Although CDG members declared any potential conflict of interest at each meeting, for some reason this was not collated in appendix 2 before completion of the draft. We would have liked to have the opportunity to comment on these details, and it is difficult to understand why these were not available in the draft. Can you explain?

The suspicion is that the conflict of interest within the CDG means that there has been a tendency to encourage the notion of ADHD as an explanation (rather than a description) of a larger variety of behavioural ‘problems’, (and potentially increase what is conceptualised as a behavioural problem), creating an increased potential market for stimulants,

Several members of the GDG have previously written papers or otherwise collaborated with the chair of the group. Should this not be declared as a potential conflict of interest? More generally, there is also the question of whether the CDG can be said to represent the broad balance of academic opinion, and how members were chosen for the CDG. Could you please tell us.

We always ensure that interests are fully declared in line with NICE’s requirements and these will be in the final full guideline available both on the net and in hard copy. We are sorry we were unable to provide the opportunity to comment on the declaration of interests in the first draft, but these had not been fully completed at this point.

GDG members declare personal non-pecuniary interests of having expressed a clear opinion on the matter under consideration. This may include recent papers.

When this guideline started in 2006, the process of recruitment differed from current procedures. Individuals were recruited through Royal Colleges, professional networks, through NICE and patient/carer national groups, with the aim of producing a GDG membership that is representative and including a range of professional fields that are actively involved in treating the disorder (such as paediatrics, psychiatry, psychology, general practice, nursing, etc.). The final make-up of the GDG was determined by consensus/consultation/advice, in line with NICE guideline processes. This aspect of guideline recruitment has now changed and for future revisions of this guideline we will be advertising and interviewing as appropriate.

The bald statement that ADHD is a valid condition will not do. The term "valid" can be misleading. There is little evidence that any functional psychiatric diagnosis, let alone ADHD, is valid, in the sense that there are syndromes defined by natural boundaries. It is essential that the guideline does not imply some scientific respectability for the concept of ADHD. We are not saying the concept is meaningless, but it is important to remember that ADHD is an abstract concept and there is a danger of "insidious reification". Once a diagnostic concept becomes reified and accepted, the category can take on "a life of its own". It is actually the duty of a guideline to protect against this potential harm.

Even on the evidence, issues such as the cross-cultural validity, the gender disparity, social class distribution have conveniently been largely ignored. Despite this, using the Washington diagnostic criteria (why were these criteria chosen beyond any other transparent, structured approach? – see below), you present fairly convincingly that there is simply insufficient and inconsistent evidence, which makes your conclusion regarding validity incompatible with even the evidence you present. If you really think you have produced evidence for a natural boundary, you have achieved a major scientific advance! You need to read the literature such as Kendell & Jablensky (2003), otherwise you are opening yourself up to providing a fundamentally flawed basis for the guideline.

Chapter 5 gives a detailed account and explanation for why the conclusion of validity was reached

We can’t help wondering if the Washington criteria were used partly to avoid addressing issues such as gender, cross cultural validity and class. Even if this is the case, as the GDG set out to examine the question of validity, as we have said, avoidance of these crucial issues is unacceptable. For example, ADHD in epidemiological studies shares the same gender distribution as the rest of preadolescent child psychiatry where a higher proportion of boys are found to have ADHD and other externalising disorders (i.e. behaviour problems causing concern/irritation in others) than girls. In surveys of prescribing for this an even higher proportion of boys are found to be in receipt of ADHD prescriptions. How do the GDG interpret this? As the GDG wish to conceptualise ADHD as a neurodevelopmental disorder (all be it on a continuum with normality) how do they account for this biologically? As regards class, the findings of differential class distributions in epidemiological studies in different countries, findings such as higher ADHD medication prescribing in higher social classes in US, but lower social classes in UK, require some explanation. Cross culturally two important questions need addressing; first differences in rates of diagnosis by ethnicity and countries and second the differing meanings attached cross culturally to ADHD symptoms. For example we need adequate explanations for findings such as higher rates of ADHD diagnosis in white compared to Hispanic and Afro-American communities in the US and lower prevalence rates of ADHD in epidemiological surveys in some countries. Anthropological research has also found differing meanings are attached to ADHD symptoms in differing cultures, with such symptoms being considered ‘abnormal’ and in the realm of medical pathology being a relatively recent phenomena, largely confined to Northern Europe and North America. Thus not considering the cross cultural significance of the ADHD literature risks the GDG coming out with guidelines that are institutional racist in character as they are likely to result in imposing a certain worldview about childhood and its problems onto communities who do not share such an opinion. The GDG have an ethical duty to examine the evidence in these socio-medical issues. We can make no apology for the likelihood that the task of interpreting the data in terms of validity of ADHD becomes more complex as insights from a variety of allied disciplines will have to be integrated.

The Washington criteria were selected as a previously applied set of criteria for addressing the question of validity of a diagnostic construct. The criteria lay out a set of clear questions that can be addressed by the review process and relate to the question of validity.

Gender differences are cited in the introductory chapter. Association with social class is cited in section 5.8.

Diagnostic and prescription rates show cultural variation for numerous disorders and these are important questions to be addressed to ensure consistency and equal opportunities. We have tried to clarify the questions of diagnostic differences by providing more detail on the critical definition of what would entail significant impairment for the diagnosis of ADHD and to trigger the diagnosis. The GDG note that high prevalence rates for ADHD are reported across numerous countries. We include a discussion of the main reasons for the wide range in prevalence rates cited. It is our opinion that a reasonable level of impairment is required to reach the diagnosis, that most people would agree warrants intervention. We acknowledge that the approach we have taken is based predominantly on a UK perspective of what constitutes impairment, since these guidelines are intended for use in the UK.

It may be helpful to include the full licensed indications for medication. For example for atomoxetine, the SPC says "STRATTERA is indicated for the treatment of Attention-Deficit/Hyperactivity Disorder (ADHD) in children of 6 years and older and in adolescents as part of a comprehensive treatment programme. " It goes on to describe what is typically included in a comprehensive treatment programme. For Ritalin, remedial measures alone also have to have been proved insufficient. Do you think your recommendations are within licensed indications? Is there enough emphasis on the comprehensive treatment package?

Why not use this element as a way of restricting the recommendation for medication? For example, the SPC for Ritalin suggests that a comprehensive treatment programme typically includes psychological, educational and social measures. If the recommendation about medication was restricted to those patients that also have these other measures in place, it may have more impact. Although it is a key priority of the guideline to ensure that drug treatment should always form part of a comprehensive treatment plan, the recommendations about medication could be couched much more clearly within this framework. As it stands, this so-called priority seems to be almost as an add-on in the list of key priorities. What we are suggesting is that it becomes even more of a key priority and the recommendations about medication are made in this context.

Thank you for your comments. We have considered your suggestions carefully. We have emphasised the need for prescribers to know the SPC for each drug in the introduction to the NICE guideline on page 6, second paragraph. In addition, we have reviewed the key priorities given your comments. There is always a balance to be struck between encouraging the comprehensive approach to treatment and recommendations that drugs are not the only treatments. As the key priorities now stand, we believe that the order of recommendations emphasises the correct place of psychological/behavioural, pharmacological and service-level recommendations. We have now added, nevertheless a recommendation to the list of priorities that echoes a recommendation to ensure that pharmacological treatment should always be part of a comprehensive approach, now not just for children but also for adults. We do not believe these are simple ‘add-ons’.

Although the guideline admits that, "Little empirical evidence is available to guide clinicians on questions such as the optimum duration of treatment, when it is appropriate to consider drug discontinuation and how and when to combine pharmacological and psychological treatments", there is insufficient emphasis on the fact that positive evidence for effectiveness has only been found for short-term behavioural and cognitive measures. The goal of intervention may be said to be to increase factors like learning, school retention, social functioning, and to decrease alcohol abuse, drug abuse and delinquency, but changes in these more meaningful outcomes in social and family domains are rarely measured. It does need to be emphasised that recommendations are being made on the basis of restricted evidence, rather than what is necessarily clinically meaningful. We are not suggesting you necessarily create a concept of clinical significance for ADHD trials, but it may be worth thinking about what might be seen as clinically significant improvements. For example, how large would the effect sizes need to be to be considered clinically significant?

Noted – partially amended (in Pharma chapter underline that evidence is on specific outcomes). Issues regarding the type of outcome measured, clinical significance and the level of impairment is addressed in Chapter 5, section 5.11.

The size of the standardised mean differences follow Cohen’s d:

-0.2 to -0.49 = small effect

-0.5 to -0.79 = moderate effect

> -0.80 = large effect size

Statistical significance is determined by the confidence intervals.

The GDG have shown a bias in interpreting the evidence on medication in the treatment of ADHD that puts their recommendations at serious odds with the evidence base. Unlike the issue of the validity of ADHD where the GDG chose a set of criteria (the Washington criteria), there is little in the way of a more structured approach to examining the evidence within the draft, thus it is unclear how the GDG’s recommendations connect with the evidence. This seems rather peculiar, as arguably the question of validity should have been discussed with reference to a multiplicity of positions (reflecting the fact that there are many validities to examine rather than one that can be assessed by one set of criteria such as the Washington), but with the effects of medication, where more objective comparisons are required, it should be possible to have a more structured approach. Are the GDG aware of how to use a ‘Hierarchy of Evidence’ approach? In the hierarchy of outcome measures in ADHD research, the most important is mortality, then severe adverse outcomes (e.g., hospitalizations, addiction, criminality), then long-term developmental outcomes (e.g. years of school completed, marks on exams), then short-term adverse effects such as behavioural improvement and cognitive performance tests measuring attention ability. Thus using this hierarchy the goal of interventions for a child with attention and behaviour problems are those of increase learning, school retention, social functioning, and to decrease alcohol abuse, drug abuse and delinquency. Changes in these more meaningful outcomes are rarely measured (King et al, 2006) instead, much of the literature reviewed by the GDG measure what is more accessible – cognitive skills, activity levels, or teacher, parent or laboratory ratings of social skills that are surrogates for the more meaningful achievements in social and family domains.

Thank you for your comments. When addressing treatments, whether psychological or pharmacological we use established methods to systematically search, retrieve, quality assure and meta-analyse (wherever possible) the evidence for the benefits and harms of treatments. These methods are widely regarded to be some of the most robust in the world (please see WHO evaluation of 25 national schizophrenia guidelines in which the NICE guideline on schizophrenia was regarded to be evidentially the most robust worldwide). When searching for evidence, guideline groups can only review the evidence retrieved from other primary studies and does not undertake primary research (except in rare instances such as, evaluating the experience of treatment). We are therefore limited to the outcomes that studies report which do not always live up to the hierarchy of outcomes you refer to. Nevertheless, the GDG are fully aware of the outcomes most relevant to children and their carers.

Although participants in the consensus conference may have raised the importance of thresholds for ‘severe’ ADHD, we cannot see that the guideline has adequately addressed these concerns. Where is the evidence that demonstrates that stimulants improve outcomes differentially the greater the severity?

Presumably the introduction of this threshold of severity has been introduced in an attempt to restrict indiscriminate prescribing. However, when no guidance is given on how you define 'severe', or 'impairment' or the many other subjective words used in the guideline, a likely implication is that we may find a shift taking place as those who prescribe simply add 'severe' before ADHD or 'moderate' but not responsive to other measures to justify that prescribing. The only clue on how to make these subjective distinctions is that it should be done by an 'expert' - this is likely to be determined by that so-called expert’s attitude to medication, rather than anything to do with severity.

Where is the evidence that medication should be first line for severe ADHD?

Furthermore, the guideline seems to be dismissive of psychosocial perspectives. Statements such as the following will not do - , "While it is important to acknowledge the validity of the social scientific paradigm and its body of literature, in the context of the development of practical clinical guidelines, it is not possible to offer alternative processes for clinical assessment or treatment. It is accepted that the research literature reflects the dominant medical scientific paradigm and hence the nature of the evidence base."

It is not so much that we are arguing for an alternative, but a full biopsychological assessment and management approach. It is clear from statements of this sort that the guideline is implicitly encouraging a biomedical orientation. It is not surprising when the chair of the CDG has committed himself in print to increasing the number of children treated. NICE should not allow itself to be a vehicle to implement such bias.

The evidence for the differential effect of stimulants on HKD and ADHD is given elsewhere in the guidelines (section on response to stimulants).

To clarify our definition of impairment further, we have further defined the terms mild, moderate and severe as applied to impairment associated with ADHD within these guidelines.

It is the view of the GDG that the text already indicates the importance of a bio-psychosocial perspective and the need for a full assessment that takes a broad multidisciplinary approach into account when making the diagnosis. The evidence base for treatment approaches is not discussed in the chapter but appear elsewhere in the guidelines

There barely seems to be any acknowledgement that ADHD in its origins was a childhood disorder. Minimising the significance of the extension of the concept into adulthood will not do. For all the debate about the validity of the concept in childhood in the guideline, any questioning of the extension of the concept to adults is obvious by its absence. There does not seem to be any attempt to discuss, for example, how the symptoms of impulsivity, hyperactivity and inattention may be developmental problems, and how adult ADHD may not make much sense as a psychiatric diagnosis in terms of developed personalities in adults.

It seems essential that the CDG seeks expert opinion from a sufficient range of adult psychiatrists, before extending its recommendations about ADHD to adults. We cannot see that this has happened.

Clearly ADHD is defined as a childhood onset disorder and this is described in the guidelines. There is now substantial evidence from longitudinal follow-up studies that the disorder persists into adulthood as the full disorder (meeting current criteria designed for children) in a minority of cases and in partial remission in a majority of cases; and this is referenced in the text.

We entirely agree that the relationship of ADHD to adult psychiatric categories and traits, such as development of personality problems, is hardly touched on in the current literature and there are key nosological questions to be addressed – this is one of the research recommendations.

The need for transitional and adult services for adults with ADHD was unanimously supported by the guideline development group members and is well supported by the current literature on the diagnosis and treatment of ADHD in adults.

Are there any advantages to having a specialist ADHD team within CAMHS? What’s the evidence?

The GDG’s recommendations are based on good practice guidance and expert opinion. The recommendations in section 6.8.1.1 are that Trusts should "consider" developing specialist teams or clinics. Particular advantages of specialist teams or clinics include optimising the delivery of evidence based treatments, service development, development of protocols, shared-care arrangements, clear communication across professionals, and the future development of transition services.

Is the preference given to parent training in the guideline over other psychological interventions evidence based? Just because the evidence does not exist for family therapy in terms of meta-analysis does not mean that this approach should be undermined.

The review of evidence on psychological interventions for ADHD included family therapy and was not limited to meta-analytic approaches (see for example the section of psychological interventions for adults with ADHD in chapter 7). However, only three RCT’s of family therapy interventions were identified by the literature search and none of these studies could be included: two compared alternative approaches to family therapy without an untreated comparison group (where studies lack an untreated control group we cannot establish whether an intervention is beneficial for children with ADHD); a third was a German language paper and was therefore not included according to NICE policy on foreign language papers (in any event the paper did not refer to randomisation). An additional search for non-experimental studies did not identify any studies using comparative designs that investigated the effect of family therapy compared with an untreated control condition.

NICE guidelines are evidence based where evidence is available and the guideline recommends parent training as there is good evidence of effectiveness. In the absence of evidence on the effectiveness of family therapy we could not recommend it as an alternative to an intervention that has good evidential support.

In spite of the fact that the NIMH Multimodal Treatment study of children with combined type ADHD (MTA) lacked a placebo control and was unblinded, in the absence of methodologically sound, well-reported RCTs, it offers useful medium-term data to date on ADHD interventions. Curiously the GDG’s interpretations of the 3 year follow up findings are even more optimistic about medication than the authors of the MTA themselves. By 3 years, according to the criteria used by the MTA all four experimental groups had shown significant improvement, but there were no significant differences between groups on any measure. Nor did any of the treatment arms confer protection from delinquent behaviour or substance abuse. Further analysis excluded potential explanations for the lack of benefit from aggressive use of medication. For example, convergence between the groups in medication use did not explain the outcome, and nor did a self-selection hypothesis (that an ‘association of severity and long-term medication use would result in selective long-term treatment of the most severe cases, potentially masking beneficial long-term effects of medication’). As a result of the publication of the three year follow-up, we now know that MTA protocol drug treatment has been shown not to be better than behavioural therapy or ‘community care’ at three years follow-up. This was predictable if the findings at 14 months were viewed as being due to enhanced placebo effects given that the behavioural arm of the treatment was completed several months before the active medication management arm, and around two thirds of those in the group with the poorest outcome group at this point, the community care group, were prescribed a stimulant. Unfortunately because of the lack of non-treatment controls, we do not know if at 3 years the outcome is any better than no treatment at all. Of more than 20 systematic reviews of ADHD medication treatment, five are of good quality (Jadad et al, 1999; King et al, 2006; Klassen et al, 1999; McDonagh and Peterson, 2005; Schachter et al, 2001) The most recent, the Health Technology Assessment (HTA) review (King et al, 2006), found 65 papers that met its inclusion criteria. The majority had less than 60 subjects, and followed patients for less than two months. Like its predecessors, the HTA review notes the inadequate reporting of study methodology, possible bias, limited reliability of results, inadequate data regarding adverse events, and a lack of evidence for superiority of one drug over others. It concludes however that there is short term effect on ‘core symptoms’ defined as attention difficulties and overactivity. But short-term improvement in behaviour is not a good measure of true benefit, using the Hierarchy of Evidence approach. Amongst commonly measured parameters, only academic achievement is a substantial outcome. The MTA study measured academic achievement with reading, mathematics, and spelling subscales from the Wechsler Individual Achievement Test. None showed any substantial advantage to drug treatment. At 14 months, there were no differences in mathematics or spelling, but combined treatment subjects made statistically significantly greater gains in reading than their behaviourally treated peers. This advantage had disappeared by 2 years. Reduction in disruptive behaviour and increased capacity to focus may be desirable in themselves and if medicating children for ADHD were a benign process, free from harms for the child and family, then improvement of the child’s current behaviour might be justified in the absence of more important benefits. But we do not have cause to be complacent about drug-induced harms. The MTA study demonstrated worrying effects of stimulant medication on growth rates. By the three year follow-up, those children who had been medicated were on average 2-4 cm shorter and 2-3.5 kgs lighter than their non-medicated peers. Case reports of sudden cardiac deaths led to an FDA advisory committee to recommend black box warnings for stimulants. These have been added for dextroamphetamine but not for methylphenidate. In addition several studies show that stimulants increase systolic and diastolic BP by up to 4 mm Hg, and heart rate by 3-6 bpm. Finally, there has not been sufficient follow-up to exclude negative effects of stimulant use on the growing brain. This lack of long term data on the effects of drug treatment are particularly important given that children on ADHD medications are usually prescribed them for many years – much longer than the periods for which there is an evidence base on effectiveness and possible risks. Whilst our current evidence suggests that the rate adverse event attributed to stimulants are not of sufficient frequency to preclude the use of a drug if there were proven substantive benefits such as improved academic performance in standardized exams or reduced delinquency, the evidence that stimulants do provide such benefits is lacking. While stimulants improve behaviour and cognitive task completion, there is little support for the clinical significance of this response. There is no RCT evidence for improvement in academic performance, reduced delinquency rates, improved school completion or improved social outcomes. There is no good evidence that children suffer less significant adverse outcomes in the short or long term if they take stimulants and there is some cause to worry that they may suffer more. Using such a structured approach brings recommendations that are at odds with the GDG. There is no evidence that stimulants are more effective in those with ‘severe’ symptoms of ADHD. It is unclear why in their research recommendation the GDG recommend examining discontinuation of treatment effects after 18 months (by which time a child would be exposed to withdrawal effects if they discontinued treatment) when the systematic review literature has not established that stimulants positive effects last longer than a few weeks. Given that we have not established that stimulants have clinically meaningful effects, even in the short term, using a Hierarchy of Evidence approach, then the only properly evidence recommendation is that stimulants should not be used as a first line treatment and that a research recommendation should be for a long term RCT which includes a placebo group. The current muddled recommendations are simply not acceptable.

Thank you; your comments have been noted.

We have been clear in the guideline that stimulant treatment is effective in improving ADHD core symptoms and conduct problems in children and adults with ADHD. This is an evidence-based guideline and our recommendations are based on this evidence. The MTA study is a complex study which we have tried to take into account but because of the study design this could not be meta-analysed. Of course, putting the whole picture together does raise issues regarding long-term use both in terms of effectiveness and possible harm and we have taken this into account in making our recommendations.

You should also note that we have not recommended psycho stimulants or other drug treatments as ‘first-line treatments’ except for children with severe impairment/symptoms (hyperkinetic disorder with severe impairment) and then in combination with parent training. All recommendations need to take into account the benefits and the risks of treatment based upon an analysis of the full data set available, and by comparison to other treatments that are available. We believe we have done this for all the treatments considered in this guideline and as a result do see that psycho stimulants have a place which we specify very carefully and based upon the evidence.

One way of looking at the MTA study is that it suggests any short term benefits of stimulants over behaviour therapy (and the benefits were few) are lost at three years. We are of course aware of the loss of randomisation at 14 months. However, at face value the study at least suggests the case for stimulants is still unclear. We are concerned that the guideline appears to minimise the potential impact of the MTA evidence. Also, the interpretation of the growth deficit could be said to be overfavourable. An interpretation of MTA is that it showed a 4cm growth deficit in those on stimulants compared to ADHD children not on stimulants and whether this is caught up after they are stopped is unclear from current evidence

Thank you, we have noted your comments

In general, there is too little on the methodological problems of clinical trials. The systematic review conducted for the US Agency for Health Care Policy and Research (Jadad et al 1999) found overall that the quality of the methodological reporting of the trials was poor, only 18 of the 77 RCTs were given 3 or more points on the Jadad methodological quality scale. You could work out the Jadad score for the trials included in your meta-analysis.

The critique of Michael Schlander (2007) Health Technology Assessments by the National Institute for Health and Clinical Excellence needs to be incorporated. The short duration of treatment vs placebo in many of the trials needs to be emphasised.

There are also issues of unblinding which may bias the interpretation of trials, and as far as we know, little attempt has been made to measure the degree of unblinding in ADHD trials.

Nor does there seem to be any discussion about whether any short-term positive effect found in clinical trials may not necessarily be specific to ADHD. Are the same short-term effects found in "normals"?

Thank you for drawing attention to this. We have noted your comments.

See chapter 3 Methods. While we agree there might be methodological problems with clinical trials, we decided that randomised controlled trials were the best study design to answer effectiveness or other impact of an intervention. Deciding on the best design type to answer a specific clinical or public health question does not mean that studies of different design types addressing the same question were discarded. We gave priority to RCT’s and where information was missing, we considered other types of designs.