Attention-deficit hyperactivity disorder (ADHD) is a syndrome comprising inattention, impulsivity and overactivity in variable proportions, sufficient to cause impairment in functioning. It was initially understood as a disorder of childhood, but it is now recognised as affecting adults also (Kooij et al. Reference Kooij, Bejerot, Blackwell, Caci, Casas-Brugué, Carpentier, Edvinsson, Fayyad, Foeken and Fitzgerald2010), and is well established within conventional psychiatry. However, psychiatry itself is currently changing rapidly. A new approach to diagnosis, Research Domain Criteria (RDoC) for research purposes has been proposed (National Institute for Mental Health, 2016), separate from the traditional Diagnostic and Statistical Manual (DSM) and International Classification of Diseases (ICD) systems, which also continue to evolve; DSM version 5 (DSM-5) having been published in 2013, and ICD-11 due in 2018. Attempts to automate diagnosis are gathering pace. Molecular genetics, epigenetics and neuroimaging are converging to enable accurate accounts of brain function in a range of mental states. The drive towards evidence-based practice has led to a much more systematic consideration of published research about treatment effectiveness. Meanwhile, the critical psychiatry movement continues to argue that the construct of ADHD exemplifies why the diagnoses employed by psychiatry are wrong and harmful. While some attacks on ADHD have simply been mendacious (Barrett, Reference Barrett2015), legitimate criticism has come from two directions. Some argue that ADHD is a medicalising social construct, originating in the United States but being globalised (Conrad & Bergey, Reference Conrad and Bergey2014). Others claim, using participant observation, that the evidence on which the diagnosis of ADHD is justified is viewed through a distorting cultural lens, where the knowledge that is privileged is chosen to support a dominant power structure, ‘the biomedical framework’ (Moncrieff & Timimi, Reference Moncrieff and Timimi2013). This leads to concerns that a diagnosis-led approach to the care of ADHD is, in important respects, dehumanising (Gambrill, Reference Gambrill2014) and delivered for the benefit of vested interests, for example, the pharmaceutical industry (Phillips, Reference Phillips2006). This paper therefore considers how these new advances affect our understanding, diagnosis and treatment of ADHD.
Diagnosis
RDoC
Though not currently intended to replace the clinical systems of ICD and DSM, RDoC greatly expands the traditional approach of symptoms and signs, to capture the changes in our understanding of psychiatric disorder arising from the insights achieved through genomics and imaging. As a newly developed research framework for general use, a detailed discussion of it is beyond the scope of this paper. For ADHD, it should help clarify the set of gene–behaviour associations that remain currently unmapped, despite overall heritability of 0.7–0.8, including those which also affect disorders other than ADHD. For example, genetic abnormalities common to multiple disorders suggest a general vulnerability to psychopathology, irrespective of diagnostic type (Cross-Disorder Group of the Psychiatric Genomics Consortium, 2013). However, many of the RDoC domains have not been validated in children. While validation research is ongoing (Doyle, Reference Doyle2015) clinicians working with ADHD in children should be cautious about extrapolating RDoC-based research that they read.
DSM-5
The introduction of DSM-5 involved considerable controversy (Strakowski & Frances, Reference Strakowski and Frances2012) some of it involving ADHD. There was a general concern that the boundaries of psychopathology were being extended too far: for ADHD, there were also specific concerns that modifying the age of onset criterion would impair the specificity of the diagnosis. Data are now accumulating to address these concerns. There appears to be a bimodal distribution for child and adult ADHD: very few child onset cases have symptoms by 38 years, and most adult cases developed symptoms after 12 years (Moffitt et al. Reference Moffitt, Houts, Asherson, Belsky, Corcoran, Hammerle, Harrington, Hogan, Meier, Polanczyk, Poulton, Ramrakha, Sugden, Williams, Rohde and Caspi2015): the childhood group also had more marked cognitive and behavioural difficulties, though ADHD symptomatology met similar criteria. This has subsequently been replicated with another cohort (Agnew-Blais et al. Reference Agnew-Blais, Polanczyk, Danese, Wertz, Moffitt and Arseneault2016). Unlike age of onset, the 18 symptomatic criteria for ADHD were carried through largely unchanged. All were found to have discriminatory power and predict impairment, though not all to the same degree (Rosales et al. Reference Rosales, Vitoratou, Banaschewski, Asherson, Buitelaar, Oades, Rothenberger, Steinhausen, Faraone and Chen2015). The domain model of RDoC thus offers a framework to research these commonalities and differences.
Diagnostic technologies
Traditional methods of diagnosing ADHD are resource-intensive, leading to significant delays in treatment, even when streamlined by guidelines (Foreman, Reference Foreman2010). The oldest technology available is rating scales, which typically are used for screening in ADHD, as they tend to over-identify (Foreman et al. Reference Foreman, Dack and Ford2008). Structured clinical interviewing provides reliable assessments (Goodman et al. Reference Goodman, Ford, Richards, Gatward and Meltzer2000). Advances in information processing have enabled computer-supported methods to be developed, which offer potential advantages in time, either through not requiring the diagnostician to be present for data collection (Foreman et al. Reference Foreman, Morton and Ford2009), or using combinations of tests with computerised algorithms and cognitive testing to provide diagnoses (Hall et al. Reference Hall, Walker, Valentine, Guo, Kaylor-Hughes, James, Daley, Sayal and Hollis2014), though these last are still in development. This range of approaches offers an opportunity to test one prediction arising from ADHD diagnostic criticisms. Were the diagnosis arising from a cultural lens, rates would vary significantly across different cultures and times, given their different ‘distances’ from the dominant viewpoint. Also, combining culture-independent cognitive tests with the conventional diagnosis would reduce diagnostic reliability, as the cultural bias would be diluted. A recent systematic review and meta-regression analysis were able to distinguish between the methodological and geographical or study year variance components of prevalences (Polanczyk et al. Reference Polanczyk, Willcutt, Salum, Kieling and Rohde2014). Contrary to the cultural lens hypothesis, it found significant differences between studies could only be explained by differences in their methodologies. The cognitive test used most frequently in relation to ADHD diagnosis is the Continuous Performance Test (CPT) for attention, while activity has been measured using actigraphy. Unfortunately, most studies have used these as independent measures to predict diagnosis, using quite small sample sizes. The results found were variable (Hall et al. Reference Hall, Valentine, Groom, Walker, Sayal, Daley and Hollis2015) consistent with the variable discriminatory power of different individual symptoms of ADHD identified by Rosales et al. (Reference Rosales, Vitoratou, Banaschewski, Asherson, Buitelaar, Oades, Rothenberger, Steinhausen, Faraone and Chen2015). The cultural lens hypothesis predicts that diagnostic reliability should be reduced when such tests are added to existing clinical assessments, as clinical support. Little research on this has been undertaken, but a combination of CPT and actigraphy measurement appeared to improve ADHD repeat reliability (Vogt & Shameli, Reference Vogt and Shameli2011) and also improve discriminant validity between ADHD and Autism (Groom et al. Reference Groom, Young, Hall, Gillott and Hollis2016). These results are not consistent with the claim that the diagnostic category of ADHD results from cultural bias.
Genetics
Though the high heritability of ADHD has been known for some time, unequivocal evidence, unconfounded with potential environmental effects was identified in 2010, when an international population with ADHD was shown to have a greater proportion of copy number variants than controls (Williams et al. Reference Williams, Zaharieva, Martin, Langley, Mantripragada, Fossdal, Stefansson, Stefansson, Magnusson, Gudmundsson, Gustafsson, Holmans, Owen, O’Donovan and Thapar2010). It also identified the genetic overlap between ADHD and other disorders that have since been replicated (Cross-Disorder Group of the Psychiatric Genomics Consortium, 2013) and so contributed to the development of the RDoC project discussed above. Conceptually, it has meant that the difficulties associated with ADHD diagnosis are no longer grounded in questions of social construction (Quinn & Lynch, Reference Quinn and Lynch2016) but of measurement, as it now seems clear that a purely phenomenological account of ADHD (as provided by DSM or ICD) is inadequate for determination of aetiology. However, a recent meta-analysis (Middeldorp et al. Reference Middeldorp, Hammerschlag, Ouwens, Groen-Blokhuis, Pourcain, Greven, Pappa, Tiesler, Ang, Nolte, Vilor-Tejedor, Bacelis, Ebejer, Zhao, Davies, Ehli, Evans, Fedko, Guxens, Hottenga, Hudziak, Jugessur, Kemp, Krapohl, Martin, Murcia, Myhre, Ormel, Ring, Standl, Stergiakouli, Stoltenberg, Thiering, Timpson, Trzaskowski, van der Most, Wang, Nyholt, Medland, Neale, Jacobsson, Sunyer, Hartman, Whitehouse, Pennell, Heinrich, Plomin, Smith, Tiemeier, Posthuma and Boomsma2016) has suggested that either dichotomous or dimensional measures of ADHD capture the genetically common phenotype of the disorder, while differing symptom profiles reflect different proportions of single nucleotide polymorphisms, which is consistent with the phenomenological observations of Rosales et al. (Reference Rosales, Vitoratou, Banaschewski, Asherson, Buitelaar, Oades, Rothenberger, Steinhausen, Faraone and Chen2015). So, the diagnostic approach adopted by ICD and DSM, while inadequate for achieving full construct validity, is sufficient for predictive validity.
Evolutionary genetics and ADHD
Darwin’s principles of natural selection imply that either the ADHD phenotype itself conveys a reproductive advantage in our environment or the genetic variability associated with it does. A recent study of very low levels of ADHD symptomatology attempted to explore this (Greven et al. Reference Greven, Merwood, Meer, Haworth, Rommelse and Buitelaar2016), finding, for this group, low heritability with high non-shared and shared environmental influences, with significantly associated advantage. While evolutionary theories about the value of the ADHD phenotype have been proposed, none have been validated (Thagaard et al. Reference Thagaard, Faraone, Sonuga-Barke and Østergaard2016). This evidence points towards the associated genetic variability being associated with advantage, rather than the phenotype itself. So, taking a strongly biological and empirical stance in relation to ADHD leads to a conservative position on either eugenic or genetically engineered attempts to eradicate the disorder, as we are currently ignorant of the negative impact of reducing our genetic variability in this way. Advances in the genetics of ADHD thus serve to prevent, rather than promote, potentially dehumanising programmes of care.
Neuroimaging
By analogy with the Human Genome Project, the Human Connectome Project (HCP) (National Institutes for Health, 2016) seeks to provide an atlas of the human ‘connectome’, that is, a comprehensive map of neural connections in the brain. It employs a twin/sibling design, thus facilitating genetic studies, and can achieve unparalleled levels of resolution. Its measures have been specifically chosen to map onto RDoC. Unfortunately, at present no paediatric HCP exists, but clearly will be needed if the advances this project offers are to be directed towards ADHD (Baroni & Castellanos, Reference Baroni and Castellanos2015).
At present, imaging offers a persuasive, but not conclusive, an account of the neurological basis for ADHD phenomenology. This is probably due to methodological problems, in particular, the small sample sizes used in studies employing this very expensive technology. This underpins the future value of HCP-based studies. In the meantime, meta-analyses can increase effective sample size. They may also propagate error, which can be greater in smaller studies, so their own results tend to be over-inflated, but only modestly (Button et al. Reference Button, Ioannidis, Mokrysz, Nosek, Flint, Robinson and Munafò2013). So, carefully designed meta-analyses can provide more accurate assessments of imaging findings, with an acceptable margin of error.
Functional neuroanatomy of ADHD
From meta-analytic studies, there is reliable evidence that the right-lateralised ventral attention network (Corbetta Lab, 2016), and the fronto-parietal central executive networks (Menon, Reference Menon2015) are hypoactive in individuals with ADHD (Hart et al. Reference Hart, Radua, Nakao, Mataix-Cols and Rubia2013), which correspond to the attentional and impulsive phenomenology of ADHD, respectively. The co-heritability of both ADHD traits and lower levels of measured executive function (Crosbie et al. Reference Crosbie, Arnold, Paterson, Swanson, Dupuis, Li, Shan, Goodale, Tam, Strug and Schachar2013) provides convergent validity. Imaging also confirms fronto-striatal dysfunction (Castellanos & Proal, Reference Castellanos and Proal2012) and some differences between children and adults, though interpreted developmentally at the time (Frodl & Skokauskas, Reference Frodl and Skokauskas2012) converge with the findings of adult–child differences discussed above. Both developmental and cross-sectional differences in the ventral striatum have also been identified, consistent with the abnormal reward sensitivity typical of ADHD (Baroni & Castellanos, Reference Baroni and Castellanos2015).
Lack of evidence for a biological substrate to ADHD has been a key component in arguing that the diagnosis is inappropriate medicalisation of a social phenomenon (Lindstrøm, Reference Lindstrøm2012). The last critical review of neuroimaging in ADHD was published 7 years ago (Leo & Cohen, Reference Leo and Cohen2009) and, while correctly identifying the problems of individual neuroimaging studies, did not consider meta-analytic methodology. The convergence of these meta-analyses with genetic, neurocognitive and phenomenological descriptions of ADHD confirm the existence of a biological substrate to the phenomenological syndrome, which has not yet been characterised fully.
Drug treatment of ADHD
Following studies such as the multimodal treatment for ADHD (Jensen et al. Reference Jensen, Arnold, Richters, Severe, Vereen, Schiller, Hinshaw, Elliou, Conners, Wells, Swanson, Wigal, Cantwell, Abikoff, Hechtman and Newcorn1999) it became generally accepted that there was sufficient evident to be confident in prescribing stimulant medication, most frequently methylphenidate, to children with ADHD. However, this has become contested, following a recent Cochrane Systematic Review (Storebø et al. Reference Storebø, Ramstad, Krogh, Nilausen, Skoog, Holmskov, Rosendal, Groth, Magnusson, Moreira-Maia, Gillies, Buch Rasmussen, Gauci, Zwi, Kirubakaran, Forsbøl, Simonsen and Gluud2015), which has been challenged by other expert groups (Banaschewski et al. Reference Banaschewski, Gerlach, Becker, Holtmann, Döpfner and Romanos2016). The disagreement is a technical one between expert meta-analysts, and beyond the scope of this article. However, these two groups have provided competing estimates of effect size, and it is instructive to compare the difference between them. For teacher-rated reduction in ADHD symptoms, the effect sizes are estimated as 0.77 (Storebø et al. Reference Storebø, Ramstad, Krogh, Nilausen, Skoog, Holmskov, Rosendal, Groth, Magnusson, Moreira-Maia, Gillies, Buch Rasmussen, Gauci, Zwi, Kirubakaran, Forsbøl, Simonsen and Gluud2015) and 0.89 (Banaschewski et al. Reference Banaschewski, Gerlach, Becker, Holtmann, Döpfner and Romanos2016), giving a difference of 0.12. This corresponds to a 4% difference in treatment efficacy, estimated as a group-level improvement, where the lower estimate of efficacy would be just below 79% (Coe, Reference Coe2002). So, while the debate is of importance regarding appropriate standards for meta-analysis and future research, the findings themselves do not greatly modify prior conclusions regarding the efficacy of stimulant medication for ADHD.
The critical account of ADHD does not deny the efficacy of the medication. Instead, it argues that, because the medication is being used as a means of social control, the adverse effects of this must outweigh any benefits, as there is no psychopathology to treat. Therefore, measures of symptomatic change do not address their argument. More appropriate are measures of quality of life (QoL). These broad measures of well-being are sensitive to many everyday aspects of life, not necessarily directly connected with symptomatology (Bai & Lazenby, Reference Bai and Lazenby2015). The Cochrane review just discussed included three QoL studies: it reported the effect size as being 0.61, equivalent to an average group-level improvement of around 73%. However, only one, parent-rated measure was used, the Child Health Questionnaire (CHQ). Though the CHQ has been validated for ADHD (Rentz et al. Reference Rentz, Matza, Secnik, Swensen and Revicki2004) the use of a parent-rating scale is open to challenge: a systematic review of QoL rating scales in ADHD has found variable self-reports in both ADHD-related QoL impairment, and in correlation with parent-rated scores (Danckaerts et al. Reference Danckaerts, Sonuga-Barke, Banaschewski, Buitelaar, Döpfner, Hollis, Santosh, Rothenberger, Sergeant, Steinhausen, Taylor, Zuddas and Coghill2009). One subsequent paper reported a self-rated QoL improvement associated with an improvement in ADHD control due to change of medication (Kordon et al. Reference Kordon, Stollhoff, Niederkirchner, Mattejat, Rettig and Schäuble2011) but the study was drug-company sponsored.
It seems that there is no more than suggestive evidence that self-rated QoL is improved by medication management of ADHD, the uncertainty arising from methodological difficulties in QoL measurement. However, the critical model of medication management implies that children’s QoL would decrease, and no evidence for this was found.
Non-drug treatments for ADHD
In 2013, our understanding of non-drug treatments for ADHD changed. A systematic review (Sonuga-Barke et al. Reference Sonuga-Barke, Brandeis, Cortese, Daley, Ferrin, Holtmann, Stevenson, Danckaerts, van der Oord, Döpfner, Dittmann, Simonoff, Zuddas, Banaschewski, Buitelaar, Coghill, Hollis, Konofal, Lecendreux, Wong and Sergeant2013) reported effect sizes which included only (probably) blinded studies, as well effect sizes that included unblinded ones. It focussed on core ADHD symptoms, that is, inattention, impulsivity and restlessness, but not behavioural problems, in relation to dietary and a variety of psychological interventions. It found that, when appropriate blinding was used, a small effect for free fatty acid supplementation (0.16) and modest effect size for artificial food colour exclusion (0.42) remained, but the effect of any psychological intervention upon core symptoms of ADHD was non-significant. This contrasted with effect sizes of 0.4–0.64 for psychological treatments when unblinded studies were included. An overlapping review of just cognitive training for ADHD, which also distinguished between blinded and unblinded ratings, obtained similar results (Rapport et al. Reference Rapport, Orban, Kofler and Friedman2013). To date, no study has appeared which has conclusively altered these findings: those with positive effects were unblinded for ADHD core symptoms; blinded studies found no effects. A possible exception was a sleep intervention study (Hiscock et al. Reference Hiscock, Sciberras, Mensah, Gerner, Efron, Khano and Oberklaid2015) which identified improvement in both unblinded ADHD assessment and blinded working memory assessment 6 months after the intervention. However, both the working memory and ADHD symptomatic improvement could simply be related to the cognitive benefits of better sleep. Consistent with this, the authors reported a post-hoc mediation analysis which ascribed 33% of the ADHD effect to better sleep. Exercise has also been used for ADHD treatment, and a recent meta-analysis has reported a moderate effect size for executive function (Vysniauske et al. Reference Vysniauske, Verburgh, Oosterlaan and Molendijk2016). Unfortunately, significant publication bias and poor study quality means this result is almost certainly exaggerated: only three additional negative studies would be required to turn the effect size non-significant. Interest in non-invasive brain stimulation for ADHD is increasing, but so far positive results have been at case study level (Rubio et al. Reference Rubio, Boes, Laganiere, Rotenberg, Jeurissen and Pascual-Leone2016).
The definitive critical textbook on ADHD (Timimi & Leo, Reference Timimi and Leo2008), consistent with the critical formulation of ADHD as a social construct, recommends a range of psychosocial treatments, and these recommendations have not changed since. It can be seen that these are unlikely to be effective on core symptoms of ADHD, and reported benefits are likely to result from observer effects.
Conclusions
Recent advances have clarified the epidemiology, genetics and neurobiology of ADHD, so claims that ADHD is solely a social construct can be conclusively refuted. While the phenomenological approaches of DSM and ICD are no longer sufficient to provide construct validity for ADHD, they provide sufficient predictive validity to be valuable as guidance for clinicians, informing effective treatment plans. Furthermore, the alternative of formulating ADHD as a medicalisation of a social predicament (Taylor, Reference Taylor1979) leads naturally to a psychosocial intervention, which will not improve the core symptoms of the disorder. However, showing that the critical formulation is wrong and misguided does not address all the concerns that this response to ADHD raises. At present, the evidence we use is based on group-level data (Baroni & Castellanos, Reference Baroni and Castellanos2015). However, the treatments we deliver for diagnosed ADHD are to individuals; symptom profiles differ between them and are confounded with the individuals’ other characteristics. The implications of this were illustrated by Oliver Sack’s account of Witty Ticcy Ray (Sacks, Reference Sacks1992), who regulated the level of his medication according to his social setting. These considerations point towards a sensitive use of outcome measures, in particular, QoL. These measures appear less well developed than symptom scores, and when they are employed (e.g. as Patient-Rated Outcomes), clinicians appear less responsive to them than to symptom changes (Greenhalgh, Reference Greenhalgh2009). This is an area where the critical psychiatry movement’s concern with sensitivity to individual differences, and concern to protect patients’ humanity, could gain useful traction.
Conflicts of Interest
None.
Financial Support
No funding body was involved in the preparation of this manuscript.
Ethical Standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committee on human experimentation with the Helsinki Declaration of 1975, as revised in 2008.
