). The FDA Prescriber Information warns about the risk of serious cardiovascular events (FDA 2017, section 5.1).
The published literature on observational studies assessing the effects of methylphenidate on various functional outcomes have reported ambiguous results. Studies using a before‐after medication design have reported on beneficial methylphenidate effects on outcomes such as serious traffic accidents (Chang 2014) and criminal behaviour (Lichtenstein 2012). In contrast to these studies, long‐term observational follow‐up studies from the Multimodal Treatment of ADHD (MTA) study in children (MTA 1999) have not found beneficial effects on outcomes such as hospital admissions, delinquency, or academic performance for those treated with methylphenidate compared to no treatment after three to eight years of follow‐up (Jensen 2007; Molina 2007 Molina 2009). At 16 years of follow‐up, the subgroup of those consistently taking methylphenidate was on average 5 cm shorter than the subgroup with the lowest methylphenidate intake (Swanson 2017). A systematic review (Krinzinger 2019) assessing the risk of psychiatric harms of methylphenidate from long‐term treatment assessed in various study designs reported that the evidence base is weak and that more studies are needed. Generally, confounding and bias are inadequately addressed in observational research in psychiatry (Munkholm 2020) and one should be cautious about drawing inference from such uncontrolled study designs.
Why it is important to do this review
In the UK, NICE recommends methylphenidate as the first‐line drug for ADHD in adults (NICE 2018, section 1.7.11). The prescription rate of ADHD drugs in the UK for adults between 25 and 45 years of age rose 16‐fold from 3.4 per 10,000 in 2000 to 53 per 10,000 in 2015 (Renoux 2016). Across 14 countries, the annual average increase in ADHD medication use was 18% between 2001 and 2015 (Raman 2018).
The World Health Organization (WHO) has twice rejected applications to include methylphenidate on the WHO Model list of Essential Medicines. The first application was submitted by faculty at the Icahn School of Medicine in 2018 (Moscibrodzki 2018). The second application by Faraone 2020b focused on methylphenidate for children and adolescents, and was additionally signed by ADHD key opinion leaders. In both rejections, the WHO argued that there were uncertainties about the benefits and harms, and that the quality of the evidence was low (WHO 2019b; p 441; Storebø 2021; WHO 2021a; WHO 2021b). In the first rejection, the WHO also highlighted that a Cochrane Review of immediate‐release methylphenidate for ADHD in adults (Epstein 2014) was retracted in 2016, since the review authors were unable to respond to criticism of the methodological limitations of the included trials (Epstein 2016; Boesen 2017a). The review on immediate‐release methylphenidate was updated by a new author team (Cândido 2021).
Previous systematic reviews have assessed methylphenidate compared to placebo (Faraone 2004; Peterson 2008; Koesters 2009; Meszaros 2009; Faraone 2010; Castells 2011; McDonagh 2011; Castells 2013; McDonagh 2015; Cunill 2016; Cortese 2018, Elliott 2020). The most recent of these reviews have reported on small to moderate effects of methylphenidate with a high degree of uncertainty about the quality of the evidence (McDonagh 2015; Cunill 2016; Cortese 2018; Elliott 2020). Important limitations to these reviews were their reliance on mostly published literature and a focus on ADHD symptom scales, which might not be the most patient‐relevant outcomes. A systematic review (Schatz 2015) of patient preference studies included three surveys of adult ADHD patients (Mühlbacher 2009; Mühlbacher 2010; Glenngård 2013), all of which ranked work and school performance and social contacts as most important.
Due to substantial selective reporting (McDonagh 2011), and a dearth of assessed patient‐reported outcomes in previous reviews, the benefits and harms of methylphenidate for adult ADHD remain uncertain. This uncertainty was particularly emphasised in the sister review on immediate‐release methylphenidate for adult ADHD (Cândido 2021).
Objectives
To assess the beneficial and harmful effects of extended‐release formulations of methylphenidate in adults diagnosed with ADHD.
Methods
Criteria for considering studies for this review
Types of studies
Double‐blind, parallel‐group, randomised controlled trials (RCTs).
Types of participants
Adults diagnosed with ADHD according to DSM (APA 1987; APA 2000; APA 2013) or ICD (WHO 1992; WHO 2019a) criteria. Adults had to be diagnosed according to the criteria of ADHD in childhood, before the introduction of the ADHD diagnosis in adolescence and adulthood with the DSM‐5 (APA 2013) and ICD‐11 (WHO 2019a).
Types of interventions
Methylphenidate in any extended‐release formulation, at any dose, compared with placebo or an active control, or both. We permitted co‐interventions, such as cognitive behavioural therapy and pharmaceutical interventions, provided they were delivered to both groups.
Types of outcome measures
Primary outcomes
-
Functional outcomes (functional capacity), such as academic and job adherence, measured as days of lost work or study activities (Philipsen 2008).
-
Self‐rated ADHD symptoms, measured on validated rating scales such as the Connors' Adult ADHD Rating Scale (Conners 1999).
-
Serious adverse events, defined according to the International Council for Harmonisation (ICH) guideline (ICH 2003).
Secondary outcomes
-
Quality of life, measured on validated scales such as the Quality of Life Enjoyment and Satisfaction Questionnaire‐Short Form (Mick 2008).
-
Investigator‐rated (trialist, physician, other healthcare professional) ADHD symptoms, measured on a validated rating scale.
-
Peer‐rated (spouse, family or friends) ADHD symptoms, measured on a validated rating scale.
-
Severe psychiatric adverse events, i.e. any psychiatric adverse event classified as 'severe' or 'serious', or any psychiatric event leading to dose reduction or to dropout.
-
Cardiovascular variables (i.e. blood pressure and heart rate).
-
Adverse events other than serious adverse events.
We used data with the longest possible follow‐up within three outcome periods: short‐term (up to six months); medium‐term (up to 12 months); and long‐term (more than 12 months).
Search methods for identification of studies
We ran the first searches in October 2017. We ran top‐up searches of all databases in March 2019 and February 2021, apart from the Database of Abstracts of Reviews of Effects (DARE), which was last published in April 2015. We searched the databases and trial registries listed below using the strategies in Appendix 1. We did not limit the searches by publication date or language.
Electronic searches
We searched the following electronic sources.
-
Cochrane Central Register of Controlled Trials (CENTRAL; Issue 2, 2021), in the Cochrane Library, which includes the Cochrane Developmental, Psychosocial and Learning Problems Specialised Register (searched 16 February 2021).
-
MEDLINE Ovid (1946 to February Week 1 2021).
-
MEDLINE In‐Process & Other Non‐Indexed Citations Ovid (1946 to 12 February 2021).
-
MEDLINE Epub Ahead of Print Ovid (12 February 2021).
-
Embase Ovid (1974 to 12 February 2021)
-
APA PsycINFO Ovid (1806 to February Week 2 2021).
-
Cochrane Database of Systematic Reviews (CDSR; Issue 2, 2021), part of the Cochrane Library (searched 16 February 2021).
-
Database of Abstracts of Reviews of Effects (DARE; Issue 2, 2015, final issue), part of the Cochrane Library (searched 3 October 2017).
-
CINAHL Plus EBSCOhost (Cumulative Index to Nursing and Allied Health Literature; 1937 to 16 February 2021).
-
Conference Proceedings Citation Index ‐ Science Web of Science (CPCI‐S; 1990 to 16 February 2021).
-
LILACS (Latin American and Caribbean Health Science Information database; lilacs.bvsalud.org/en/; searched 16 February 2021).
-
ClinicalTrials.gov (clinicaltrials.gov; searched 16 February 2021).
-
EU Clinical Trials Register (clinicaltrialsregister.eu; searched 16 February 2021).
-
ISRCTN Registry (www.isrctn.com; searched 16 February 2021).
-
WHO International Clinical Trials Registry Platform (apps.who.int/trialsearch; attempted access 16 February 2021 but site unresponsive because of high traffic generated by the pandemic)
Searching other resources
Drug regulatory agency databases
We searched the following 12 drug regulatory agencies for publicly‐available trial data up to 12 May 2020. To our knowledge, there have been no new methylphenidate formulations approved for adult use since that date, so we did not update these searches. Some agencies host multiple databases; full details of these and descriptions of the searches are reported in Appendix 2.
-
The US Food and Drug Administration (FDA; www.accessdata.fda.gov/scripts/cder/daf)
-
The European Medicines Agency (EMA; www.ema.europa.eu/en/medicines)
-
The Heads of Medicines Agencies (HMA; mri.cts-mrp.eu/Human/)
-
Health Canada (Drug Product Database; Summary Basis of Decision; Clinical Study Reports)
-
The Australian Therapeutic Goods Administration (TGA; Australian Register of Therapeutic Goods; Public Assessment Reports)
-
The German Bundesinstitut für Arzneimittel und Medizinprodukte (BfArM; www.pharmnet-bund.de/)
-
The Dutch Medicines Evaluation Board (MEB; www.geneesmiddeleninformatiebank.nl/en)
-
The British Medicines and Healthcare products Regulatory Agency (MHRA; products.mhra.gov.uk/)
-
The French Agence Nationale de Sécurité du Médicament et des Produits de Santé (ANSM; archiveansm.integra.fr/Mediatheque/Publications/Rapports-Syntheses-Rapports-publics-d-evaluation)
-
The Irish Health Product Regulatory Authority (HPRA; hpra.ie/homepage/medicines/medicines-information/find-a-medicine)
-
SwissMedic (www.swissmedic.ch/swissmedic/en/home/humanarzneimittel/authorisations/swisspar.html)
-
The Japanese Pharmaceuticals and Medical Devices Agency (PMDA; www.pmda.go.jp/english/review-services/reviews/approved-information/drugs/0003.html)
Pharmaceutical company databases
We also searched methylphenidate manufacturer websites and databases for clinical trials. See Appendix 2 for details.
-
Yale University Open Data Access (YODA; yoda.yale.edu/), which contains trials sponsored by Johnson & Johnson, including Janssen.
-
Novartis’ database (www.novctrd.com/).
-
Purdue Pharma’s website (www.purduepharma.com/healthcare-professionals/clinical-trials/).
-
Medice’s website (www.medice.de/the-company).
Author correspondence
We contacted all corresponding authors for each included trial. We wrote to all authors or contact people up to three times. If we did not receive an answer the first time, we searched Google to identify other email addresses. We approached some authors already at the protocol stage.
Searching other systematic reviews and guidelines
We searched national clinical ADHD guidelines (Sundhedsstyrelsen 2017; NICE 2018) and recently‐published systematic reviews of ADHD medications (Cortese 2018; Elliott 2020).
Cross‐referencing of trial identifiers
Finally, we cross‐referenced trial registry identifiers, grant funding numbers, and searched the reference lists of included published reports and documents to find additional information.
Data collection and analysis
Selection of studies
Two review authors (KB and ASP) independently screened titles and abstracts for eligibility, using Covidence. We compared full‐text reports of potentially eligible studies against the inclusion criteria (Criteria for considering studies for this review), resolving disagreements by discussion with a third review author (KJ). We listed the reasons for exclusion of full‐text reports in the PRISMA diagram (Moher 2009), and they may be seen in Characteristics of excluded studies.
Data extraction and management
Two review authors (KB and ASP) independently extracted study data on methodology, population characteristics and outcomes using a prespecified data extraction sheet, and compared the data. We resolved disagreements by discussion with a third review author (KJ). One review author (KB) entered the data into Review Manager 5 (Review Manager 2020), and a second review author (ASP) double‐checked them for accuracy.
Assessment of risk of bias in included studies
Two review authors (KB and ASP) independently assessed each trial for risks of bias using the original Cochrane risk of bias tool (Higgins 2011). Consensus was sought, and disagreements were resolved through discussion with a third review author (KJ). The risk of bias tool consists of seven domains, with each domain given a rating of low, unclear or high risk of bias.
-
Random sequence generation (selection bias): Was the sequence generated adequately (e.g. computer‐generated) or inadequately (e.g. using the day of study enrolment) to allocate participants?
-
Allocation sequence concealment (selection bias): Was the implementation of the randomisation sequence adequate (e.g. central allocation by a third party) or inadequate (e.g. open allocation or using non‐opaque envelopes)?
-
Blinding of participants and personnel (performance bias): Methylphenidate’s effects may compromise blinding of participants and investigators. We rated the trials at unclear risk of bias, unless the blinding was tested and confirmed. We applied the same method to trials that added active components to the placebo pill, e.g. folic acid in Levin 1998a and Levin 1998b, to mimic adverse effects like stomach‐ache and decreased appetite.
-
Blinding of outcome assessment (detection bias): We applied the same method for this domain as described above for blinding of participants and personnel.
-
Incomplete outcome data (attrition bias): We based our assessment on four variables:
-
dropout rates in each arm;
-
dropout rates in each arm due to adverse events;
-
the statistical method used to account for missing data ‐ acceptable methods would include multiple imputations, whereas improper methods would include completers’ analysis and last observation carried forward (LOCF) (Lachin 2016);
-
other aberrations, such as analyses including more participants than actually completed the trials. There is no dropout level that eliminates the risk of bias, unless it is zero, and we did not define a lower threshold to guide our assessments.
-
-
Selective outcome reporting (reporting bias): We rated a trial at high risk of bias if it prespecified one of our primary or secondary outcomes in a trial registry, protocol, or published report, but did not report it. We categorised as reporting bias any adverse event‐reporting threshold (i.e. that a certain percentage of participants must experience an adverse event before it is reported).
-
Other potential sources of bias: Was the study free of other sources of bias, such as baseline differences?
Overall risk of bias assessment for individual trials
We made an overall risk of bias assessment for each trial, based on the individual domain ratings, as recommended in the Cochrane Handbook (Higgins 2011). We considered all bias domains as critical for the overall risk of bias, so if one or more domains were rated as high risk we rated the trial as having an overall high risk of bias.
Specific limitations to the generalisability (external validity)
We assessed three study design characteristics related to the generalisability of ADHD drug trials. We depicted these three design characteristics as additional items within the Cochrane risk of bias tool, and rated them being at high, low or unclear risk of limiting the external validity of each trial. These design characteristics should not be mistaken for additional risk of bias domains (which describe internal validity). Rather, they informed our assessment of 'indirectness' when we graded the certainty of the evidence (see Data synthesis).
1. Psychiatric comorbidity
Adults with an ADHD diagnosis are often also diagnosed with other psychiatric diagnoses, such as depression, anxiety or substance abuse (Kessler 2006, Jacob 2007, Sobanski 2007a, Cumyn 2009, Grosz‐Lesch 2013, Philipsen 2014). The exclusion of participants with psychiatric comorbidity may reduce the trials’ generalisability (Surman 2010). It has also been suggested that such exclusions could lead to an overestimation of the treatment effect (Pliszka 1989; Sobanski 2007b, MHRA 2010).
-
Low risk: Participants diagnosed with other psychiatric diagnoses than ADHD were not excluded, unless methylphenidate was contraindicated (e.g. bipolar affective disorder or suicidality).
-
Unclear risk: There was no description, or an unclear one, whether participants with psychiatric comorbidity were allowed.
-
High risk: Participants with psychiatric comorbidities other than ADHD or other psychiatric symptoms, were excluded from participation in the trial.
2. Responder selection ('enriched design')
Some patients do not benefit from or tolerate methylphenidate, and are referred to as "non‐responders" (Elia 1991; De Sonneville 1994; Cho 2007). It is ethically unacceptable to include participants who are known to experience harm or not to benefit from the intervention. However, the exclusion of such participants will lead to an overestimation of benefits and an underestimation of harms compared to a treatment‐naïve population, especially if only participants known to benefit from the drug are included. Selecting participants based on previous exposure to the drug is called an 'enriched design' (FDA 2019).
-
Low risk: Participants were treatment‐naïve.
-
Unclear risk: There was no description, or an unclear one, of exclusion criteria or of previous stimulant use.
-
High risk: Previous exposure to methylphenidate or other stimulants was allowed, or 'non‐responders' (or similar) were excluded before randomisation, or both.
3. Withdrawal effects
Discontinuation of psychotropic drugs can cause acute and protracted withdrawal symptoms (Lerner 2019). The literature pertaining specifically to methylphenidate withdrawal is scarce, but evidence related to other CNS stimulants, most commonly cocaine, amphetamine, and methamphetamine, suggests that withdrawal symptoms may persist for weeks or months (Lago 1994; Baker 2004; Cruickshank 2009; Phillips 2014). Due to methylphenidate's status as a Schedule‐II drug (US DEA 2021), FDA's Prescriber Information on methylphenidate products warns that chronic use can lead to dependence and tolerance (FDA 2017). It seems plausible that if participants take CNS stimulants upon enrolment in a methylphenidate trial and stop treatment before randomisation, they might experience withdrawal effects. If the taper period (often called a “washout”) is not of sufficient duration, withdrawal or rebound symptoms may occur during the trial (Cox 2008, Récalt 2019). To our knowledge, there are no published guidelines on how to taper methylphenidate. Consensus documents (Kooij 2010; Kooij 2019) and British (NICE 2018, section 1.10, 'Review of medication and discontinuation') and German (DGPPN 2021) national ADHD guidelines contain no guidance on how to taper methylphenidate. A systematic review of CNS stimulant discontinuation in children and adolescents with ADHD did not identify any concrete guidance (Lohr 2021). To assess the risk of withdrawal effects, we therefore made an individual trial‐by‐trial assessment of whether the stated washout periods were of sufficient duration to avoid withdrawal effects.
-
Low risk: Participants were treatment‐naïve, or the stimulant‐free period before randomisation was of such duration that the risk of withdrawal effects was considered minimal.
-
Unclear risk: There was no description, or an unclear one, of the participants’ treatment status, or it was uncertain if the stimulant‐free period prior to randomisation was sufficient to avoid withdrawal effects.
-
High risk: Stimulants were stopped before randomisation, with a washout phase of insufficient duration, or without a washout.
Measures of treatment effect
Dichotomous data
We summarised dichotomous data as risk ratios (RRs) with corresponding 95% confidence intervals (95% CIs).
Continuous data
For continuous data and rating scales, we extracted mean change or endpoint (preferred) data and the corresponding standard deviations (SDs) or standard errors (SEs), in order to calculate a mean difference (MD). If the outcome was reported on different scales, we used a standardised mean difference (SMD) with a 95% CI.
Unit of analysis issues
Cross‐over trials
We excluded cross‐over trials due to the risk of carry‐over stimulant effects and the likely loss of blinding. In a trial of immediate‐release methylphenidate (Gualtieri 1985), all eight participants were able to correctly guess their allocation, probably due to the adverse events.
Studies with multiple time points
We split up outcome periods into short‐term (up to six months), medium‐term (up to 12 months), and long‐term (more than 12 months). If a trial reported multiple sets of study data within the same predefined outcome period, we extracted data with the longest follow‐up time.
Studies with multiple treatment groups
In studies with multiple methylphenidate groups and a single comparator arm, we pooled the groups to make a single, pair‐wise comparison (Higgins 2021).
Dealing with missing data
A drug class review from 2011 reported that a considerable amount of data were missing in ADHD drug trials (McDonagh 2011). As well as searching for unpublished and regulatory data, as described elsewhere, we took the following precautions.
-
We highlighted unreported trial outcomes in the ‘selective outcome reporting’ bias domain of the risk of bias tables.
-
We imputed missing standard deviations (SDs) based on trials using the same scale and of similar size.
-
We extracted and meta‐analysed the overall dropout rates. We used the full population sizes rather than restricted populations, e.g. per‐protocol populations or ‘safety‐populations’.
-
We highlighted each trial's statistical method for dealing with missing data, such as last observation carried forward (LOCF), in the ‘attrition bias’ domain of the risk of bias tables.
-
We preferred analyses based on imputed rather than completer's data, and made it clear if we made such a choice.
Assessment of heterogeneity
We assessed clinical heterogeneity related to differences in the trial populations, interventions or settings. We assessed methodological heterogeneity using the Cochrane risk of bias tool and the three domains related to the trials’ generalisability (psychiatric comorbidity, responder selection, and withdrawal effects; see Assessment of risk of bias in included studies). We assessed statistical heterogeneity using the I2 statistic and Chi2 test (Deeks 2021), and explored reasons for heterogeneity through subgroup (see: Subgroup analysis and investigation of heterogeneity) and sensitivity analyses (see: Sensitivity analysis) when the I2 statistic was greater than 50%.
Assessment of reporting biases
We describe elsewhere in detail how we assessed outcome reporting bias (Assessment of risk of bias in included studies).
We assessed publication bias and other small‐study effects using funnel plots, with estimated effect sizes plotted against their standard error (Egger 1997), provided 10 or more studies reported a given outcome (Page 2021).
Data synthesis
We summarised data using random‐effects meta‐analyses with inverse‐variance weighting due to anticipated clinical, methodological and statistical heterogeneity (McDonagh 2011; Storebø 2015). We considered all studies eligible for meta‐analysis regardless of their risk of bias assessment and methylphenidate dosage, as there is no established dose‐response relationship (NICE 2008; Huss 2017).
Subgroup analysis and investigation of heterogeneity
We prespecified subgroup analyses when the I2 was greater than 50%.
-
Psychiatric comorbidity: trials that excluded participants with psychiatric comorbidity versus trials that did not.
-
Responder selection: trials with responder selection ('enriched design') versus trials without responder selection.
-
Withdrawal effects: trials with a high risk of withdrawal effects versus trials with a low risk of withdrawal effects.
-
Conflicts of interest: industry‐sponsored trials versus other trials.
Sensitivity analysis
We also prespecified sensitivity analyses when the I2 was greater than 50%.
-
Restricting the analyses to trials with low risk of attrition bias according to our bias assessment.
Summary of findings and assessment of the certainty of the evidence
We created a summary of findings table for our three primary outcomes and four of our secondary outcomes comparing extended‐release methylphenidate with placebo. Rather than including all individual adverse events in the summary of findings table, we chose the ‘proportion of participants experiencing adverse event’ as an overall outcome for harms in the table. We chose to include short‐term data only, due to the scarcity of medium‐ and long‐term follow‐up. We used GradePro GDT to construct the table.
Using the GRADE approach, two review authors (KB and ASP) independently assessed the certainty of evidence for each outcome included in the table. They assigned ratings of 'high', 'moderate', 'low', or 'very low' certainty according to the presence of the following five elements; 'study limitations' (i.e. risk of bias), 'inconsistency' (i.e. statistical heterogeneity), 'indirectness' (i.e. low generalisability), imprecision (i.e. wide confidence intervals), and 'publication bias' (Schünemann 2013).
Results
Description of studies
Results of the search
We ran searches in October 2017, March 2019 and February 2021 of databases and trial registries, which returned a total of 11,104 records; 6182 duplicates were removed in EndNote/Covidence, leaving 4922 records. We found 68 records by searching drug regulatory agency databases and pharmaceutical company databases (May 2020), corresponding with trial authors (continuously from 2016 to 2021), and by cross‐referencing all documents, giving a total of 4990 records to be screened. We excluded 4648 of these records and assessed 342 reports in full text. We excluded 144 full‐text reports, leaving 198 reports for inclusion. We therefore included 189 reports, drug regulatory documents, trial registry entries, and other documents, describing 24 randomised, placebo‐controlled trials; 2 reports describing one ongoing trial; and 7 reports awaiting classification (Figure 1). We listed the reasons for exclusion of full‐text records in the Characteristics of excluded studies.
Flow diagram.
BfArM= Bundesinstitut für Arzneimittel und Medizinprodukte, FDA= US Food and Drug Administration, MEB= Medicines Evaluation Board, MHRA= Medicines and Healthcare products Regulatory Agency, TGA= Therapeutic Goods Administration, YODA= Yale University Open Data Access.
Three trials appeared in trial registries only (Combine Substudy 2005; College Study 2009; Methacan 2017), while the remaining 22 trials also appeared in published reports. We included regulatory documents from 11 drug applications reporting data from eight trials: four OROS methylphenidate applications reported data from three pivotal trials (Medori 2005; Adler 2006; Casas 2008); three long‐acting methylphenidate applications reported data from one trial (Huss 2010); two controlled‐release methylphenidate applications reported data from one trial (Weiss 2014); one dex‐methylphenidate extended‐release application reported data from one trial (Spencer 2003a); and one extended‐release methylphenidate application reported data from two trials (Rösler 2004; Retz 2008). We have assessed these drug applications further in a separate research project (Boesen 2021a; Leufkens 2021). See the appendices for more information on clinical trial registries, pharmaceutical company databases, cross‐referenced records, and other resources (Appendix 3), drug regulatory documents (Appendix 4), and author correspondence (Appendix 5). We have made our full dataset available (see Contributions of authors).
Included studies
We included 24 randomised, parallel‐group, placebo‐controlled trials of extended‐release methylphenidate (5066 participants), conducted between 1998 and 2019 (Table 1). All 24 trials compared extended‐release methylphenidate with placebo, and two trials also included active comparators, bupropion (Levin 1998a) and atomoxetine (Weisler 2009) respectively.
Twenty‐one trials reported outcome data for our review, and three did not: the Combine Substudy 2005 appears to be unpublished; the College Study 2009 was apparently never conducted; and Asherson 2016 reported group‐level data including participants younger and older than 18 years of age, which made the data ineligible for our review. The COMPAS study (Compas 2007 GPT/Compas 2007 CM) used a factorial study design with four arms testing two comparisons (methylphenidate versus placebo; group psychotherapy versus clinical management). To allow pairwise comparisons we split up the trial into two: methylphenidate versus placebo for those receiving group psychotherapy (Compas 2007 GPT) and methylphenidate versus placebo for those receiving clinical management (Compas 2007 CM).
Key trial characteristics
The median sample size was 224 (range 20 to 725) and the median trial length was eight weeks (range 2 to 52 weeks). The trials assessed six different extended‐release methylphenidate formulations; osmotic‐controlled release oral delivery system (OROS) (14 trials), sustained‐release (SR) (4 trials), extended‐release (ER) (3 trials), long‐acting (LA) (1 trial), controlled‐release (CR) (1 trial), and dex‐methylphenidate extended‐release (1 trial). Fourteen trials used a dose‐titration design, five trials used a fixed‐dose design and included multiple methylphenidate groups, three trials used a fixed‐dose design with one methylphenidate group, and for two trials there was missing information. Twelve trials (3836 participants; 76% of total population) were industry‐sponsored, four trials (708; 14% of population) were publicly‐funded with industry involvement, seven trials (502; 10% of population) were publicly‐funded, and in one trial (20 participants) funding was unclear. See also Table 1 below.
Table 1. Key trial characteristics
| Trial | Formulation | Dose design (fixed/titration) | Study start date to study end date | Sponsor | Sponsor categorisation | Number of participants | Duration in weeks | Other comments |
|---|---|---|---|---|---|---|---|---|
| Sustained‐release methylphenidate | Titration (40 mg to 80 mg) | February 1998 ‐October 2004 | National Institute on Drug Abuse (NIDA) | Publicly funded | 98 | 10 | Number of participants includes the bupropion group | |
| Sustained‐release methylphenidate | Titration (40 mg to 60 mg) | April 1998 ‐ March 2004 | National Institute on Drug Abuse (NIDA) | Publicly funded | 106 | 13 | ‐ | |
| Sustained‐release methylphenidate | Fixed (20 mg) | Not reported | Not reported | Not reported | 20 | 4 | ‐ | |
| Osmotic Controlled‐release Oral Delivery System (OROS) methylphenidate | Titration (36 mg to 1.3 mg/kg, max 144 mg) | June 2003 –August 2007 | McNeil/Janssen | Industry trial | 227 | 6 | ‐ | |
| Dex‐methylphenidate extended release | Fixed (20 mg to 30 mg to 40 mg) | April 2003 ‐September 2003
| Novartis | Industry trial | 221 | 5 | ‐ | |
| OROS methylphenidate | Titration (36 mg to 90 mg) | December 2004 ‐December 2006
| University of Maryland and McNeil | Publicly funded, with industry involvement | 20 | 2 | According to clinicaltrials.gov, the sponsor was the University of Maryland and McNeil was a collaborator. According to the published report, the trial was “supported by” McNeil. | |
| Extended‐release methylphenidate | Titration (10 mg to 60 mg) | November 2004 ‐May 2006 | Medice | Industry trial | 359 | 24 | ‐ | |
| Sustained‐release methylphenidate | No data | December 2005 ‐April 2007 (estimated) | National Institute on Alcohol Abuse and Alcoholism (NIAAA) | Publicly funded | No data | 12 | ‐ | |
| OROS methylphenidate | Fixed (18 mg to 36 mg to 72 mg) | April 2005 ‐August 2006 | Janssen | Industry trial | 402 | 5 | ‐ | |
| OROS methylphenidate | Titration (18 mg to 72 mg) | November 2005 ‐March 2008 | National Institute on Drug Abuse (NIDA) | Publicly funded, with industry involvement | 255 | 11 | Janssen was listed as collaborator on clinicaltrials.gov. | |
| OROS methylphenidate | Titration (36 mg to 108 mg) | April 2006 ‐December 2006 | McNeil | Industry trial | 229 | 7 | ‐ | |
| OROS methylphenidate | Titration (18 mg to 72 mg) | February 2006 ‐June 2007 | Addiction Centre Stockholm | Publicly funded | 24 | 12 | ‐ | |
| Compas 2007 (Compas 2007 CM; Compas 2007 GPT) | Extended‐release methylphenidate | Titration (10 mg to 60 mg, or up to 1.3 mg/kg) | April 2007 ‐August 2011 | German Federal Ministry of Education and Research | Publicly funded, with industry involvement | 433 | 52 | Medice’s medical director was involved in designing the trial and collecting data according to the published reports. |
| OROS methylphenidate | Fixed (72 mg) | May 2007 ‐ April 2010 | Karolinska University Hospital | Publicly funded | 30 | 5 | ‐ | |
| OROS methylphenidate | Titration (18 mg to 180 mg) | April 2007 ‐September 2011 | Addiction Centre Stockholm | Publicly funded | 54 | 24 | ‐ | |
| OROS methylphenidate | Fixed (54 mg to 72 mg) | February 2008 ‐ April 2009 | Janssen | Industry trial | 279 | 13 | ‐ | |
| Extended‐release methylphenidate | Titration (40 mg to 120 mg) | September 2008 ‐ January 2010 | Medice | Industry trial | 162 | 8 | ‐ | |
| OROS methylphenidate | No data | No data | University of Pittsburgh | Publicly funded, with industry involvement | No data | 8 | Study was seemingly never conducted. Ortho‐McNeil‐Janssen was listed as collaborator on clinicaltrials.gov. | |
| OROS methylphenidate | Titration (18 mg to 72 mg) | July 2009 ‐February 2010 | Janssen | Industry trial | 357 | 6 | ‐ | |
| OROS methylphenidate | Fixed (54 mg) | May 2009 ‐January 2010 | Janssen | Industry trial | 216 | 6 | Number of participants includes the atomoxetine group, but not the bavisant groups | |
| Long‐acting methylphenidate | Fixed (40 mg to 60 mg to 80 mg) | November 2010 ‐ August 2012 | Novartis | Industry trial | 725 | 9 | ‐ | |
| OROS methylphenidate | Titration (18 mg to 72 mg) | March 2011 ‐ April 2012 | Janssen | Industry trial | 284 | 8 | ‐ | |
| Controlled‐release methylphenidate | Fixed (25 mg to 45 mg to 70 mg to 100 mg) | October 2014 ‐January 2015 | Purdue | Industry trial | 375 | 4 | ‐ | |
| OROS methylphenidate | Titration (18 mg to 72 mg) | May 2016 ‐July 2019 | King’s College London | Publicly funded | 190 | 8 | Number of participants includes those aged 18 years and older only |
Setting and participants
Twenty‐one trials had an outpatient setting and three trials were conducted in prison (Ginsberg 2007; Konstenius 2007; Asherson 2016). The industry trials were homogeneous by setting, population and trial design. The five publicly‐funded trials that reported outcome data for this review either included participants with a comorbid substance abuse diagnosis (Levin 1998a; Levin 1998b; Konstenius 2006), or were conducted in prison with inmates (Ginsberg 2007), or both (Konstenius 2007). We summarise trial setting and participant characteristics in Table 2 below.
Table 2. Key trial settings and participant characteristics
| Trial ID | Location, number of sites | Mean age in years | % of male | Required co‐diagnosis | Other trial characteristics |
|---|---|---|---|---|---|
| USA, 5 sites | 39 | 57 | Opiate dependence | All participants were maintained on a methadone program during the trial. Assessed substance use outcomes as well | |
| USA, 2 sites | 37 | 83 | Cocaine dependence | Assessed substance use outcomes as well | |
| USA, 1 site | 37 | 63 | No | None | |
| USA, 1 site | 36 | 55 | No | None | |
| USA, 18 sites | 39 | 57 | No | None | |
| USA, 1 site | 40 | 0 | No | All participants were mothers who had daughters also diagnosed with attention deficit hyperactivity disorder | |
| Germany, 28 sites | 35 | 50 | No | None | |
| USA, 1 site | No data | No data | Alcohol dependence | None | |
| 13 European countries, 48 sites | 34 | 54 | No | None | |
| USA, 6 sites | 38 | 57 | No | All participants received a transdermal nicotine patch as part of a smoking cessation programme. Assessed smoking‐related outcomes as well | |
| USA, 27 sites | 39 | 56 | No | None | |
| Sweden, 1 site | 37 | 79 | Amphetamine dependency | Assessed substance use outcomes as well | |
| Compas 2007 (Compas 2007 CM; Compas 2007 GPT) | Germany, 7 sites | 35 | 50 | No | Factorial study design (methylphenidate versus placebo; group psychotherapy versus clinical management) |
| Sweden, 1 site | 34 | 100 | No | Conducted in a prison setting with male inmates | |
| Sweden, 1 site | 42 | 100 | Amphetamine dependence | Conducted in prison setting with male inmates; first 2 weeks of trial in prison before release, and remaining 22 weeks outside prison. Assessed substance use outcomes as well | |
| 11 European countries, 42 sites | 36 | 52 | No | None | |
| Germany, 11 sites | 37 | 47 | No | None | |
| USA, 1 site | No data | No data | No | Study not conducted | |
| USA, 35 sites | 36 | 53 | No | None | |
| USA, 37 sites | 33 | 62 | No | None | |
| 9 countries (Belgium, Sweden, Denmark, Germany, Norway, Columbia, South Africa, USA, Singapore), 67 sites
| 35 | 55 | No | None | |
| Japan, 39 sites | 34 | 49 | No | None | |
| USA and Canada, 38 sites | 36 | 47 | No | None | |
| UK, 3 sites | No data | 100 | No | Conducted in a prison setting with male inmates |
Excluded studies
We excluded 144 reports after full‐text reading (Figure 1). Fifty‐seven reported on an ineligible study design (e.g. no blinding, or no control group); 33 reported on an ineligible intervention (e.g. immediate‐release methylphenidate); 22 were duplicates; 16 included ineligible patient populations (e.g. children with ADHD or adults diagnosed with psychiatric diagnoses other than ADHD); and 16 were irrelevant (e.g. not reporting in clinical trials).
We report, in detail, 38 full‐text reports, falling into four categories:
-
extended‐release methylphenidate trials using a cross‐over design (9 reports);
-
non‐blind or industry‐sponsored open‐label extension phases of already‐included trials (9 reports);
-
immediate‐release methylphenidate trials (we only list trials not already included in the immediate‐release methylphenidate reviews by Cândido 2021 or Epstein 2014, or when it was difficult to determine the used formulation) (9 reports);
-
trials with other design issues, such as randomisation to an intervention in which both groups received methylphenidate (11 reports).
See Characteristics of excluded studies tables for details.
Studies awaiting classification
We were unable to retrieve seven records from our systematic searches. All seven were conference proceedings, and probably not reporting on trials of relevance to this review. See details in Characteristics of studies awaiting classification.
Ongoing studies
We identified one French publicly‐funded study (Methacan 2017), which is scheduled to be completed in May 2022. It includes adults diagnosed with ADHD and comorbid cannabis dependence.
Risk of bias in included studies
We assessed the risks of bias for the 21 studies reporting outcome data (Figure 2). See the individual assessments in the Characteristics of included studies tables. We did not assess the risk of bias for the three trials (College Study 2009; Combine Substudy 2005; Asherson 2016) from which we had no data, or from the ongoing trial (Methacan 2017).
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
Fifteen studies described adequate methods for randomisation and allocation concealment and were rated at low risk of selection bias, while six studies reported insufficient information and were rated at unclear risk (Levin 1998a; Levin 1998b; Levin 2001; Spencer 2003a; Chronis‐Tuscano 2004; Rösler 2004).
Blinding
We rated all 21 trials at unclear risk of bias for blinding of participants and personnel and blinding of investigators. Thirteen trials used a dose titration schedule and eight trials used a fixed‐dose trial design (Levin 2001; Spencer 2003a; Medori 2005; Ginsberg 2007; Casas 2008; Weisler 2009; Huss 2010; Weiss 2014). One trial (Konstenius 2007) reported testing for unblinding. The trial used a dose titration design and tested whether the participants could guess their allocation during the titration phase. They reported that 48% of the participants in each group correctly guessed their allocation, 11% and 26% were wrong, and 41% and 26% were uncertain. Thus, the blinding was less than perfect. The remaining trials did not report whether or not they tested for unblinding.
Incomplete outcome data
We rated four trials at low risk of attrition bias. Two of these trials had no dropouts, Chronis‐Tuscano 2004 lasted two weeks only and Ginsberg 2007 was conducted in a prison setting, and two industry‐sponsored trials had low dropout rates (Retz 2008; 5% for methylphenidate, 4% for placebo) and (Takahashi 2011; 6% and 4%), respectively). We judged the remaining 17 trials to be at high risk of attrition bias, as the total dropout rates ranged from 10% to 63% in the methylphenidate groups and 6% to 55% in the placebo groups. Some trials had similar dropout rates between the groups, but most trials had more dropouts in either the methylphenidate or the placebo group; see our meta‐analysis (Analysis 3.1). Eleven of these trials stated that they conducted intention‐to‐treat analyses by using the last observation forward (LOCF). Fifteen trials also reported dropouts due to adverse events, generally favouring placebo (see meta‐analysis Analysis 3.2). We noted aberrations in two trials (Casas 2008; Huss 2010); the combination of high dropout rates and the use of the LOCF method meant that more participants were categorised as 'responders', i.e. benefiting from the drug, than the actual number of participants completing the trials in the high‐dose groups.
Selective reporting
We rated two trials as having a low risk of reporting bias (Compas 2007 CM/Compas 2007 GPT; Ginsberg 2007), two trials as having unclear risk (Medori 2005; Casas 2008), and all other trials as having a high risk. Harms were particularly poorly reported. Nineteen trials reported on adverse events to various degrees, but we had access to the full lists without reporting thresholds for two trials only, Compas 2007 (Compas 2007 GPT/Compas 2007 CM) and Ginsberg 2007. Adverse events reporting thresholds were prevalent also in drug regulatory documents, and thresholds varied for different documents describing the same trials.
Other potential sources of bias
We rated one trial (Retz 2008) at high risk of other bias due to baseline differences in gender distribution and self‐rated symptom severity. We rated one trial at unclear risk of other bias (Spencer 2003a), since it was noted in the FDA review that an audit of trial sites had found a miscoding of a participant's reasons for dropping out, which meant we could not judge the overall impact. We rated the remaining trials at low risk of other bias.
Overall risk of bias assessment
We rated 20 studies at an overall 'high risk' of bias and one trial (Ginsberg 2007) at an overall 'unclear risk' of bias. We rated the blinding domains as 'unclear risk' for all trials, and we identified substantial biases in the attrition and reporting bias domains.
Generalisability (external validity) assessment
We considered all 21 trials to have at least one design characteristic rated at high risk of limiting the external validity (Figure 2). We rated 14 trials (3539 participants, 70% of total population) at high risk regarding all three design characteristics, 9 of which were industry trials. See the individual assessments in the Characteristics of included studies tables.
Restrictive inclusion criteria
We rated one trial at low risk (Konstenius 2007), one at unclear risk (Ginsberg 2007), and 19 trials at high risk due to restrictions of allowed psychiatric comorbidity. The limited external validity of these trials was acknowledged in several reports (Chronis‐Tuscano 2004; Adler 2006; Compas 2007 GPT/Compas 2007 CM; Casas 2008; Huss 2010; Takahashi 2011).
Responder selection ('enriched design')
We considered two trials to be at unclear risk (Levin 2001; Biederman 2003) and 19 trials to be at high risk; 16 of these trials described how participants were excluded based on previous treatment response; for example, due to a “history of poor response” (Spencer 2003a), were “known non‐responders” (Casas 2008) or “non‐responsive” (Ginsberg 2007). Three trials reported how many of the participants had previously been exposed to methylphenidate (Chronis‐Tuscano 2004; Rösler 2004; Retz 2008).
Withdrawal effects
We rated one trial (Goodman 2009) at low risk since the inclusion criteria stipulated stimulant treatment was not allowed within five years of enrolment. We rated four trials at unclear risk, either due to a lack of information (Levin 2001; Biederman 2003; Weisler 2009) or because of a washout phase of six months (Compas 2007 GPT/Compas 2007 CM). We rated 16 trials at high risk due to washout or CNS stimulant‐free periods between one week (Levin 1998b), one to two weeks (Adler 2006; Weiss 2014), one to four weeks (Spencer 2003a; Huss 2010; Takahashi 2011), two weeks (Levin 1998a; Rösler 2004; Konstenius 2006; Ginsberg 2007; Konstenius 2007; Casas 2008; Retz 2008) and one month (Medori 2005; Winhusen 2005), respectively. Chronis‐Tuscano 2004 did not specify the washout period, but the trial was designed in a way that exposed all participants to the methylphenidate before randomisation.
Effects of interventions
See: Summary of findings 1 Extended‐release methylphenidate compared to placebo for ADHD
Extended‐release methylphenidate versus placebo
Twenty‐one trials reported outcome data for this comparison. All 21 trials reported short‐term outcome data between two and 26 weeks of follow‐up, and one trial also reported outcome data at 52 weeks of follow‐up (Compas 2007 CM/Compas 2007 GPT). Due to large dropout rates at 52 weeks, we analysed the symptom‐rating scale data using both the 'completers’ and 'last mean carried forward' datasets, as reported by the authors. No trials with a continuous randomised, placebo‐controlled phase reported long‐term follow‐up of more than 52 weeks.
Primary outcomes
Functional outcomes
Short‐term follow‐up (up to six months)
One trial (two comparisons; Compas 2007 CM/Compas 2007 GPT) reported a mean difference (MD) of −0.15 days (95% CI −2.11 to 1.81; I2 = 68%; 1 trial, 409 participants; very low‐certainty evidence) missed at work/school/training at 13 weeks of follow‐up (Analysis 1.1), and −0.42 days (95% CI −2.48 to 1.64; I2 = 39%; 1 trial, 2 comparisons, 329 participants; very low‐certainty evidence) between 13 and 26 weeks (Analysis 1.2).
Three trials specified functional outcomes we could not analyse: College Study 2009 defined Grade Points Average (i.e. final grades for college students) but the trial was not conducted. Konstenius 2007 defined “relapse to crime” (e.g. readmission to prison) and Asherson 2016 prespecified “number of adjudications for antisocial behaviour” and “positive incentive and earned privileges” in prison, but the outcomes were not reported.
Medium‐term follow‐up (at 52 weeks)
The MD was 0.71 days (95% CI −1.61 to 3.03; I2 = 0%; 1 trial, 2 comparisons, 292 participants; very low‐certainty evidence) missed from work/school/training in favour of placebo between weeks 26 and 52 using completers' data (Analysis 1.3). We asked the corresponding author for baseline to week 52 data, but did not receive the information.
Self‐rated ADHD symptom severity
Short‐term follow‐up (between 2 and 26 weeks)
Sixteen trials (10 reporting endpoint values, six change scores) showed a small‐to‐moderate SMD −0.37 (95% CI −0.43 to −0.30; I2 = 0%; 16 trials, 17 comparisons, 3799 participants; very low‐certainty evidence) in favour of methylphenidate (Analysis 1.4). The funnel plot was symmetrical (Figure 3). Ginsberg 2007 was a small study (30 participants) conducted in a prison setting. The trial reported an SMD of 1.6, which we considered to be an outlier and removed it from the meta‐analysis. The overall result did not change.
Funnel plot. Extended‐release methylphenidate versus placebo: ADHD symptoms (self‐rated).
The MD was −3 points (95% CI −4.34 to −1.76; I2 = 13%; 6 trials, 7 comparisons, 1189 participants) on the most commonly‐used DSM‐IV Total ADHD 18‐item scale (range 0 to 54 points) (Analysis 1.5).
Medium‐term follow‐up (at 52 weeks)
The MD was −1.59 points (95% CI −4.26 to 1.09; I2 = 20%; 1 trial, 2 comparisons, 234 participants; very low‐certainty evidence) on the DSM‐IV Total ADHD scale (range 0 to 54), using completers’ data (Analysis 1.6). The MD was −2.23 points (95% CI −3.50 to −0.95; I2 = 2%; 1 trial, 2 comparisons, 419 participants; very low‐certainty evidence) on the ADHD Index (range 0 to 36) using the last mean carried forward data (Analysis 1.7).
Serious adverse events
Short‐ and medium‐term follow‐up (between 5 and 52 weeks)
There was no evidence of a difference between the groups on this outcome (RR 1.43, 95% CI 0.85 to 2.43; I2 = 0%; 14 trials, 15 comparisons, 4078 participants; very low‐certainty evidence) (Analysis 1.8). The funnel plot was symmetrical (Figure 4). See Appendix 6 for the full list of events.
Funnel plot of comparison: 1 Extended‐release methylphenidate versus placebo ‐ primary outcomes (short and medium term), outcome: 1.8 Serious adverse events.
Secondary outcomes
Quality of life
Short‐term follow‐up (between 4 and 26 weeks)
Six trials (3 reporting endpoint values, 3 change scores) showed a small SMD of −0.15 (95% −0.25 to −0.05; I2 = 0%; 6 trials, 7 comparisons, 1888 participants; very low‐certainty evidence), favouring methylphenidate (Analysis 2.1). Two studies did not report prespecified quality‐of‐life outcomes (Biederman 2003; Adler 2006), one trial did not report data for adults aged 18 years and older separately (Asherson 2016), and one trial reported subscales only but not a total score (Goodman 2009).
The MD was −0.9 points (95% CI −2.17 to 0.31; I2 = 0%; 4 trials, 5 comparisons, 1234 participants) on the most commonly‐used scale, Q‐LES‐QSF (range 14 to 70 points) (Analysis 2.2). The minimal important difference on this scale has been estimated to be 3 points (Mick 2008).
Medium‐term follow‐up (at 52 weeks)
The MD was −0.66 points (95% CI −2.97 to 1.64; I2 = 36%; 1 trial, 2 comparisons, 419 participants; very low‐certainty evidence; Analysis 2.3).
Investigator‐rated ADHD symptom severity
Short‐term follow‐up (between 2 and 26 weeks)
Eighteen trials (15 reporting endpoint values, 3 change scores) showed a small‐to‐moderate SMD of −0.42 (95% CI −0.49 to −0.36; I2 = 0%; 18 trials, 19 comparisons, 4183 participants; very low‐certainty evidence), favouring methylphenidate (Analysis 2.4). The funnel plot was symmetrical around the large trials (Figure 5). As with self‐rated symptoms (Analysis 1.4), we removed Ginsberg 2007 from this analysis. The trial reported an SMD of 2.1, which seemed to be an outlier. Had we retained Ginsberg 2007, the funnel plot would be symmetrical also around the smaller trials, but it does not affect our result.
Funnel plot of comparison: 2 Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short‐ and medium‐term), outcome: 2.4 ADHD symptoms (investigator‐rated) (short‐term).
The MD was −4.74 points (95% CI −5.65 to −3.84; I2 = 15%; 11 trials, 12 comparisons, 3121 participants) on the most commonly‐used rating scale, DSM‐IV Total ADHD 18‐item (range 0 to 54 points) (Analysis 2.5).
Medium‐term follow‐up (at 52 weeks)
The MD was −1.29 points (95% CI −3.38 to 0.80; I2= 0%; 1 trial, 2 comparisons, 243 participants; very low‐certainty evidence) on the DSM‐IV Total ADHD rating scale (range 0 to 54) using completers' data (Analysis 2.6). The MD was −2.18 points (95% CI −3.55 to −0.80; I2 = 15%; 1 trial, 2 comparisons, 419 participants; very low‐certainty evidence) on the ADHD Index (range 0 to 36) using the last mean carried forward data (Analysis 2.7).
Peer‐rated ADHD symptom severity
Short‐term follow‐up (between 4 and 9 weeks)
Three trials reported on this outcome. The raters were called "subject‐designated raters" (Goodman 2009), "assessments made by friend, family members or colleague" (Huss 2010), and "consistent observer" (Weiss 2014). They found a small‐to‐moderate SMD of −0.31 (95% CI −0.48 to −0.14, I2 = 18%; 1005 participants; very low‐certainty evidence), favouring methylphenidate (Analysis 2.8).
Severe psychiatric adverse events
Short‐term follow‐up (up to six months)
The components of this outcome were inconsistently reported. We did not perform a meta‐analysis, due to scarcity and heterogeneity of data, but all available data are presented in Appendix 7.
Cardiovascular parameters
Short‐ and medium‐term follow‐up (between 5 and 52 weeks)
Eleven trials (5 reporting endpoint values, 6 change scores) showed a difference in pulse of 4.64 beats per minute (95% CI 3.13 to 6.15; I2 = 65%; 11 trials, 12 comparisons, 2558 participants; very low‐certainty evidence), higher for methylphenidate (Analysis 2.9). There was no difference for systolic blood pressure (Analysis 2.10), whereas 10 trials (5 reporting endpoint values, 5 change scores) showed a difference in diastolic blood pressure of 1.24 mm Hg (95% CI 0.04 to 2.44; I2 = 60%; 10 trials, 11 comparisons, 2276 participants; very low‐certainty evidence) higher for methylphenidate (Analysis 2.11). We noted that Konstenius 2007 did not report data on diastolic blood pressure from the placebo group (and it was not included in the analysis), but the methylphenidate group data were the same before and after treatment, down to the decimal place.
Adverse events
Short‐ and medium‐term follow‐up (between 2 and 52 weeks)
For overall dropout, there was no evidence of a difference between the groups (RR 1.01, 95% CI 0.84 to 1.21; I2 = 59%; 21 trials, 22 comparisons, 4769 participants; very low‐certainty evidence; Analysis 3.1). For dropouts due to adverse events, the RR was 2.36 (95% CI 1.62 to 3.43; I2 = 16%; 15 trials, 3963 participants; very low‐certainty evidence), favouring placebo (Analysis 3.2). For participants experiencing any adverse event, the RR was 1.27 (95% CI 1.19 to 1.37; I2 = 69%; 14 trials, 15 comparisons, 4214 participants; very low‐certainty evidence), favouring placebo (Analysis 3.3). For number of adverse events, three trials (951 participants) reported on 2183 events in 586 participants on methylphenidate and 1216 events in 365 participants on placebo (see full dataset; Contributions of authors). For participants experiencing any psychiatric adverse event, the RR was 1.83 (95% CI 1.43 to 2.34; I2 = 68%; 8 trials, 9 comparisons, 2737 participants, very low‐certainty evidence), favouring placebo (Analysis 3.4); note, only one study (Biederman 2003) reported the exact criteria for this outcome. Weight was −1.78 kg (CI 95% CI −2.36 to −1.20, I2 = 81%; 10 trials, 11 comparisons, 2302 participants; very low‐certainty evidence), lower for those on methylphenidate (Analysis 3.5).
We estimated the RR for 12 specific adverse effects (see Table 3, and Analysis 3.6 to Analysis 3.17 for further detail). Ten events occurred more frequently for those on methylphenidate, with RRs ranging from 1.35 (95% CI 1.18 to 1.54) for headache to 5.52 (95% CI 2.77 to 10.99) for feeling jittery. Two events, depression and irritability, did not occur more frequently for those on methylphenidate. We rated the certainty of the evidence as very low for all effect estimates.
Table 3. Adverse events table
| Adverse event | Number of trials (participants) | Risk ratio (95% Confidence Interval) | I2 | Analysis |
|---|---|---|---|---|
| Decreased appetite | 16 trials, 17 comparisons (4491 participants) | 4.08 (3.34 to 4.98) | 0% | |
| Dry mouth | 16 trials, 17 comparisons (4491 participants) | 3.36 (2.61 to 4.34) | 16% | |
| Headache | 16 trials, 17 comparisons (4238 participants) | 1.35 (1.18 to 1.54) | 0% | |
| Palpitations | 15 trials, 16 comparisons (4274 participants) | 3.67 (2.42 to 5.57) | 12% | |
| Insomnia | 16 trials, 17 comparisons (4435 participants) | 1.97 (1.51 to 2.57) | 5% | |
| Sexual dysfunction | 9 trials, 10 comparisons (3057 participants) | 2.66 (1.31 to 5.39) | 0% | |
| Anxiety | 13 trials, 14 comparisons (3829 participants) | 2.23 (1.40 to 3.57) | 28% | |
| Depression | 9 trials, 10 comparisons (2791 participants) | 1.47 (0.87 to 2.49) | 39% | |
| Irritability | 11 trials, 12 comparisons (3377 participants) | 1.39 (1.00 to 1.93) | 0% | |
| Feeling jittery | 7 trials (2254 participants) | 5.52 (2.77 to 10.99) | 0% | |
| Agitation | 10 trials, 11 comparisons (3166 participants) | 2.16 (1.25 to 3.71) | 9% | |
| Aggression | 4 trials, 5 comparisons (1319 participants) | 2.10 (1.32 to 3.33) | 0% |
Subgroup and sensitivity analyses
Six outcomes met our predefined threshold for conducting subgroup and sensitivity analyses, i.e. I2 above 50%. We considered it inappropriate to analyse four of these outcomes, due to probable selective reporting: pulse was reported by 11 trials (Analysis 2.9); diastolic blood pressure by 10 trials (Analysis 2.11); participants experiencing psychiatric adverse events by eight trials (Analysis 3.4); and body weight by 10 trials (Analysis 3.5). We analysed two outcomes: 'overall dropout rate' (I2 = 59%) (Analysis 3.1) and 'participants experiencing any adverse event' (I2 = 69%) (Analysis 3.3).
Overall dropout rate
The four subgroup analyses did not show any differences for 'psychiatric comorbidity' (Analysis 4.1), 'responder selection' (Analysis 4.2, 'withdrawal effects' (Analysis 4.3), or 'conflicts of interest' (Analysis 4.4). We did not conduct the sensitivity analysis based on attrition bias.
Participants experiencing any adverse events
There were no differences in the two subgroup analyses for psychiatric comorbidity (Analysis 4.5) and 'withdrawal effects' (Analysis 4.6), whereas the subgroup analysis of 'conflicts of interest' showed a higher proportion of participants experiencing adverse events in industry‐funded trials compared to publicly‐funded trials with industry involvement (Analysis 4.7). We did not conduct the 'responder selection' subgroup analysis, since all trials were classified at high risk of being enriched. The sensitivity analysis based on attrition bias showed no evidence of a difference (Analysis 4.8).
Extended‐release methylphenidate versus atomoxetine
Weisler 2009 compared OROS methylphenidate 54 mg per day (68 participants) and atomoxetine 80 mg per day (74 participants) at six‐week follow‐up. The trial also included a placebo comparison (these results are reported above).
Primary outcomes
Functional outcomes
There were no data for this outcome.
Self‐rated ADHD symptom severity
There was no difference between OROS methylphenidate and atomoxetine for this outcome (MD −0.66, 95% CI −6.36 to 5.04; 1 trial, 142 participants; very low‐certainty evidence; Analysis 5.1) when assessed with the CAARS:DSM IV total ADHD scale (range 0 to 54 points).
Serious adverse events
There was no difference between OROS methylphenidate and atomoxetine (RR 3.26, 95% CI 0.14 to 78.72; 1 trial, 142 participants; very low‐certainty evidence; Analysis 5.2).
Secondary outcomes
There were differences favouring atomoxetine on two outcomes: initial insomnia (RR 5.44, 95% CI 1.65 to 17.98; 1 trial, 142 participants; very low‐certainty; Analysis 5.3) and weight change (MD −0.91 kg, 95% CI −1.49 to −0.33; 1 trial, 142 participants; very low‐certainty evidence; Analysis 5.4).
There were no differences for the other secondary outcomes of investigator‐rated ADHD symptoms (Analysis 5.5), overall dropout rates (Analysis 5.6), proportion of participants experiencing adverse events (Analysis 5.7), dropouts due to adverse events (Analysis 5.8), headache (Analysis 5.9), decreased appetite (Analysis 5.10), irritability (Analysis 5.11), dry mouth (Analysis 5.12), decreased libido (Analysis 5.13), feeling jittery (Analysis 5.14), or palpitations (Analysis 5.15).
Extended‐release methylphenidate versus bupropion
Levin 1998a compared sustained‐release methylphenidate (32 participants) and bupropion 400 mg per day (33 participants) at 10‐week follow‐up.
Primary outcomes
Functional outcomes
There were no data for this outcome.
Self‐rated ADHD symptom severity
There was no difference between SR methylphenidate and bupropion for self‐rated ADHD symptom severity (MD 3.46, 95% CI −1.40 to 8.32; 1 trial, 65 participants; very low‐certainty evidence), assessed with the Adult ADHD Rating Scale (range 0 to 54 points) (Analysis 6.1).
Serious adverse events
There were no data for this outcome.
Secondary outcomes
There were no reported differences between the groups for overall dropout rates (Analysis 6.2), or dropouts due to adverse events (Analysis 6.3); 1 trial, 65 participants.
Certainty of the evidence
We rated the certainty of the evidence for all outcomes and for all comparisons as very low. We downgraded our certainty in the evidence in two domains; 'risk of bias' and 'indirectness'. We judged 20 of the 21 trials that reported outcome data to be at high risk of bias overall, and one trial (Ginsberg 2007) at unclear risk of bias overall. Several trial design characteristics contributed to the downgrading in the 'indirectness' domain, such as short trial duration, exclusion of participants with psychiatric comorbidity, responder selection ('enriched design'), and withdrawal effects. See also (Assessment of risk of bias in included studies).
Discussion
Summary of main results
We included 24 clinical trials (5066 participants) using six different extended‐release methylphenidate formulations for adults with ADHD. Twenty‐one trials reported data for this review, of which 12 were sponsored by industry.
We found very low‐certainty evidence that the reported beneficial effects of methylphenidate over placebo assessed with symptom‐rating scales by the participants, investigators, or peers such as family members, were small to moderate. The effect on quality of life was small, and there was no effect on 'days missed at work'. In comparing the two groups, we found no increase in the risk of serious adverse events or overall dropout among those receiving methylphenidate, whereas the risk of experiencing any adverse event, dropping out due to an adverse event, and the risk of experiencing any psychiatric adverse event were higher in the methylphenidate group. The relative risk for 10 of the 12 assessed adverse events were higher in the methylphenidate group than placebo, including headache, dry mouth, palpitations and psychiatric adverse events such as anxiety, agitation and feeling jittery, whereas there was no increased risk for two events, depression and irritability. We considered 20 out of 21 trials to be at high risk of bias, particularly due to an uncertainty related to the blinding of participants and investigators, high attrition rates, and selective outcome reporting. Eighteen of the trials had a follow‐up between 2 and 13 weeks, two trials at 24 weeks, and one trial at 52 weeks. The evidence is therefore mainly applicable to short‐term treatment, with long‐term effects largely unknown. Finally, we identified important limitations about the generalisability of the results.
Two trials, Levin 1998a and Weisler 2009, reported on comparisons of extended‐release methylphenidate versus other ADHD medications, bupropion and atomoxetine. The trials had modest sample sizes (98 and 216 participants) and the direct comparative evidence is therefore sparse. There were no differences between the drugs on most reported outcomes, and no conclusions can be drawn. We also note with concern that we did not identify trials comparing immediate‐release with extended‐release methylphenidate, nor different extended‐release formulations compared with each other. We are currently raising the issue related to the lack of direct comparisons in ADHD drug authorisations (Boesen 2021b; Boesen 2021c).
Overall completeness and applicability of evidence
We searched drug regulatory agency databases, and included more than 1100 pages of regulatory documents from drug approval reviews, public assessment reports, and other documents. Despite these efforts, our results are limited by selective reporting. Important patient‐relevant outcomes were not reported in a number of the included trials: 14 of 21 trials reported on serious adverse events, 10 reported on cardiovascular variables; six of 10 trials reported on prespecified quality‐of‐life outcomes (of which, only one study, Takahashi 2011, appeared in a journal publication; the other results were obtained from regulatory documents), and only one trial reported a functional outcome. Additionally, only seven of our 23 attempts to contact authors were fruitful in obtaining more information or more outcome data from the clinical trials (Appendix 5).
The rates of important psychiatric harms are uncertain. In the British drug regulator's OROS methylphenidate public assessment report (British Medicines and Healthcare products Regulatory Agency; MHRA), we noted that “psychosis/mania” was reported to occur in 3% of participants on methylphenidate versus 1% in the placebo group during the double‐blind placebo‐controlled phase (MHRA 2010, p 63). This estimate, based on three pivotal trials (Medori 2005; Adler 2006; Casas 2008), is 30 times higher than the 0.1% risk of “new psychotic or symptoms” according to the FDA’s Prescribing Information (FDA 2017, section 5.2). We also noted large differences across trials that were not explained by obvious trial differences. For example, Rösler 2004 reported that 11% of participants on methylphenidate experienced "reduced libido", compared to 1% in the pooled analysis of Medori 2005, Adler 2006, and Casas 2008 (MHRA 2010, p 61). The Dutch drug regulatory agency (Dutch Medicines Evaluation Board; MEB) noted in their public assessment report of long‐acting methylphenidate (which they did not approve for market authorisation) that the strict exclusion criteria probably reduced the incidence of reported harms in the trials:
“Psychiatric AEs were common (i.e. total of 43%, including insomnia 11%, anxiety 7%, irritability 7%, and restlessness 7%) and some were serious. Since these events were more common in patients with a psychiatric history and since elaborate psychiatric exclusion criteria were implemented in the study, which prevented inclusion of patients with more serious psychiatric comorbidity, it is concluded that the rate of psychiatric AEs would have been even higher without these exclusions” (MEB 2016, p 36).
In terms of applicability, there is, as far as we know, limited empirical evidence on minimal important differences in rating scales used in ADHD drug trials. It is therefore uncertain whether the reported effects are clinically relevant. For the quality‐of‐life scale, Quality of Life Enjoyment and Satisfaction Questionnaire‐Short Form (range 14 to 70 points), the minimal important difference is estimated to be three points (Mick 2008); however, we found a difference between methylphenidate and placebo of only one point (Analysis 2.2). The lack of direct comparisons between methylphenidate and other ADHD drugs also limits the clinical applicability of the evidence.
Quality of the evidence
We rated the overall certainty of the evidence as 'very low' due to high risk of bias and limitations in the external validity. This means that further research is likely to change the effect estimates. The statistical heterogeneity was low for most effect estimates, which was unsurprising, since most participants were enrolled in homogeneous, industry‐sponsored trials.
A major limitation to the evidence is the limited generalisability due to several problematic trial design characteristics; exclusion of participants with psychiatric comorbidity, responder selection based on previous treatment response, potential withdrawal effects, short trial durations, and a lack of patient‐relevant outcomes. In MHRA's public assessment report on OROS methylphenidate, they noted, “Current psychiatric morbidity and a history of psychiatric morbidity appeared to reduce the effect size (excluding the 72mg dose in study 3013). No obvious pattern of differences was seen in whether the adult was MPH naïve or not” (MHRA 2010, p 51). However, in order to properly address the impact of such study designs we would need access to Clinical Study Reports and individual patient data. Ideally, new and less biased trials should be conducted.
Potential biases in the review process
There are several biases and limitations in the review. An important limitation may be the omission of potentially relevant data. As stated above, we included a large amount of regulatory data, but we were also denied access to clinical data. In 2016, we asked Medice, the Marketing Authorisation Holder of Medikinet in Europe, for access to trial data from their two pivotal trials (Rösler 2004 and Retz 2008). Medice’s medical director declined our request, with the argument that Cochrane Reviews are “full of errors and biased” and then cited an article by Banaschewski 2016, which criticised a Cochrane Review of methylphenidate for children and adolescents with ADHD (Storebø 2015). In 2019, we asked the German drug regulatory agency, Bundesinstitut für Arzneimittel und Medizinprodukte (BfArM) about access to the same trial data. The German regulator consulted with Medice, which again declined the request. We recently raised this rejection with the Heads of Medicines Agencies (Boesen 2021d; Boesen 2021e), which is a European umbrella organisation consisting of the heads of the European national drug regulatory agencies. This rejection of access to Clinical Study Reports conflicts with the European Medicines Agency's transparency policies 0043 (EMA 2018) and 0070 (EMA 2019) that grant access to data from centrally‐authorised drugs (Gøtzsche 2011; CURIA 2020; Dyer 2020).
The composite outcome of 'severe psychiatric adverse event' was not well‐suited to estimate psychiatric harms, and one should always be careful with composite outcomes. It was also a limitation that we did not specify which adverse events to assess at protocol stage. For future reviews of psychiatric drugs, we recommend prespecifying individual harms, preferably in collaboration with patients. It would also be preferable to pool adverse events according to their Medical Dictionary for Regulatory Activities (MedDRA) terms, rather than analysing individual events. Miscoding and splitting up events into separate terms may dilute potential signals of harms and make the analyses unreliable (Maund 2014). Ideally, our high‐level outcomes of harm (i.e. aggregate outcomes such as 'dropout due to adverse events'), should be defined at the protocol stage also, see Differences between protocol and review.
We did not prespecify a hierarchy for extracting data from multiple resources. Such an approach is recommended in complex reviews to avoid selectively reporting available results (Mayo‐Wilson 2017; Mayo‐Wilson 2018). We included the data that were judged as most complete and in best accordance with our specified criteria (see 'Dealing with missing data’). To be as transparent as possible, we share the full dataset (see link in Contributions of authors). Separate efforts would be required to systematically assess potential differences (Mayo‐Wilson 2017a; Jørgensen 2020; Paludan‐Müller 2021).
In most included trials, participants were allowed (to varying degrees) to continue psychopharmacological treatments, such as antidepressants and anxiolytics. How these drugs affected the reported outcomes is not known. We would probably need detailed information at an individual participant level to perform reliable subgroup analyses based on treatment status. We decided not to go into this question in detail in the current review. However, it seems to be an important research question that needs to be addressed, provided there are sufficiently granulated data.
To our knowledge, this is the first review to systematically assess external validity issues in trials of methylphenidate for ADHD in adults. It should be highlighted that our attempt to operationalise these design characteristics is open for discussion. We hope our review may increase the focus on external validity issues, which may be as pertinent from a clinical point of view as the standard risk of bias assessment when assessing the benefits and harms of drugs like methylphenidate.
Agreements and disagreements with other studies or reviews
Our results for adverse effects, cardiovascular effects, and weight changes correspond with previous systematic reviews of ADHD mediations for children and adolescents (Storebø 2015; Liang 2018; Catalá‐López 2017; Storebø 2018) and for adults (Mick 2013). Our quality assessment also agrees with Cochrane Reviews on methylphenidate for ADHD in children and adolescents (Storebø 2015; Storebø 2018), immediate‐release methylphenidate for adults (Cândido 2021), and amphetamines for ADHD in children and adolescents (Punja 2016) and adults (Castells 2018). However, it is difficult to compare our results directly with non‐Cochrane Reviews on methylphenidate for adult ADHD (Faraone 2004; Peterson 2008; Koesters 2009; Meszaros 2009; Faraone 2010; Castells 2011; Castells 2013; Cunill 2016; Cortese 2018). These reviews used different methodologies and included different drugs and trials. One of the most recent systematic reviews (Cortese 2018) reported similar effects on ADHD symptom scales, dropout rates due to adverse events, and overall dropout rates for methylphenidate. However, the review did not assess functional outcomes, quality of life or individual adverse events, and the review authors rated the certainty of the evidence for some of the outcomes as ‘moderate’ (clinician‐rated ADHD symptoms) or ‘high’ (dropout due to adverse events), whereas we rated the certainty of the evidence as ‘very low’ for all outcomes. The most recent systematic review by Elliott 2020 identified clinical trial limitations, such as external validity issues, short trial duration, and a lack of patient‐reported outcomes, similar to our own findings. The updated sister review to this review on immediate‐release methylphenidate for adult ADHD (Cândido 2021) concluded, “Clinicians should consider whether it is appropriate to prescribe IR methylphenidate, given its limited efficacy and increased risk of harms. Future RCTs should explore the long‐term efficacy and risks of IR methylphenidate, and the influence of conflicts of interest on reported effects”.
Flow diagram.
BfArM= Bundesinstitut für Arzneimittel und Medizinprodukte, FDA= US Food and Drug Administration, MEB= Medicines Evaluation Board, MHRA= Medicines and Healthcare products Regulatory Agency, TGA= Therapeutic Goods Administration, YODA= Yale University Open Data Access.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Funnel plot. Extended‐release methylphenidate versus placebo: ADHD symptoms (self‐rated).
Funnel plot of comparison: 1 Extended‐release methylphenidate versus placebo ‐ primary outcomes (short and medium term), outcome: 1.8 Serious adverse events.
Funnel plot of comparison: 2 Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short‐ and medium‐term), outcome: 2.4 ADHD symptoms (investigator‐rated) (short‐term).
Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 1: Days missed at work/school/training (week 13)
Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 2: Days missed at work/school/training (weeks 13‐26)
Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 3: Days missed at work/school/training (weeks 26‐52)
Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 4: ADHD symptoms (self‐rated) (short‐term)
![Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 5: ADHD symptoms (self‐rated) [subgroup MD] (short‐term)](/es/cdsr/doi/10.1002/14651858.CD012857.pub2/media/CDSR/CD012857/image_n/nCD012857-CMP-001.05.svg)
Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 5: ADHD symptoms (self‐rated) [subgroup MD] (short‐term)
Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 6: ADHD symptoms (self‐rated) (medium‐term)
![Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 7: ADHD symptoms (self‐rated) [LMCF data] (medium‐term)](/es/cdsr/doi/10.1002/14651858.CD012857.pub2/media/CDSR/CD012857/image_n/nCD012857-CMP-001.07.svg)
Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 7: ADHD symptoms (self‐rated) [LMCF data] (medium‐term)
Comparison 1: Extended‐release methylphenidate versus placebo ‐ primary outcomes (short‐ and medium‐term), Outcome 8: Serious adverse events
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 1: Quality of life (self‐rated) (short‐term)
![Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 2: Quality of life (self‐rated) [subgroup MD] (short‐term)](/es/cdsr/doi/10.1002/14651858.CD012857.pub2/media/CDSR/CD012857/image_n/nCD012857-CMP-002.02.svg)
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 2: Quality of life (self‐rated) [subgroup MD] (short‐term)
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 3: Quality of life (self‐rated) (medium‐term)
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 4: ADHD symptoms (investigator rated) (short‐term)
![Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 5: ADHD symptoms (investigator‐rated) [subgroup MD] (short‐term)](/es/cdsr/doi/10.1002/14651858.CD012857.pub2/media/CDSR/CD012857/image_n/nCD012857-CMP-002.05.svg)
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 5: ADHD symptoms (investigator‐rated) [subgroup MD] (short‐term)
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 6: ADHD symptoms (investigator‐rated) (medium‐term)
![Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 7: ADHD symptoms (investigator‐rated) [LMCF data] (medium‐term)](/es/cdsr/doi/10.1002/14651858.CD012857.pub2/media/CDSR/CD012857/image_n/nCD012857-CMP-002.07.svg)
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 7: ADHD symptoms (investigator‐rated) [LMCF data] (medium‐term)
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 8: ADHD symptoms (peer‐rated) (short‐term)
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 9: Cardiovascular: pulse (beats per minute)
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 10: Cardiovascular: systolic blood pressure
Comparison 2: Extended‐release methylphenidate versus placebo ‐ secondary outcomes (short and medium term), Outcome 11: Cardiovascular: diastolic blood pressure
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 1: Dropout (overall)
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 2: Dropout (due to adverse events)
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 3: Adverse events (proportion experiencing any adverse event)
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 4: Psychiatric adverse events (proportion of participants)
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 5: Weight (change in kilograms)
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 6: Adverse event: decreased appetite
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 7: Adverse event: dry mouth
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 8: Adverse event: headache
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 9: Adverse event: palpitations
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 10: Adverse event: insomnia
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 11: Adverse event: sexual dysfunction
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 12: Adverse event: anxiety
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 13: Adverse event: depression
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 14: Adverse event: irritability
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 15: Adverse event: feeling jittery
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 16: Adverse event: agitation
Comparison 3: Extended‐release methylphenidate versus placebo ‐ secondary outcomes: adverse events (short‐ and medium‐term), Outcome 17: Adverse event: aggression
Comparison 4: Extended‐release methylphenidate versus placebo: subgroup and sensitivity analyses, Outcome 1: Overall dropout rate: psychiatric comorbidity (subgroup analysis)
Comparison 4: Extended‐release methylphenidate versus placebo: subgroup and sensitivity analyses, Outcome 2: Overall dropout rate: 'enriched design' (subgroup analysis)
Comparison 4: Extended‐release methylphenidate versus placebo: subgroup and sensitivity analyses, Outcome 3: Overall dropout rate: withdrawal effects (subgroup analysis)
Comparison 4: Extended‐release methylphenidate versus placebo: subgroup and sensitivity analyses, Outcome 4: Overall dropout rate: funding (subgroup analysis)
Comparison 4: Extended‐release methylphenidate versus placebo: subgroup and sensitivity analyses, Outcome 5: Proportion experiencing any adverse event: psychiatric comorbidity (subgroup analysis)
Comparison 4: Extended‐release methylphenidate versus placebo: subgroup and sensitivity analyses, Outcome 6: Proportion experiencing any adverse event: withdrawal effects (subgroup analysis)
Comparison 4: Extended‐release methylphenidate versus placebo: subgroup and sensitivity analyses, Outcome 7: Proportion experiencing any adverse event: funding (subgroup analysis)
Comparison 4: Extended‐release methylphenidate versus placebo: subgroup and sensitivity analyses, Outcome 8: Proportion experiencing any adverse event: dropout rates (sensitivity analysis)
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 1: ADHD symptoms (self‐rated)
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 2: Serious adverse events
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 3: Adverse event: initial insomnia
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 4: Weight
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 5: ADHD symptoms (investigator‐rated)
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 6: Overall dropout
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 7: Adverse events (AE): proportion experiencing any AE
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 8: Adverse events (leading to dropout)
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 9: Adverse event: headache
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 10: Adverse event: decreased appetite
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 11: Adverse event: irritability
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 12: Adverse event: dry mouth
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 13: Adverse event: decreased libido
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 14: Adverse event: feeling jittery
Comparison 5: Extended‐release methylphenidate versus atomoxetine, Outcome 15: Adverse events: palpitations
Comparison 6: Extended‐release methylphenidate versus bupropion, Outcome 1: ADHD symptoms (self‐rated)
Comparison 6: Extended‐release methylphenidate versus bupropion, Outcome 2: Overall dropout
Comparison 6: Extended‐release methylphenidate versus bupropion, Outcome 3: Dropouts due to adverse events
| Extended‐release methylphenidate compared to placebo for ADHD | ||||||
| Patient or population: adults diagnosed with ADHD | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with placebo | Risk with extended‐release methylphenidate | |||||
| Functional outcome:days missed at work/ school/ training
Time of assessment: 13 weeks | The mean number of days missed at work, school or training in the control group was 2.4 | The mean number of days missed at work, school or training was 0.15 days fewer | ‐ | 409 | ⊕⊝⊝⊝ | ‐ |
| ADHD symptoms (self‐rated)
Time of assessment: median 8 weeks (range 2 ‐ 26) | ‐ | The mean ADHD score in the intervention groups was 0.37 SDlower | ‐ | 3799 | ⊕⊝⊝⊝ | An SMD of 0.37 may be considered a small‐to‐moderate effect size |
| Serious adverse events
Time of assessment: median 8 weeks (range 4 ‐ 52) | Study population | RR1.43 | 4078 | ⊕⊝⊝⊝ | ‐ | |
| 12 per 1000 | 17 per 1000 | |||||
| Quality of life (self‐rated)
Time of assessment: median 6 weeks (range 4 ‐ 52) | ‐ | The mean quality of life score in the intervention groups was 0.15 SD lower (0.25 lower to 0.05 lower) | ‐ | 1888 | ⊕⊝⊝⊝ | An SMD of 0.15 may be considered a small effect size |
| ADHD symptoms (investigator‐rated)
Time of assessment: median 8 weeks (range 2 ‐ 26) | ‐ | The mean investigator‐rated ADHD score in the intervention groups was 0.42 SD lower | ‐ | 4183 | ⊕⊝⊝⊝ | An SMD of 0.42 may be considered a small‐to‐moderate effect size |
| ADHD symptoms (peer‐rated)
Time of assessment: median 7.5 weeks (range 6 ‐ 9) | ‐ | The mean peer‐rated ADHD score in the intervention groups was 0.31 SD lower | ‐ | 1005 | ⊕⊝⊝⊝ | An SMD of 0.31 may be considered a small‐to‐moderate effect size |
| Adverse events (proportion experiencing any adverse event)
Time of assessment: median 8 weeks (range 4 ‐ 52) | Study population | RR1.27 | 4214 | ⊕⊝⊝⊝ | ‐ | |
| 641 per 1000 | 814 per 1000 | |||||
| *The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aWe downgraded two levels due to risk bias; primarily due to blinding issues caused by methylphenidate's effects, high attrition (dropout) rates, and selective reporting of the included trials. | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Days missed at work/school/training (week 13) Show forest plot | 2 | 409 | Mean Difference (IV, Random, 95% CI) | ‐0.15 [‐2.11, 1.81] |
| 1.2 Days missed at work/school/training (weeks 13‐26) Show forest plot | 2 | 327 | Mean Difference (IV, Random, 95% CI) | ‐0.42 [‐2.48, 1.64] |
| 1.3 Days missed at work/school/training (weeks 26‐52) Show forest plot | 2 | 292 | Mean Difference (IV, Random, 95% CI) | 0.71 [‐1.61, 3.03] |
| 1.4 ADHD symptoms (self‐rated) (short‐term) Show forest plot | 17 | 3799 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.37 [‐0.43, ‐0.30] |
| 1.5 ADHD symptoms (self‐rated) [subgroup MD] (short‐term) Show forest plot | 17 | Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 1.5.1 DSM‐IV Total ADHD (18 items, 0‐54 points) | 7 | 1189 | Mean Difference (IV, Random, 95% CI) | ‐3.05 [‐4.34, ‐1.76] |
| 1.5.2 ASRS (18 items, 0‐72 points) | 2 | 953 | Mean Difference (IV, Random, 95% CI) | ‐6.25 [‐8.25, ‐4.25] |
| 1.5.3 CAARS: short version (26 items, 0‐78 points) | 5 | 1451 | Mean Difference (IV, Random, 95% CI) | ‐5.21 [‐7.14, ‐3.29] |
| 1.5.4 CAARS: long version (66 items, 0‐90 points) | 1 | 162 | Mean Difference (IV, Random, 95% CI) | ‐28.40 [‐48.28, ‐8.52] |
| 1.5.5 CAARS: unspecified subscale | 2 | 44 | Mean Difference (IV, Random, 95% CI) | ‐8.23 [‐17.79, 1.32] |
| 1.6 ADHD symptoms (self‐rated) (medium‐term) Show forest plot | 2 | 234 | Mean Difference (IV, Random, 95% CI) | ‐1.59 [‐4.26, 1.09] |
| 1.7 ADHD symptoms (self‐rated) [LMCF data] (medium‐term) Show forest plot | 2 | 419 | Mean Difference (IV, Random, 95% CI) | ‐2.23 [‐3.50, ‐0.95] |
| 1.8 Serious adverse events Show forest plot | 14 | 4078 | Risk Ratio (IV, Random, 95% CI) | 1.43 [0.85, 2.43] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 Quality of life (self‐rated) (short‐term) Show forest plot | 7 | 1888 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.15 [‐0.25, ‐0.05] |
| 2.2 Quality of life (self‐rated) [subgroup MD] (short‐term) Show forest plot | 7 | Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 2.2.1 Q‐LES‐QSF (14‐70 points) | 5 | 1234 | Mean Difference (IV, Random, 95% CI) | ‐0.93 [‐2.17, 0.31] |
| 2.2.2 AIM‐A total score | 1 | 279 | Mean Difference (IV, Random, 95% CI) | ‐0.50 [‐0.99, ‐0.00] |
| 2.2.3 AAQOL | 1 | 375 | Mean Difference (IV, Random, 95% CI) | ‐6.10 [‐10.67, ‐1.53] |
| 2.3 Quality of life (self‐rated) (medium‐term) Show forest plot | 2 | 419 | Mean Difference (IV, Random, 95% CI) | ‐0.66 [‐2.97, 1.64] |
| 2.4 ADHD symptoms (investigator rated) (short‐term) Show forest plot | 19 | 4183 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.42 [‐0.49, ‐0.36] |
| 2.5 ADHD symptoms (investigator‐rated) [subgroup MD] (short‐term) Show forest plot | 19 | Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 2.5.1 DSM‐IV Total ADHD (18 items, 0‐54 points) | 12 | 3121 | Mean Difference (IV, Random, 95% CI) | ‐4.74 [‐5.65, ‐3.84] |
| 2.5.2 DSM‐5 Total ADHD (18 items, 0‐54 points) | 1 | 375 | Mean Difference (IV, Random, 95% CI) | ‐4.38 [‐7.37, ‐1.39] |
| 2.5.3 WRAADDS (28 items, 0‐56 points) | 2 | 521 | Mean Difference (IV, Random, 95% CI) | ‐6.06 [‐8.48, ‐3.63] |
| 2.5.4 CGI (range 0‐7) | 2 | 36 | Mean Difference (IV, Random, 95% CI) | 0.09 [‐0.67, 0.86] |
| 2.5.5 TAADS (7 items, 0‐28 points) | 1 | 106 | Mean Difference (IV, Random, 95% CI) | ‐3.50 [‐7.85, 0.85] |
| 2.5.6 CAARS: unspecified scale | 1 | 24 | Mean Difference (IV, Random, 95% CI) | 0.10 [‐10.21, 10.41] |
| 2.6 ADHD symptoms (investigator‐rated) (medium‐term) Show forest plot | 2 | 243 | Mean Difference (IV, Random, 95% CI) | ‐1.29 [‐3.38, 0.80] |
| 2.7 ADHD symptoms (investigator‐rated) [LMCF data] (medium‐term) Show forest plot | 2 | 419 | Mean Difference (IV, Random, 95% CI) | ‐2.18 [‐3.55, ‐0.80] |
| 2.8 ADHD symptoms (peer‐rated) (short‐term) Show forest plot | 3 | 1005 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.31 [‐0.48, ‐0.14] |
| 2.9 Cardiovascular: pulse (beats per minute) Show forest plot | 11 | 2558 | Mean Difference (IV, Random, 95% CI) | 4.64 [3.13, 6.15] |
| 2.10 Cardiovascular: systolic blood pressure Show forest plot | 10 | 2276 | Mean Difference (IV, Random, 95% CI) | 1.11 [‐0.06, 2.27] |
| 2.11 Cardiovascular: diastolic blood pressure Show forest plot | 10 | 2276 | Mean Difference (IV, Random, 95% CI) | 1.24 [0.04, 2.44] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 3.1 Dropout (overall) Show forest plot | 22 | 4769 | Risk Ratio (IV, Random, 95% CI) | 1.01 [0.84, 1.21] |
| 3.2 Dropout (due to adverse events) Show forest plot | 15 | 3963 | Risk Ratio (IV, Random, 95% CI) | 2.36 [1.62, 3.43] |
| 3.3 Adverse events (proportion experiencing any adverse event) Show forest plot | 14 | 4214 | Risk Ratio (IV, Random, 95% CI) | 1.27 [1.19, 1.37] |
| 3.4 Psychiatric adverse events (proportion of participants) Show forest plot | 7 | 2737 | Risk Ratio (IV, Random, 95% CI) | 1.83 [1.43, 2.34] |
| 3.5 Weight (change in kilograms) Show forest plot | 10 | 2302 | Mean Difference (IV, Random, 95% CI) | ‐1.78 [‐2.36, ‐1.20] |
| 3.6 Adverse event: decreased appetite Show forest plot | 16 | 4491 | Risk Ratio (IV, Random, 95% CI) | 4.08 [3.34, 4.98] |
| 3.7 Adverse event: dry mouth Show forest plot | 16 | 4491 | Risk Ratio (IV, Random, 95% CI) | 3.36 [2.61, 4.34] |
| 3.8 Adverse event: headache Show forest plot | 16 | 4238 | Risk Ratio (IV, Random, 95% CI) | 1.35 [1.18, 1.54] |
| 3.9 Adverse event: palpitations Show forest plot | 15 | 4265 | Risk Ratio (IV, Random, 95% CI) | 3.67 [2.42, 5.57] |
| 3.10 Adverse event: insomnia Show forest plot | 16 | 4435 | Risk Ratio (IV, Random, 95% CI) | 1.97 [1.51, 2.57] |
| 3.11 Adverse event: sexual dysfunction Show forest plot | 7 | 3057 | Risk Ratio (IV, Random, 95% CI) | 2.66 [1.31, 5.39] |
| 3.12 Adverse event: anxiety Show forest plot | 13 | 3829 | Risk Ratio (IV, Random, 95% CI) | 2.23 [1.40, 3.57] |
| 3.13 Adverse event: depression Show forest plot | 7 | 2791 | Risk Ratio (IV, Random, 95% CI) | 1.47 [0.87, 2.49] |
| 3.14 Adverse event: irritability Show forest plot | 11 | 3377 | Risk Ratio (IV, Random, 95% CI) | 1.39 [1.00, 1.93] |
| 3.15 Adverse event: feeling jittery Show forest plot | 7 | 2254 | Risk Ratio (IV, Random, 95% CI) | 5.52 [2.77, 10.99] |
| 3.16 Adverse event: agitation Show forest plot | 8 | 3166 | Risk Ratio (IV, Random, 95% CI) | 2.16 [1.25, 3.71] |
| 3.17 Adverse event: aggression Show forest plot | 2 | 1319 | Risk Ratio (IV, Random, 95% CI) | 2.10 [1.32, 3.33] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 4.1 Overall dropout rate: psychiatric comorbidity (subgroup analysis) Show forest plot | 22 | 4769 | Risk Ratio (IV, Random, 95% CI) | 1.01 [0.84, 1.21] |
| 4.1.1 Low risk | 1 | 54 | Risk Ratio (IV, Random, 95% CI) | 0.74 [0.53, 1.03] |
| 4.1.2 Unclear risk | 1 | 30 | Risk Ratio (IV, Random, 95% CI) | Not estimable |
| 4.1.3 High risk | 20 | 4685 | Risk Ratio (IV, Random, 95% CI) | 1.04 [0.86, 1.25] |
| 4.2 Overall dropout rate: 'enriched design' (subgroup analysis) Show forest plot | 22 | 4769 | Risk Ratio (IV, Random, 95% CI) | 1.01 [0.84, 1.21] |
| 4.2.1 Unclear risk | 2 | 247 | Risk Ratio (IV, Random, 95% CI) | 1.02 [0.26, 3.98] |
| 4.2.2 High risk | 20 | 4522 | Risk Ratio (IV, Random, 95% CI) | 0.99 [0.83, 1.19] |
| 4.3 Overall dropout rate: withdrawal effects (subgroup analysis) Show forest plot | 22 | 4769 | Risk Ratio (IV, Random, 95% CI) | 1.01 [0.84, 1.21] |
| 4.3.1 Low risk | 1 | 357 | Risk Ratio (IV, Random, 95% CI) | 0.91 [0.61, 1.34] |
| 4.3.2 Unclear risk | 5 | 822 | Risk Ratio (IV, Random, 95% CI) | 0.90 [0.62, 1.30] |
| 4.3.3 High risk | 16 | 3590 | Risk Ratio (IV, Random, 95% CI) | 1.07 [0.85, 1.34] |
| 4.4 Overall dropout rate: funding (subgroup analysis) Show forest plot | 22 | 4769 | Risk Ratio (IV, Random, 95% CI) | 1.01 [0.84, 1.21] |
| 4.4.1 Industry funded | 13 | 3782 | Risk Ratio (IV, Random, 95% CI) | 1.14 [0.88, 1.47] |
| 4.4.2 Publicly funded with industry involvement | 3 | 688 | Risk Ratio (IV, Random, 95% CI) | 0.73 [0.59, 0.91] |
| 4.4.3 Publicly funded | 5 | 279 | Risk Ratio (IV, Random, 95% CI) | 0.99 [0.70, 1.40] |
| 4.4.4 Unclear funding | 1 | 20 | Risk Ratio (IV, Random, 95% CI) | 0.33 [0.04, 2.69] |
| 4.5 Proportion experiencing any adverse event: psychiatric comorbidity (subgroup analysis) Show forest plot | 14 | 4214 | Risk Ratio (IV, Random, 95% CI) | 1.27 [1.19, 1.37] |
| 4.5.1 High | 13 | 4184 | Risk Ratio (IV, Random, 95% CI) | 1.26 [1.18, 1.36] |
| 4.5.2 Unclear | 1 | 30 | Risk Ratio (IV, Random, 95% CI) | 2.40 [1.12, 5.13] |
| 4.6 Proportion experiencing any adverse event: withdrawal effects (subgroup analysis) Show forest plot | 14 | 4214 | Risk Ratio (IV, Random, 95% CI) | 1.27 [1.19, 1.37] |
| 4.6.1 High | 11 | 3310 | Risk Ratio (IV, Random, 95% CI) | 1.28 [1.18, 1.38] |
| 4.6.2 Unclear | 2 | 555 | Risk Ratio (IV, Random, 95% CI) | 1.21 [0.95, 1.53] |
| 4.6.3 Low | 1 | 349 | Risk Ratio (IV, Random, 95% CI) | 1.46 [1.22, 1.74] |
| 4.7 Proportion experiencing any adverse event: funding (subgroup analysis) Show forest plot | 14 | 4214 | Risk Ratio (IV, Random, 95% CI) | 1.27 [1.19, 1.37] |
| 4.7.1 Industry funded | 12 | 3545 | Risk Ratio (IV, Random, 95% CI) | 1.32 [1.23, 1.41] |
| 4.7.2 Publicly funded with industry involvement | 2 | 669 | Risk Ratio (IV, Random, 95% CI) | 1.09 [1.05, 1.14] |
| 4.8 Proportion experiencing any adverse event: dropout rates (sensitivity analysis) Show forest plot | 14 | 4214 | Risk Ratio (IV, Random, 95% CI) | 1.27 [1.19, 1.37] |
| 4.8.1 High | 10 | 3363 | Risk Ratio (IV, Random, 95% CI) | 1.21 [1.14, 1.29] |
| 4.8.2 Unclear | 1 | 375 | Risk Ratio (IV, Random, 95% CI) | 1.46 [1.12, 1.89] |
| 4.8.3 Low | 3 | 476 | Risk Ratio (IV, Random, 95% CI) | 1.56 [1.35, 1.81] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 5.1 ADHD symptoms (self‐rated) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 5.2 Serious adverse events Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.3 Adverse event: initial insomnia Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.4 Weight Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 5.5 ADHD symptoms (investigator‐rated) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 5.6 Overall dropout Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.7 Adverse events (AE): proportion experiencing any AE Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.8 Adverse events (leading to dropout) Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.9 Adverse event: headache Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.10 Adverse event: decreased appetite Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.11 Adverse event: irritability Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.12 Adverse event: dry mouth Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.13 Adverse event: decreased libido Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.14 Adverse event: feeling jittery Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 5.15 Adverse events: palpitations Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 6.1 ADHD symptoms (self‐rated) Show forest plot | 1 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| 6.2 Overall dropout Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
| 6.3 Dropouts due to adverse events Show forest plot | 1 | Risk Ratio (IV, Random, 95% CI) | Totals not selected | |
