Generalisability
The trial targeted a sample that could be generalised to other prison populations of young males aged 16–25 years, meeting diagnostic criteria for ADHD, in the UK. To increase generalisability, a screening approach was taken to ensure that as many prisoners as possible who met diagnostic criteria were identified and invited to take part in the trial. Exclusion criteria were kept to a minimum and focused on excluding the minority of participants who posed a high risk to researchers, or who had coexisting disorders for which medical treatment of ADHD was a strong caution or contraindicated. Other criteria ensured that participants were able to understand the trial procedures and provide their own fully informed consent to the take part in the trial. The first step of the protocol was therefore a screening step in which all offenders entering the prisons were screened for ADHD, prior to being invited to complete detailed diagnostic assessments and review by a psychiatrist trained in the diagnosis of ADHD and other common mental health disorders. It is unknown if those prisoners who refused consent to be screened or did not attend the diagnostic interview for ADHD differ in any way from the trial participants.
The site HMP YOI Isis in London is a category C prison for sentenced prisoners, including male offenders aged 18–25 years deemed to be relatively low risk with short- to medium-length sentences. HM YOI Polmont in Scotland differs because it is the only prison in Scotland for young persons aged 16–21 years; therefore, there was no selection on type of crime, sentence or level of risk. Nevertheless, the range of crimes and sentencing was largely similar to HMP YOI Isis. The trial did not include any remand (pre-trial) or convicted but unsentenced prisoners, who generally show higher rates of comorbidity than sentenced prisoners.90 The sample was all male, so these findings cannot be generalised to the female prison population.
The focus in this study and in the pilot study, as well as the previous Swedish study,33 was male prisoners; therefore, little is known about the efficacy of pharmacological treatments in treating female prisoners with ADHD. It is unknown whether or not there would be similar safety concerns and similar levels of behavioural problems, mental health disorders and substance misuse. It is expected that the profile of comorbid disorders might differ between male and female offenders, and their effects on pharmacological treatment in female offender populations remains largely unknown. The best evidence that similar effects may arise comes from pharmacoepidemiogical data from Sweden, although the number of females with ADHD at risk of committing criminal offences is far smaller.91
Regarding ethnic background, at HMP YOI Isis, a clinical audit conducted in June 2019 found that ethnic minorities were approximately 70%, whereas the ADHD service had a case load of 38% ethnic minorities and 63% non-ethnic minority (white) prisoners. This indicated a considerable bias in the prisoners being referred for diagnosis and treatment by the health-care and prison staff within the London prison. These figures contrast with the proportion of people from ethnic minority backgrounds in London, reported in the 2011 UK National survey,92 of 59.8% from white, 18.4% from Asian, 13.3% from black and 8.5% from mixed and other backgrounds. This clearly demonstrates the well-established finding that ethnic minority groups are greatly over-represented in the prison population, which is an issue of general national and international concern.93 The sample from which we recruited in London is representative of the prison population, but is not representative of the general population of youth with ADHD in the London area.
Regarding recruitment to the trial, we found that, at HMP YOI Isis, 34.8% of the participants came from white, 25.2% from other (including Asian and mixed background) and 40% from black backgrounds. The figure of 34.8% participants from a white background is close to the estimated proportion of 30% in the prison, indicating no bias in the selection for this trial based on ethnicity, in keeping with the screening approach we adopted for this trial. For Scotland, the 2011 survey reported a low rate of ethnic minority populations, of around 4%. At the time of writing, we had no information on the proportion of different ethnic groups residing at HM YOI Polmont, but, among participants in this trial, we found that > 98% were from white British backgrounds, providing a good representation of the ethnic mix in Scotland.
Previous research has also shown high rates of comorbidity in children, adolescents and adults diagnosed with ADHD in both clinical and epidemiological samples. A 2017 review7 of population surveys for adults with ADHD found that 52% met criteria for one or more comorbidities: oppositional defiant disorder (as children) or ASPD, 8%; intermittent explosive disorder, 13%; an anxiety disorder, 34%; a mood disorder, 22%; and a drug or alcohol problem, 11%. This compares with the findings in the current trial of 82% who had one or more comorbid mental health disorders: 19% with a mood disorder, 19% with an anxiety disorder, 75% with ASPD, 45% with problematic alcohol use (AUDIT-C score of ≥ 16) and 64% with moderate- to high-risk drug use (see Table 5 and Chapter 5, Could the trial outcome be affected by including participants with high levels of drug and alcohol usage?). This suggests that the prison population of young adults meeting diagnostic criteria for ADHD that we investigated here have a similar profile for anxiety and mood disorders, but higher rates of ASPD and problem alcohol and drug use. For the overall male convicted prison population, it would be expected that 7% had a psychosis, 40% a neurotic disorder, 63% a history of alcohol abuse, 43% a history of drug dependence and 63% a personality disorder.90 Internationally, a meta-analysis of > 33,000 prisoners found that 3.6% of men had a psychosis and 10.2% had major depression.94
Interpretation of primary and secondary outcomes
This report describes the results of a RCT of the short-term effects of OROS-MPH on ADHD symptoms and behavioural outcomes in young male prisoners aged 16–25 years with ADHD. The trial was completed within the planned time scale, meeting all recruitment targets at HMP YOI Isis (London, UK) and HM YOI Polmont (Falkirk, UK), with randomisation of 101 participants to the drug treatment arm and 99 to the placebo arm of the trial. The aim of the trial was to establish whether or not the efficacy of a long-acting formulation of MPH in this population was similar to the effects observed in comparable trials in non-prison (community) populations. The rationale for this trial consisted of arguments supporting both smaller and larger effects than previous studies in non–offender populations, reflecting the potential complexity of the diagnosis and response to ADHD drug treatments in young adult prisoners.
The primary and secondary outcomes of this trial failed to show statistically significant differences between the OROS-MPH and placebo treatment groups in the outcome assessment at 8 weeks. Any differences between the mean or median scores for the two arms were negligible for all of the outcome measures and are unlikely to be clinically meaningful even if found to be significant in a larger trial sample. To check for the possibility of a hidden effect in the trial data for any reason, we conducted a further analysis to look for two distinct clusters of outcomes in the data. We would expect to see a bimodal distribution if there were any drug effects, reflecting the existence of two response groups. We checked the distribution in the change of CAARS-O scores from baseline to week 8 by plotting the distribution of change scores, but found only a single smooth function that gave no hint of two distinct responder groups within the data (see Figure 5).
The lack of any benefit of the active drug against placebo was unexpected. MPH has been used in numerous previous RCTs of ADHD in children and adults, and shows a consistent moderate to large effect across the various studies using outcome measures that are the same or closely similar to the CAARS-O investigator-rated scale used in this trial. In a 2018 comprehensive network meta-analysis32 comparing the effects of different pharmacological treatments for ADHD, the data for MPH were found to have a SMD of 0.78 (95% CI 0.62 to 0.93) in children and 0.59 (95% CI 0.35 to 0.64) in adults with ADHD. We therefore need to consider possible reasons for seeing such robustly negative findings in this study of young male adult prisoners with ADHD.
Trial medication
To rule out any errors with the trial medication that could have affected the trial findings, a request was made to investigate this with Modepharma Ltd, which liaised with Piramal Healthcare UK Ltd regarding the trial kits. Batch records were reviewed and there was no evidence of any errors. Spot checks were also done on the randomisation data against the batch records. The trial kits were stored under temperature-controlled conditions by the manufacturer, during transit and in the pharmacy. Following randomisation, the trial medication was stored in the house block in the prisons, and, even though temperature was not controlled, there was no indication that temperature would have been > 30 °C, hence degradation of OROS-MPH is not expected. The decision not to monitor the temperature in the house block resulted from advice from the MHRA and was confirmed by the sponsor.
Adherence to medication
An important consideration is adherence to the trial medication, which was much poorer than completion of trial outcomes. For the ITT analysis of all randomised participants, we estimated a group difference for the primary outcome of 0.57 (95% CI –2.41 to 3.56; p = 0.71) (see Table 10). This is clearly not significant and shows a very small difference in change in CAARS-O score between the groups that is far from a clinically meaningful effect size. To address the potential effect of poor adherence to trial medication, a sensitivity analysis was repeated using only those participants who had complied with treatment, defined as taking prescribed trial medication on 75% of the days during which they were in the 8-week trial, and also not having withdrawn from treatment or the trial. Only 83 out of 200 (41.5%) participants met this criterion, of which 34 were taking OROS-MPH and 49 were taking placebo. The estimated group difference remained very small, at 0.29 (95% CI –2.90 to 7.01; p = 0.91) and was very similar to the ITT analysis. This finding, therefore, lends no support to the possibility that poor compliance with trial medication could explain the negative findings in this trial.
Given that MPH relative to placebo was shown to significantly reduce ADHD symptoms in previous trials of non-prison populations,32 the finding that the per-protocol and ITT analyses were so similar was itself surprising and emphasised the need to explore the effects of adherence to trial medication further. For example, it may be that adherence to trial medication was greater during the titration phase, with medication effects at 5 weeks that were lost by 8 weeks, or that the rapid onset and offset of symptom control with OROS-MPH led to differences in reporting of ADHD symptoms, depending on the proximity of taking trial medication to completion of the outcome measures. To explore this further, we conducted two additional subgroup analyses. First, we conducted an analysis of the effects of OROS-MPH versus placebo on the CAARS-O outcome at week 5, using a definition of good compliance as the group that had taken trial medication on 75% of the days before the end of week 5. Second, we investigated the possible impact of the short-term effects of medication influencing the accuracy of reports of ADHD symptoms by including only those who took trial medication on the 2 days prior to the outcome assessment at weeks 5 and 8, for the week-5 and week-8 outcomes, respectively. Of these additional analyses, the largest difference was seen for the week-5 outcome for a group of 97 participants who had taken one or more capsules on both days immediately preceding the assessment, with a difference in outcome between the trial arms of 2.30 (95% CI –2.28 to 6.88; t = 1.00; p = 0.32) (see Chapter 5, Effects of poor adherence).
Taken together, these analyses indicate that, although adherence to medication was poor, there was no apparent impact of adherence to trial medication on the trial outcomes from the initial sensitivity analyses or additional analyses.
Optimal dosing
The optimal dose of MPH used in the treatment of adult ADHD is highly variable; hence, we adopted a titration procedure to individually titrate each participant to the optimal dose of trial medication. A similar titration protocol using the same trial medication (OROS-MPH) in a previous community trial of adults with ADHD reported large drug/placebo differences for ADHD symptoms, although they titrated to higher doses, with a mean dose of 80.9 mg ± 31.8 mg.59 However, as noted in the pilot study for this trial, the prison population titrated to relatively low average doses.29 This means that the titration protocol we followed might lead to underdosing of participants, potentially accounting for smaller effects of the drug, compared with placebo. The reasons for such low dosing were not formally investigated in the previous study,29 but it might be due to high sensitivity to common adverse effects. For example, during the current trial, on the AES, 20 participants taking OROS-MPH, versus 10 taking placebo, reported dry mouth; 20 versus 8, respectively, reported sweating; and 35 versus 19, respectively, reported appetite loss. Although sleep disturbance is a common problem, reported by 51% of participants at baseline, differences between the OROS-MPH and placebo arms were small during the trial (56 vs. 50 participants, respectively).
To explore the potential effects of dose on the trial outcome, we completed subgroup analyses for low-dose (titrated to one or two capsules) and high-dose (titrated to three or four capsules) groups (see Chapter 5, Could the trial outcome be affected by underdosing of participants?). Around 19% were in the low-dose group and, overall, dosing seemed to be a reasonable reflection of standard clinical practice. However, the difference in ADHD symptoms between the OROS-MPH and placebo arms was slightly smaller in the high-dose group than the low-dose group; no clinically meaningful differences were seen between the two arms of the trial for either dosage group.
Overall, these findings do not suggest that low dosing could explain the lack of significant drug effects in this trial, compared with placebo. However, it remains possible that high levels of illicit drug use in this population might cause brain changes that reduce the effects of MPH. This was suggested by a 24-week randomised clinical trial95 of OROS-MPH in 54 released prisoners with ADHD and amphetamine dependence. In this trial,95 they titrated up to 180 mg per day, which is considerably higher than the dose range used in the current trial. They found that, compared with the placebo group, the OROS-MPH group showed significantly greater improvement in the CAARS self-report scale (ADHD symptoms) (95% CI −14.18 to −3.28; p = 0.002), with 17 out of 26 participants in the OROS-MPH group showing a symptoms reduction of at least 30%, compared with only 7 out of 26 in the placebo group. This trial95 therefore suggests that much higher doses of OROS-MPH than used in the current trial may be required to see clinically significant effects of OROS-MPH.
Another question that arises in the current trial is whether or not dosing during the titration procedure was driven by the experience of adverse effects of medication, rather than treatment effects. This was suggested by the finding of a lower average prescribed dose for OROS-MPH than for placebo (3.20 vs. 3.49, respectively) at the start of week 6 (see Table 27) and a smaller proportion of prescribed medication taken (44% vs. 57%, respectively) at week 8 (see Table 28), as well as a higher rate of common adverse effects in the OROS-MPH arm (see Chapter 4, Adverse events). To investigate this, we compared a measure of adherence to medication in a group with and a group without the most common side effects. However, no significant differences were identified, providing no evidence to support this explanation (see Chapter 5, Could adverse effects be driving adherence to the trial medication?). Nevertheless, no firm conclusions can be drawn from this analysis and sensitivity of prisoners to minor adverse effects remains a possibility.
We also asked participants and researchers at week 8 to which treatment arm they believed they had been assigned. In the OROS-MPH arm, 44 participants (50.0%) and 47 researchers (52.8%) guessed correctly, suggesting random guessing, whereas, for the placebo arm, 72 participants (78.3%) and 68 researchers (73.1%) guessed correctly. Overall, this indicated some ability for both researchers and participants to distinguish between the trial arms and suggests that participant blinding might not have been maintained (p = 0.001) (see Chapter 4, Treatment arm prediction).
Diagnostic accuracy and comorbidity
Another possible explanation could be poor accuracy for the diagnosis of ADHD in the population of young male adult prisoners who took part in this trial. This is a young male adult population, aged 16–25 years, with below average general cognitive ability, with comorbidities, and histories of childhood trauma and drug use, in which the diagnosis was based largely on self-report of symptoms and behaviours reflecting the diagnosis of ADHD. Assessment of ADHD was based on a face-to-face diagnostic interview with participants, and, in most cases, did not make use of more extensive investigations, such as informant report or school reports. This was different from the Swedish study,33 in which participants were extensively evaluated, which included neuropsychological testing; interviewing parents or significant others; and collecting information through school reports and previous medical records from child and adolescent psychiatry, adult psychiatry and forensic psychiatry, when applicable. This raises the possibility that participants over-reported ADHD symptoms, or that the symptoms they experienced might be better explained by alternative mental health, psychosocial or neurodevelopmental conditions.
A further complication could be the unmeasured use of drugs in the prison, such as spice, which could affect the mental state, producing ADHD-like symptoms at baseline. To explore the impact that diagnostic certainty might have on the trial outcome, we completed a subgroup analysis that used a high threshold of ADHD symptoms, derived from the DIVA 2.0 assessment. We considered that, the larger the number of symptoms of ADHD clustered together, both in middle childhood and currently, the more likely it was that they had the typical neurodevelopmental form of ADHD. We selected a group with seven or more symptoms in both the inattentive and hyperactive/impulsive symptom domains, in both childhood and adulthood. A total of 102 participants met these higher symptom thresholds, yet the improvement seen in the OROS-MPH versus placebo arms was no different from the sample overall (see Chapter 5, Was the trial outcome affected by poor diagnostic accuracy in this prison population?).
Further subgroup analyses investigated the potential role of comorbidities identified from the MINI. To ensure that mental health comorbidities were in sufficiently large numbers, specific mental health disorders were grouped into categories: mood disorders (major depression, suicidality, mania, hypomania), anxiety disorders (panic, agoraphobia, social anxiety, obsessive–compulsive disorder, PTSD) and ASPD. As expected, 164 (82%) participants in the sample had one or more comorbidities; only 36 (18%) had no comorbidities. Although the subgroup with no comorbidities was small, there was no evidence for a clinically significant difference in the primary outcome measure of ADHD symptoms, suggesting that it is unlikely that the presence of comorbidity could explain the lack of any significant effects in this trial. Further subgroup analyses investigated a group with and a group without anxiety and mood disorders, and found no evidence for any clinically meaningful effects in any of these subgroups. (see Chapter 5, Could the trial outcome be affected by including participants with comorbid disorders?)
Regarding comorbid drug and alcohol abuse, both are associated with ADHD and, as discussed in Chapter 1, Scientific background and rationale, might have an impact on the treatment response. Although RCTs do show effects on ADHD symptoms in patients with ADHD and drug dependency in previous studies, these use higher doses than the study reported here. For example, in the review from Cunill et al.,25 the average dose of MPH used was 62.2 mg per day, and the overall consensus is that higher than standard doses may be required to see a clinical effect26,95 and to maintain compliance to taking MPH.27 This could arise because of brain changes as a long-term consequence of drugs and/or alcohol, reducing the potential effects of medication for ADHD or requiring higher dosing. A history of drug and alcohol abuse is also likely to be an indicator of risk for ongoing (unmeasured) use of drugs in the prison during the trial, which might have a more direct impact on ADHD symptoms and the clinical response to OROS-MPH.
The availability of new psychoactive substances, particularly synthetic cannabis, known as spice, became highly prevalent in prisons during the preparation for this report and is now considered the most serious threat to the safety and security of the prison system as identified by prison inspectors.96 Yet little is known about the effects of spice on ADHD symptoms and the treatment response to MPH, and, in our trial, very few participants reported using spice before the trial. Furthermore, although we did not include routine drug testing in this trial, available tests do not detect spice.
We attempted to account for drug use in additional subgroup analyses. According to the AUDIT-C data, 89 (44.5%) participants met criteria for problematic alcohol use, and neither the group at high risk nor the group at low risk showed significant differences in outcome between the two arms of the trial (see Chapter 5, Could the trial outcome be affected by including participants with high levels of drug and alcohol usage?). Similarly, there were no differences seen in outcome for the group with no or low risk of substance involvement according to the NIDA Quick Screen of SUD, although the subgroup analysis of 73 participants who had no or very low use of any of opiates, spice, cannabis or stimulants showed an estimated score difference between the OROS-MPH and placebo arms of 4.26 points (95% CI –9.31 to 0.78 points; t = –1.69; p = 0.10). This was below our threshold of 5 points and not statistically significant (see Chapter 5, Could the trial outcome be affected by including participants with high levels of drug and alcohol usage?).
Finally, it might be difficult to separate ADHD from BPD or ASPD on clinical grounds alone.97 For example, in previous research,46 we found that women meeting diagnostic criteria for BPD also show high levels of inattentiveness in their daily lives, in addition to shared symptoms such as emotional dysregulation. Similar problems are also expected to arise for men with BPDs or ASPDs. The presence of personality disorder as a comorbidity might reduce the overall effect of medication, particularly if some of the symptoms of personality disorder mimic those of ADHD. To address this question, we investigated a group with and a group without ASPD and BPD, and found that 149 (74.5%) participants in the trial met ASPD criteria, but only 15 met BPD criteria. As the numbers of those with BPD and those without ASPD were so small, we did not investigate these groups further. For ASPD, we found only very small differences between those with and those without ASPD, suggesting that the presence of this comorbidity did not affect the trial findings (see Chapter 5, Could the trial outcome be affected by including participants with borderline personality disorder? and Could the trial outcome be affected by including participants with comorbid disorders?).
Overall, we found no evidence supporting the idea that removing (or including) specific comorbidities that might confound the diagnostic assessment, or the response of ADHD symptoms to OROS-MPH, played any role in explaining the outcome of this trial.
Other potential factors
We considered a number of additional factors that might explain the trial outcome.
Temporal effects might play a role if there was a systematic change at some time point during the trial, such as unidentified changes in conduct of the trial procedures. However, we are not aware of any potential changes during the trial procedures that could explain the lack of difference between the two trial arms and the analysis of groups of 50 participants divided by consecutive recruitment to the trial showed no temporal changes on differences in outcome between the trial arms (see Chapter 5, Was the trial outcome effected by a systematic change to trial procedures?).
Attention deficit hyperactivity disorder severity might also affect results if, in this population, only the most severe cases show a significant effect of medication, compared with placebo. This is possible, as the previous RCT of ADHD in a prison population in Sweden33 showed a very large effect with no observed change in the placebo arm. The cases in this study were a highly selective group of prisoners with severe ADHD, and lengthy sentences, who were kept in a special prison wing for the trial. In contrast, the cases in the CIAO-II RCT had relatively short-term sentences and were screened on entry to the prison, giving a population that included less severe cases regarding both their ADHD and criminal offences. It is feasible that the non-specific effects of the trial protocol could lead to similar improvements in both trial arms for less severe cases, but show greater drug versus placebo differences in more severe cases. To explore this question, we conducted a subgroup analysis that involved 123 participants who met the criteria for severe ADHD with a score on the baseline CAARS-O of ≥ 35, equivalent to a t-score of 70 (see Chapter 5, Was the trial outcome affected by including less severe cases of attention deficit hyperactivity disorder in the trial?). The results showed that there was a greater improvement in the OROS-MPH arm than in the placebo arm, but the difference of 2.81 (95% CI –1.11 to 6.73) was too small to be clinically meaningful, and was not significant even using the nominal significance value (p = 0.16).
Another index of severity could be emotional dysregulation, which varied across study participants. Emotional dysregulation is a transdiagnostic phenomenon that could reflect comorbid conditions, leading to an overall reduced effect of medication. Alternatively, emotional dysregulation could be an index of ADHD severity, predicting a larger effect in the group with higher levels of emotional dysregulation. Dividing the sample into a group with low and a group with high levels of emotional dysregulation, based on a median value for the WRAADDS, 113 (56.5%) participants in the sample had high levels of emotional dysregulation. However, the effect was again small and non-significant (see Chapter 5, Could the trial outcome be affected by including participants with different levels of emotional dysregulation?). Overall, there was no evidence that severity of ADHD or emotional dysregulation had a meaningful impact on the trial results.
Finally, we investigated the potential role of childhood trauma using findings from the baseline CTQ. The association of adverse childhood experiences (emotional abuse and neglect, physical abuse and neglect, and sexual abuse) is widely discussed in the literature,98 but has not been investigated as a potential confounder of ADHD medication treatment effects. However, the role of severe childhood trauma in ADHD and other mental health disorders, and its impact on response to pharmacological treatments is a frequent topic among clinicians. It is possible that severe childhood trauma might give rise to mental health disorders that overlap symptomatically with ADHD, may not respond to ADHD medications and require psychosocial treatment approaches.99,100 In the trial sample, we did find high rates of reported childhood trauma. In particular, 21% reported childhood emotional abuse, 24% physical abuse, 84% emotional neglect and 98% physical neglect. Again, none of the subgroup analyses found differences between the arms that were of potential clinical significance, although we were greatly underpowered to investigate groups with low levels of childhood abuse. Therefore, it remains possible that high levels of childhood trauma and neglect might negate the effects of OROS-MPH in this population, requiring further studies in high-risk populations such as young offenders (see Chapter 5, Could the trial outcome be affected by including participants with high levels of childhood trauma?).
Comparison of pre–post effects from the pilot study
Prior to this trial, we conducted an open single-arm study to investigate the feasibility of diagnosing and treating young adult prisoners with ADHD. The pre–post findings from this trial were considerably greater than those from the RCT reported here. For the primary CAARS-O outcome measure in the pilot study, there was a mean difference from baseline to the 8-week outcome of 22.6 points in the ITT analysis,29 compared with 8.1 points in the current trial (OROS-MPH arm, 8.4 points; placebo arm, 7.9 points) (see Table 8). The pre–post effect in the pilot study was considerably larger than the pre–post effect for either the OROS-MPH or placebo arms in the current trial, which requires some explanation.
Overall, the pilot was conducted along closely similar lines to those of the current trial. Indeed, the protocol for this trial was based on the pilot study. Regarding selection criteria and outcome measures, these were similar for the screening and diagnostic step, subsequent titration of OROS-MPH, followed by outcome assessments after 8 and 12 weeks, including the primary CAARS-O outcome used the current study. There are, however, three key differences from the current study. First, the primary outcome was at 12 weeks for the pilot study, whereas it was at 8 weeks for the current trial. However, we also collected 8-week outcome data in the pilot study, enabling a direct comparison. Second, the titration protocol was to a higher value of 90 mg of OROS-MPH in the pilot study; however, as only 4% of the sample titrated to the highest dose, this had little impact on the results of the pilot study. Third, participants who did not comply with the trial medication left the pilot trial and did not provide follow-up data. Thus, the pre–post findings from this single-arm study can be considered equivalent to a per-protocol analysis of the current trial. Exclusion from follow-up usually followed a period of 3 consecutive days of not taking medication, after which this was discussed with the participant to try and identify why they had missed doses and problem-solve the difficulties. If participants then continued to be non-compliant for a further 7 days, their participation in the trial was stopped in most cases.
To provide a more accurate comparison of the two trials, we summarised the pre–post effects, using a similar definition for retention in the trial for the group treated with OROS-MPH at HMP YOI Isis (as the pilot was conducted at this site only). To achieve this, we report on the 8-week outcome data, and excluded those who did not take medication for a continuous period of 10 days from the current trial, to best match the pilot study data set. The comparison of the baseline data shows that the two studies were similar regarding the severity of ADHD: mean CAARS-O scores were 37.6 points for the subset from HMP YOI Isis that did not take trial medication for a continuous period of 10 days, compared with 36.5 points for the pilot study (see Table 19). Furthermore, the study samples were similar for other baseline measures, including age, IQ estimate, WRAADDS (emotional dysregulation) scores, number of critical incidents and the CGI global clinical severity score (see Table 20). Drug and alcohol use appeared to be broadly similar between the two studies, although different measures were used: cannabis use was 92% in the pilot study and 95% in the current trial, and other drug use was 37% in the pilot study and 42% in the current trial. One potential difference was in educational outcomes, although this is thought to be due to an increase in qualifications obtained while in prison in the current trial, compared with the pilot study.
However, despite the similarities in the patient populations in the two trials, the comparison of the CAARS-O outcome scores at 8 weeks showed considerable differences in the pre–post effects, with differences from baseline to the 8-week outcome of 8.1 points across the current trial overall (8.4 points in the OROS-MPH arm and 7.9 points in the placebo arm) and 11.08 points in the subset at HMP YOI Isis that did not take medication for a continuous period of 10 days, compared with 25.0 in the pilot study (see Table 19). This large difference is difficult to explain, but suggests that there were considerable differences in the rater bias effects when conducting the two trials of young adult prisoners with ADHD. We can only speculate what these might be, but would point towards the following potential factors. The pilot study was an open trial, so both the participants and investigator knew that participants were on an active medication. Furthermore, although similar diagnostic and selection procedures were involved in both the pilot study and this trial, there was only one research assistant throughout the pilot study, who also had a high degree of contact with participants and a vested interest in generating good outcomes. Potentially, this could have led to an informal source of supportive counselling or coaching, with a possible effect on reported ADHD symptoms. Together, these effects from both participants and the researcher might have generated much greater rater bias effects than envisaged, leading to the large difference in pre–post ADHD symptom scores between the two studies.
We further considered other aspects of the prison health-care setting that might have changed. For example, the responsibility and staffing at HMP YOI Isis changed because the service changed from being run by Health Care UK in the pilot study to being run by Oxleas NHS Foundation Trust in the current trial. However, the models of care were similar and we could not identify any significant differences in the nature or quality of the service delivered, particularly regarding the involvement of broader health care and the availability of specialist psychiatric care. Overall, we conclude that rater bias effects in prison studies of ADHD are far greater than we envisaged, with, potentially, a positive downwards bias (to lower CAARS-O outcome scores) in the open trial, and a positive upwards bias (to increase CAARS-O outcome scores) in the randomised clinical trial.
Comparison with Swedish study
The results of the current trial are also in marked contrast to those reported for the smaller Swedish prison study of OROS-MPH in 30 long-stay prisoners, which showed a very large effect on CAARS-O outcome scores, with a SMD for OROS-MPH versus placebo of 2.17 points.33 Of particular note, unlike the current trial, and other trials of MPH compared with placebo in children and adults with ADHD, during the RCT there was a very small difference in the placebo group from baseline to the 5-week outcome measure, reflected in the large difference between the two arms of the trial at the 5-week outcome. The authors reported a large decrease from baseline to the 5-week outcome for the CAARS-O score of 19.6 points (95% CI 14.7 to 24.5 points; p < 0.001) in the OROS-MPH group, compared with a non-significant decrease of 1.9 points (95% CI –0.4 to 4.2 points) in the placebo group.
The Swedish study33 had a fixed titration protocol, starting at 36 mg per day for 3 days, 54 mg for 3 days, and then 72 mg for the remaining 4 weeks, potentially explaining the greater difference in ADHD symptom between OROS-MPH and placebo arms at the 5-week outcome. This view is further supported by the finding that, in the open-label extension of the Swedish study, the mean end-point dosage was a daily dose of 108 mg. The average dose achieved in the Swedish trial was therefore higher than that in the current trial, in which only 42 out of 101 participants randomised to the OROS-MPH arm reached the highest possible dose of 72 mg (see Table 26). The findings in the Swedish study, therefore, suggest that the overall large effect of medication versus placebo on ADHD symptoms may have been driven by the small response to placebo, as well as the use of a rapid fixed-dose titration to 72 mg of OROS-MPH.
There are some crucial differences in the trial designs that might explain this. The Swedish study33 was completed in Norrtälje prison, which is a high-security prison (highest security class A) hosting long-term convicts, a very different group to that of the current trial. Participants were drug free for at least 3 months prior to the trial and were drug tested at the start and throughout the trial. In the Swedish study, all participants who consented were moved to a mental health ward in the prison, which was rebranded as an ADHD ward for the duration of the trial. By staying on this wing, they were agreeing to limited contact with other inmates during the period of the RCT. Consequently, there were younger inmates who declined to participate since they prioritised other activities (e.g. playing football with other inmates instead of taking part in the trial). Participants on the ward were also much more closely monitored on a daily basis than participants in the current trial, and there were regular tests for drug abuse, thereby ensuring low or absent levels of drug use during the trial. This selection was entirely different from that of the current trial, in which participants were required to remain in prison only for the duration of the 8-week trial, and the large majority were on short-term sentences with an average of 14 months, with a subgroup being released early. The Swedish study33 also had participants that were older, with an average age of 34 years. There were also participant differences in the severity of ADHD. Although baseline CAARS-O scores were broadly similar to those of the current trial, averaging 40.0 points in the OROS-MPH arm and 39.9 points in the placebo arm, compared with 36.4 and 37.2 points in the OROS-MPH and placebo arms, respectively, of the current trial, the severity of the illness was markedly different. There was a CGI score in the Swedish sample of 5.9,33 compared with 3.9 in the current trial, and, in the Swedish study, there were higher rates of anxiety and mood disorders (73%), all 30 participants met criteria for ASPD, and just under one-quarter met criteria for autism spectrum disorder (not formally evaluated in the current trial).
The strong selection in the Swedish study33 was reflected in the high completion rate for the 5-week initial randomised placebo-controlled trial. All participants were told in advance that they would enter an open trial with flexible dosing only if they first completed the 5-week RCT, potentially leading to greater tolerance of short-term adverse effects and ensuring 72-mg dosing in all cases. Furthermore, by using forced titration over 1 week, participants did not have to wait long before experiencing effects from treatment if they were on MPH. The rapid titration to 72 mg in all cases had the potential to be more obvious to participants. By comparison, in the current trial, if participants received too low a dose for several weeks, any effects may be less obvious and less distinct from baseline.
These differences might have influenced participant reports of symptoms and behaviours. The absence of a treatment effect in the placebo arm is unusual in trials of MPH in ADHD, suggesting that general effects of being engaged in the trial had no role to play in the Swedish study.33 This could arise, for example, if participants perceived no potential benefits of reporting change or they received no greater benefits from being in the trial than usual. In contrast, in the current trial, prisoners might have felt that reporting improved behaviour might lead to benefits such as a shorter time in prison or an increase in privileges, perhaps minimising differences in the reporting of ADHD symptoms in the two arms.
Another potential issue could be the suitability of the CAARS-O to detect changes in symptoms of ADHD in the UK prisons, compared with the Swedish prison, because of differences in the daily routine and rehabilitation programme. Some of the items, such as losing things necessary for tasks or activities or making mistakes during tasks because of a lack of attention, may have less relevance in an environment where prisoners are locked up for long periods, follow a monotonous routine and have low organisational demands, biasing the results in the UK prison environment towards a negative result. However, most participants were engaged in an active programme of education and training and were able to give examples of ADHD symptoms based on their daily experience, for example losing items in their cell because of untidiness, difficulties following conversations with inmates or teachers and forgetting to attend (for medication and other activities), as well as improvements in these and other functions, such as sustained focus when reading and writing, and feeling less restless.
Another difference was that there were no medication compliance issues in the Swedish sample, which is thought to be due to the small sample and very close monitoring in a specialised facility for the trial. However, as we have seen, in the current trial there is no evidence that adherence to medication had a significant impact on the trial findings, with similar effects in subgroups that showed good adherence to medication and the complete ITT sample. Although the majority of the inclusion and exclusion criteria were similar to those of the Swedish study, a higher level of ASPD and anxiety and mood disorders was noted in the Swedish study. It seems unlikely that this could explain the difference in the trial outcomes. Finally, in both studies, there were high levels of reported substance misuse, but in the Swedish study any evidence of substance misuse 3 months prior to baseline was an exclusion criterion. Furthermore, under the close monitoring in the ‘ADHD wing’ it would not be feasible for prisoners to access drugs such as spice and the Swedish prisoners had repeated supervised urine drug tests. This potentially had a significant impact in the current trial, but is an unmeasured risk, and experience and recent data suggest that higher dosages of MPH are required if there is ongoing substance misuse.26–28
Limitations
There were a number of limitations that should be considered when interpreting the findings reported here.
Throughout the trial, the research teams had weekly telephone conference calls during which there was an opportunity to discuss the diagnosis of individual cases prior to randomisation. However, to reduce the potential for biasing outcome assessment data by the rater (who was part of the trial team in the prison), post-randomisation outcome assessments were not discussed. Initial training was provided for the primary outcome measure, but further training and evaluation throughout the trial to ensure that all staff continued to conduct the primary outcome assessments in a reliable way were not conducted. This might have led to a reduction in inter-rater reliability, although, in most cases, the same investigator conducted both the baseline and outcome assessments on the primary outcome measure, limiting the effects of using different raters on within-individual change in the primary outcome measure over the period of the trial.
Drug usage in prisons is a well-reported and documented problem. However, it was decided early on in the design phase of the protocol development that there would be no recording of drug usage during the trial because this would be unreliable if based on participants’ accounts of drug use in the prison. We considered the use of drug-testing as part of the trial, but one of the most common drugs found in the prison (spice) has no test, and we were concerned by the impact on recruitment and on patient selection that would be introduced, minimising the generalisability of the findings. During the trial, 18 participants were found to have been using drugs (cannabis or spice), or suspected to have been using drugs, by prison or health-care staff or disclosed drug use to the researchers during the trial, where it was recorded as an AE. However, there were probably other undetected cases. A high level of undetected drug use by participants might account, in part, for the lack of differences between the two trial arms, if, for example, drug use generates short-term effects that are similar to the ADHD symptoms captured in the primary and secondary outcome measures.
A potential confounder not measured in this trial was traumatic brain injury (TBI), which is seen in around 60% of prisoners.101 In a study from co-applicant Susan Young,102 43% of prisoners were deemed to have suffered TBI using a strict definition, with a mean age at first head injury of 12.5 years. A further study103 reported that 60% of prisoners self-reported head injuries, with 16% reporting moderate to severe TBI. Adults with TBI were also younger when entering the criminal justice system.103 Although there may be some clinical effects of MPH in ADHD secondary to TBI, this is not yet well established and the efficacy in this context remains uncertain.104
Poor adherence to treatment in the current trial might be related to the aspects of the prison environment including a limited time window for receiving medication, safety concerns regarding gang relations exposing them to be targeted during dispensing, and inconsistent release of prisoners from cells to receive medication. These potential risks were not measured; however, in additional analyses we found no evidence for an effect in a subgroup with good adherence to treatment.
In addition to pharmacological treatments, some participants in this trial might also have received additional psychological therapies from prison health care or general practitioners (GPs), although this information was not recorded. Although such treatments could potentially reduce reported levels of ADHD symptoms, it is expected that these were evenly distributed across the two arms of the trial, and previous trials have found significant effects of MPH, but no effects of ADHD-tailored psychological interventions, compared with clinical management as usual.105 Nevertheless, because this was not recorded, it is possible that psychological treatments were not evenly distributed between the two arms of the trial.
The number of behavioural problems recorded in the prison records were very small at baseline, and the prison officer ratings of aggression on the MOAS-P and BRC-R were both very low at baseline, meaning that these measures were largely uninformative. This could reflect a high level of tolerance by prison officers for minor to moderate levels of disruptive behaviour in this population.
The relationship between participants and research and prison staff might also play a key role in influencing participants’ reports of symptoms and behaviours, and could potentially have led to reported improvements in the placebo arm that swamped any potential effects of OROS-MPH. For this reason, we included information from prison records and behavioural reports from prison and education staff, although these were largely uninformative and lacked power to detect any potential differences. We also considered the use of more objective tests, such as QbTest, which measures both cognitive performance during a continuous performance task and motor activity, measured as head movement. However, in the pilot study, during which we did collect QbTest data, the data were found to be of too poor quality to report as a result of poor engagement with the test procedures by young prisoners with ADHD. Given the results of the present trial, it would be important to consider what other objective tests of clinical improvement could be used in future studies of young offenders with ADHD.
There are several potential sources of bias on ratings of the primary and secondary outcomes during the trial. For example, participants might have been very keen to please the researcher for a number of reasons. They may have looked forward to their assessments as a change from their usual prison routine, and a chance to discuss a broader range of problems related to their diagnosis and treatment in the prison. It should also be noted that the staff working on the trial at both sites were mostly female, potentially enhancing the motivation to appear to be doing well to please the research staff. This may have led to social desire bias,106 which might have had a similar impact on reporting of symptoms and behaviours in the two arms of the trial, thereby minimising potential differences in outcome.
Another potential bias could be the halo effect. These are young men on relatively short tariffs and quite a different sample to the older sample in Sweden who were on longer tariffs. They are likely to be aware that they might get a reduced tariff for good behaviour, and there may also be an expectation of greater privileges in the prison through participation in certain prison programmes. A further motivation might be the possibility that taking part in the trial and reporting good outcomes might increase the possibility of earlier release. This knowledge of the system might mean that they were highly motivated to communicate improved functioning or reduced impulsive or disruptive behaviours equally in the two arms of the trial. Although it was made clear that this was a research project that would have no influence on outcome, they may still have perceived that reporting good outcomes can only help their situation.
Another potential bias could be the Hawthorne effect,107 in which being aware that you are being observed can lead to changes in behaviour; this may have potentially affected both arms of the trial. This effect might be greatly exaggerated in prisons, where participants are aware that they are constantly watched by both health-care and prison staff, and intense relationships are formed.
Related to these potential effects of the participant–rater interactions, participation in the trial may also have increased the sense of self-worth of participants, as some commented that they had never been asked to consent to anything before. They liked the idea that they were given a free choice whether or not to participate. Because of our screening procedure, many of the participants diagnosed with ADHD for this trial may not otherwise have engaged with the mental health team at all and the constant contact with the researchers, who had more time to spend than the health-care staff, may also have had a positive effect on participants’ well-being. It is also feasible that improvements in a general sense of well-being as a result of the research procedures might have led to lower clinical and behavioural ratings in both arms equally. Although the research staff were trained to enquire about specific examples of current symptoms and behaviours to minimise these potential effects on the primary outcome, it would not be difficult for participants to report symptom improvements when they are asked the same questions at each research assessment because they would come to understand, to an extent, the questions they were being asked.
Regarding the recording of prison record data, it was discovered early on during the trial that, at HMP YOI Isis, the prison record data for participants who were transferred, released or deported would no longer be accessible to the research team. These were accessible with assistance from HMPPS. It was also found that the adjudication data may not be a true reflection of all of the critical incidents that took place, as only issues that had been witnessed were logged by prison officers. There has also been a reduction in prison officers and increased staff turnover during the preceding few years that might have led to an overall lack of sensitivity to reporting critical incidents. This reduction was considerable. The HMPPS budget was reduced from £3.47B in 2010/11 to £2.71B in 2016/17; as a consequence, there was a 30% reduction in front-line prison officers.108–110
At HM YOI Polmont, it was discovered during the data-checking stage at the end of the trial that the educational, vocational training sessions scheduled and attended were different on the trial database (using data collected at the time) from the prison database (using data at the end of the trial, which had changed from the time the data had been collected). This issue was explored and it was discovered that there was no audit trail on this information within the prison system, and that the prison database could have been entered or modified at any stage. This was discussed between the chief investigator, statistician, sponsor representative and the trial manager, and it was agreed that the data obtained at the time of data-gathering at baseline and week 8 should be used, as these were records of the prison database at those times.
The system in place to give IEPs for negative and positive behaviour was abandoned at HM YOI Polmont just before the start of the project, meaning that these data were available at the HMP YOI Isis site only. For this reason, IEPs were dropped from the SAP.
No pre tests were completed to compare the placebo with the active medication in relation to shape, size, colour, taste or smell. Even though the outer shell between the placebo and active medication was identical, once opened, there is a slight visible difference in size, with the distinctive black ink mark on the active medication that was not possible to replicate on the placebo capsules. This means that a participant who hid the medication in his mouth while being observed to swallow it could potentially open the shell and then establish whether or not he had received OROS-MPH. However, it is not likely that this had a significant impact on the trial outcome because adherence to treatment and dosing was lower in the OROS-MPH arm, guessing of trial medication arm was similar for participants and researchers (for OROS-MPH guessing was close to 50%, indicating random awareness for the active medication) and dosing might have been driven by the presence of adverse effects.
General considerations
General considerations that arise from the conduct of the trial protocol are useful to consider for future trials of mental health and behavioural problems in prison settings.
In the trial protocol, we indicated that, based on experience from the pilot study, we envisaged that adherence to the treatment regime would be challenging for around 20% of participants. However, in the current trial, adherence in terms of the percentage of prescribed tablets taken was around 50% overall by week 8, suggesting that the likely adherence rate was overestimated in the trial protocol. Apart from patient choice, several factors affected adherence to medication that were specific to the prison setting. The participants were able to receive the medication during set times only, and one of the prisons required the participants to remember to bring their identification cards with them. At times, there were issues with participants not being let out of their cells to receive trial (or any other) medication. We were able to manage some of these difficulties by using research staff to support the process of delivering trial medication, and adherence did appear to be greater when research staff were present during the dispensing times. For example, they were able to remind participants about the medication, to remind them to bring their identification cards and to ensure that prison officers let participants out of their cells to get to the medication dispending hatch. At HMP YOI Isis, this approach was implemented a few months into the trial to increase compliance and researchers made themselves available 4 days a week during the dispensing times in the morning. This indicates that, when working with prisoners with ADHD, there needs to be additional support from both prison and health-care staff to ensure that prisoners receive their required medication.
Compliance with some other medications appears to be greater than for the trial medication, but this level of information is not routinely collected in the prison setting. As a snapshot, during a week at HM YOI Polmont, five prisoners prescribed ADHD medication adhered to the treatment 40% of the time, in contrast with 77% adherence to daily general medication by 13 prisoners. Related to this, there were concerns about the type and use of mental health medications being prescribed. Pharmacy staff were particularly concerned by the high rate of prescribing and high doses requested for sedative medications such as mirtazapine and sedative antipsychotics. These drugs are likely to have an internal market in the prison that relates to the high demand for sedation among prisoners, as well as high levels of irritable and aggressive behaviour and sleep problems being managed in this way. The Inspectorate of Prisons report96 states that prisoners are more likely to use depressants than stimulants to counter the boredom and stress of prison life. In contrast, as seen in both our trial data and from the treatment of non-trial patients, there was generally intermittent use of OROS-MPH and dosing was low, indicating that participants were not pushing for higher doses in the same way they do for sedative medications. This also indicates a low abuse potential for OROS-MPH, which is difficult to use in a way that gets you ‘high’, and the potential internal market among prisoners for this medication appears to be low.
More generally, although procedures were in place for the delivery of critical medications to prisoners, the health-care staff often reported difficulties in delivering pharmacological treatments to prisoners. For example, Inspectorate of Prisons reports state that the clinical rooms (for dispensing of medication) were very crowded, which made the risk of error more likely,111 and lack of effective supervision of medicine queues compromised patient confidentiality and increased the potential for medication diversion.112 Therefore, it is worth considering the processes and procedures around dispensing medication in the prison setting to fully be able to evaluate the efficacy of a medication. It may be that mental health well-being is not considered high priority, and there needs to be greater support around the dispensing of medication by prison staff. Regarding the prescribing of stimulant medications for ADHD, because these are controlled medications, they require two people to issue the medication to prisoners, unlike other psychotropic medications. This needs to be explored further to be able to generalise these findings and consider the support needed for the delivery to prisoners of effective treatments for mental health problems.
The majority of the participants had short-term sentences and many probably faced unknown arrangements once released, such as no fixed abode and not being signed up with a GP, which would be required for any future medication. Therefore, many faced the serious prospect of having no follow-up procedures in place following release and, in the absence of an agreed clinical pathway, jeopardising their future ADHD treatment. This might have affected the participants and made them feel as if it would be pointless and not make any difference to their lives to engage fully with treatments offered in the prison setting. This contrasted with the Swedish study,33 in which there was a much longer period for planning treatment following the trial and continued treatment was assured either with the regular prison psychiatrists or by direct referral to an outpatient clinic. Future studies could focus not only on outcomes in prisons, but, critically, on wider clinical and social care pathways, including the support given in the community around housing and employment, as well as the focus on treating mental health disorders.
There was a noticeable difference in compliance with medication between the Swedish study33 and the current trial, reflecting a different approach in relation to the dispensing between the prison settings. In the Swedish study, participants could receive their medication at any point until lunchtime, and it was dispensed by the prison officers. In the CIAO-II RCT, the dispensing was by health-care staff during heavily restricted time slots, which was also affected by the prison regime tactics such as ‘lock down’ and ‘split regimes’. This flexible approach in the Swedish study was considered to be a major factor that contributed to the excellent adherence to medication and the direct involvement of prison officers in delivering mental health treatments may also have played a critical role. These factors could be explored further in future studies to evaluate the full impact that strict dispensing practice has, not only on treatment of ADHD, but on mental health disorders more generally.
It would be beneficial to look at the effect of prisoners attending behaviour programmes, both individually or in a group, on recidivism and critical incidents to fully understand the impact such measures can have and to contrast outcomes in prisoners who meet criteria for ADHD with those of prisoners who do not meet criteria for ADHD. For example, an evaluation of one prison programme found a statistically significant reduction of 6 per cent in the 1-year reconviction rate, compared with the reconviction rate of a matched comparison group, and a statistically significant reduction in the frequency of reoffending of 60 recordable offences per 100 released prisoners.113
It would also be useful for future studies of ADHD among prisoners if they had controls in the prison setting who did not meet the ADHD criteria to evaluate the effects of alcohol and substance misuse, childhood trauma and other factors in the prison setting on this group, compared with those diagnosed with ADHD.
Prison services for ADHD are thought to be highly underdeveloped.21 For example, there was no access to ADHD diagnosis and treatment prior to the current trials at HMP YOI Isis. A basic survey of existing services for prisoners with ADHD would be useful to understand what provision currently looks like nationally. Prisoners are meant to receive the same services as they would in the community, in accordance with the principle of equivalence.114
Conclusions and recommendations for future research
In conclusion, the trial is robustly neutral and does not support the routine treatment of young adult offenders with MPH.
Although this conclusion is disappointing, there are several lessons to be considered for future research and clinical practice. We divide these into the following: two recommendations with the highest priority, questions that address limitations of the current trial, general considerations for mental health research in prisons, and the future of clinical care for ADHD among prisoners.
High-priority research
Following the findings of this trial, there are two main objectives we recommend as having the highest priority: dosing and community studies.
Dosing
We cannot rule out the potential effects of poor adherence to trial medication and inappropriate dose titration. Overdosing, resulting in intolerable side effects, and underdosing, resulting in limited efficacy, are both common reasons for poor adherence and may have affected the results of this trial. Future studies of prisoners with ADHD should consider using a fixed-dose titration and ensuring that adequate doses are maintained.
Community studies
We cannot rule out the potential impact of the prison environment and trial procedures in this context on the reporting of ADHD symptoms. Future studies are, therefore, still required to investigate the treatment of ADHD in offenders in community settings.
Addressing limitations of the current trial
Multimodal treatment
The poor response to medication in this trial might reflect high levels of coexisting mental health disorders and exposure to high levels of trauma. Future studies could investigate the impact of a more comprehensive multimodal approach to treatment, including psychosocial and psychological treatments alongside medication for ADHD.
Early detection and prevention
The poor response in this trial might reflect the severity of ADHD and entrenched behaviours that have developed over many years. Future research should focus on early detection and prevention of ADHD, and the associated development of conduct disorder and ASPD.
Informant and use of objective measures
The possibility that participants exaggerated their symptoms in the ADHD screening test and DIVA might have led to a dilution of the number of participants with ADHD, leading to a negative result. One way to counteract this in future studies would be to require greater reliance on collateral history or past school reports in making the diagnosis, the use of tests of cognitive performance for deficits associated with ADHD such as sustained attention and response inhibition deficits, and measurement of motor restlessness.
General recommendation for investigations of mental health disorders and neurodevelopmental disabilities in prisoners
Effects of substance abuse
Active misuse of substances is a persisting issue in the two prisons that took part in this trial, as in others in the UK and elsewhere. The substances used by prisoners, such as spice, may be of lower cost and quality than in the community, and the effects of the use of such substances on this trial are essentially unknown. Future research needs to investigate the impact of spice and other drug use on ADHD symptoms, related conditions and the response to medication. We also need to understand whether or not, and how, the endemic use of substances in our prisons can be improved, given what is known about the adverse mental state effects of widespread uncontrolled substance misuse.
Models of care for individuals with complex and multiple conditions
A model is required (and should be subsequently evaluated) for the multidisciplinary and multiagency review of people who present with complex and multiple clinical conditions, including ADHD (e.g. the coexistence of ADHD, substance misuse and head injury). Although this framework is already meant to exist for those with serious mental illnesses through the Care Programme Approach in England and Wales,115 a wider framework including health and social care components may be useful for all prisoners who present with one or more conditions.
Future management of attention deficit hyperactivity disorder in prisoners
Clinical services for attention deficit hyperactivity disorder
Although the findings of this trial do not support the routine treatment of ADHD with OROS-MPH in prisoners, given the existing literature on treatment of ADHD, these findings should not exclude a role for health-care services in managing prisoners with ADHD. This may be particularly important for cases with greater severity and diagnostic certainty. In line with developments in community psychiatry, services for ADHD should ideally be integrated within existing prison mental health inreach teams, and considered alongside the treatment of substance abuse, and comorbid mental health and neurodevelopmental conditions. Examination of clinical pathways for ADHD is still indicated, with a focus on community mental health team engagement and onward GP registration for those with severe disorders, a good response to treatment of ADHD, and consideration of suitable options for employment and accommodation.
Barriers to treatment
Qualitative research is needed to find out from the prisoners what barriers there are to the management of ADHD in the community and in prisons. Qualitative research is also needed to find out what mental health and prison staff think about the diagnosis and management of ADHD. Given that few community or prison services currently include ADHD treatment components, it is likely that those who would benefit from the diagnosis and treatment of ADHD will not be receiving the assessment, care or treatment that they require.