NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Peterson BS, Trampush J, Maglione M, et al. ADHD Diagnosis and Treatment in Children and Adolescents [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2024 Mar. (Comparative Effectiveness Review, No. 267.)
We identified a large body of evidence contributing to the knowledge base on attention deficit hyperactivity disorder (ADHD) diagnostic tools, treatment outcomes, and monitoring strategies. We included studies dating back to 1980, marking the advent of modern diagnostic criteria for ADHD and the introduction of long-acting forms of stimulant medication. The questions addressed in our review were informed by Key Informants and supported by a Technical Expert Panel. A dedicated systematic review team with content experts conducted a detailed synthesis of existing research, including over 500 studies in this systematic review.
Despite the large number of publications included, our review has limitations in its scope due, in part, to decisions about which studies to include in the review. For example, we required intervention studies to treat participants for at least four weeks to ensure that the studies assessed sustained, and not merely temporary, effects on outcomes. This decision excluded some early studies of ADHD treatment that have contributed to the development of the field. We also required studies to be either large or to report a power analysis to ensure that they were sufficiently powered to detect effects. This criterion ensured the reader would not be left guessing whether a study was either underpowered to show effects or genuinely showed the absence of evidence of an effect. This criterion, however, also excluded studies that have contributed historically to the evidence base. We furthermore limited treatment studies to youth with a clinical diagnosis of ADHD, which excluded studies that evaluated interventions in broader populations. Finally, we restricted publications to the English language, which may have excluded other important studies that have contributed to the evidence base.
Findings in Relation to the Decisional Dilemma(s)
The following text discusses findings in the context of the decisional dilemmas the review set out to address.
Diagnostic Approaches for ADHD
Studies of diagnostic approaches most commonly report sensitivity (true positive rate) and specificity (true negative rate) for a given diagnostic threshold applied to the measure being assessed. Sensitivity and specificity, however, depend on the diagnostic threshold selected, and their values are inherently a trade-off, such that varying the diagnostic threshold to increase either sensitivity or specificity reduces the other. Interpreting diagnostic performance in terms of sensitivity and specificity is therefore difficult. Investigators instead often report performance for sensitivity and specificity in terms of Receiver Operating Characteristics (ROC) curves because the area under the curve (AUC) provides an overall, single index of performance that does not depend on the diagnostic threshold for the tool being assessed. AUC values range from 0.5 (corresponding to the y=x diagonal of the ROC curve, and indicating that the tool provides no information above chance for classification) to 1.0 (corresponding to the x=0 vertical line, which indicates that the test can correctly classify all participants as having ADHD, and all non-ADHD participants as not having it – a perfect test). AUC values are commonly interpreted as follows: 90 to 100 represents excellent performance; 80 to 90 good; 70 to 80 fair; 60 to 70 poor; and 50 to 60 indicates failed performance. Our assessment of performance of the various tools was specifically for clinical diagnosis compared with a diagnosis made by expert mental health clinicians, distinct from any other clinical utility the tools may have.
Many diagnostic studies in this review aimed to distinguish ADHD youth from neurotypical controls, which is of limited clinical relevance: in clinically referred youth, most parents, teachers, and clinicians are reasonably confident that something is wrong, but they are unsure whether the cause of their concern is ADHD. The more clinically relevant and difficult question, therefore, is how well the measures distinguish ADHD youth from youth who have other emotional and behavioral problems. Moreover, studies that simply discriminate ADHD youth from neurotypical controls cannot discern whether diagnostic performance is determined by the presence of ADHD or by the presence of any other characteristics that accompany clinical “caseness,” such as the presence of comorbid illnesses or effects of chronic stress or current or past treatment.
Overall, AUCs for parent rating scales ranged widely from poor339 to excellent,632 with a low strength of evidence due to imprecision and study limitations. Analyses restricted to the Child Behavioral Checklist (CBCL) (the most commonly evaluated scale) yielded more consistent good AUCs differentiating youth with ADHD from others in clinical samples, but the number of studies contributing data was small. One study reported moderate rater agreement between mothers and fathers for inattention, hyperactivity, and impulsivity. Internal consistency for rating scale items was generally high across most rating scales. Reported test-retest reliability was substantial, but only two studies reported on the measure.
AUCs for teacher rating scales ranged from failed performance (distinguishing ADHD from either neurotypical controls or other patients491) to good (distinguishing ADHD from either neurotypical controls or clinic patients359) with a low strength of evidence, primarily due to imprecision. The internal consistency for scale items was generally high. Teacher ratings demonstrated very low rater agreement with the corresponding parent rating scales, suggesting either a problem with the instruments or a large variability in symptom presentation that depended on environmental context (home or school).
Clinicians likely need ratings from both parents and teachers to yield a more complete representation of symptom expression across informants or settings. We found only two studies, however, that formally combined ratings from parents and teachers to diagnose ADHD, with one study reporting limited specificity when using the Conners to distinguish ADHD from other clinically referred youth,18 and a machine learning study reporting excellent diagnostic accuracy when using the Behavior Rating Inventory of Executive Function (BRIEF) to distinguish ADHD youth from typically developing controls.467
Though data are limited, self-reports from youth seem to perform less well than corresponding parent and teacher reports, with AUCs ranging from failed for Child Behavioral Checklist/Achenbach System of Empirically Based Assessment (CBCL/ASEBA) distinguishing ADHD from other patients491 to good for the Strengths and Weaknesses of ADHD Symptoms and Normal Behavior Scale (SWAN) distinguishing ADHD from neurotypical controls.168, 297
Studies employing combined approaches, such as integrating diagnostic aids with clinician impressions, were limited. One study reported increased sensitivity and specificity when an initial clinician diagnosis was combined with an electroencephalogram (EEG) biomarker for that patient (the reference standard was a consensus diagnosis from a panel of ADHD experts).27 These findings were not independently replicated, and no test-retest reliability was reported.
AUCs for all blood biomarkers ranged from 0.68 (serum miRNAs)635 to excellent (erythropoietin and erythropoietin receptors levels)309 in differentiating ADHD from neurotypical youth, but with a low strength of evidence. None have been independently replicated, and no test-retest reliability was reported.
Diagnostic Accuracy for Youth Younger Than 7 Years of Age
We found only a small number of studies in youth younger than seven year of age. Only three of the studies assessed the performance of rating scales: the CBCL ADHD Problems Scale to distinguish ADHD (co-occurring with a disruptive behavior disorder) from a disruptive behavior disorder alone with “good” AUC;167 or the total score for the Disruptive Behavior Diagnostic Observation Schedule to distinguish ADHD (with or without a comorbid disruptive behavior disorder) from typically developing youth also with “good” AUC;167 or the BRIEF to distinguish ADHD from typically developing controls (average diagnostic accuracy was excellent). The other studies assessed imaging or EEG measures, with AUCs ranging from fair to excellent. The findings provide very little evidence for the utility of any diagnostic approach in youth younger than age seven, though the two studies of rating scales suggest that performance may be comparable to performance of similar scales in youth older than seven.
Comparative Diagnostic Accuracy of EEG, Imaging, or Executive Function Measures for Youth Aged 7 Through 17
Most studies used machine learning for classification based on EEG measures. AUCs ranged from poor197 to excellent in differentiating ADHD youth from neurotypical controls.412 Strength of evidence is low due to large variations in diagnostic performance across studies, and often the methods for classification were not well described. The intraclass correlation coefficient (ICC) for the Theta/Beta ratio, based on repeated measures on two different visits, was 0.83.27
In the neuroimaging studies, AUCs ranged from “poor” for distinguishing ADHD youth without co-occurring disorders from healthy controls1188 to “excellent” for distinguishing ADHD youth from healthy controls.581 Most studies relied on machine learning to develop the diagnostic algorithms, and none assessed test-retest reliability or the independent reproducibility of findings.
Many machine learning studies have been reported to date. Machine learning has usually been applied retrospectively to pre-existing datasets or repositories. AUCs generally were not reported for machine learning studies. Using EEG data, sensitivity ranged from 80 percent (with equal specificity)179 to 98 percent (also with high corresponding specificity).157, 172 Using MRI data, sensitivity showed a wider range from 61 percent (with a corresponding specificity of 68%)1188 to almost perfect sensitivity and specificity.581 Most studies attempted to discriminate ADHD youth from healthy controls retrospectively in pre-existing datasets, not from other clinical populations and not prospectively. In addition, reporting of final mathematical models or algorithms differentiating the diagnostic groups was limited. The overall strength of evidence is low.
Most of the EEG and imaging studies have employed leave-one-out cross validation and have rarely assessed performance in independent samples not contributing to generation of the diagnostic algorithm -- a serious overall weakness. No independent replication studies using the same marker/measure have been conducted, and very few have assessed test-retest or inter-rater reliability. No clinical effectiveness studies have been performed using these measures or diagnostic algorithms in the real world. Thus, biomarker, EEG, imaging, and machine learning algorithms do not seem remotely close to being ready for clinical application.
Studies evaluating neuropsychological tests yielded AUCs ranging from poor24, 263 to excellent140 in differentiating ADHD youth from both neurotypical controls and other patients, with a low strength of evidence. Many studies used idiosyncratic combinations of cognitive measures, including various measures from continuous performance tests (e.g., errors of omission, errors of commission, response time, response time variability, and detectability) to differentiate ADHD from control participants. These idiosyncratic combinations make the results of meta-analyses difficult to interpret. Extracting specific, comparable measures of inattention and impulsivity from continuous performance tests yielded diagnostic performance ranging from poor to excellent in differentiating ADHD youth from neurotypical controls, and fair in differentiating from other patients.21, 24, 162 Only one diagnostic study assessed test-retest reliability, which was poor. No studies provided an independent replication of diagnosis using the same measure. Strength of evidence for continuous performance tasks (CPT) measures was low; thus, despite the widespread use of neuropsychological testing in the evaluation of youth suspected as having ADHD, often at considerable expense, indirect comparisons of AUCs suggest that the performance of neuropsychological test measures in the diagnosis of ADHD is comparable to the diagnostic performance of ADHD rating scales from a single informant, and the overall strength of evidence for estimates of that diagnostic performance is low. Moreover, in head-to-head comparisons, the diagnostic accuracy of parent rating scales is typically better than neuropsychological test measures.467, 732
Variation in Diagnostic Accuracy by Clinical Setting or Patient Subgroup
We did not identify studies that directly compared diagnostic accuracy in head-to-head comparisons across different clinical settings. Instead, we had to compare performance indirectly, across studies. In addition, the reporting of diagnostic accuracy data was limited, and therefore analyses had to be performed on estimates as reported by the original authors, precluding meta-analytic modeling. Indirect comparisons nevertheless indicated that the setting is an effect modifier for diagnostic performance. The range of reported diagnostic sensitivities was narrower in non-clinical samples, indicating that the detection of true positive cases was more consistent across studies in the community when compared to clinical settings, perhaps because ADHD youth identified in community samples are much less complex and less heterogeneous in their presentations than those presenting in clinical settings. We also found that specificity (the rate of identifying true negatives) was significantly lower when diagnosing ADHD youth in community settings compared with clinical settings. A lower true negative rate indicated that youth in the community who did not have ADHD were mistakenly diagnosed as having ADHD, perhaps because they had symptoms that were confused with those of ADHD. We also found that diagnostic specificity was significantly lower when differentiating ADHD youth from other patients than from neurotypical controls, likely because patients with other clinical problems have symptoms that overlap those of ADHD. Thus, the diagnostic group being differentiated from ADHD – whether it is neurotypical “healthy” controls, or youth who have a different emotional/behavioral/psychiatric disorder – and the setting in which the diagnostic tool is being applied has a critical role in diagnostic performance. We also found some indication that diagnostic performance was better for youth who were older compared with younger than 7 years of age (Figure 14), but effects were not statistically significant. Hence, we analyzed studies of mixed samples together and reported on the diagnostic performance by diagnostic test modality, rather than by age group.
Adverse Effects of Being Labeled Correctly or Incorrectly as Having ADHD
We did not identify any study that addressed the consequence of correctly or incorrectly receiving a diagnosis of ADHD.
Safety and Effectiveness of Pharmacologic and Nonpharmacologic Treatments
Analyses that included studies of all therapeutic interventions, regardless of treatment modality, provided strong evidence for the significant efficacy of treatments in improving ADHD outcomes. We conducted extensive analyses to understand which classes of interventions produced significant therapeutic responses in various clinical outcome domains. We can compare the magnitude of those therapeutic responses (effect sizes) across interventions, as well as within and across outcome measures, using the standardized mean difference (SMD) for the active compared with control intervention. SMD values of 0.2 to 0.5 are considered small, 0.5 to 0.8 medium, and above 0.8 are large. We will use the descriptive terms in summarizing the magnitude of treatment responses here, but the precise numerical values can be found in the Results section.
We note that many of the studies for psychosocial interventions, parent support, neurofeedback, and nutritional and supplement therapies, compared the active intervention against either wait list controls, treatment as usual, or another passive intervention group, and therefore they did not adequately control for the effects of parent or therapist attention and other non-specific effects of therapy. In addition, many of these studies were unable to blind either the youth undergoing treatment, their patents, the treating clinician, or study assessors to treatment assignment and study hypotheses,1195, 1196 predisposing assessment of outcomes to reporter bias, particularly as parents and teachers often have an allegiance to non-medication interventions.1197, 1198 These limitations in study design considerably undermines the strength of evidence for psychosocial, parent, neurofeedback, and nutritional interventions.
With these caveats noted, numerous classes of intervention yielded significant effects on measures of ADHD symptom severity. These included: Food and Drug Administration (FDA)-approved medications collectively; stimulant medications collectively, and methylphenidate and amphetamines separately; nonstimulant medications collectively, and norepinephrine reuptake inhibitors (NRIs) and alpha agonists separately; psychosocial treatment collectively; neurofeedback; nutrition or supplements; and parent support. All had small to medium effect sizes, except stimulants, which had large effect sizes, especially amphetamines, which also had highly variable effect sizes. Augmentation of ongoing stimulant treatment with non-stimulant medication (alpha agonists) yielded small but statistically significant improvements in ADHD symptoms compared to augmentation with placebo. Half the neurofeedback studies were at high risk of bias, and when those studies were excluded, effects on ADHD symptoms were no longer significant. Seven omega 3 studies, a subset of nutritional supplements, did not yield significant effects on ADHD symptoms or other outcomes. A newer stimulant medication (not approved for ADHD treatment), modafinil, produced significant improvement in ADHD symptoms in each of four studies, though in aggregate the medium effect size for improvement was not statistically significant due to effect size heterogeneity. The strength of evidence for effects on ADHD symptoms is high for FDA-approved medications collectively and for stimulant and non-stimulant medications separately; strength of evidence is moderate for psychosocial interventions, and low for neurofeedback, parent support, the group of nutritional interventions, and non-stimulant augmentation of ongoing stimulant therapy.
For broadband measures, FDA-approved medications and stimulants collectively yielded significant, medium-sized effects, with comparable effects for amphetamine and methylphenidate derivatives, though amphetamines yielded much more variable effects across studies. Only one stimulant study included children younger than six years of age.109 Non-stimulants collectively, and NRIs and alpha agonists separately, also improved broadband scale scores, with a moderate effect size. Only one non-stimulant study included children younger than six years old.378 Parent support had significant small effects across a small number of studies and low strength of evidence, and cognitive training had medium effects across an equally small number of studies with low strength of evidence. For disruptive behaviors, significant improvement was observed with FDA-approved medications and parent support, with moderate effect size, and with cognitive training and nutrition or supplements, both with small effect sizes and low strength of evidence. For functional impairments, only FDA-approved medications, as well as stimulant and non-stimulant medications collectively yielded significant improvement, with effect sizes that were medium, large, and small, respectively. No treatment modality yielded significant meta-analytic effects on academic performance, though only nine studies (3 psychological, 1 stimulant, 1 combined psychological plus stimulant, and 4 school interventions) assessed this as a treatment outcome. One study assessed the effectiveness of an FDA-approved medication in improving academic performance and found large, significant, and positive effects; all other individual studies yielded nonsignificant improvements of small effect size. We found only one neuromodulation study (direct current stimulation), a small number of studies assessing the effects of exercise, or the effects of complementary and alternative medicines that met our inclusion criteria; none yielded significant improvement in any ADHD outcome domains. Thus, the large number of studies combined with their medium-to-large effect sizes allow us to conclude with a high strength of evidence that FDA-approved medications collectively improve ADHD clinical outcomes in all domains we assessed – in ADHD symptom severity, broadband measures, disruptive problem behaviors, and functional impairment.
Medication therapies were associated with adverse events, including appetite suppression, with a high strength of evidence. Stimulants were associated with an increased number of participants reporting adverse events compared with placebo, with a similar but nonsignificant effect of methylphenidate and a similar though significant effect of amphetamines. Stimulants were associated with appetite suppression compared to placebo, with somewhat smaller effects for methylphenidate than for amphetamines. Non-stimulants compared with placebo were associated with an increased number of participants reporting adverse events, with comparable rates in NRI studies and alpha agonists. Non-stimulants were also associated with suppressed appetite compared to placebo, with significant appetite suppression from NRIs but weaker and non-significant effects from alpha agonists. Studies of non-pharmacological therapies rarely reported the systematic assessment of adverse effects.
The most common head-to-head comparison between two alternative medication treatment types was atomoxetine versus different methylphenidate medications, which did not detect significant differences in effects on ADHD symptoms, broadband measures, behavioral problems, functional impairment, appetite suppression, or the number of patients experiencing adverse events, though the direction of effects consistently favored methylphenidate medications. Indirect comparison of studies evaluating stimulants and non-stimulants compared to control groups showed larger reported effect sizes for stimulants providing greater improvement for ADHD symptoms and broadband measures while functional impairment, appetite suppression, and the number of participants reporting adverse events were comparable. We identified only one head-to-head comparisons of NRIs versus alpha agonists that met eligibility criteria. It reported significantly greater improvement in ADHD symptoms from the alpha agonist guanfacine over the NRI atomoxetine, with a small effect size, though indirect comparisons did not find a significant difference between alpha agonists and NRIs in their effects on any outcome domains.
We found little evidence that youth-directed psychosocial and medication interventions are better in improving ADHD outcomes when delivered in combination than as monotherapies. Most of these studies, however, compared the effects of combination therapy against the effects of medication alone, which is a high bar to surpass. Combination compared with monotherapy yielded an improvement in ADHD symptoms with a small effect size at a trend-level of statistical significance, but no evidence for improvement in other outcome domains. Furthermore, our findings suggest that combined medication and youth-directed psychosocial therapies do not improve ADHD symptoms better than either medication or behavioral therapy alone. We note, however, that few combinations have been evaluated and these analyses do not consider the possibility that exact sequencing of psychological and medication therapies may produce differential effects on outcomes.204, 471
Very few studies have evaluated the long-term effectiveness of any treatment modality for any ADHD outcome domains. For example, only one study of an FDA-approved medication (atomoxetine)164 that met our inclusion criteria evaluated effects on long-term outcomes. It found significant improvement in broadband measures, with a very large effect size, but no effects on ADHD symptoms or functional impairment, and significantly more adverse events and less weight gain, compared with placebo. Two studies of psychosocial interventions (behavioral therapy and attention training and a sleep intervention in sleep-disordered youth) produced evidence for significant long-term improvement in ADHD symptoms, with a moderate effect size;334, 523 one also evaluated treatment satisfaction, finding a small and nonsignificant effect. Three studies of parent support found negligible and non-significant long-term effects on ADHD symptoms,228, 257, 290 two studies found nonsignificant long-term effects on broadband measures110, 257 and one on functional impairment.228 Two neurofeedback studies reported long-term effects on problem behaviors and functional impairment that were small and not significant.126, 458 One of these studies reported a small but significant long-term improvement in ADHD symptoms,458 whereas the other reported small nonsignificant effects.126 Two studies of school-based interventions assessed effects on long-term outcomes.259, 531 One (a study of an intensive summer program) found no improvement in ADHD symptoms or school disciplinary problems compared to no intervention.531 The other (a school-based training intervention) found no significant improvement in impaired peer relations for ADHD youth.259 Neither intervention improved long-term academic performance. More studies assessed the long-term effects of combined pharmacological (stimulant) and behavioral treatment on ADHD outcomes; however, only one assessed long-term effects on ADHD symptoms and functional impairment, finding small, nonsignificant effects for each.343 Two assessed long-term effects on problem behaviors, with conflicting results.107, 343 One study reported small, nonsignificant long-term effects on broadband measures.107 Thus, with few exceptions, the body of evidence suggests that most interventions, including combined medication and psychological treatment, yield no significant long-term improvement in most ADHD outcomes.
Variation in Outcomes by Clinical Presentation
We found little evidence that treatment outcomes varied by ADHD presentation but available data were limited.
Risk of Medication Diversion
We found only one study that assessed the risk of medication diversion in the treatment of ADHD. It was a double-blind randomized controlled trial comparing stimulant plus cognitive behavioral therapy (CBT) vs placebo plus CBT in treating adolescents who had ADHD with comorbid substance use disorder (SUD). The stimulant arm had twice the self-reported rate of diversion than the placebo arm which, though not statistically significant, suggests that further studies of diversion and stimulant misuse is warranted, particularly in ADHD youth with SUD. Caution is indicated when prescribing stimulants to ADHD youth who have comorbid SUD.
ADHD Monitoring
We identified only 10 studies pertaining to the assessment of monitoring strategies for ADHD outcomes.
Several of the studies indicated that monitoring measures correlated poorly over time, whether the correlations were between different raters using the same rating scale545 or between different assessment modalities (e.g., rating scale with computerized performance test).173, 203 These findings suggest that assessment modalities may be more complementary than interchangeable, and that more than a single assessment modality may be required for comprehensive and effective monitoring of ADHD outcomes.173, 545 One study suggested that subjective outcome measures, such as rating scales, may be more sensitive than more objective measures, such as the continuous performance task, for detecting treatment-induced changes in ADHD.203
Three studies assessed the effects on ADHD symptoms of interventions that train pediatricians to improve either their symptom monitoring or their adherence to treatment guidelines.255, 256, 268 Despite very extensive training efforts, and even when expert support and consultation was available,256 pediatricians exhibited poor compliance in attending training programs for treating ADHD,256, 268 and even when they did attend, pediatrician compliance with treatment guidelines was poor, both in terms of monitoring treatment response and in following dosing guidelines. Use of expert consultative services and compliance with recommendations was poor.256
One study suggested that monitoring height and weight, combined with either medication holidays or caloric supplementation, may be helpful for attenuating stimulant-associated weight loss but not slowing of height velocity.274 Another study suggested that use of an electronic bottle cap may be more accurate and valid than patient reports, clinician impression, or pill counts for monitoring of medication adherence.629
Findings in Relation to Existing Research Syntheses and Practice Guidelines
The conclusions and clinical recommendations of this review are generally consistent with those of the two prior Agency for Healthcare Research and Quality reviews on ADHD.11, 53 The Key Questions of the 2011 review focused primarily on long-term (> 1 year) treatment effectiveness and adverse effects, whereas the three Key Questions of the 2018 review were nearly identical to ours. The 2018 review served as an important resource for development of the 2019 clinical practice guidelines for the evaluation and treatment of ADHD from the American Academy of Pediatrics (AAP),1199 which in turn was the primary source for the recommendations from the U.S. Center for Disease Control for the diagnosis and treatment of ADHD.1200
Our findings for diagnostic tools suggest that the clinical diagnosis of ADHD likely benefits from ratings of ADHD symptoms from multiple informants, which is consistent with the AAP guidelines that advise documentation of symptoms and impairment in more than one setting (such as home and school), with information obtained from parents, school personnel, and mental health clinicians. To these informants we would add that inquiring about symptoms from both parents, and directly from the youth, can also be helpful. The 2018 review did not assess the diagnostic performance of ADHD rating scales. That review concluded, however, that brain imaging and EEG had insufficient evidence to support their use as diagnostic tools, consistent with our conclusions, and despite the FDA approval of one EEG measure as a purported diagnostic aid.26, 27 To those conclusions we add that neuropsychological tests (including measures from continuous performance tests) and blood biomarkers also do not yet have sufficient evidence to serve as diagnostic tools.
Our treatment findings concluded that FDA-approved stimulant and non-stimulant medications had the greatest strength of evidence across all interventions for significantly improving ADHD symptoms and other outcomes. Thirty-five papers that met criteria for inclusion in the current review assessed treatment effectiveness for more than a year, which was the focus of the 2011 review. That 2011 review concluded with a low strength of evidence that methylphenidate and atomoxetine were both effective long-term, though the average effect sizes after a year were somewhat lower than those for the short-term studies included in the present review. The 2018 review did not restrict the time frame for treatment, but nevertheless found insufficient evidence to modify conclusions for the effectiveness of FDA-approved medications. The present review adds to these prior reviews by providing mean effect sizes for comparisons of FDA-approved medication with placebo on improving not only ADHD symptoms, but a range of other important outcomes as well, at least for short-term outcomes. The current review also showed that stimulant and most non-stimulant medications yielded comparable effects on key effectiveness outcomes when these medications were compared head-to-head. Clinical guidelines advise starting treatment for youth older than six years of age with FDA-approved medications, which the findings of this review support.
The current review did not find that combination therapies of medication plus psychosocial therapies produce better results than medication alone. Moreover, we found that the effect sizes for parent therapies tended to be smaller than those for other interventions in improving ADHD outcomes. The 2011 review found larger effect sizes than we found for parent training for preschool youth with ADHD or disruptive behavioral disorders, but the prior review included studies that did not meet criteria for inclusion in our review. The 2018 review also found that parent training improved ADHD symptoms, but the review did not provide a mean effect size. Neither of the prior reviews assessed the effectiveness of combination treatment. The AAP clinical guidelines for preschool children advise treatment with parent training and/or classroom behavioral interventions as the first line of treatment, if available. These recommendations remain supported by the present review, particularly given the paucity of prior medication studies for preschool children. The guidelines also recommend the combination of parent training, classroom interventions, or behavioral interventions with medication therapy for older youth with ADHD, though no evidence suggests that this combination of therapies is better than monotherapy, and some evidence from head-to-head comparison studies suggests that the combination is not better than monotherapy.
The 2018 review found some evidence that cognitive training, and insufficient evidence that neurofeedback, improve ADHD symptoms. Our report includes substantially more studies and we found low strength of evidence that cognitive training does not improve ADHD symptoms, and some evidence that neurofeedback does, although the strength of evidence is low. We also found, with low strength of evidence, that the group of nutritional supplements and dietary interventions improve ADHD symptoms and problem behaviors. However, the approaches were very diverse, and approaches assessed in more than one study did not show an effect. The evidence for specific nutritional or supplement interventions is still too low to suggest their routine use.
The 2018 review found no papers pertaining to the assessment of monitoring strategies for youth with ADHD, whereas our current review identified ten such papers. The APA and Centers for Disease Control and Prevention (CDC) clinical guidelines do not include recommendations for monitoring strategies.
Implications
Our review points to the complementary nature of rating scales from multiple informants – from both parents if possible and from teachers, and even from the youth as well – since the scores tend to correlate poorly with one another and because ADHD symptoms in the same child can vary across settings. No single informant is a gold-standard. Multiple informants will provide a more complete clinical picture for how symptoms are expressed and perceived in different settings, and they will accordingly inform clinical judgement when making a diagnosis. Similarly, neuropsychological test measures of executive functioning, such as the CPT, may help inform a clinical diagnosis, but they are not definitive either in ruling in or ruling out a diagnosis of ADHD. Rating scales and neuropsychological tests are more helpful in diagnosis when the clinical question is whether a youth has ADHD or is healthy, rather than when the clinical question is whether a youth had ADHD or another mental health or behavioral problem, which tends to incorrectly identify youth with other clinical conditions as having ADHD. Biomarkers, EEG, and magnetic resonance imaging (MRI) are not yet close to being ready to aid clinical diagnosis. Ultimately, a valid and reliable diagnosis of ADHD requires the judgment of a clinician who is experienced in the evaluation of youth with and without ADHD, with the aid of standardized rating scales and input from multiple informants across multiple settings, including parents, teachers, and the youth themselves.
An increasing number of treatment modalities have been shown to significantly improve ADHD symptoms, and with comparable effect sizes when delivered as monotherapies. These include stimulant medications (methylphenidate and amphetamine), non-stimulant medications (particularly the NRIs atomoxetine and viloxazine, as well as the alpha agonists clonidine and guanfacine), individual psychosocial treatments, neurofeedback, and nutritional interventions, though very few of the non-medication studies have employed precisely the same interventions, which precluded an assessment of which specific interventions within each of these treatment categories were most effective. Psychosocial interventions, parent support, neurofeedback, and nutrition and supplements may exert considerably weaker effects on ADHD symptoms than the other interventions. Strength of evidence is high for medications and moderate for the other treatment modalities. The absence of head-to-head studies comparing the effectiveness of these monotherapies precludes recommendations regarding which is most likely to be helpful and should be tried first. Stimulant and NRI medications, separately and in head-to-head comparisons, have shown effectiveness and similar rates of side effects, including appetite suppression. The combination of treatment modalities, including combined medication plus psychosocial therapy, has minimal evidence for improving ADHD outcomes, and in fact a moderate strength of evidence indicates that combined therapy is no better than monotherapy. Treatment guidelines that recommend combination therapy1199, 1201, 1202 should consider that successful combinations showing clear superiority still need to be explored and identified. A further finding of this review is that only FDA-approved medications have been shown to statistically significantly improve broadband symptoms and functional impairment.
Findings from studies that attempted to train pediatricians in better adherence to ADHD monitoring and treatment guidelines suggest that training established pediatricians to adhere more closely to the guidelines does not work and that either much stronger incentives are needed for established pediatricians (such as including training and demonstrated compliance in criteria for maintenance of board certification), or else demonstrable guideline adherence should be included in pediatric residency training programs.
Strengths and Limitations
A major strength of this review is its inclusiveness, incorporating publications from 1980 and yielding more than 500 separate studies that informed our findings. Other strengths include: a review of evidence for the utility of biomarkers, EEG, and neuroimaging measures in the diagnosis of ADHD; parsing of non-pharmacological therapies by the target of the therapy (the youth, parent, or school); and the parsing of ADHD outcome measures to provide more clarity on the functional domains that treatments affect.
Despite the large number of included studies, we restricted this review to studies that reported on children with a clinically confirmed diagnosis of ADHD, excluding studies with broader samples (such as evaluations of psychosocial programs that were not specific to youth with a clinical diagnosis). In addition, although studies of children of all ages were eligible for inclusion in the report, the number of studies exclusively addressing younger children with ADHD were relatively few. The median minimum age in included studies was six years old. Samples were predominantly male, and the median number of girls included in the studies was only 25 percent. Furthermore, smaller studies were not included unless they demonstrated a power analysis, which may have excluded studies of more intensive treatments. We also excluded studies documenting very short-term treatment effects by requiring studies to report on a minimum treatment duration of four weeks. This requirement may have excluded relevant brief interventions, or very intense psychosocial interventions delivered in a short time period. Furthermore, this synthesis was focused on outcomes selected with the help of an expert panel, and it should be noted that individual interventions may show effects on other outcomes. Because few studies compared treatment effects in direct, head-to-head comparisons, we had to explore modifiers indirectly, across studies. Finally, despite a very comprehensive search, few monitoring studies were available to inform this report.
Future Research
One of the most important potential uses of this systematic review would be the identification of effect modifiers for both the performance of diagnostic tools and therapeutic interventions – for example, determining whether a diagnostic tool performs better or worse, or a treatment is more or less effective, in one patient subgroup than another (Key Question [KQ] 1c and KQ2a), such as in younger or older patients, in ethnic minorities, in those experiencing material hardship, in patients with a comorbid illness, or in those with a specific ADHD presentation. These analyses are essential for improving clinical assessments and treatment planning. Future studies of ADHD should more systematically address the modifier effects of these patient characteristics. More research is needed on the performance of diagnostic tools, the consequences of being misdiagnosed as either having or not having ADHD, the real-world effectiveness and long-term outcomes of medication and other therapies, and effectiveness of monitoring strategies. Much more research is needed on the diagnosis and treatment of preschool children who have ADHD.
Future Research on ADHD Diagnosis
Future studies of diagnostic tools should include assessment of how well the tools distinguish ADHD youth not simply from typically developing youth, but especially from youth who have other emotional and behavioral problems. They should also assess the potential adverse consequences of youth being incorrectly diagnosed with or without ADHD. Research is needed to identify consensus algorithms that combine rating scale data from multiple informants to improve the clinical diagnosis of ADHD, which at present is unguided, ad hoc, and suboptimal.
Despite the theoretical promise and a large number of prior studies of the use of continuous performance tests, EEG, or imaging to diagnose ADHD, conclusions about these potential diagnostic tools was severely limited by the use of different diagnostic measures within each test modality, differing diagnostic thresholds applied to those measures across studies, and differing algorithms that combine those variables to reach a diagnostic decision, and the frequent failure to clearly report those study elements in the publication. Therefore, to support future efforts at synthetic analyses, diagnostic studies should report sufficient detail of their measures and diagnostic algorithms -- precise operational definitions and measurements of the variable(s) used for diagnosis, any diagnostic algorithm employed, the chosen statistical cut-offs, and the number of false positives and false negatives the diagnostic tool yields.
Studies of diagnostic tools should include ROC analyses to support comparison of test performance across studies that are independent of diagnostic threshold for the tool. Studies should also include assessment of test-retest reliability to help discern whether variability in measures and test performance across settings is a function of setting or is a consequence of measurement variability across time. Future studies should address the role of co-occurring disorders in the diagnostic process and their influences on their performance of the diagnostic tools. In addition, more studies are needed that compare the diagnostic accuracy of different test modalities head-to-head.
Making available in public repositories the raw, individual-level data, as well as the algorithms or computer code, for diagnostic tools is important to aid future efforts at replication, synthesis, and new discovery. Independent replication of performance measures of diagnostic tools in real-world settings is essential prior to FDA approval and before recommendations for widespread clinical use.
Finally, the “diagnostic tests” that are most often used clinically, usually at considerable financial expense, are neuropsychological measures of “executive functioning”. These include, among others, measures of working memory and errors of omission on continuous performance tests (thought to represent the clinical construct of inattention) and measures of impulsive responding on continuous performance tests (thought to represent the clinical construct of impulsivity). These and other objective, quantitative neuropsychological test measures of executive functioning notoriously correlate only weakly with the clinical constructs of inattention, impulsivity, and hyperactivity that are based on observation of real-world behavior and that define ADHD.173 Many youth with ADHD have normal executive functioning profiles on neuropsychological testing, and many who have impaired executive functioning on neuropsychological tests do not have ADHD.1203 A major open question for future research is how these two constructs—neuropsychological test measures of executive functioning and the real-world functional problems that define ADHD—map on to one another, and how the correspondence of that mapping can be improved.
Future Research on ADHD Treatment
More trials are needed that compare alternative interventions head-to-head or that compare combination treatments with monotherapy. Future studies of psychosocial and parent interventions should employ study designs that support more valid causal inferences and higher strength of evidence for the effectiveness of the interventions assessed, including active attention comparator conditions and effective blinding of participants and assessors to study interventions and hypotheses.1195, 1196 More and higher quality studies with independent replication are needed to assess the effectiveness of individual complementary and alternative therapies, as well as exercise. Much more research is needed to assess long-term treatment compliance, long-term treatment effectiveness across a wide array of interventions, patient-centered outcomes beyond ADHD symptom improvement, medication diversion, and adverse effects associated with treatment (including non-pharmacological interventions).
Studies evaluating ADHD interventions should address the role of patient characteristics as modifiers of treatment effects. This effort will help to identify which treatments are most effective for which patients, to aid in the development of personalized treatments for youth with ADHD. To aid discovery and confirmation of these modifiers, future treatment studies should make publicly available all individual-level demographic, clinical, comorbidity, treatment, and all available outcome data (not only the primary outcomes), together with a detailed data dictionary. Patient-centered outcomes that assess functional domains other than ADHD symptoms, such as functional impairment and academic performance, should be acquired in clinical trials and shared publicly.
Future Research on ADHD Monitoring
Much more research is needed that compares the utility of various strategies for monitoring outcomes and tracking response to treatment over time in ADHD youth. The temporal stability of outcome measures and their sensitivity to change in response to treatment should be assessed to support ADHD monitoring strategies.
Future synthetic studies should also consider reviewing studies of long-term outcomes in ADHD youth, even if not in the context of comparing monitoring strategies, as the findings will be of interest to patients, parents, and clinicians and will critically inform treatment decisions.
Applicability
Several included studies reported multiple exclusions for eligible participants, which limited the generalizability of findings. Diagnostic performance, as well as treatment effects in clinical practice, may not translate from the favorable effects shown in the documented research to real world practice. In addition, the number of girls included in the identified studies was small and several studies did not include any female participants, potentially limiting the applicability of the findings.
- Discussion - ADHD Diagnosis and Treatment in Children and AdolescentsDiscussion - ADHD Diagnosis and Treatment in Children and Adolescents
Your browsing activity is empty.
Activity recording is turned off.
See more...