Nov 23, 2008
Like all psychiatric disorders, ADHD is diagnosed based on the presence of particular behavioral symptoms that are judged to cause significant impairment in an individual’s functioning, and not on the results of a specific test. In fact, recently published ADHD evaluation guidelines from the American Academy of Pediatrics (AAP) explicitly state that no particular diagnostic test should be routinely used when evaluating a child for ADHD.
While most ADHD experts would agree that no single test could or should be used in isolation to diagnose ADHD, there are several important reasons why the availability of an accurate objective test would be useful.
First, many children do not receive a careful and comprehensive assessment for ADHD but are instead diagnosed with based on evaluation procedures that are far from optimal.
Second, although AAP guidelines indicate that specific diagnostic tests should not be routinely used, many parents are concerned about the lack of objective procedures in their child’s evaluation. In fact, many families do not pursue treatment for ADHD because the the absence of objective evaluation procedures leads them to question the diagnosis. You can read a review of an interesting study on this issue at www.helpforadd.com/2006/january.htm
For these reasons an accurate and objective diagnostic test for ADHD could be of value in many clinical situations. Two important conditions would have to be met for such a test to be useful.
First, it would have to be highly sensitive to the presence of ADHD, i.e., individuals who truly have ADHD as determined by a comprehensive evaluation should score positive for ADHD on the test. If the test were 100% sensitive, every individual who has ADHD based on current diagnostic criteria would score positive on the test. As the sensitivity of a test drops, the number of “false negatives” — normal test results in individuals who truly have the disorder increase and its utility goes down.
Second, individuals who don’t have ADHD should never score positive on the test, i.e., a positive result should occur only for individuals with ADHD and no one else. When a diagnostic test has high specificity, individuals without the condition rarely score positive on the test. When specificity is low, many individuals without the condition will score positive and may be incorrectly diagnosed as a result. This is referred to as a “false positive”.
Although many psychological tests yield different results, on average, for individuals with and without ADHD, they are not sensitive or specific enough to be particularly useful when making individual diagnostic decisions. For example, a widely used objective test in ADHD evaluations are Continuous Performance Tests (CPTs). These tests provide a computerized measure of a child’s ability to sustain attention and refrain from impulsive responding. Although average performance on CPTs for children with ADHD is below that of peers, and CPT data can be helpful when thoughtfully integrated with other diagnostic information, these tests yields too many false positives and false negatives to be useful as an “objective” diagnostic test for ADHD.
- Is there anything better? -
Several past issues of Attention Research Update have reviewed Quantitative EEG, i.e., QEEG, as a diagnostic aide for ADHD. The use of QEEG is based on findings that individuals with ADHD have a distinctive pattern of brain electrical activity that is often referred to as “cortical slowing”; this is characterized by an elevation of low frequency theta waves and a reduction of higher frequency beta waves in the prefrontal cortex. Theta wave activity is associated with an unfocused and inattentive state while beta activity is associated with more focused attention. Thus, an elevated theta/beta ratio reflects a less alert and more unfocused state.
In a QEEG testing, EEG data is collected from a child or adult in a non-invasive procedure that requires about 30 minutes. The EEG data is digitized and computer scored so that an individual’s theta/beta ratio can be computed; this ratio is then compared to what is typical for individuals of similar age. When this ratio is sufficiently elevated ratio — the cut-off typically used is 1.5 standard deviations above average which corresponds to the highest 7% of the population — the individual is considered to have the EEG marker for ADHD.
In past studies, roughly 90% of individuals diagnosed with ADHD based on a comprehensive evaluation tested positive for this EEG marker. In contrast, about 95% of normal controls tested negative. Thus, while not a perfectly reliable indicator, the sensitivity and specificity of QEEG in identifying ADHD was extremely strong. You can review these studies at www.helpforadd.com/2001/april.htm and www.helpforadd.com/yr2000/april.htm.
The important limitation of this work was that QEEG was tested using individuals known to have ADHD and normal controls without any disorder. Differentiating between ADHD and no disorder, however, is not the situation that clinicians typically face. Instead, a child is referred because of attention and/or behavior problems and the clinician must determine whether these problems reflect ADHD, are better explained by another disorder, or do not rise to the level where any diagnosis is appropriate. Thus, for QEEG to be useful in ADHD evaluations, it must also accurately distinguish between ADHD and other disorders.
One recently published preliminary study indicated promising findings in this regard. Twenty-six children and adolescents referred to an outpatient psychiatry clinic for attention and behavior problems received a thorough ADHD assessment conducted by a team of child psychiatrists. They also received a QEEG evaluation. Sixteen of the 26 were determined to meet DSM-IV criteria for ADHD by the psychiatric team while 10 were diagnosed with other conditions. Of the 16 diagnosed with ADHD, 15 showed the QEEG marker for ADHD; in contrast, none of the 10 diagnosed with other conditions showed the QEEG marker. Thus, the QEEG test performed extremely well. A comprehensive review of this study can be found at www.helpforadd.com/2007/november.htm
While these results were encouraging, the sample was small and from a single clinical site. This raises important questions about the generalizability of the findings that need to be addressed in a study that incorporates a larger sample drawn from multiple clinical sites. Recently, a study meeting these criteria was published; I believe it is one of the most interesting and important studies I have seen in several years [Snyder et al. (2008). Blinded, multi-center validation of EEG and rating scales in identifying ADHD within a clinical sample. Psychiatry Research, 159, 346–358.]
- Methods -
Participants — Participants were 159 6–18-year olds (101 males and 58 females) taken by parents to 1 of 4 pediatric and psychiatric clinics because of concerns related to attention and behavior problems. One hundred and fifteen were children (6–11 years old) and the remainder were adolescents. There was a good representation of African Americans in the sample (37%).
Psychiatric Exam — At each site, children received a standardized psychiatric evaluation that included a semi-structured interview (the KSADS-PL) and measures of functional impairment and disorder severity. Both parents and children/adolescents were interviewed, the optimal procedure for covering both externalizing and internalizing disorders. Other clinical procedures included taking a medical history, a developmental history, and providing a physical exam.
Using results of these interviews and associated measures, the clinical team performed a complete differential diagnosis for the presence of ADHD, comorbid conditions, and other childhood/adolescent disorders. Diagnostic decisions about ADHD followed a standard protocol to determine whether strict DSM-IV diagnostic criteria were met. The team’s determination about the presence or absence of ADHD was considered the “gold standard” against which diagnostic decisions based on the results from standardized behavior rating scales and QEEG were compared.
Rating Scales — Parents and teacher completed 2 behavior rating scales that are widely used in the assessment of ADHD — the Conners Rating Scale and the ADHD-IV Rating Scale. Children were considered to be positive for ADHD if their scores on these scales exceeded the recommended cut-off for identifying ADHD. This enabled the researchers to determine how well diagnostic decisions derived fro rating scales agreed with results from the comprehensive psychiatric exam.
QEEG — EEG data was collected on each child using standard collection procedures by trained EEG technicians. The theta/beta ratio computed for each child and compared to values for age matched controls from a large normative data base. Participants whose theta/beta ratio was at 1.5 standard deviations above the average score, i.e., roughly the top 7%, were considered to show the EEG marker for ADHD.
It is important to note that diagnostic decisions from the psychiatric evaluation were made without the team having any access to rating scale or QEEG data. Thus, decisions made about the presence or absence of ADHD from the psychiatric evaluation was not influenced in any way by knowledge of these other results.
- Results -
The logic of this study is simple and straight forward. The authors treated results of the psychiatric exam as the “gold standard” for determining which of the 159 participants met diagnostic criteria for ADHD. Then, they examined how well results based on the behavior rating scales and the EEG exam matched this standard.
Ninety-seven of the 159 children and adolescents (61%) were diagnosed with ADHD by the psychiatric evaluation. Sixty-four of these children were diagnosed with at least 1 other disorder, 35 had at least 2 additional disorders, and 11 had at least 3 additional disorders. The most common comorbidities were one of the disruptive behavior disorders (Oppositional Defiant Disorder or Conduct Disorder) which occurred in 66 of the 97, followed by an anxiety disorder (46 of 97), learning disorder (33 of 97), and mood disorder (23 of 97).
Of the 62 participants not meeting criteria for ADHD, all but 8 were diagnosed with one of these other disorders while 8 had no diagnosis.
Overall, therefore, this was a diverse clinical sample that included the full range of psychiatric difficulties that clinicians are called on to address.
- How accurate were behavior rating scales at identifying ADHD? -
The behavior rating scales did not perform well. The results were as follows:
ADHD-IV Parent — 28% false negatives, 67% false positives, 56% overall accuracy.
ADHD-IV Teacher — 62% false negatives, 39% false positive, 47% overall accuracy.
ADHD-IV P&T combined — 45% false negatives, 57% false positives, 50% overall accuracy.
Conners Parent — 22% false negatives, 86% false positives, 55% overall accuracy.
Conners Teacher — 33% false negatives, 59% false positives, 58% overall accuracy.
Conners P&T combined — 28% false negatives, 81% false positives, 53% combined.
As can be seen, the rate of false negatives (children diagnosed with ADHD based on the psychiatric evaluation that scored below the recommended ADHD cut-off on the rating scale) ranged from 28% for the parent version of the Conners Rating Scales to 62% for the teacher version of the ADHD-IV rating scale.
False positive rates (children without ADHD based on the psychiatric evaluation who scored positive for ADHD on the rating scale) ranged from 39% to 86%. The extremely high false positive rate for the parent version of the Conners indicates that parents tended to rate their child high on ADHD symptoms even when ADHD was not judged to be present.
Overall classification accuracy — how often rating scale results agreed with psychiatric evaluation results — was below 60% for every scale. Thus, the level of agreement was not much better than chance.
- QEEG Accuracy -
The accuracy of QEEG as a diagnostic test was much higher — the false negative rate was only 13% and the false positive rate was only 6%; this resulted in an overall accuracy rate of 89%. These figures indicate the following:
- 87% of children diagnosed with ADHD by the psychiatric evaluation showed the EEG marker for the disorder;
- 94% of children without ADHD screened negative for the EEG marker;
- If ADHD diagnosis was based strictly on the presence or absence of the EEG marker, it would match decisions based on the psychiatric evaluation almost 90% of the time.
Overall, these results are far superior to the classification accuracy using rating scales.
- Extending the findings to different subgroups and comorbidities -
Because participants represented a diverse clinical sample, the researchers could test whether QEEG accuracy was similar for children vs. adolescents, whites vs. blacks, and males vs. females. Across these different demographic groups, overall accuracy rates ranged from 87% to 95%. Thus, QEEG worked well within all demographic groups.
The authors also examined whether diagnostic accuracy of the QEEG was consistent depending on whether or not other disorders were present. When the psychiatric evaluation indicated ADHD, the QEEG was equally likely to be positive regardless of what other psychiatric conditions were diagnosed. Similarly, when ADHD was not diagnosed, the likelihood that the QEEG marker was negative did not depend on what other conditions were present. Thus, the classification accuracy of QEEG as a diagnostic test for ADHD was not influenced by the presence or absence of other psychiatric conditions.
- Summary and Implications -
I believe these findings are very important. One clear implication is that results from behavior rating scales must be used cautiously in ADHD evaluations. Making diagnostic decisions based on counting symptoms and/or determining whether a child’s rating scale results falls in a clinically elevated range for ADHD will lead to high rates of misdiagnosis in comparison to what would emerge from a comprehensive psychiatric evaluation. In particular, data from this study suggests that many children who do not truly have ADHD would be erroneously diagnosed with the disorder.
I should emphasize that ADHD evaluation guidelines from the American Academy of Pediatrics and the American Academy of Child and Adolescent Psychiatry clearly indicate that rating scales should never be used in isolation to diagnose an individual with ADHD. Thus, using rating scales in this way is inconsistent with best practice guidelines. However, given the limited time available in many primary care settings to conduct comprehensive evaluations, it would not be surprising if rating scale results are sometimes given greater emphasis than is recommended.
This is where QEEG results can be so helpful. In regards to matching ADHD diagnostic decisions that result from a comprehensive evaluation, this test performed quite well — overall accuracy rates were nearly 90%. The false positive rate of only 6% means that few children and adolescents for whom ADHD is ruled out by a comprehensive psychiatric exam would be diagnosed if QEEG findings were used to make the decision. This is a striking contrast to false positive rates that exceeded 80 % for some of the rating scale measures. Although adjusting how rating scale data is used to make decisions about ADHD might improve classification accuracy somewhat, it is difficult to imagine that overall accuracy rates would ever approach that found for QEEG.
It is important to underscore that despite the strong results found for QEEG, this tool is not a substitute for a comprehensive diagnostic exam and should not be used as a stand alone test for ADHD. One essential reason for this is that diagnostic evaluations for ADHD should go beyond simply deciding whether ADHD is present and gather other information that is critical for developing an optimal treatment plan. While QEEG may help with the former, it does not contribute to the latter, beyond identifying individuals for whom medication treatment would be an appropriate option to consider. I have heard experienced users of QEEG discuss that other data from the procedure can be useful in more expansive treatment planning, but I am not familiar with research that supports this broader use.
Given this limitation, what value would there be to routinely incorporating QEEG into ADHD diagnostic evaluations. Several things come to mind including the following:
1. In primary care settings where a comprehensive psychiatric exam can be difficult to provide, results from a 30–40 minute QEEG procedure can identify with reasonably high accuracy individuals for whom ADHD is likely to be an appropriate diagnosis.
2. Because false positive rates are so low, QEEG could reduce the number of individuals who are perhaps inappropriately treated with ADHD medication if physicians referred individuals with negative results for further evaluation.
3. In cases where parents are reluctant to pursue treatment for their child because of concerns that objective evaluation procedures were lacking, QEEG provides an objective biological marker of ADHD that can increase parents’ confidence in their child’s evaluation.
4. For people who continue to doubt that ADHD is an actual condition with important biological underpinnings, these findings highlight that the vast majority of individuals meeting DSM-IV criteria for ADHD have a distinctive pattern of brain EEG activity.
Overall, findings from this carefully conducted study make an important contribution to documenting the utility of QEEG as an objective test to assist in the diagnosis of ADHD. If this procedure were to become more widely used, these data suggest that the number of children and adolescents who are inappropriately diagnosed and treated for the disorder would diminish substantially.
– Dr. David Rabiner is a child clinical psychologist and Director of Undergraduate Studies in the Department of Psychology and Neuroscience at Duke University. He publishes Attention Research Update, an online newsletter that helps parents, professionals, and educators keep up with the latest research on ADHD, and teaches the online course How to Navigate Conventional and Complementary ADHD Treatments for Healthy Brain Development.
For related reading, you may enjoy: