Identifying New Symptoms for Diagnosing ADHD in Adulthood
Russell A. Barkley, Ph.D. and Kevin R. Murphy, Ph.D.
Reprinted with permission from ADHD Report, Guilford Publications, Inc., New York, NY,
April 2006, Volume 14 No. 4
To subscribe to ADHD Report
The current symptom list for ADHD in the DSM-IV (American Psychiatric Association, 2000) was developed on children and was only field tested using children (Lahey, Applegate, McBurnett, Biederman, Greenhill et al., 1994; Spitzer, Barkley, & Davies, 1989). The utility of extending that list to adults with ADHD is therefore an open question. This article addresses the important issue of whether or not better symptoms could be identified for the adult stage of this disorder than those 18 childhood symptoms currently represented in the DSM-IV.
We began by making a list of the most common complaints that we had heard from adults presenting at the Adult ADHD clinic at the University of Massachusetts Medical Center where more than 100 adults were evaluated each year. We also went back through previous charts of adults seen at this clinic to identify such symptom items. We also used the theory of executive functioning developed by Barkley (1997) and extended to understanding ADHD in order to generate potential symptoms that deal with each of the five executive components of his model: response inhibition, nonverbal working memory and sense of time, verbal working memory, emotional/motivation self-regulation, and planning (generativity or reconstitution). The results of our work on this new symptom list will appear early next year in a new book presenting the results of an original research project (Barkley & Murphy, 2007). This project constitutes one of the most comprehensive evaluations of adults with ADHD. In this project, we extensively evaluated 146 adults with ADHD on numerous measures of adaptive functioning across many domains of major life activities. We compared them to both a community control group of 109 adults and a clinical control group of 97 adults seen at the same ADHD Clinic but not diagnosed with the disorder. These adults had a mean age of 32-37 years, depending on the group, with 47-68% of each group being male.
A New Item Pool of Potential Symptoms for ADHD in Adults
We developed a list of 91 new items that might have some potential for being associated with and predictive of ADHD at the adult stage of its development. We included items that further elaborated on the problems with behavioral and cognitive inhibition that are thought to be a core feature of ADHD (Barkley, 1997, Nigg, 2001) yet which are represented by only three items in the current DSM-IV list, most of which may reflect verbal impulsiveness. And so we added items dealing with impulsive decision making, making impulsive comments to others, poor delay of gratification, doing things without considering their consequences, and so forth, that may better reflect this construct. Other items deal with working memory (holding information in mind that is guiding behavior), the sense and use of time thought to be related to it, emotional self-regulation, and planning and forethought, all of which derived from Barkley's theory. Still other items of a less theoretical nature were included because they were often voiced by adults with ADHD or had been identified as problematic for them in previous studies, such as excessive speeding while driving, poor management of money, motor clumsiness, poor handwriting, and a proneness to accidents (see Barkley, 2006). Because most of these symptoms originated in Barkley's theory of executive functioning (EF), we consider this list to largely reflect that construct.
These 91 items were collected in a structured interview with the participants, in which case they had to be endorsed as occurring “often” or more frequently to be considered a positively reported symptom. The results for these items would have the greatest bearing on any effort to develop new symptoms to be listed in DSM-V for ADHD in adults as they would be of the same binary or dichotomous nature as those in the current DSM-IV symptom list. We also collected them with reference to the same time period stipulated in the DSM-IV, that being the previous 6 months as reported by participants, and the same descriptor of symptom frequency as in DSM-IV, that being the word “often.”
Obviously, the symptom list served its purpose because all items occurred significantly more often in the ADHD gropy than in the Community control group. In that sense, all 91 potential ADHD symptoms were problematic for the ADHD group, supporting the developmental inappropriateness of their severity. However, all but one of these items also occurred in more of the Clinical control adults than in the Community controls. To reduce this item set down to those likely to have the greatest promise for characterizing ADHD in adults, we imposed two criteria. First, the item had to occur in at least 65% (roughly two-thirds) of the ADHD group. Second, it had to occur in significantly more of the ADHD group than in the Clinical control group. There were 43 such items. We then threw out the four items that we believed were too close in wording to those found in the DSM-IV and therefore likely to be redundant with them, which left 39 items.
These 39 items constituted the pool of those symptoms offering the greatest potential for characterizing ADHD in adults. They were analyzed further for their ability to accurately discriminate among the groups using logistic regression. The items that best discriminated the ADHD cases from those in the Community control group were:
These five items had an impressive overall classification accuracy of 99%, with accuracy of 99% for the Community control group and 99% for the ADHD group.
However, it was just as important to determine those items that differentiated the ADHD group from the Clinical control group. Six items did so:
These 6 items had an overall classification accuracy of 77%, with accuracy of 65% for the Clinical control group and 85% for ADHD group. This overall classification accuracy is superior to the items from the DSM-IV symptom list (supportive data will appear in the new book). The first two of these new EF items are the same items that proved so effective at discriminating the ADHD and Community control group above. Readers also should not be surprised at the last symptom here (speeding) given that past studies of driving performance in adults and teens with ADHD have repeatedly identified speeding with a motor vehicle as a significant problem (see Barkley, 2004; Barkley & Cox, in press, for reviews).
In view of these findings, we can consider the following 9 symptoms largely reflecting executive functioning as a potential item set worthy of further testing for use in adult ADHD:
Factor Structure of New EF Symptoms and Old DSM-IV Symptoms
Do these new items reflecting EF represent a different dimension of symptoms than those already identified as characterizing the DSM-IV symptom list? As we will show in this new book, the DSM-IV symptom list may best be characterized as involving three dimensions or factors in adults rather than the two presented in DSM-IV and based on children. These three factors were inattention, hyperactive-impulsive behavior, and verbal impulsivity. To find out how these new EF symptoms relate to the existing 18 symptoms from DSM-IV, we conducted a factor analysis involving all 27 symptoms.
The new EF symptoms did not constitute a new dimension of ADHD symptoms but mapped on to the already existing three-dimensional structure noted above. The first factor was one of sustained attention-working memory-distractibility and explained 44% of the variance. It contained most of the new EF symptoms. The second factor was one of hyperactive-impulsive behavior and accounted for 7% of the variance. It also contained some of the new EF symptoms reflecting impulsive cognitive decision-making and perseverative behavior. And the third factor was a verbal impulsiveness factor reflecting the three pre-existing DSM items. It explained just 4% of the variance.
New List of Adult ADHD Symptoms for DSM-V
Rather than add 9 new EF symptoms to the 18 already in the DSM-IV, is it possible to reduce this pool of 27 items down to a more manageable size for use in the clinical diagnosis of adults? We tried to do so by using logistic regression once again. We first examined those symptoms that might best distinguish the ADHD group from the Community control group. Just 1 item was able to achieve an overall classification accuracy of 97% for both groups. This DSM item was “being easily distracted” and proved an excellent discriminator from normal cases. Three other items were able to add another 3% to this classification accuracy and these were: difficulty sustaining attention (DSM), difficulty organizing tasks and activities (DSM), and poor follow-through on promises or commitments I may make to others. The latter three, however, are not necessary for clinical purposes as the single item of being easily distracted worked the best of all items.
Where the new EF items had greater value was in helping to distinguish the ADHD group from the Clinical control group. Our analysis found 7 symptoms were needed to maximize group discrimination, with just 1 being from the DSM list and the remaining 6 from the new list of EF symptoms:
Often leave seat in classroom or in other situations in which remaining seated is expected (DSM). Noteworthy here was that this item was reverse-weighted, meaning that it was significantly related to being in the Clinical control group rather than in the ADHD group.
The overall classification accuracy for these 7 symptoms was 77%, with accuracy of 88% for the ADHD cases and 64% for the Clinical control group. The one DSM symptom making it into this classification analysis (Often leave seat . . .) actually characterized the Clinical control group better than the ADHD group and served to rule out ADHD rather than rule it in. It should therefore not be considered a symptom of adult ADHD as it is an indication of likely not having it but having some other disorder instead.
We can combine the one DSM-IV item that best differentiated adults with ADHD from the Community control group (being easily distracted) with the six new EF items that served to best differentiate adults with ADHD from the Clinical control group to get a list of the 7 best symptoms for identifying cases of ADHD. If desired, the additional two DSM items that made a nominal increase in classification accuracy between the ADHD and Community group could also be added to lengthen the list to 9 symptoms. These would be the nine items to be recommended to the DSM-V committee when considering adopting a specific item set for the diagnosis of ADHD in adults.
We fully realize that such a list incorporates no symptoms of hyperactivity, perhaps violating the very conceptualization of ADHD or at least how it is currently subtyped (inattentive type, hyperactive-impulsive type, and combined type). If empirical grounds are the main basis for constructing a symptom list, then our analyses show that symptoms of hyperactivity do not contribute significantly to identifying adults with ADHD when evaluated in the context of the entire set of DSM symptoms that include the inattention items or in the context of the best executive function items. Such a single set of symptoms, loading as it does on a single factor or dimension of inattention-executive function symptoms, would certainly preclude using the current DSM subtyping approach to ADHD as it would eliminate the hyperactive subtype. That is not such a bad thing, however, considering that it is not a very common subtype diagnosed in adults with ADHD. It also declines markedly in prevalence with age, with most such children placed in this category ultimately moving on to the combined type within a few years (see Barkley, 2006). In short, it is not hyperactivity that distinguishes adults with ADHD from normal adults or those having other disorders but distractibility, impulsive decision-making, and poor executive functioning. So why mislead clinicians and the public otherwise by keeping symptoms of hyperactivity in any adult symptom list? We see little merit to the idea, either for clinical diagnosis or for subtyping ADHD in adults.
As we noted above, if all one wants to do in clinical diagnosis is determine if someone is normal (Community control) or not, then a single symptom (being easily distracted) does that quite well. But most adults coming to clinics for assistance are not normal and usually have some disorder. Given this circumstance, it appears from our work here that the inclusion of new EF symptoms increases the ability of this symptom set to accurately distinguish ADHD from non-ADHD clinical cases beyond that shown for the DSM items alone. While it would be nice if these symptoms could classify cases into these two groups as well as the symptom sets have done between ADHD and normal (Community) cases (97-100%), this is asking for too much. Considering that other clinical disorders are likely to have some effects on attention, executive functioning, and inhibition, this makes it more difficult for symptom presence alone to result in perfect differential diagnosis of ADHD from other disorders. Here again, it will be the inclusion of additional diagnostic criteria (onset, impairment, etc.) and clinical training concerning the nature of other non-ADHD disorders that facilitates this differential diagnostic process.
New Symptom Threshold for Adult ADHD for DSM-V
If the new symptom list generated above for diagnosing ADHD in adults were to be adopted in DSM-V, it would require a new threshold for the number of symptoms needed for diagnosis. Recall that the current DSM-IV recommends a threshold of 6 out of the nine symptoms on either the inattention list or the hyperactive-impulsive list. To examine this issue, we created a symptom summary score for the 9 best symptoms listed above. We then examined the distributions of these scores for each of the three groups of participants. Our results show that a threshold of 6 symptoms would work reasonably well. Only 1% of the Community control group would meet or exceed that threshold while 92% of the ADHD group would do so. But nearly half of the Clinical control group would do so as well (47%). Dropping the threshold to 5 symptoms would falsely identify no more Community cases than the previous 1% and would capture an additional 4% of ADHD cases (96% true positives) but would also raise the false positive rate in the Clinical group by 18% to 65% meeting this threshold. We do not recommend doing so. Our preference would be to adopt the list of nine symptoms and to apply the threshold of 6 out of 9 symptoms in view of the above findings. Of course, our results should be cross-validated against another sample of ADHD and control adults before they should be considered for clinical adoption, but our study does provide a starting point--it is an initial DSM-V field trial for guiding subsequent efforts at criteria development. Added to this procedure should be establishing the onset of symptoms producing impairment by the more generous age span of 14-16 years of age (as demonstrated in our new book). All of these criteria are shown in Table 1 (see original article in ADHD Report) in a format similar to that used in DSM-IV. This would serve as our recommendation for the DSM-V committee to consider for a set of criteria for ADHD in adults.
This paper raised the important question of whether or not the symptoms as currently presented in DSM-IV, and developed solely on children, are the best that can be developed for the evaluation of ADHD in adults. Here we attempted to address this critical question by presenting results from an analysis of a large item pool of originally 91 symptoms mainly comprised of difficulties with executive functioning. While results indicated that more adults with ADHD were likely to have problems with all of these executive symptoms than were Community control adults, we were able to reduce this pool down to 39 symptoms that were present in at least 65% of ADHD cases while also being significantly more likely to be present in those cases than in the Clinical control group. Regression analyses were used to reduce this item pool down to the 9 best executive function symptoms for discriminating among these groups. These 9 EF symptoms appeared to load on the same dimensions (factors) of inattention and hyperactive-impulsive behavior as was found to exist for the DSM symptom list in Chapter 3. This suggests that these two symptom dimensions actually assess a broader domain of cognitive functioning than their names imply, most likely that of executive functioning.
We then analyzed these 9 EF symptoms in the context of the existing 18 DSM items to see if they were redundant with them or were in fact better at discriminating among these three groups. Our results showed that 6 of these EF symptoms and just 1 DSM inattention symptom (easily distracted) would be the best symptom list for identifying ADHD in adults. Adding 2 additional inattention symptoms would slightly enhance group classification, thus yielding a set of 9 symptoms that could be recommended for consideration for use in DSM-V for the diagnosis of adult ADHD.
We further found that a total of 6 out of these 9 best symptoms would be a useful diagnostic threshold accurately classifying 99% of Community controls, 92% of ADHD cases, and 53% of Community cases. Along with a review of childhood recollected DSM symptoms and the use of a revised age of onset of 14-16 years for symptoms producing impairment, these 9 best symptoms and the threshold of 6 of these 9 would comprise a better diagnostic algorithm for ADHD in adults than do the current DSM-IV criteria. We are therefore prepared to recommend the criteria set forth in Table 1 as being better for identifying ADHD in adults than the current DSM-IV criteria.
An interview with the child offers the clinician the opportunity to observe the child's behavior and can yield valuable information as to the child's social and emotional adjustment, feelings about themselves and others, attitudes about school and other aspects of their life. However, even children with ADD often behave well during such interviews. Therefore, observations of a child's behavior, level of activity, attentiveness, or other compliance made during the interview sessions should not be taken as true of the child in other settings. Normal behavior in a one-on-one setting does not diminish the likelihood of the child having ADD.
American Psychiatric Association. (2000). Diagnostic and Statistical Manual of Mental Disorders (4th ed., text revision). Washington, DC: Author.
Barkley, R. A. (1997). ADHD and the Nature of Self-Control New York: Guilford.
Barkley, R. A. (2004). Driving impairments in teens and young adults with ADHD. In T. J. Spencer (Ed.), Psychiatric Clinics of North America, June, Vol. 27, #2 (pp. 233-260). New York: Moseby.
Barkley, R.A. (2006). Attention deficit hyperactivity disorder: A handbook for diagnosis and treatment (3rd ed.) New York: Guilford.
Barkley, R. A., & Cox, D. (2006). A review of driving risks and impairments associated with Attention-Deficit/Hyperactivity Disorder and the effects of stimulant medication on driving performance. Journal of Safety Research. In press.
Lahey, B. B., Applegate, B., McBurnett, K., Biederman, J., Greenhill, L., Hynd, G. W., Barkley R. A., Newcorn, J., Jensen, P., Richters, J., Garfinkel, B., Kerdyk, L., Frick, P. J., Ollendick, T., Perez, D., Hart, E. L., Waldman, I., & Shaffer, D. (1994). DSM-IV field trials for attention deficit/hyperactivity disorder in children and adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 151, 1673-1685.
Nigg, J. T. (2001). Is ADHD an inhibitory disorder? Psychological Bulletin, 125, 571-596.
Spitzer, R. L., Davies, M., & Barkley, R. A. (1990). The DSM-III-R field trial for the Disruptive Behavior Disorders. Journal of the American Academy of Child and Adolescent Psychiatry, 29, 690-697.