Time to cessation
In our sample, the majority of participants prescribed medication at the age of 16 had stopped taking it 2 years later. The rate of cessation that we report remains greater than the estimated rate of decline of symptoms from epidemiological studies. Faraone et al.’s meta-analysis found the persistence of ADHD symptoms meeting the full criteria for the condition 1 year later to be 83%, whereas the probability of medication persistence in our study was 63%. Faraone et al. place the persistence of symptomatic ADHD at age 20 at 69%, and a more conservative estimate of persistence of the full ADHD syndrome at 28%; both in excess of the probability of remaining on medication by age 20 in this study (21%) [
1].
In contrast, Beau-Lejdstrom and colleagues analysed UK ADHD prescribing in under-16 s between 1992 and 2013 and found that more than three-quarters of children were still taking medication after 1 year and 60% were still on medication after 2 years [
17]. Our findings are more similar to estimates from McCarthy et al.’s earlier study where approximately 40% of those starting medication in childhood or adolescence remained on medication at age 18 [
6].
However, there is some evidence that prescribing patterns may be changing. Whilst a direct comparison should not be made with earlier studies, the populations are similar enough to note that our estimates of the rate of cessation from this analysis are slightly lower than those reported in the CADDY study, which covered the period from 1999 to 2006 [
5]. Furthermore, those born later in our cohort who reached the age of 16 between 2009 and 2011 had a reduced hazard of cessation compared with those born earlier, with the lowest likelihood of cessation amongst those turning 16 in 2011. These findings may reflect changing prescribing practice in the UK over the transition period, the development of adult services and an increasing acceptance that ADHD continues to merit pharmacological intervention after the age of 16, possibly influenced by guidance from NICE in 2008 and the British Association for Psychopharmacology [
7,
18].
Predictors of cessation
Being referred to adult psychiatry at any point was associated with a markedly reduced likelihood of medication cessation. This may reflect a greater willingness of GPs to continue prescribing ADHD medication with the specialist oversight recommended by guidelines [
18]. It is also likely that such patients would have more severe or persistent ADHD symptoms, which would mean they were both more likely to be referred and to continue medication.
The prescription of other non-ADHD psychotropic medication at the age of 16 or over was also associated with a reduced likelihood of medication cessation. This factor may be a marker for severity of ADHD or for other patient factors influencing help seeking behaviour and engagement with services [
19]. We did not find an association between cessation and gender (unadjusted HR for females of 0.93; 95% CI 0.78–1.10), in contrast to the CADDY study [
5]. This finding may have been influenced by the younger starting age in CADDY, or by secular changes. The gender distribution in our sample was in line with that from other studies of ADHD prescribing—12% of our sample were female, compared to 15% in Beau-Lejdstrom’s recent study [
17] and 9% in the older CADDY study [
5].
The only comorbid psychiatric diagnosis associated with a lower likelihood of cessation was ASD. Explanations for this finding are varied; people with ASD may be more likely to adhere to medication, or to experience greater impairment [
20]. Due to their dual neurodevelopmental disorders they may also be managed in services such as paediatrics or learning disability, where transition may occur later or where there may be a greater acceptance of the use of medication. A similar explanation could apply to the association between learning disability and remaining on ADHD medication.
Strengths and limitations
The chief strength of this analysis is the use of high quality and recent data from a national database capturing primary care prescribing until the end of 2013. The dataset covered a period which included the introduction of the NICE guidance on prescribing for adults and an expansion in awareness of adult ADHD. Use of shared care protocols means that in the UK, GPs undertake the majority of prescribing in ADHD, with specialist oversight [
21]. Consequently, primary care prescriptions are likely to provide the fullest available picture of prescribing without using data linkages. Alternative sources such as dispensing records may offer more limited details on the diagnoses of the patient and other prescriptions which may be issued to them [
10].
Nonetheless, primary care records will not capture prescriptions issued in specialist services which are not then passed on to primary care to continue, for example, in shorter-term trials of medication, or highly severe and complex cases. Where prescriptions are initiated outside primary care then transferred over to the GP, the length of time that an individual has been on medication may be underestimated, although the date of cessation (which was central to this analysis) would be unaffected. It is also important to consider that the outcome in this study was cessation of prescribing; some young people may not have been adherent to their medication and may have stopped taking it at an earlier date, if they took it at all [
22]. With no standard definition of how long without prescribing constitutes ‘cessation’ of medication, we also had to decide upon a definition of cessation; the choice of a minimum 6-month period without prescribed medication was taken to allow for errors in estimating prescription length, medication breaks, and administrative delays.
There were fewer than 200 individuals remaining in the survival analysis after 4 years of follow-up, with greater uncertainty around the probability of cessation beyond the age of 20. Subjects censored during the study period may also have affected the results if the censoring was informative. This could apply to individuals whose registration with a CPRD practice was terminated before the end of the study period and who were prescribed medication until the point that their registration finished. This termination of registration could be related to moving away, which feasibly could be related to stopping medication, if it is not continued elsewhere. Alternatively, it could be due to moving to higher education and indicate a propensity to continue medication for the purpose of studying. Nonetheless, when these subjects were removed in a sensitivity analysis, there was no significant effect on the findings, suggesting minimal influence of informative censoring.
The main limitation of using this primary care dataset was the lack of clinical detail coded in the records, such as measures of ADHD symptoms and severity. It was not possible to determine for each case whether cessation took place due to remission of ADHD, patient choice, or ineffectiveness of medication; or whether it was due to service-related factors. Instead, the results should be considered in the light of other research into transition to provide context and explanations for exploration in future studies.
Implications
Despite the existence of guidance on the management of ADHD in young people in transition, research findings imply that these recommendations are not always implemented in UK clinical practice. Hall and colleagues report wide variation in the provision of services and in prescribing arrangements for people with ADHD in transition in England [
10,
11]. Professionals also report a lack of confidence in prescribing for ADHD in over-18s [
9,
23,
24]. Given these potential barriers to ongoing prescribing, it is, therefore, plausible that a proportion of the young people in our study may be stopping medication from which they could still benefit, and may, therefore, be at greater risk of experiencing the adverse outcomes associated with ADHD [
25].
The experiences of clinicians including paediatricians, psychiatrists and GPs are important both in understanding how these results reflect what is happening in everyday practice; and in considering how to target barriers and facilitators to optimising ADHD management in primary care and specialist services. Beliefs and knowledge will be influential in determining whether and how commissioners and clinicians implement the guidance on managing young people with ADHD in transition and, therefore, warrant further investigation through interview or survey studies.
Prescribing for young adults will also be heavily influenced by transition pathways and by the availability of services where ADHD medication can be monitored. There are added complications due to the UK prescribing regulations on controlled ADHD drugs, which may be implemented differently across localities and create further complexity in commissioning and in shared care arrangements [
18]. Furthermore, prescribing is only one part of the approach to managing young people with persisting ADHD symptoms. There are various models of delivering care for older adolescents and adults with ADHD including extension of child mental health services, youth services and specific adult ADHD services, and recent recommendations on transition have been made by the UK Adult ADHD Network (UKAAN) [
26‐
28]. However, the evidence regarding the effectiveness of different service models in improving outcomes needs to be strengthened—a systematic review in 2015 by Paul and colleagues concluded that there was currently no ‘high quality’ evidence to support the use of any particular transition care models [
29]. Studies of adherence suggest that young adults may view stopping medication as an exertion of their autonomy, or may perceive that the negatives of medication outweigh the positive [
22]. Consequently, there is a related question to be answered around what interventions or delivery models are most acceptable to young people, and would increase their engagement with managing their ADHD, and with services at this vulnerable time.
In conclusion, we detected high levels of cessation of prescribing for young people reaching transition age, which in combination with epidemiological data suggest that some young people may be stopping pharmacological treatment when they could still benefit from it. Robust evaluation of the cost-effectiveness, outcomes and acceptability of different service configurations is, therefore, necessary and would support and inform investment by commissioners in an environment where resources are scarce.