The study by Asherson and colleagues

Asherson et al^{Reference Asherson2} conducted the largest double-blind RCT assessing the effects of methylphenidate in prison inmates with ADHD. The study included 200 participants aged 16–25 from two prisons (one in England and the other in Scotland), 101 randomised to osmotic-release oral system methylphenidate (OROS-methylphenidate) and 99 to placebo. At the study end-point (8 weeks), the authors observed what they defined as – and many in the field would agree – an ‘unexpected’ finding: there were no significant differences between participants randomised to OROS-methylphenidate and placebo on the primary (ADHD symptom severity on the observer-rated Conners’ Adult ADHD Rating Scale, CAARS-O) and on any of the 13 secondary outcomes, including measures of emotion dysregulation, psychopathology, mind-wandering, attitudes towards violence, global impression of therapeutic effect, reports from prison and educational staff and the number of critical incidents and education sessions attended reported in prison records. Also, no significant differences were noted in terms of blood pressure and heart rate, which raises concerns on subjectively reported adherence. The lack of separation of methylphenidate from placebo is at odds with findings from a previous smaller double-blind RCT^{Reference Ginsberg and Lindefors3} showing a striking superiority of OROS-methylphenidate (Cohen's d = 2.17) in improving ADHD symptom severity on the CAARS-O in a group of 30 prison inmates, aged 21–61, from a high-security prison in Sweden.

Why did methylphenidate not separate from placebo?

Asherson and colleagues were very thorough in assessing possible reasons for these unexpected findings. As only 41.5% of those assigned to OROS-methylphenidate took the medication on at least 75% of the days on which it was prescribed, the authors performed a sensitivity analysis focused on those with high adherence; results did not change substantially. This highlights the challenge of increasing adherence, rather than the concerns about overuse/misuse of methylphenidate, in the prison setting.

Also, as a substantial proportion of participants presented with comorbid mental health conditions or substance use disorder, additional sensitivity analyses were conducted on the subgroup of participants without comorbid mental health or drug and alcohol use disorders, but again, methylphenidate did not separate from placebo. Furthermore, borderline personality disorder, childhood trauma, and reactive and proactive aggression did not significantly moderate the response to the medication. Additionally, the final medication dose, self-reported drug use and diagnostic certainty were not significantly associated with medication response. The Hawthorne effect, i.e., in this case, the possibility that prison inmates could have reported improvements in symptoms to please the assessors, was ruled out by non-significant findings when considering ratings from educational and prison staff.

Should prison inmates with ADHD be treated with methylphenidate?

So, should this large and well-conducted trial lead to the conclusion that response to methylphenidate in prisoners with ADHD is poor and this medication should not be recommended in this population? In my view, the answer is no.

First, it would be unexpected for a single study to drive clinical recommendations. We would need additional double-blind RCTs from other research groups and ideally meta-analytic evidence. It would be important for possible future meta-analyses to assess the effect of substance use disorder and psychiatric comorbidities, which are highly prevalent in this clinical population.

Second, even though RCT is the preferred design to test the effects of an intervention, it does have limitations, and data from RCTs should be considered alongside evidence from observational studies that include outcomes that are rarely included in RCTs. Although observational studies are hampered by the lack of randomisation, a particular design, referred to as within-individual design, has been successfully used in the field of ADHD pharmacotherapy to control for at least some of the bias related to the lack of randomisation. In this design, outcomes of interest are compared during periods on and off medication in the same person, thus controlling for confounding of indication. One such study (summarised in Faraone et al^{Reference Faraone, Banaschewski and Coghill1}) included 25 656 Swedish individuals, aged 15 or older, with ADHD. Around 54% of them had taken ADHD medications (including methylphenidate, the stimulant most commonly used in Sweden) and 37% had been convicted for at least one crime during follow-up in the study period (2006–2009). The authors found that, compared with non-medication periods, the rate of criminal acts was decreased by 32% (stratified Cox regression hazard ratio 0.68; 95% CI 0.63–0.73) for men and 41% (hazard ratio 0.59; 95% CI 0.50–0.70) for women when on medication. The public health relevance of these data should not be overlooked.

Third, the maximum dose of OROS-methylphenidate used by Asherson and colleagues (72 mg/day; final mean dose 54 mg/day) may not have been enough to effectively tackle ADHD symptoms. It is also possible that the use of illicit drugs by a substantial proportion of the sample may have increased resistance to usual doses of methylphenidate, but this hypothesis should be empirically tested. Of note, currently the British National Formulary (bnf.nice.org.uk/drug/methylphenidate-hydrochloride.html) recommends a maximum of 108 mg/day of OROS-methylphenidate in both adults and children, even though the maximum licensed dose in children in the UK is 54 mg/day. Overall, the issue of possible advantages in using doses beyond the licensed ones, and even beyond the maximum recommended ones, is far from clear and requires rigorous testing in future studies. This seems to be particularly important in relation to the use of ADHD medications in prison inmates.

Fourth, the conclusion that participants in the Asherson et al study had a poor response to methylphenidate does not seem to be accurate. Nearly half of them (48.3%) responded to OROS-methylphenidate, when defining response as a reduction of 20% in the CAARS-O score. Rates of response to OROS-methylphenidate vary across previous RCTs, in part owing to different thresholds to define response (ranging from 20% to 50% reduction, more commonly 30%). The rate of response in the Asherson et al RCT is somehow lower than the figure (~60%) reported in another trial of methylphenidate in adults with ADHD using the same threshold to define response.^{Reference Biehl, Merz and Dresler4} However, the main problem is that, in the Asherson et al study, the percentage of participants who responded to placebo (47.8%) was very close to the rate of response to OROS-methylphenidate. This is in contrast to previous RCTs in which response to placebo has generally been smaller compared with the rates of response to methylphenidate (and other stimulants). Interestingly, placebo and nocebo effects have been assessed for other agents (e.g. antidepressants) in psychiatry, but are less well understood for ADHD medications. However, a recent meta-analysis^{Reference Faraone, Newcorn and Cipriani5} of 128 RCTs, encompassing 10 578 children/adolescents and 9175 adults, found that, when considering a rating of ‘improved’ or ‘very much improved’ on the Clinical Global Impression (CGI) scales, the pooled estimate of response was 60% for stimulants, 47% for non-stimulants and 25% for placebo. Of note, except for self-ratings of ADHD symptom severity, each of the other scales (clinician-rated, parent-rated and teacher-rated) showed a statistically significant positive correlation between the baseline-to-end-point placebo effect and the baseline-to-end-point active medication effect. These correlations suggest that the active drug effect is accounted for by the improvement specifically due to the active drug plus the improvement related to placebo effects. Therefore, psychosocial contextual factors such as those related to expectation of benefit may contribute to the beneficial effects of the placebo as well as of the active drug. The question then arises as to why the placebo effect was so high in the trial by Asheron and colleagues. The authors thoughtfully speculated that ‘in an environment impoverished of meaningful interactions with others in a caring role, there was an enhanced placebo effect contributing to the outcome of this study’. Future RCTs on ADHD medications, in the general population of individuals with ADHD as well as in prison inmates with the disorder, should endeavour to better understand the factors underlying placebo effects.