Home > Clinical Guides > Attention Deficit Hyperactivity... > COMBINING AND COMPARING PSYCHOLOGICAL...

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Mental Health (UK). Attention Deficit Hyperactivity Disorder: Diagnosis and Management of ADHD in Children, Young People and Adults. Leicester (UK): British Psychological Society (UK); 2009. (NICE Clinical Guidelines, No. 72.)

11COMBINING AND COMPARING PSYCHOLOGICAL AND PHARMACOLOGICAL INTERVENTIONS

11.1. INTRODUCTION

This chapter reviews the evidence on the use of combined interventions where medication and psychological therapies are used together to treat ADHD. As well as the possibility of increasing treatment effects through the use of the two modalities of intervention together, the potential value of combined treatment for ADHD is an area of interest because it might lead to beneficial effects in different domains – with medication targeting core ADHD symptoms such as inattention and psychological interventions targeting secondary problems and coexisting conditions associated with ADHD. Combining pharmacological and psychological approaches may also have the potential to deliver both immediate effects on ADHD symptoms through medication along with more long-lasting effects through the development of behavioural and cognitive skills and strategies. Another area of interest in relation to combined treatment is the potential to minimise the risks of adverse effects of medication if combined treatment can achieve treatment effects comparable with medication treatment alone but with a lower dose of medication.

This chapter also reviews the evidence from trials that allow direct comparisons to be made between the effectiveness of psychological therapies and pharmacological interventions for the treatment of ADHD.

Evidence on combined treatment for adults with ADHD

None of the included studies investigated the effectiveness of combined interventions for adults with a diagnosis of ADHD or compared the effectiveness of psychological therapies delivered to a group not receiving medication for ADHD with those receiving stimulant medication in an adult population.

11.2. COMBINED INTERVENTIONS FOR CHILDREN WITH ADHD

11.2.1. Introduction

There are several reasons why non-pharmacological treatment, usually psychological, might be combined with pharmacological treatment. These are listed below.

  • In severe presentations of ADHD, the impairment is such that medication when combined with psychological therapy might offer the prospect of a more rapid improvement than with psychological interventions alone, which are likely to take longer to work. This may be particularly necessary if there is marked social dysfunction present, if there is severe pressure on family or marital relationships or if the child is faced with imminent exclusion from school.
  • Even if a psychological intervention is the preferred option, some young people have such severe clinical presentations that they and/or their parents may not be in a position to make use of psychological techniques. The potential for medication to deliver an initial rapid improvement in the early weeks of a combined intervention might enable them to benefit from psychological techniques.
  • It has been argued that stimulants may enhance conditionability, a key element of behavioural learning (Eysenck & Rachman, 1971; Sprague & Werry, 1971). In other words, stimulants may enhance the effectiveness of psychological interventions that employ behavioural and social learning principles.
  • Combining stimulants with a psychological intervention may be a way of reducing the dosage and duration of medication treatment, and thus may address concerns about the use of medication.
  • It has been suggested that there may be complementary benefits in combining approaches (Gitteleman-Klein et al., 1976a) in that stimulants may enhance attentional processes and reduce impulsive responding, whereas social reinforcement may help the child to internalise the value of appropriate behaviours.
  • There is little evidence that stimulant medication alters the relatively poor long-term outcome for many of those with ADHD (Weiss & Hechtman, 1993). Adding psychological and other therapies might therefore yield better long-term outcomes.
  • There are concerns that stimulants alone may not bring symptoms within the normal clinical range and have limited effects on other problems associated with ADHD such as prosocial behaviour (Buhrmester et al., 1992) and cognition (Pelham, 1986). Linked to this it is recognised that ADHD rarely presents with just the core symptoms of ADHD. A range of additional problems across multiple domains are usually present, which are likely to require a range of interventions (Wells et al., 2000).

Current practice

Current practice in the treatment of ADHD varies. Psychological, educational and pharmacological interventions may all be used; the decision is driven by the symptoms presented, the needs of the child and family and the local availability of services.

11.2.2. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline can be found in Table 32 (further information about the search for health economic evidence can be found in Section 11.5). Studies were included if they were RCTs that compared combined treatment for ADHD (where medication and psychological interventions are determined by the study protocol) with medication only delivered according to the same protocol as used as for the combined intervention.

Table 32. Databases searched and inclusion/exclusion criteria for clinical effectiveness of psychological interventions.

Table 32

Databases searched and inclusion/exclusion criteria for clinical effectiveness of psychological interventions.

11.2.3. Studies considered33

From the primary RCT search, the review team identified trials comparing combined treatment with medication only. Only trials that compared groups receiving true combined interventions (that is, medication for ADHD and a concurrent psychological intervention, with both interventions determined by the study protocol) with groups receiving medication alone (according to the same protocol as for the combined treatment group) were included in the review. Analyses comparing combined treatments with psychological therapies alone or with no treatment control conditions were not undertaken. The reason for this was that the analysis that directly compared pharmacological and psychological interventions (see 11.3 below) clearly favoured medication. If combined treatments were compared with psychological therapies alone or no treatment control conditions, findings favouring combination treatment might therefore only provide evidence of the effects of medication, rather than providing support for combined treatment per se.

Trials where participants received a psychological intervention as an adjunct to medication that they were receiving as part of their ongoing usual care were considered separately (see Chapter 7). This approach to the analysis was adopted because where participants in a trial continue to receive medication as usual the medication and/or protocol determining the medication regimen may not be uniform for all participants, and if that is the case the trial could not be considered to be a true test of a combined treatment strategy. It is also likely that adherence to medication would be higher in a clinical trial context, with the consequence that the effects of medication received as part of a clinical trial might be greater than the effects of continuing to receive medication as usual. Hence, if medication modifies the response to psychological therapies it is possible that findings from true trials of combined treatment might differ from findings from trials that include participants continuing with their usual medication.

An additional analysis is reported that compares an intensive and comprehensive combined intervention for ADHD with standard care that may include medication. This analysis, based on data from the MTA study (MTA 1999), was performed in order to provide a comparison of what might currently be considered the best possible care for ADHD with the more standard level of care provided in routine clinical practice.

Seven trials met the eligibility criteria set by the GDG, providing data on 544 participants. All were published in peer-reviewed journals between 1981 and 2004. In addition, 20 studies were excluded from the analysis. The most common reasons for exclusion were that the paper reported no appropriate data or the intervention was inappropriate (further information about both included and excluded studies can be found in Appendix 17).

11.2.4. Clinical evidence for combined treatment for ADHD versus medication only

Evidence from important outcomes and overall quality of evidence are presented in Table 33. The full evidence profiles and associated forest plots can be found in Appendix 19 and Appendix 18, respectively.

Table 33. Study information and evidence summary table for trials of combined interventions versus stimulant medication.

Table 33

Study information and evidence summary table for trials of combined interventions versus stimulant medication.

Evidence from included trials of treatment for children with ADHD that compare a combined intervention with receipt of the medication component of the intervention alone indicates that there is little or no advantage of any type of combined intervention over medication alone. Compared with medication there is no evidence of an added effect of combined treatment on measures of core ADHD symptoms, emotional state or self-efficacy.

The only evidence of a benefit of combined treatment over medication alone is for parent ratings of conduct problems at the end of treatment, however, the benefits of combined treatment on this outcome are only weak because the effect size is at the lower end of the small effect size range and no benefit of combined treatment was detected at later follow-up times.

The MTA study (MTA 1999) is the largest trial of combination treatment for ADHD and although the MTA data suggests that there was a small benefit from combined treatment over medication management alone on parent ratings of conduct problems at the end of treatment, the effect did not reach the magnitude of a small effect size.

11.2.5. Clinical evidence for intensive combined treatment versus usual care for children with ADHD

Comparison of the MTA combined intervention (medication management plus an intensive multimodal psychological intervention for ADHD that involved interventions with the child and parent, and a classroom intervention) with the MTA community care group allows comparison of an intensive and comprehensive approach to care with standard care (MTA 1999) (see Table 34).

Table 34. Study information and evidence summary table for the MTA trial of combined interventions versus community care.

Table 34

Study information and evidence summary table for the MTA trial of combined interventions versus community care.

The MTA study of combined intervention provides an example of what might be considered fully comprehensive care for ADHD; this is ongoing protocol-led management of stimulant medication coupled with a complex psychological intervention, that is, a multicomponent psychological intervention that continues for a year or more, includes components directed at the child, the parent and the teacher/classroom, and has intensive components (the summer camp in the case of the MTA psychological intervention). The MTA trial participants in the community care group received routine clinical care for ADHD; two-thirds of this group received medication for ADHD and community care participants might also have received non-pharmacological interventions. It is also important to note that the MTA study was US based. Standard care for ADHD in the US may differ from routine care in the UK, with the potential that a higher proportion of the children with ADHD in the community care group received medication than would be the case in a similar UK sample.

The MTA combined intervention was generally favoured over usual care on parent and teacher ratings of ADHD symptoms and conduct problems. According to the composite measure of teacher and parent ratings of core ADHD symptoms that is reported for later follow-up assessments, comprehensive care continues to be favoured over routine care 10 months after the end of the intervention, but the effect is only small. Twenty-two months after the end of the intervention neither comprehensive care nor routine care is favoured according to the composite measure of core ADHD symptoms. However, measures of conduct problems point to an unequivocal advantage of comprehensive care over routine care. At the end of the intervention both parent and teacher ratings of conduct behaviour favour comprehensive care over routine care, with a moderate effect size. The composite score for parent and teacher ratings of conduct behaviour reported for the later follow-up assessments indicates that the beneficial effect of comprehensive care over routine care reduced to a small effect 10 months after the end of treatment but increased to a large effect 22 months after the end of treatment.

Parent and teacher ratings of social skills at the end of the intervention also point to small gains from comprehensive care over routine care, but these weak effects disappear at the later follow-up assessments according to a composite measure that combines parent and teacher ratings of social skills. In contrast, parent ratings of their child’s emotional state point to a weak advantage of routine care over comprehensive treatment at the end of the intervention, but teacher ratings at the end of the intervention do not favour comprehensive treatment or routine care.

Taking all these findings into consideration there appears to be some benefit of a comprehensive intervention for ADHD over routine care. Measures of core ADHD symptoms at the end of the intervention indicate that comprehensive care is moderately more effective for core ADHD symptoms than community care, and comprehensive care may be particularly beneficial for conduct problems. It may be that the main factor generating the positive effects of the combined intervention is the medication management component. In any event, the comparison between outcomes for the MTA combined intervention ‘comprehensive care’ group and the community care group does not provide a consistent indication that comprehensive care is more effective than routine care that may include medication for ADHD. The advantage of comprehensive treatment over routine care should also be considered in the context of the lack of evidence of benefit from combined treatment approaches over active protocol-determined medication regimens (see Section 11.2.4).

11.2.6. Clinical evidence summary

Evidence from trials comparing combined treatment with medication interventions alone does not point to any added benefit of adding a psychological intervention to a protocol determined medication regimen. The data therefore suggests that if medication treatment for ADHD has already been instigated and the child has responded positively to treatment, then the addition of a psychological intervention to treat ADHD (whether a parent training programme or child-directed therapy) is unlikely to provide any added benefit in terms of reduced ADHD symptoms or improved behaviour, emotional state or self-esteem.

The findings on the effects of combined treatment therefore indicate that beneficial effects of psychological interventions for ADHD are not dependent on effective pharmacological treatment that allows the child with ADHD to be able to reap the benefits of a psychological intervention. It may be the case that in combined treatment trials the study-determined medication regimen has a large beneficial impact on outcomes such that any additional beneficial effects of a psychological intervention cannot be detected as there is no potential for any further improvement.

However, it should be noted that psychological interventions are effective as an adjunct to usual care medication (see Chapter 7). This may be because medication is less effective in routine clinical practice than in the context of a clinical trial. It is also the case that the MTA study (MTA1999) suggests that combinations of interventions may be helpful in targeting different problems and promoting some outcomes. Offering combination interventions may therefore allow children and parents to participate in treatment decisions and make choices about their own health outcomes (Taylor et al., 2004).

11.3. COMPARING PSYCHOLOGICAL AND PHARMACOLOGICAL INTERVENTIONS FOR CHILDREN WITH ADHD

11.3.1. Introduction

Direct comparison of the effectiveness of psychological and pharmacological interventions for ADHD is possible where RCTs include a group receiving a psychological intervention without medication and a group receiving medication only. Studies that allow this comparison are potentially informative as they allow a direct head-to-head comparison of effectiveness between psychological and pharmacological interventions.

11.3.2. Databases searched and inclusion/exclusion criteria

Information about the databases searched and the inclusion/exclusion criteria used for this section of the guideline can be found in Table 35 (further information about the search for health economic evidence can be found in Section 11.5). Studies were only included where both the medication and psychological interventions were determined as part of the study protocol.

Table 35. Databases searched and inclusion/exclusion criteria for clinical effectiveness of psychological interventions.

Table 35

Databases searched and inclusion/exclusion criteria for clinical effectiveness of psychological interventions.

11.3.3. Studies considered34

From the primary RCT search, the review team identified trials comparing medication for ADHD with a psychological intervention.

Six trials met the eligibility criteria set by the GDG, providing data on 462 participants. All studies were published in peer-reviewed journals between 1976 and 1999. In addition, four studies were excluded from the analysis; two because they were case studies, one because of insufficient data, and one because of methodological problems (further information about both included and excluded studies can be found in Appendix 17).

11.3.4. Clinical evidence for psychological interventions versus protocol-managed medication for children with ADHD

There is only sparse clinical trial evidence allowing direct comparison of the clinical effectiveness of psychological and pharmacological interventions for ADHD. Of the six trials that meet inclusion criteria, five are relatively small, with the medication or psychological intervention group sizes ranging from nine to 30. The MTA study (MTA1999) was relatively large, having 120 participants in the medication group and 119 in the psychological intervention group.

For individual outcomes, the quality of the evidence was generally moderate to high. Overall, for children with ADHD the evidence from trials that compare stimulant medication (predominantly methylphenidate) with a psychological intervention delivered to a group not receiving medication for ADHD generally favours stimulant medication, although where they reach statistical significance the effects are not large (see Table 36).

Table 36. Study information and evidence summary table for trials of stimulant medication versus psychological interventions.

Table 36

Study information and evidence summary table for trials of stimulant medication versus psychological interventions.

For both teacher and parent ratings of core ADHD symptoms and conduct problems at the end of treatment, stimulant medication delivers better outcomes than psychological interventions, with effect sizes in the small to moderate range. However, the benefits of stimulant medication over psychological therapies for core ADHD symptoms and conduct problems in general do not appear to be sustained at later follow-up assessments (3–6 months, 7–12 months and 13–24 months after the end of treatment). The MTA study (MTA1999) found a benefit of medication over the complex MTA psychological intervention on the composite parent- and teacher-rated measure of core ADHD symptoms at 10 months after the end of the intervention, but the effect did not reach the magnitude of a small effect size.

Stimulant medication also appears to be more effective than psychological interventions at improving social skills as rated by teachers, but this effect was small at the end of treatment, was not sustained at later follow-up assessments, and was not reflected in parent ratings of social skills, which does not point to any benefit of stimulant medication over psychological therapies at the end of treatment or any time thereafter. For emotional state (as represented by depression, anxiety, emotional adjustment and internalising symptoms) there was also a benefit of stimulant medication over psychological interventions at the end of treatment, but the effect size was small and for this outcome limited to parent ratings, with no effect on teacher ratings detected.

The lack of evidence for the sustained superiority of medication over psychological interventions for ADHD is, however, difficult to interpret. At longer follow-up time points outcomes may be influenced by the treatment a child has received since the end of the period of the experimental intervention. In particular, children who received a psychological intervention and were not on medication for ADHD during the period of the trial while the intervention was delivered may have subsequently begun to receive stimulant medication for ADHD. Notably, in the MTA trial, by the follow-up time point at 10 months after the end of the experimental intervention, 44% of the group that only received the MTA behavioural intervention had commenced medication. At 22 months after the end of the experimental intervention, the proportion of the MTA behavioural intervention group that was using medication at high levels was little changed at 45% (Jensen et al., 2007). In contrast 71% of MTA study participants who were in the medication management and combined intervention groups were using medication at high levels 22 months after the end of the experimental intervention, a decrease from 91% at the end of the intervention.

11.3.5. Clinical evidence summary

While there is no evidence that psychological interventions are favoured over stimulant medication for any outcome, or at any time point, it is also the case that medication does not appear to be strongly favoured over psychological interventions.

11.4. THE MTA STUDY: IMPLICATIONS FOR TREATMENT DECISIONS

11.4.1. Further considerations with respect to the treatment of ADHD – additional evidence from the MTA study

A number of publications have reported on sub-analysis and secondary analysis of data gathered as part of the MTA study (for example, MTA Co-operative Group, 2004a; Jensen et al., 2007). Only primary outcome data reported for the end of treatment and 14 and 22 months post-treatment follow-up have been used in the analysis of the effectiveness of interventions for ADHD, but some of the further analysis reported by the MTA Co-operative Group may help inform choices made in the treatment of ADHD.

Longer term impact of MTA interventions and the relation to substance use and delinquency

The MTA study group has reported follow-up outcome data at time points beyond the end of the MTA intervention (aMTA Co-operative Group, 2004a; Jensen et al., 2007). The initial primary outcome data was collected at the end of the MTA interventions (14 months after interventions were commenced), with follow-up data reported to date for 10 months after the end treatment and 22 months after the end treatment (or 24 and 36 months after the interventions were commenced). It is important to note that after the end of the MTA interventions the participants in the trial returned to usual care.

According to the outcome data collected 22 months after the end of the MTA interventions, the MTA treatment allocation groups could no longer be distinguished on any measure according to primary analysis (Jensen et al., 2007). On average MTA participants maintained some of the gains made at the end of treatment time point on measures of both ADHD and oppositional defiant disorder symptoms. The behavioural treatment and community care groups maintained the gains they had made at the end of treatment. In contrast the combined treatment and medication management groups lost their end of treatment advantage over the behavioural treatment and community care groups, although they maintained gains over baseline that approximated to the sustained gains made by the behavioural and community care groups. These findings are, however, based on the comparison with baseline data for each group, not on a comparison with an untreated control group, and hence it is not possible to conclude that any of the MTA interventions have long-term beneficial effects over no treatment. Indeed, at 22 months after the end of treatment, prognostic factors for ADHD were found to have more of an impact on outcomes than treatment group allocation – girls and those of higher socioeconomic status fared better than boys.

Follow-up data gathered 10 months after the end of the experimental MTA interventions is in line with the data gathered at 22 months after the end of treatment (MTA Co-operative Group, 2004a). Ten months after the end of the MTA interventions, the combined treatment and medication management groups, which showed the greatest improvement compared with baseline at the end of the intervention, show some deterioration, whereas the behavioural treatment and community care groups maintain gains made in comparison with baseline during the period of the trial intervention.

Jensen and colleagues (2007) suggest that factors that may contribute to the convergence of outcomes for the four MTA study intervention groups at longer-term follow-up compared with outcomes at the end of treatment include: a decrease in ADHD symptoms related to age independent of treatment; changes in the intensity of medication use; and different degrees of starting and stopping medication in the different treatment allocation groups that occurred after the end of the MTA interventions. Other factors may also be involved. There is a degree of convergence across the four groups in terms of their use of medication for ADHD at follow-up. Medication use in the group allocated to behavioural treatment increased from 14% at the end of the MTA intervention to 45% 22 months later, whereas among MTA participants who received the medication management intervention (including the combined treatment group) medication use decreased from 91% to 71%. In the community care group medication use was near unchanged: 60% at end of treatment and 62% 22 months later. Further support for the inference that changes in medication use may have mediated the convergence between outcomes across the groups at follow-up is provided by analysis indicating that the subgroup that reported stopping taking medication 10 months after the end of treatment showed the greatest deterioration (MTA Co-operative Group, 2004b).

Substance use at 22 months after the end of the MTA interventions was lower in the MTA participants who received intensive behaviour therapy (members of the combined treatment and behavioural intervention groups) compared with those who did not (members of the medication management and community care groups) (Molina et al., 2007). However, the data did not point to there being any associations between treatment allocation and early substance use, growth of delinquency over time and the level/seriousness of delinquency. Seriousness of offences was associated with self-selected use of prescription medication, and Molina and colleagues (2007) speculate that this could be reactive in that there may be a tendency to opt for medication in response to increased symptom severity.

The analysis by Molina and colleagues (2007) did not point to either a protective or adverse effect of medication for ADHD (whether study allocated or self-selected in community care participants) on the initiation of substance use in MTA participants. However, it should be noted that at the post-treatment follow-up at 22 months the mean age of participants was still relatively young (most were between 11 and 13 years of age).

Factors associated with treatment effects according to data gathered at the end of the MTA interventions

Analysis of the MTA data points to some impacts of socioeconomic status on treatment outcomes at the end of the intervention (Rieppi et al., 2002). For children from better educated families, combination treatment may be more effective than medication management alone for ADHD symptoms whereas for low socioeconomic status families, combination treatment may be more effective for oppositional aggressive symptoms.

Other analysis found that response to treatment in the MTA study did not differ significantly by ethnicity after controlling for public assistance (Arnold et al., 2003). However, at the end of the intervention medication doses reached a higher level for African American children receiving medication management only compared with the average for the group allocated to this intervention. As this was not the case for children from ethnic minorities receiving combination treatment, Arnold and colleagues (2003) suggest that it is possible that the behavioural intervention may have neutralised adverse effects of low socioeconomic status that might otherwise exacerbate symptoms and lead to a need for a higher medication dose. A speculative inference from the analysis is that white middle class children without comorbid anxiety or disruptive behaviour may not gain from adding behavioural treatment to medication, but children of low socioeconomic status, or with comorbid anxiety and disruptive behaviour, especially if from a minority ethnic group, may gain added benefit from combining behavioural treatment with medication.

Another analysis looking at potential moderators of the response to treatment in the MTA study found no moderators of response to behavioural treatment or community care (Owens et al., 2003). However, the analysis indicated that parental depression decreased treatment effectiveness in the medication management group but not in the behavioural treatment group. This finding led Owens and colleagues (2003) to speculate that the parental components of the behavioural intervention may in effect treat the parents to some degree, thus mitigating negative impacts of parental depression on the outcomes for the child. Owens and colleagues (2003) also found that a high initial severity of ADHD symptoms decreased the treatment effects from the medication management and combined treatment interventions, but as the analysis used a measure of response to treatment it is possible that this may reflect the need for those with more severe symptoms at the outset to improve more so as to be classed as responding to treatment.

A further finding unrelated to behavioural treatment reported by Owens and colleagues (2003), was that for those on medication management (that is, participants receiving the medication management or combined treatment interventions), participants in the subgroup with parental depression and a higher severity of symptoms responded better to medication if they had a higher starting IQ compared with those in this subpopulation with lower IQ. In terms of implications for treatment, Owens and colleagues (2003) suggest that their analysis indicates that treatment of parental depression may be important in order to get a positive response to treatment of ADHD using medication. They also speculate that it might be important to intervene early with medication management or combined treatment before ADHD severity increases and a positive response to treatment becomes less likely.

Analysis looking at outcomes at the end of treatment for subgroups with comorbid anxiety and disruptive behaviour (oppositional defiant disorder or conduct disorder) pointed to some impacts on treatment effects (Jensen et al., 2001). All MTA interventions including community care were found to be effective in the subgroup with ADHD and comorbid anxiety. For subgroups with ADHD only or ADHD and disruptive behaviour (oppositional defiant disorder or conduct disorder), medication was favoured – whether alone or in combination with behavioural treatment – but behavioural treatment alone may be contraindicated. For the subgroup with ADHD and both anxiety and disruptive behaviour, there was evidence of an advantage of combined treatment, particularly with respect to overall impairment and functioning.

Earlier analysis looking at the impact of comorbidity suggested that MTA participants with comorbid disruptive behaviours (oppositional defiant disorder and conduct disorder) did not benefit from the addition of behavioural treatment (that is, combined treatment) over medication management alone at the end of treatment time point (bMTA Co-operative Group, 1999b). A further tentative inference from the data gathered at the end of treatment is that the intensive MTA behavioural intervention may have had similar effects to routine medication because the majority (66%) of the community care group received medication for ADHD and the behavioural intervention group did not differ significantly from the community care group for end of treatment outcomes. It must, however, be noted that the absence of a statistical difference between the groups does not prove that there is no difference between the effects of the behavioural intervention and continued community care.

Secondary analysis looking at treatment response found that twice as many children met criteria for successful treatment at the end of treatment time point in the groups receiving medication management (medication management and combined treatment groups) compared with the behavioural intervention and community care groups (Swanson et al., 2001). The authors infer that if medication management was adopted in usual care the number of cases successfully treated would effectively double from 30% to 62%. The analysis also suggests that the addition of intensive psychological treatment in combination with medication management would result in 12% more children being successfully treated as the response rate was 56% in the medication management group compared with 68% in the combined treatment group (equivalent to a 20% increase in the success rate through the addition of intensive psychological treatment).

A further analysis of the MTA study data gathered at the end of treatment time point indicates that the more severe subgroup meeting criteria for hyperkinetic disorder showed a larger decrease in symptoms with medication than with behaviour therapy, and a larger medication advantage than those not meeting criteria for hyperkinetic disorder (Santosh et al., 2005). Accordingly, as they show a greater response to medication than the less severe non-hyperkinetic disorder subgroup, Santosh and colleagues suggest that for those with hyperkinetic disorder medication management is favoured as a first-line treatment. As the response of the non-hyperkinetic disorder subgroup to medication was in the same direction, albeit to a lesser degree, the data also suggest that stimulants may be indicated for some children with ADHD who do not meet criteria for hyperkinetic disorder.

11.5. HEALTH ECONOMICS EVIDENCE

11.5.1. Systematic literature review

The systematic literature search identified two economic studies that compared the cost effectiveness of pharmacological, psychological and combination therapies in children with ADHD (Lord & Paisley, 2000; Zupancic et al., 1998), plus an economic analysis of the interventions assessed in the MTA study (Jensen et al., 2005; Foster et al., 2007). In addition, the economic modelling undertaken to support NICE guidance on the use of methylphenidate, atomoxetine and dexamfetamine in children with ADHD incorporated a sub-analysis that compared combination therapies with the evaluated medications (King et al., 2006). Details on the methods used for the systematic search of the economic literature are described in Chapter 3. The economic analysis of the MTA study is described in a separate sub-section in this chapter. Information on the methods used and the results reported in all economic studies included in the systematic literature review are presented in the form of evidence tables in Appendix 14.

Lord and Paisley (2000) conducted an economic analysis to compare the cost effectiveness of combination therapy, consisting of methylphenidate plus behavioural therapy, with behavioural therapy alone for children with ADHD in the UK. The perspective of the analysis was that of the NHS. The study, based on a decision-analytic model, utilised clinical-effectiveness data from the MTA study. Resource use estimates were based on expert opinion and reflected clinical practice in the UK. Costs consisted of drug acquisition and pharmacotherapist costs. Costs of behavioural therapy were omitted from the analysis, as these were common in the two strategies assessed. The measure of outcome was the SMD in the SNAP-IV score between the two treatment options. The time horizon of the analysis was 14 months, the length of the MTA study. According to the results of the analysis, the ICER of combination therapy versus behavioural therapy alone was £1,596 per SMD in the SNAP-IV score (1999 prices). This ratio ranged in sensitivity analysis from £694 to £4,545 per SMD in the SNAP-IV score. One limitation of the analysis was the use of the change in SNAP-IV scores as the primary outcome measure, which could not capture the HRQoL of children with ADHD. In addition, the study utilised clinical data from the MTA study, which was conducted in the US and examined interventions that were more intensive than typical interventions in the UK. On the other hand, the resource use estimates by Lord and Paisley (2000) referred to UK clinical practice, and therefore the results of the economic analysis should be interpreted with caution.

Zupancic and colleagues (1998) assessed the cost effectiveness of methylphenidate, dexamfetamine, pemoline, psychological/behavioural therapy and combination therapy (consisting of psychological/behavioural therapy and methylphenidate) in comparison with no treatment from the perspective of a third-party payer in Canada. Details on the methodology of the study are reported in Chapter 10. The meta-analysis of clinical studies included in the systematic literature review indicated that psychological/behavioural therapy, either alone or as an adjunct to pharmacological therapy, was not effective. The economic analysis demonstrated that methylphenidate dominated both psychological/behavioural therapy and combination therapy. The limitations of the analysis are described in Chapter 10. Additional limitations specific to the evaluation of psychological/behavioural and combination therapies were: the rather poor quality and the insufficient power of clinical studies assessing these two strategies; the assumptions regarding duration of therapy (daily provision of drugs versus 16-hour provision of psychological/behavioural therapy), which, according to the authors, might have biased the results against psychological/behavioural and combination therapies; and, finally, the choice of the outcome measure, that is, the change in CTRS scores, which might have underestimated the efficacy of psychological/behavioural therapy alone or in combination, given that this therapy has been shown to be more effective in enhancing academic performance and improving conflicted peer relations rather than improving core ADHD symptoms.

The economic analysis of the NICE TA on the use of methylphenidate, atomoxetine and dexamfetamine for children and adolescents with ADHD (NICE, 2006b) incorporated a sub-analysis assessing the cost effectiveness of combination strategies relative to strategies involving only sequences of medications (King et al., 2006). Details on the methodology adopted in the study analysis are provided in Chapter 10. The sub-analysis including combination therapies assessed 37 strategies in total: 18 strategies consisting of all possible three-line sequences of the medications reviewed, 18 respective strategies of three-line sequences of combined treatment, and a strategy of no treatment. After excluding all strategies ruled out by dominance, two options remained: a combination strategy consisting of behavioural therapy plus first-line dexamfetamine, second-line atomoxetine and third-line MR methylphenidate administered every 8 hours, and a medication strategy consisting of first-line dexamfetamine, second-line IR methylphenidate and third-line atomoxetine. The ICER of the first versus the second strategy was £1,241,570/QALY; consequently, the authors concluded that combination strategies were not cost effective from the perspective of the NHS. However, the available clinical data for this analysis were very limited (based on a single trial comparing IR methylphenidate alone versus in combination with behavioural therapy) and no firm conclusions could be drawn.

Overall, the existing evidence reported in Zupancic and colleagues (1998) and King and colleagues (2006) suggests that combination and psychological therapies may not be cost-effective treatment options compared with medication for children with ADHD. However, there were considerable limitations in the clinical-effectiveness data used in the economic analyses, as described above. The study by Lord and Paisley (2000) used resource use estimates representing UK routine clinical practice, and clinical data from the MTA study, which evaluated intensive interventions in the US. Considering also that the primary measure of outcome in the analysis was the SMD of SNAP-IV scores rather than a dichotomous outcome, it is evident that no safe conclusions can be made by this analysis either.

11.5.2. Economic modelling

Objective

The choice of treatment strategy among various types of interventions available to children with ADHD was identified by the GDG and the health economist as an area with potential major resource implications. The existing economic evidence in this field was limited and was characterised by considerable uncertainty; therefore a decision-analytic model was developed for this guideline to examine the relative cost effectiveness of pharmacological, psychological and combination therapies for children with ADHD.

Treatment strategies examined

The treatment strategies examined were medication versus behavioural therapy versus combined therapy (that is, behavioural therapy provided concurrently with medication). Medication was represented by use of methylphenidate in the economic model, for three reasons: methylphenidate was the only drug examined in the clinical trials comparing pharmacological with psychological and/or combined therapies that were included in the guideline systematic literature review; it is the most commonly used medication in clinical practice; finally, indirect clinical evidence suggests that it is likely the most effective drug in improving core symptoms in children with ADHD. Nevertheless, recommendations based on the results of the economic analysis refer to medication as a treatment option, and are not intended to be specific to the use of methylphenidate.

Medication was defined as use of IR methylphenidate at an average daily dose of 25 mg for 4 weeks (titration period), followed by use of MR methylphenidate at an average daily dose of 36 mg. Children taking medication had regular contacts with healthcare professionals (psychiatrists or paediatricians and nurse specialists), with higher intensity during the titration period. Behavioural therapy was defined as ten 1-hour meetings of clinical psychologists with groups of ten parents of children with ADHD. In addition, clinical psychologists provided telephone support to parents when needed, and had two meetings with children’s teachers at school lasting 30 minutes each, in order to provide advice. Combined treatment consisted of both medication and behavioural therapy.

Methods

Model structure An economic model in the form of a decision tree was developed to estimate total costs and benefits associated with provision of medication, behavioural therapy and combined treatment to children with ADHD. According to the model structure, hypothetical cohorts of children with ADHD were started on one of the three treatment options under assessment. If children receiving behavioural therapy or medication did not respond to treatment following completion of 8 weeks of therapy (in accordance with the duration of clinical trials that provided efficacy data), they were switched to medication or behavioural therapy, respectively, or to combined treatment. However, children not responding to combined therapy after 8 weeks were not then offered medication or behavioural therapy alone, as it was assumed that none of the ‘monotherapy’ interventions would be effective following unsuccessful combination therapy. It must be noted that the model assumed that non-response to a treatment option did not affect effectiveness (that is, response rates) of subsequent treatments; this means that response rates of any treatment option a child might have received were independent from each other. Children on medication or combined intervention who stopped treatment because of intolerable side effects were switched to behavioural therapy. Children who switched to combined treatment because of non-responsiveness after 8 weeks of medication were assumed not to experience intolerable side effects from combined treatment, given that they had not experienced intolerable side effects from medication alone. Children completing medication or combination therapy could also experience (tolerable) side effects that did not affect continuation of therapy. Children not responding after two lines of treatment (or one, if they completed 8 weeks of combined treatment and did not respond to it), were assumed to receive ‘other treatment’. This consisted of further management of children with ADHD, including contacts with healthcare professionals, unlicensed medications, inpatient care or no treatment.

The time horizon of the analysis was 1 year. Children responding to any of the treatment options assessed were assumed to continue successful treatment beyond 8 weeks (with 100% compliance) and remain responsive (that is, retain improved symptoms) until the end of the analysis. Children non-responsive to treatment who moved to ‘other treatment’ remained on it until the end of the analysis. It is acknowledged that the time horizon of 1 year is rather limited and does not allow estimation of the overall, long-term costs and benefits associated with treatment of children with ADHD; however, there is no sufficient evidence to allow modelling for longer periods of time, as long-term harms and benefits of the examined interventions have not been adequately explored. A schematic diagram of the decision tree is provided in Figure 5.

Figure 5. Schematic diagram of the structure of the economic model SE = side effects; BT = behavioural therapy.

Figure 5

Schematic diagram of the structure of the economic model SE = side effects; BT = behavioural therapy.

Costs and health benefit measures included in the analysis The analysis adopted the perspective of the NHS. Health service costs consisted of intervention costs, costs of monitoring children who responded to treatment and therefore remained in receipt of any of the treatments assessed for the whole time horizon of the analysis, as well as costs of ‘other treatment’ for children not responding to the treatment options assessed.

Costs of treating side effects were not separately considered in the analysis because the majority of side effects from medication (anorexia, nausea, insomnia, headache, increased irritability, and so on) are routinely managed by healthcare professionals responsible for the monitoring of children receiving medication and were estimated not to incur extra costs.

Costs of personal social services and education services were not included in the analysis owing to lack of relevant data, but it is unlikely that these differ significantly across children receiving different types of treatment over the period of 1 year. Other societal costs, such as social benefit payments and productivity losses of carers of children with ADHD were not considered as they were beyond the scope of this analysis. Benefits were expressed in the form of QALYs. Results are reported as ICERs.

Effectiveness data and other clinical input parameters As already discussed in the economic sections of Chapter 7, for the economic analyses undertaken in this guideline, it was decided to utilise data only from studies reporting outcomes as response rates, with response defined in a way that the GDG found both clinically meaningful and significant.

The guideline systematic review identified four studies that examined pharmacological versus psychological and/or combination therapies for children with ADHD and reported outcomes as response rates (ABIKOFF2004B; GITTELMAN-KLEIN1976B; KLEIN1997B; MTA1999B). GITTELMAN-KLEIN1976B and KLEIN1997B compared methylphenidate with behavioural therapy and combined treatment. ABIKOFF2004A compared methylphenidate with a combination of methylphenidate and a multimodal psychosocial treatment that included parent training and counselling, social skills training, psychotherapy and academic assistance. MTA1999B compared intensive medication management with intensive behavioural therapy and a combination of the two. The definitions of the term ‘response’ established by the studies were considered appropriate by the GDG. For the base-case analysis, it was decided to utilise data from GITTELMAN-KLEIN1976B and KLEIN1997B, as the studies examined the interventions of interest in this economic analysis; data from ABIKOFF2004A were considered in a sensitivity analysis; data from MTA1999B were examined separately, as the study involved interventions of high intensity.

The study population in GITTELMAN-KLEIN1976B and KLEIN1997B consisted of school-age children with pervasive symptoms of ADHD and parent reports for hyperactivity or behavioural problems at home. Both studies defined response as a final score of 1 to 3 on the CGI-I scale (that is, completely well, much improved, or improved). More details on the study characteristics can be found in Appendix 17.

Analysis of efficacy data from the above trials was based on intention-to-treat (that is, response rates were calculated taking into account the number of all children participating in each arm at the start of the trials and not completers only); other important input parameters for the economic model, such as rates of children dropping out of treatment because of intolerable side effects as well as rates of side effects in each treatment arm were not reported in these studies. For this reason, the proportions of children who stopped medication or combined treatment because of intolerable side effects were derived from the guideline meta-analysis of studies comparing methylphenidate with placebo, including comorbid and non-comorbid populations of children with ADHD. The attributable risk of stopping methylphenidate because of intolerable side effects was calculated by subtracting the overall rate of stopping placebo because of side effects from the respective rate for methylphenidate. This attributable risk estimate was applied to children who received medication or combined treatment in the analysis, while children receiving behavioural therapy were assumed not to experience intolerable side effects that would lead to discontinuation of treatment.

Rates of side effects for children under medication or combined therapy were based on the same dataset of studies (that is, placebo-controlled studies of methylphenidate in children with ADHD including comorbid and non-comorbid populations) and were estimated in a similar way. However, the existing studies did not report an overall side-effect rate, but rather provided rates of specific side effects; it was not possible to estimate an overall side-effect rate from these data as some children might have experienced more than one side effect. In order to overcome this problem, it was decided to proxy the overall rate of side effects for methylphenidate using data on the rate of appetite loss; this was selected because it is a common, statistically significant side effect of methylphenidate and also it is deemed to substantially reduce the quality of life of children with ADHD. The attributable risk of side effects for methylphenidate was therefore calculated as the difference between rate of appetite loss for methylphenidate and rate of appetite loss for placebo; this estimate was subsequently applied to children receiving medication or combined treatment; it was assumed that, for the proportion of children experiencing side effects, these persisted for the entirety of the time period when medication or combined treatment was provided. Children receiving behavioural therapy or ‘other treatment’ did not experience side effects from treatment.

Discontinuation of treatment for reasons other than intolerable side effects was not considered in the analysis owing to lack of data appropriate to inform the economic model: GITTELMAN-KLEIN1976B and KLEIN1997B reported very small discontinuation rates that were insignificant; moreover, it was not clearly reported from which arms of the trials children dropped out. The only other available data came from MTA1999B, which referred to intensive interventions, and therefore respective data did not reflect discontinuation of treatment options assessed in this analysis. In addition, such data could only be applied to first-line treatment, as children completing treatment without response, as well as their parents, were thought to demonstrate different attitudes towards second-line treatment, which would not be reflected in discontinuation rates characterising initiation of treatment.

The proportions of children moving to combined treatment following failure of medication or behavioural therapy was based on a trial comparing medication with behavioural therapy in which proportions of children not fully responding to the interventions assessed were subsequently switched to combined treatment (Döpfner et al., 2004).

Estimation of response rate of ‘other treatment’ was based on a published meta-analysis of follow-up studies on children with ADHD; the study reported the annual probability of continuation of residual ADHD symptoms in the population of people with ADHD (Faraone et al., 2006), which was interpreted for the purposes of this analysis as no response. From this annual rate, it was possible to estimate the response rates of children that remained under ‘other treatment’ for varying time periods.

Effectiveness data and other clinical input parameters utilised in the base-case economic analysis are presented in Table 37.

Table 37. Response rates and other clinical input parameters utilised in the base-case economic analysis of pharmacological versus psychological versus combined interventions for children with ADHD.

Table 37

Response rates and other clinical input parameters utilised in the base-case economic analysis of pharmacological versus psychological versus combined interventions for children with ADHD.

Utility data and estimation of QALYs As already discussed in the economic section of Chapter 7, two sets of utility scores were used for the economic analyses undertaken in this guideline involving children with ADHD: base-case analyses utilised the scores reported by Coghill and colleagues (2004), generated from EQ-5D; utility scores provided by Secnik and colleagues (2005b), produced by SG technique using vignettes of health states of children with ADHD in the UK, were used in a sensitivity analysis.

One limitation of using Coghill and colleagues’ (2004) utility scores in the current analysis was that they did not take into account any decrement in quality of life resulting from the presence of side effects. Nevertheless, this was an important parameter to consider in this analysis, since children under medication or combined therapy could experience side effects and a subsequent reduction in HRQoL, in contrast with children under behavioural therapy, who did not experience side effects. For this reason, a decrement in utility resulting from the presence of side effects was estimated from Secnik and colleagues (2005b) and was applied to the base-case utility scores to create additional scores for responders and non-responders experiencing side effects. Regarding the sensitivity analysis that tested the data from Secnik and colleagues (2005b), it was assumed that utility scores reflecting no medication/untreated ADHD expressed utility of children receiving behavioural therapy.

Utility scores used in the economic analysis of pharmacological versus psychological versus combined treatment for children with ADHD are provided in Table 38.

Table 38. Utility scores included in the economic model of pharmacological versus psychological versus combined interventions for children with ADHD.

Table 38

Utility scores included in the economic model of pharmacological versus psychological versus combined interventions for children with ADHD.

It was assumed that HRQoL in children initially responding to treatment improved linearly over 8 weeks starting from the utility score of non-responders and reaching the utility score for responders (8 weeks was the duration of interventions in the clinical trials considered in the economic analysis), and that it remained at this value for the rest of the time of the analysis. Decrement in quality of life owing to the presence of side effects was modelled from initiation of respective treatment. Once side effects occurred, they were assumed to remain over the whole period over which medication or combined therapy was provided. Children who stopped treatment because of intolerable side effects faced a decrement in quality of life for 2 weeks, after which the intolerable therapy was discontinued.

Resource utilisation and cost data

Owing to lack of patient-level cost data, deterministic costing of all treatment options assessed was undertaken. Relevant healthcare resource use was estimated and subsequently combined with unit prices to provide total costs associated with medication, behavioural therapy and combined treatment. Resource utilisation estimates reflected, as closely as possible, resource use described in the clinical studies utilised in the economic analysis (GITTELMAN-KLEIN1976B; KLEIN1997B). Where relevant information on resource use was lacking (for example, resource use beyond the duration of the trials) or was clearly unrepresentative of British routine practice, then estimates were produced/modified based on the expert opinion of the GDG.

For children receiving medication, the GDG estimated the average optimal daily dose of methylphenidate during titration and post-titration, which was, overall, consistent with doses reported in the clinical studies that provided efficacy data. Titration was estimated to last 4 weeks, over which time children received IR methylphenidate. MR methylphenidate was administered post-titration, according to routine clinical practice in the UK. Children were attended by a psychiatrist or a paediatrician during titration. Those responding to medication were assumed to continue receiving methylphenidate until the end of the analysis, being monitored by a psychiatrist, paediatrician or a nurse at regular intervals. Children stopping medication because of intolerable side effects were assumed to receive methylphenidate for 2 weeks before discontinuing and to spend half of the total estimated time (during titration) with a psychiatrist or paediatrician.

Behavioural therapy in GITTELMAN-KLEIN1976B and KLEIN1997B was provided to parents of children with ADHD on a one-to-one basis. However, existing evidence indicated that clinical effectiveness of psychological interventions for children with ADHD did not depend on the mode of delivery and was similar in individual and group-based therapies. Given that the intervention costs of group-based therapies are spread to a number of families, group-based behavioural therapy dominates individually delivered therapy because it produces the same clinical outcome at a lower cost. For this reason, group-based behavioural therapy has been modelled in the base-case analysis; the cost effectiveness of individual behavioural therapy versus medication and combination therapy, indicated under special circumstances, has been explored in a sensitivity analysis.

According to average resource use described in clinical trials of psychological interventions for children and confirmed by GDG expert opinion, behavioural therapy was modelled as ten meetings (lasting 1 hour each) of clinical psychologists with groups of parents of children with ADHD. Every group comprised 10 families. Clinical psychologists were assumed to spend an extra hour for training and preparation. In addition, based on resource use data reported in GITTELMAN-KLEIN1976B and KLEIN1997B, these sessions were augmented by an average of 1 hour of telephone contacts with each family. Clinical psychologists also visited the teachers of the children at school and provided advice; two visits of 30 minutes each were assumed. Following completion of the intervention, parents of children responding to behavioural therapy attended three individual booster sessions with psychologists lasting 30 minutes each, in order to maintain children’s response for the remaining time of the analysis.

Resource use in combined treatment was the sum of resource use of medication and behavioural therapy, given that the two interventions are led by different types of healthcare professionals and no overlap in services provided occurs.

Regarding costs of ‘other treatment’, no data on average annual costs associated with management of children with ADHD in the UK are available. King and colleagues (2006) gave an overall estimate of £14 million spent on follow-up care of children with ADHD by health, social and education services in England and Wales (initial specialist assessment was excluded from these costs). Using this estimate, a prevalence of ADHD equalling 3.62% in boys and 0.85% in girls (Ford et al., 2003), and the population of boys and girls aged 5–18 in 2006 in England and Wales (Office for National Statistics, 2007), it was estimated that a child with diagnosed ADHD incurred on average a cost of £67 annually. This estimate may seem low, but it is likely to reflect the fact that some children with ADHD may not receive any treatment for this condition.

Unit prices were taken from the BNF 55 (Joint Formulary Committee, 2008), and the Unit Costs for Health and Social Care 2006 (Curtis & Netten, 2006); 2006 prices were used. The reported unit costs for clinical psychologists did not include qualification costs, owing to lack of relevant data; it was therefore decided to exclude qualification costs from the unit costs of all healthcare professions included in this analysis, for consistency purposes. Discounting was not applied, as costs and benefits were measured over a period of 1 year.

Resource use estimates and unit costs as well as total costs of interventions assessed over the 1 year of the analysis are reported in Table 39.

Table 39. Cost data utilised in the base-case economic analysis of pharmacological versus psychological versus combined interventions for children with ADHD.

Table 39

Cost data utilised in the base-case economic analysis of pharmacological versus psychological versus combined interventions for children with ADHD.

Sensitivity analysis

A sensitivity analysis was undertaken to investigate the robustness of the results under the uncertainty characterising input parameters of the model. The following scenarios were tested in a one-way sensitivity analysis:

  1. Changes in response rates to treatment:
    • use of the 95% CI of the RR of response rates of combined treatment to behavioural therapy (mean RR = 2.04; 95% CI, 1.46 to 2.86)
    • use of the 95% CI of the RR of response rates of medication to behavioural therapy (mean RR = 1.55; 95% CI, 1.06 to 2.27)
    • inclusion of data from ABIKOFF2004a in the meta-analysis.
  2. Utility scores obtained from Secnik and colleagues (2005b).
  3. Use of one line of treatment only; children responding to treatment remained on it for the rest of the time of the analysis; children not responding were switched to ‘other treatment’.
  4. Changes in resource use estimates for behavioural therapy (or the behavioural therapy component of combined treatment):
    • Group-based CBT, more appropriate for school-age children, provided by clinical psychologists, consisting of ten sessions lasting 1 hour each with parents and ten sessions lasting 1 hour each with children (ten parents and ten children in each group, respectively), including 1 hour of telephone calls per family, plus 2 extra hours for training and preparation and two 0.5 hour individual meetings with the children’s teachers, reflecting effective, optimal routine practice for school-age children (expert opinion of the GDG). In addition, three individual booster sessions, lasting 30 minutes each, were offered to parents of children responding to treatment, in order to maintain children’s response for the remaining time of the analysis. The cost of this intervention was £371 per child.
    • Individual behavioural therapy, consisting of ten weekly sessions with a clinical psychologist, lasting 1 hour each, in cases where group-based programmes are not a suitable option. The remaining components of the intervention (telephone contact with parents, visiting children’s teachers at school and booster sessions for children responding to treatment) were the same as with the group intervention. The total cost of this intervention was £894 per child. This scenario explored the cost effectiveness of individual behavioural therapy under a number of alternative hypotheses, such as: use of the upper and lower 95% CIs of the RRs as described above; inclusion of data from ABIKKOF2004a in the meta-analysis of clinical studies; use of utility scores obtained from Secnik and colleagues (2005b); use of one line of treatment only; and provision of behavioural therapy by health visitors instead of clinical psychologists (at a unit cost of £61 per clinic hour excluding qualification costs, according to Curtis and Netten, 2006).

In addition to the above scenarios, threshold analyses were carried out to identify the values of selected parameters at which the conclusions of the cost-effectiveness analysis would be reversed. The following parameters were tested:

  1. rate of side effects from medication (or combined therapy)
  2. rate of stopping medication (or combined therapy) because of intolerable side effects
  3. decrement in utility scores because of side effects
  4. response to ‘other treatment
  5. cost of ‘other treatment’.

Results

Base-case analysis Combined therapy resulted in greatest health benefits but at the same time it was the most expensive treatment option. Group-based behavioural therapy was the least effective and cheapest option. Medication was dominated by extended dominance. The ICER of combined therapy versus behavioural therapy was £122,682 per QALY. This value is far beyond the cost-effectiveness threshold of £20,000 per QALY set by NICE (NICE, 2006c). This means that, according to base-case results, group-based behavioural therapy is the most cost-effective treatment option among those assessed. Full results of the base-case analysis are presented in Table 40.

Table 40. Cost-effectiveness of pharmacological versus psychological versus combined treatment for children with ADHD – results of the base-case analysis over 1 year.

Table 40

Cost-effectiveness of pharmacological versus psychological versus combined treatment for children with ADHD – results of the base-case analysis over 1 year.

Sensitivity analysis Group-based behavioural therapy remained the most cost-effective option under the vast majority of scenarios tested in the sensitivity analysis. The only scenario that affected conclusions of economic modelling was use of the upper 95% CIs of the RR of response rate of medication to behavioural therapy. In this case the ICER of medication versus behavioural therapy fell at £4,652 per QALY, thus medication became more cost effective than group-based behavioural therapy. In all other scenarios, either the ICERs of combined therapy and/or medication versus group behavioural therapy were very high, beyond the cost-effectiveness threshold of £20,000 per QALY, or group behavioural therapy dominated the two other options.

Individual behaviour therapy was not cost effective compared with medication under any sub-analyses tested. In many scenarios it was dominated by medication (that is, it was less effective and more costly). In none of the scenarios explored was combined treatment found to be cost effective, even when it included group-based behavioural therapy. Results concerning either group or individual psychological therapies were not sensitive to any of the parameters examined in threshold analysis.

Full results of the one-way sensitivity analysis are shown in Table 41 and Table 42.

Table 41. Results of one-way sensitivity analysis for group-based behavioural therapy.

Table 41

Results of one-way sensitivity analysis for group-based behavioural therapy.

Table 42. Results of one-way sensitivity analysis for individual behavioural therapy.

Table 42

Results of one-way sensitivity analysis for individual behavioural therapy.

Limitations of the economic analysis

The results of the economic analysis were based on a simple decision-analytic model developed to estimate total costs and health benefits associated with provision of medication, behavioural therapy or combined treatment over a period of 1 year. Clinical evidence was derived from two trials that reported outcomes in the form of response to treatment. The total number of participants in these two trials was small (N = 125). Nevertheless, further evidence coming from studies reporting outcomes in the form of changes on scales measuring ADHD symptoms that were included in the guideline systematic review and meta-analysis supported clinical evidence utilised in this analysis.

Long-term harms and benefits of the treatment options assessed have not been explored in depth. Identifying potential harms of medication in the long term is likely to reduce its cost effectiveness relative to non-pharmacological interventions and in fact may raise other concerns over its use. Owing to lack of relevant data, the time horizon of the analysis was only 1 year. Despite the short duration of the analysis, a number of assumptions were still required at the development of the economic model. Children were assumed to remain improved, following initial response to treatment, over the rest of the time of the analysis up to 1 year, provided that they continued medication under monitoring if they had responded to medication, or that they attended a number of booster sessions if they had responded to behavioural therapy. In both cases full compliance for all children was assumed, and no deterioration was modelled. Responsiveness to treatment was assumed to be independent of non-responsiveness to previous treatment provided. In reality, lack of response to one type of treatment could be related to improved or, conversely, reduced responsiveness to another type of treatment. Acceptability of the treatment to children and their carers reflected in overall continuation rates associated with pharmacological or psychological interventions for ADHD, was not considered, owing to lack of relevant data. However, this is an important factor that may significantly affect the relative cost effectiveness of an intervention.

Estimated costs consisted of intervention costs only; potential cost savings to the healthcare, social and education services resulting from improvement in ADHD symptoms of children were not considered owing to lack of evidence. It is therefore likely that the relative cost effectiveness of the interventions assessed for children with ADHD is different from that suggested by the results of the analysis. It is expected that including potential cost savings would alter the cost-effectiveness results in favour of more effective interventions (that is, mainly combined therapy and, at a lower degree, medication).

Estimates of healthcare resource use reflected, as closely as possible, resource use described in the clinical studies utilised in the analysis; these estimates were consistent with optimal resource use in the UK, according to GDG expert opinion. Nevertheless, the clinical studies described only vaguely some aspects of resource use, and obviously they did not provide any relevant data for resource use beyond the duration of the trials (that is, beyond 8 weeks of treatment). It is unknown whether the number of booster sessions modelled for families receiving psychological interventions or the frequency and type of monitoring assumed for children taking medication are adequate to retain a positive outcome over a year, and this is a further limitation of the analysis.

Utility scores used in the base-case analysis were based on EQ-5D questionnaires completed by parents of children with ADHD in England (Coghill et al., 2004). EQ-5D is a generic measure of HRQoL and as such, it has been recommended by NICE for use in economic evaluation. However, the full methods used to convert EQ-5D scores into utility scores were not reported in the study. In addition, the GDG expressed concerns about the appropriateness of using a generic measure to capture aspects of quality of life in children with ADHD. For this reason, utility scores developed using vignettes describing health states specific to ADHD (Secnik et al., 2005b) were used in the sensitivity analysis. Utility scores used both in the base-case and sensitivity analysis were generated using parents of children with ADHD as proxy reporters of their children’s perceptions of their own HRQoL. There are concerns about using parents’ ratings as proxies to children’s experience; still, for some groups of children who are unable to reliably report their own perceptions and preferences, parent proxies may be appropriate (Wallander et al., 2001; de Civita et al., 2005). In the area of ADHD, no data on HRQoL preferences directly reported by children, rather than by their parents, are currently available.

Behavioural therapy was assumed to be delivered in groups of parents in base-case analysis, despite the fact that both GITTELMAN-KLEIN1976B and KLEIN1997B, who provided the efficacy data for the analysis, examined individually delivered behavioural therapy. Although equivalence in efficacy between group-based and individually delivered programmes has not been established in head-to-head comparisons, existing indirect clinical evidence suggests that the mode of delivery does not affect the clinical effectiveness of psychological therapies for children with ADHD. Analysis of efficacy data was based on intention-to-treat. This means that estimated clinical effectiveness took into account the fact that some children/families might drop out of treatment. On the other hand, full intervention costs were estimated, assuming that all children completed treatment (with the exception of those children stopping treatment because of side effects, who switched to another therapy). This assumption has overestimated total costs of interventions to the disadvantage of strategies that are characterised by higher drop-out rates (and therefore lower overall costs).

11.5.3. Overall conclusions from the economic analysis

The results of the economic analysis indicate that group-based behavioural therapy is more cost effective than medication and combined therapy for children with ADHD. On the other hand, medication is more cost effective than individual behaviour therapy. Combination therapy was not cost effective under any scenario explored in the analysis.

The above conclusions are subject to a number of limitations, as already discussed. Further research is needed to explore fully the long-term harms and benefits associated with the treatment options assessed, as well as to investigate in depth the perceptions of children and their carers on aspects of HRQoL associated with ADHD. Moreover, future head-to-head comparisons need to confirm the equivalence in efficacy between group-based and individually delivered behavioural therapy, so that the cost effectiveness of group-based behavioural therapy versus medication can be determined with higher certainty.

11.5.4. Economic analysis alongside the MTA study

Two studies (Jensen et al., 2005; Foster et al., 2007) assessed the cost effectiveness of the interventions examined in the MTA study (aMTA Co-operative Group 1999a, 2004a, 2007) from the perspective of a third-party payer in the US. The interventions assessed in the study were medication management, intensive behavioural treatment, combination therapy and routine community care. The economic analysis of the MTA study is discussed separately from the rest of the economic literature, because it refers to intensive interventions, which are likely to differ from pharmacological, psychological and combination therapies routinely available in the UK for children with ADHD in terms of both effectiveness and associated resource use. Details on the methods adopted in the studies, their overall limitations, and results involving the comparison between medication management and routine community care are provided in Chapter 10. Characteristics and results of the studies are summarised in the form of evidence tables in Appendix 14.

According to Jensen and colleagues (2005), intensive behavioural treatment was dominated by medication management in all sub-groups of children with/without coexisting conditions examined, as well as in the total study population. Consequently it was clearly not a cost-effective option. Combined treatment was more effective than medication management in the majority of the sub-groups examined; however, the ICER of combined treatment versus medication management was rather high, ranging from US$29,840 (ADHD plus both internalising coexisting conditions, that is, anxiety and depression, and externalising coexisting conditions, that is, conduct and oppositional defiant disorders) to US$74,560 (ADHD plus externalising disorder) per normalised child, with normalisation determined by scores on the SNAP scale. For children with ADHD plus internalising disorder, medication management was more effective and cheaper than combined treatment (dominant option). The ICER of combined treatment versus medication management for the total population of children with ADHD combined was US$55,253 per child normalised (all costs expressed in 2000 prices). Based on the findings of the analysis, the authors concluded that medication management, although not as effective as combined treatment, was likely to be the most cost-effective option for children with ADHD, in particular for those without coexisting conditions. For children with ADHD and both internalising and externalising disorders they suggested that combined treatment might be relatively cost effective. Besides cost effectiveness, the authors highlighted the need to consider additional factors when making decisions on the appropriate treatment for children with ADHD, such as the presence of side effects of medication, the comfort and satisfaction of families with the treatment approach, and the family’s overall feelings about the causes of ADHD.

Foster and colleagues (2007) demonstrated that, for the total population of children with ADHD, medication management was the most cost effective among the four interventions assessed at lower WTP for functioning improvement (from 0 to around US$55,000 per CIS ES). At higher levels of WTP, combined treatment became the most cost-effective strategy. These findings applied also to the population of children with ADHD and externalising disorder. For children with pure ADHD, medication management appeared to be cost effective at all levels of WTP. In contrast, for children with ADHD and internalising coexisting conditions, intensive behavioural treatment might be cost effective at high levels of WTP, while medication management appeared to be cost effective at low levels. Finally, in children with ADHD plus both internalising and externalising coexisting conditions, medication management was clearly cost effective at lower levels of WTP. At higher levels, the probabilities of medication management, intensive behavioural treatment and combined treatment being cost effective were similar and no clearly cost-effective option could be identified. Based on the results of their analysis, the authors stated that, for pure ADHD, medication management was certainly the most cost-effective option at all levels of WTP. In contrast, for coexisting conditions, WTP was crucial in determining the cost-effective treatment option: for lower WTP, medication management was the most cost-effective intervention; but for policy makers willing to pay more to avert future costs such as special education and juvenile justice costs, intensive behavioural treatment alone or combined with medication management (depending on the comorbidity) was likely to be the most cost-effective treatment.

As described previously, Schlander and colleagues (2006a, 2006b, 2006c) evaluated the relative cost effectiveness of the interventions examined in the MTA study in the context of four European countries, utilising the effectiveness data and resource use estimates reported in the MTA study, but applying country-specific unit costs. One of the analyses referred to the UK setting. The analysis adopted the perspective of the NHS (direct medical expenditures). Costs were calculated in UK£ and then converted to 2005 Euros (€). In addition to previous sub-groups identified, the authors provided results for children with ADHD combined type (according to DSM-IV), hyperkinetic/conduct disorder (according to ICD-10), pure ADHD (without coexisting conditions), and pure hyperkinetic disorder (without coexisting conditions). The measures of outcome used in the economic analyses were the number of children with ADHD normalised, the CIS ES and also the QALYs gained by treatment.

In most sub-populations of children, intensive behavioural treatment was dominated by medication management. The two exceptions were the sub-groups of children with internalising coexisting conditions and children with both internalising and externalising coexisting conditions, when the outcome was measured as CIS ES. In these cases, intensive behavioural therapy was shown to be more effective than medication at an incremental cost of €13,030 and €113,540 per CIS ES, respectively (£8,990 and £78,300, respectively, at a conversion rate of 1UK£ = 1.45€). Combined treatment achieved higher proportions of children normalised compared with medication management in all sub-groups of children examined. The ICER of combined treatment versus medication management per normalised child in the UK reached €66,150 for ADHD combined type, €57,600 for pure ADHD, €37,320 for hyperkinetic/conduct disorder, and €26,460 for pure hyperkinetic disorder (or £45,620, £39,720, €25,740, and £18,250 respectively, at a conversion rate of 1UK£ = 1.45€). When the measure of outcome was the CIS ES, then combined treatment was less effective than medication management in children with pure ADHD, children with pure hyperkinetic disorder and children with hyperkinetic/conduct disorder. In all these cases medication management dominated combined treatment. Medication management was dominant over combined treatment also in children with internalising coexisting conditions. The ICER of combined treatment versus medication management in the total population of children with ADHD was as high as €705,115 per CIS ES.

CEACs demonstrated that, for the majority of sub-populations examined, medication management had the highest probability of being cost effective among the treatment options compared, at least for low levels of WTP. When the WTP rose up to roughly €40,000, €60,000, and €80,000 per child normalised, then combined treatment appeared to be the most cost-effective option for children with both internalising and externalising disorders, the total population of children with ADHD, and children with externalising coexisting conditions, respectively. For children with internalising coexisting conditions, medication management was the most cost-effective treatment at any level of WTP per child normalised. Regarding functional improvement, medication management was also shown to be the most cost-effective option at lower levels of WTP. However, in children with internalising coexisting conditions intensive behavioural treatment was the most likely cost-effective option at levels of WTP of around € 15,000 per CIS ES and above.

Schlander and colleagues (2006a) did not provide ICERs expressing cost per QALY gained specific to the UK context. Instead, they reported ranges of such ICERs for the four European settings examined in the analysis. However, it was possible to estimate such ratios for the various sub-populations of children with ADHD, using the reported costs per child treated in the UK context, the proportions of children normalised in the MTA study and utility weights reported in Coghill and colleagues (2004). QALYs were estimated assuming that improvement in HRQoL occurred at time zero for responders. Decrement in HRQoL from medication was not considered in these estimates. Since intensive behavioural therapy was dominated by medication management when the measure of outcome was the proportions of children normalised, the appropriate comparison (apart from the comparison between medication management and routine community care, which has been reported in Chapter 10) was between combined treatment and medication management. The estimated ICERs from this comparison were £612,530 per QALY for ADHD combined type, £543,960 per QALY for pure ADHD, £351,780 per QALY for hyperkinetic/conduct disorder, and £248,060 per QALY for pure hyperkinetic disorder.

The above results indicate that intensive behavioural therapy and combined treatment are highly unlikely to be cost effective for children with ADHD from the perspective of the NHS, given also the cost-effectiveness threshold of £20,000 per QALY set by NICE (NICE, 2006c). Although these results refer to intensive interventions, they lead to the same conclusions as those reported in other published studies about the cost effectiveness of behavioural and combined therapies, and the results of the economic model described in the previous section in this chapter. Medication management was the most cost-effective option compared with intensive behavioural therapy and combined treatment, at least for modest levels of WTP. As reported in Chapter 10, routine community care reflecting US clinical practice might be more cost effective than medication management. However, no safe conclusions can be made, as routine clinical practice in the US may vary significantly from respective practice in the UK, and therefore the results of the analysis (which were based on US resource estimates) might not be representative of the UK healthcare setting.

11.6. FROM EVIDENCE TO RECOMMENDATIONS: TREATMENT DECISIONS AND COMBINED TREATMENT FOR CHILDREN WITH ADHD

Evidence from studies that have compared the effectiveness of stimulant medication for ADHD against the effectiveness of the use of psychological therapies for ADHD without concurrent administration of stimulant medication may help to inform the choice of first-line treatment for ADHD. Clear evidence strongly favouring one approach or another might point to an unequivocal recommendation as to which approach should always be used first, with alternatives being employed only where children do not respond to the first-line treatment.

While there is no evidence that psychological interventions are favoured for any outcome, or at any time point, it is also the case that stimulant medication for ADHD is not strongly favoured over psychological interventions, with the benefits of medication being weakest in comparison with complex psychological interventions. It also remains unclear whether the beneficial effects of stimulant medication over psychological interventions are sustained after the end of treatment. Accordingly the decision about whether to use a psychological intervention or stimulant medication for ADHD appears to be more balanced. In this context the choice of first-line intervention might be influenced by factors other than effectiveness, including possible adverse effects of medication and preferences of the child and/or parent.

Economic evidence suggests that group-based psychological interventions are likely to be more cost effective than medication (the evidential grounds for concluding that group-based psychological interventions are beneficial for children with ADHD are outlined in Chapter 7 at 7.2.14). In contrast, individually delivered psychological therapies are not cost effective compared with medication. Combined treatment is most likely not cost effective regardless of the mode of delivery of its psychological treatment component. It must be noted that because of lack of data on the long-term benefits and harms of interventions assessed, safe conclusions on the relative cost effectiveness between medication and psychological interventions in the long run cannot be drawn. Existing economic evidence indicates that intensive behavioural therapy alone or in combination with medication management is unlikely to be cost effective for children with ADHD.

11.7. RECOMMENDATION

11.7.1.1.

Drug treatment is not indicated as the first-line treatment for all school-age children and young people with ADHD. It should be reserved for those with severe symptoms and impairment or for those with moderate levels of impairment who have refused non-drug interventions, or whose symptoms have not responded sufficiently to parent-training/education programmes or group psychological treatment35.

Here and elsewhere in the guideline, each study considered for review is referred to by a study ID in capital letters (primary author and date of study publication, except where a study is in press or only submitted for publication, then a date is not used).

Here and elsewhere in the guideline, each study considered for review is referred to by a study ID in capital letters (primary author and date of study publication, except where a study is in press or only submitted for publication, then a date is not used).

This recommendation is also included as 10.18.2.1 in Chapter 10.

Footnotes

33

Here and elsewhere in the guideline, each study considered for review is referred to by a study ID in capital letters (primary author and date of study publication, except where a study is in press or only submitted for publication, then a date is not used).

34

Here and elsewhere in the guideline, each study considered for review is referred to by a study ID in capital letters (primary author and date of study publication, except where a study is in press or only submitted for publication, then a date is not used).

35

This recommendation is also included as 10.18.2.1 in Chapter 10.

Copyright © 2009, The British Psychological Society & The Royal College of Psychiatrists.

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Enquiries in this regard should be directed to the British Psychological Society.

Cover of Attention Deficit Hyperactivity Disorder
Attention Deficit Hyperactivity Disorder: Diagnosis and Management of ADHD in Children, Young People and Adults.
NICE Clinical Guidelines, No. 72.
National Collaborating Centre for Mental Health (UK).
Leicester (UK): British Psychological Society (UK); 2009.

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...