Abstract
Keywords
ADHD is a mental health problem characterized by impaired executive functioning (cf. inattentive symptoms) and impulse control (cf. hyperactivity/impulsive symptoms; Biederman et al., 2006; Wilens, Faraone, & Biederman, 2004). Historically, ADHD has been conceptualized as a childhood disorder (Faraone et al., 2006). However, more recent longitudinal and follow-up studies have found that among those diagnosed with ADHD as children, a considerable number (45%-60%) have symptoms that persist into adulthood (Biederman, Petty, Evans, Small, & Faraone, 2010; Kessler, Adler, Barkley, et al., 2005; Kessler et al., 2010; Mick, Faraone, & Biederman, 2004; Wilens et al., 2009). In addition to longitudinal findings, recent research has also indicated that ADHD symptoms can be initially present in adulthood as well (Moffitt et al., 2015).
Studies on the prevalence of ADHD among adults in the general population report ranges from 3.4% to 8.4% (de Graaf et al., 2008; Fayyad et al., 2007; Kessler et al., 2007), with the large National Comorbidity Study Replication (NCS-R) reporting an estimated prevalence of adult ADHD at 4.4% among U.S. civilians (Kessler et al., 2006). Adult ADHD-like symptomatology (henceforth referred to as adult ADHD) has been of increasing academic and clinical interest, and has spurred considerable debate in the field as to whether it is clinically different than childhood-onset ADHD (Bell, 2011; Faraone & Biederman, 2016; Faraone et al., 2006; Moffitt et al., 2015).
Adult ADHD
In a longitudinal study of 140 males who were diagnosed with ADHD prior to age 17, Biederman and colleagues (2010) found that 57% of the sample either met the criteria for ADHD or reported clinically significant ADHD symptoms at the 10-year follow-up. In addition, this study found that there was an inverse relationship between participant age at follow-up and reported symptoms: rates were highest in participants who were 18 to 21 years at follow-up, and progressively declined with age. Other longitudinal studies have shown that, in addition to unresolved childhood ADHD, a considerable number of adult ADHD cases are adult-onset (symptoms first presenting in adulthood; Agnew-Blais et al., 2016; Moffitt et al., 2015). The veracity of such late-onset cases have been questioned, with some in the field suggesting that adult-onset ADHD many actually be related to the exacerbation of previously subthreshold childhood ADHD (Faraone & Biederman, 2016). Regardless of the etiology of the disorder, ADHD in adults is a growing area of clinical inquiry.
Compared with the presentation of ADHD in children, the symptom profile of adult ADHD is characterized by a higher number of symptoms in the inattentive cluster (e.g., overlooking details, trouble staying focused), rather than the hyperactivity/impulsivity cluster (Hanson et al., 2012; Kessler et al., 2010; Solanto, Wasserstein, Marks, & Mitchell, 2012; Wilens et al., 2004). This pattern was demonstrated in another longitudinal study by Biederman, Mick, and Faraone (2000) that showed that inattentive-cluster symptoms were less likely to remit over time, relative to hyperactivity/impulsivity symptoms and that both clusters tended to decrease over time. In light of these age-related symptom differences, it has been suggested that the clinical threshold for hyperactivity/impulsive symptoms should be lower when assessing adults for ADHD (Martel, Levinson, Langer, & Nigg, 2016; Solanto et al., 2012). In addition to displaying fewer externalizing symptoms, the functional consequences of ADHD in adulthood may not be as apparent as in childhood, due in part to learned compensatory strategies (e.g., implementing highly structured schedules) or selective avoidance (e.g., avoiding situations or careers where one must sit still for extended periods of time; Doyle, 2006).
Despite potential compensatory mechanisms, the functional impairments of adult ADHD are considerable and well documented. Adults with ADHD are more likely to have more days of missed work and may have difficulty with job retention (de Graaf et al., 2008; Fredriksen et al., 2014; Kessler, Lane, Stang, & Van Brunt, 2009). Compared with healthy controls, college students with ADHD often show diminished/stunted academic performance, including lower grade point averages (GPAs) and increased risk for dropping out (Advokat, Lane, & Luo, 2011; Krauss, Russell, Powers, & Li, 2006; Norwalk, Norvilitis, & MacLean, 2009). Adults with ADHD are more likely to engage in risky driving and more likely to have driving-related accidents (Barkley, 2004; Murphy & Barkley, 1996). Several studies have also shown that adults with ADHD may demonstrate social skill deficits and have difficulty with interpersonal relationships (Adler & Cohen, 2004; Koemans, van Vroenhoven, Karreman, & Bekker, 2015; Murphy & Barkley, 1996). Finally, a number of comorbid mental health disorders, including depression, anxiety, and substance use disorders, are found to be common among adults diagnosed with ADHD (Adler & Cohen, 2004; Faraone et al., 2006; Miller, Nigg, & Faraone, 2007; Murphy & Barkley, 1996; Wilens et al., 2004; Wilens et al., 2009). Thus, although the symptoms of ADHD in adults may not be as noticeable as in children, the functional consequences remain significant, and may directly (or indirectly) affect multiple domains of everyday life.
ADHD in the Military
In a review of neuropsychological disorders in the military, French, Anderson-Barnes, Ryan, Zazeckis, and Harvey (2012) noted that, “The actual prevalence of ADHD and LD [learning disorders] within the military is not known, and while these conditions are generally disqualifying for military service, they are frequently encountered in military neuropsychological practice” (p. 198). Although ADHD symptoms are anecdotally reported, large-scale screening efforts have not been implemented in the military. This may be partially accounted for by the precarious status of ADHD as a medical condition: According to Army regulation AR 40-501 (U.S. Department of the Army, 2007), a diagnosis of ADHD is a medical disqualifier unless the individual demonstrates “ . . . passing academic performance and there has been no use of medication(s) in the previous 12 months.” (It should be noted that although this regulation is Army-specific, similar requirements exist for all Department of Defense branches.) In contrast to the many studies within the general population, few studies have assessed the rates of ADHD in contemporary military units.
One study reviewed the electronic medical records of Marine and Army personnel who sought care while deployed to Iraq (N = 1,078); the authors reported an overall ADHD rate of 2% though the diagnostic criteria used to classify cases were unspecified (Larson et al., 2011). In addition, by relying on an on-the-record diagnosis, the rates of ADHD in this study are likely an underestimate. Another study by Hanson and colleagues (2012) administered the six-item Adult ADHD Self-Report Scale Screener (ASRS-S) to 260 Army service members prior to a deployment to Iraq. Using composite scoring with a dichotomous cutoff, the positive screen rate for this sample was 10.4%, more than 5 times the rate reported in the Larson et al. (2011) study. This discrepancy in prevalence data indicates the need for additional research into the prevalence and distribution of ADHD within military populations. If the true prevalence of ADHD is in fact closer to the rates reported by Hanson, it would make adult ADHD one of the most prevalent mental disorders in the military.
This study was conducted to identify the estimated prevalence of adult ADHD symptoms in a large, active duty Army sample. As noted, the functional impairments associated with ADHD may not be readily apparent, and the current estimates vary considerably across studies. Establishing accurate benchmarks for mental health disorders is essential if the military wishes to adequately allocate resources and ensure that military personnel are operating at optimal levels. The second goal of this study was to compare two common methods of ADHD classification to (a) detect any taxonomic differences that may exist between the two techniques, and (b) increase the potential comparability of the present study’s findings to other reported rates.
Method
Participants
Participants were 21,449 active duty U.S. Soldiers who completed the Army Study to Assess Risk and Resilience in Servicemembers (STARRS) assessment battery. Data were collected between May 2011 and March 2013 as part of the Army STARRS All Army Survey (AAS). The AAS used stratified probability sampling to identify representative Army units and was designed to be a realistic cross-section of the active duty Army force. Additional details on sampling design and data collection are available elsewhere (Kessler et al., 2013; Ursano et al., 2014).
Measures
Participants completed a battery of experiential and psychological measures including the six-item ASRS-S (Kessler, Adler, Ames, et al., 2005). The full ASRS is an 18-item measure developed as part of the World Health Organization’s Composite International Diagnostic Interview (CIDI). The six-item screener was developed to optimize clinical utility, and has demonstrated psychometric properties equal to or above those of the full ASRS (Kessler, Adler, Ames, et al., 2005). The ASRS-S assesses ADHD symptom frequency using a 5-point Likert-type scale, with responses ranging from 0 (never) to 4 (very often). Four of the screener items (Items 1 through 4) assess symptoms from the ADHD inattentive symptom cluster, and two of the items (Items 5 and 6) draw from the hyperactive/impulsivity symptom cluster. The ASRS has been used in military samples (Hanson et al., 2012) and civilian samples (Kessler et al., 2007) and has demonstrated good reliability and diagnostic utility (Kessler, Adler, Ames, et al., 2005).
Scoring the ASRS-S
To allow greater comparison with other studies, two ASRS-S scoring methods were used. The first (henceforth referred to as composite scoring) involves summing the responses for the six items (range: 0-24) with scores ≥14 indicating a positive ADHD screen. The second method (referred to as item-response scoring) is based off Kessler’s original scoring criteria, which uses item-response cutoffs to establish positive screens (see Table 1 for specific item cutoffs). Although there have been several suggested cutoffs, a cutoff of four or more clinically significant items has been determined as optimal (Kessler, Adler, Ames, et al., 2005). Statistical analyses were done using SPSS v.23 and SAS v.9.4.
World Health Organization ASRS-S.
Note. All items are scored on a 5-point Likert-type response scale ranging from 0 to 4. Cutoffs are based on those recommended in the original ASRS-S paper (Kessler, Adler, Ames, et al., 2005). ASRS-S = Adult ADHD Self-Report Scale Screener; IA = inattentive symptom cluster; HI = hyperactive/impulsivity symptom cluster.
Results
The majority of the sample was male (87.6%; n = 18,790) and under the age of 30 (62.6%; n = 13,359). Of the 21,449 respondents, 21,021 (98.0%) completed all six of the ASRS-S items. There was no significant difference in sex, race/ethnicity, or age in those who completed the ASRS-S and those who did not (only complete cases were used in the analyses). Using the ASRS-S composite scoring method (Hanson et al., 2012), approximately 7.6% of the sample screened positive for adult ADHD. Using the alternative ASRS item-response method (which requires four of the six items to be marked clinically significant; Kessler, Adler, Ames, et al., 2005), approximately 9.0% of the sample had a positive ADHD screen. As would be expected, the tetrachoric correlation between the two scoring methods was high (r = .979; p < .001). Diagnostic concordance between the two scoring methods can be seen in Table 2.
Diagnostic Concordance Table for ASRS-S Scoring Methods.
Note. ASRS-S = Adult Self-Report ADHD Scale Screener.
Logistic regression analyses were conducted to examine positive screening rates as a function of race (Table 3) and age (Table 4). A binary logistic regression on race/ethnicity was run for each classification method. Dummy-coded vectors were created to represent membership in the race/ethnicity categories: Black/African American, Asian, Pacific Islander, Native American/Alaskan Native, and Other; because it was the most prevalent racial classification in the sample, White was used as the reference group. Relative to Whites, individuals endorsing Native American/Alaskan Native exhibited a higher risk of screening positive for ADHD using both the item-response scoring (OR = 1.54, 95% confidence interval [CI] = [1.21, 1.97], p < .001) and the composite scoring (OR = 1.60, 95% CI = [1.23, 2.07], p < .001). Conversely, endorsing Black/African American was found to be a small, but statistically significant protective factor (relative to White), with odds ratios (ORs) less than one for both scoring methods. Participants’ ages were also dummy coded into four groups (20-24, 25-29, 30-39, 40+), with 18- to 19-year-olds as the reference group (following our hypothesis that this group would have the highest ADHD rates). Using the composite scoring method, no significant effect of age was observed. However, using the item-response scoring method, two groups emerged as being at greater risk for ADHD screens, relative to the youngest group: 25-29 (OR = 1.47, 95% CI = [1.10, 1.96], p = .009) and 30-39 (OR =1.35, 95% CI = [1.02-1.81], p = .039). Regressions were also run to examine the effect of sex on ADHD status. Female sex was not found to have a significant effect on ADHD status using composite scoring (OR = 1.14, 95% CI = [0.98, 1.33], p = .08) but was a significant risk-factor using item-response scoring (OR = 1.16, 95% CI = [1.01, 1.33], p = .04). However, because females comprised just 11% of the total sample (n = 2,452), and given the low number of females with positive ADHD screens (208 and 249; less than 1.0% of women for both composite and item-response, respectively), these sex-related findings should be interpreted with caution.
Logistic Regression of Positive ADHD Screens by Classification Method and Race/Ethnicity.
Note. OR = odds ratio; CI = confidence interval.
Omnibus test of significance: χ2 = 17.40 (df = 5), p = .004.
Omnibus test of significance: χ2 = 19.49 (df = 5), p = .002.
p < .05.
Logistic Regression of Positive ADHD Screens by Classification Method and Age-Group.
Note. OR = odds ratio; CI = confidence interval.
Omnibus test of significance: χ2 = 26.19 (df = 4), p < .001.
Omnibus test of significance: χ2 = 25.81 (df = 4), p < .001.
p < .05.
Conclusion
The present study highlights how differing approaches to ADHD assessment within the military results in prevalence rate discrepancies. It is notable that methodological differences within a single outcome measure yielded varying prevalence estimates. In the present analysis, using the composite scoring method resulted in slightly lower ADHD rates than those reported by Hanson and colleagues (7.6% vs. 10.4%, respectively); however, these two rates are more comparable than those presented by Larson and colleagues (2011). This study is consistent with prior work demonstrating higher rates of ADHD in the military compared to demographically similar general population individuals. For comparison, Kessler, Adler, Gruber, et al. (2007) examined a subsample of NCS-R respondents between ages 18 and 44 (n = 3,199), and the rate of probable ADHD was estimated at 4.4% (as determined by the Adult ADHD Clinical Diagnostic Scale; Adler & Spencer, 2004).
The rates of ADHD presented in this article are slightly different from those reported in the Army STARRS article on 30-day prevalence of mental health disorders in the STARRS sample (7.0%; Kessler et al., 2014); however, that study analyzed a subsample of the AAS (n = 5,428). Furthermore, the publically available Army STARRS codebook reports a 6-month ADHD prevalence of 5.8%. This discrepancy is due to the inclusion of six additional ADHD items that are not part of the ASRS-S.
Perhaps the most interesting finding of this study involved the prevalence of ADHD symptoms across age-groups. The distribution of positive ADHD screens across age-groups (see Figure 1) reflects a departure from our initial hypotheses that ADHD symptoms would have an inverse relationship with age. Contrary to studies indicating that adult ADHD symptoms tend to peak in adolescence and taper off with age (Biederman et al., 2010), this study showed that, compared with younger age-groups, the age-group of 25- to 29-year-olds had the greatest estimated prevalence of ADHD symptoms and was almost 1.5 times more likely to have a positive screen than the group of 18- to 19-year-olds (using the item-response method).

Percent of positive ADHD screens as determined by two different scoring methods of the Adult ADHD Self-Report Scale Screener by age-group (N = 21,449).
Although the two scoring methods examined in this study were highly correlated and produced similar prevalence estimates, the statistical significance of several findings hinged on diagnostic method, raising the question of underlying differences in the two methods. Examination of the two scoring methods indicates that the item-response may be a more sensitive scoring method. For example, using the item-response method, one can receive a positive screen by endorsing the minimum clinical rating for four items, which would have a corresponding composite score of nine (cf. the minimum composite cutoff score of 14).
Limitations
Given that ADHD is a potentially disqualifying diagnosis for military service, it is possible that the self-reported rates of ADHD symptoms observed in this study are underestimated. Given that the self-reported rates are already higher than other mental health disorders (e.g., major depressive disorder [4.8%] and generalized anxiety disorder [5.7%]; data not shown), potential underreporting only emphasizes just how pervasive ADHD symptoms may be within the military.
Another limitation of the present study is the reliance on the ASRS-S for assessment of ADHD symptoms. Although the ASRS-S is well-validated and has been previously used to assess military samples, the clinical specificity of this (or any) screener is far less than a longer instrument or a diagnostic interview. As indicated earlier in this paper, all mention of ADHD symptoms refers to symptom-based positive screens, and not clinical diagnoses. The use of more in-depth assessments such as the Adult ADHD Clinical Diagnostic Scale (Adler & Spencer, 2004), the Conners’ Adult ADHD Rating Scale (CAARS; Conners, Erhardt, & Sparrow, 1999), or the Barkley Adult ADHD Rating Scale (BAARS; Barkley, 2011) would be more appropriate in a clinical setting and would provide greater specificity in parsing out comorbid conditions that also affect cognition. That said, the use of a screener in this population does reflect a certain practical functionality: If the military were to implement more widespread screening for ADHD, a longer survey instrument would likely be eschewed in favor of something more efficient (such as the ASRS-S).
Recommendations and Future Directions
Moving forward, it might benefit the military leadership to more fully assess ADHD symptoms in service members of all age-groups, not only the youngest cohorts. It may also be prudent for the Army to examine policies regarding diagnosis and management of ADHD. As it currently stands, soldiers can have a diagnosis of ADHD so long as they have not taken any medication in the past year. Because the first-line treatment for ADHD is pharmacotherapy (Biederman et al., 2006), this creates a seemingly paradoxical situation wherein an individual is permitted to have the disorder, but not to receive treatment to manage the condition. As the findings of this paper indicate, there are a large percentage of individuals in the Army, from all age-groups, who report problems with attention, concentration, and job performance. What remains to be shown is whether or not these individuals are operating at an acceptable, yet suboptimal, level of functioning due to untreated adult ADHD symptoms. Accurate assessment of symptoms and functioning would be a first step toward ensuring optimal performance among service members.
Footnotes
Authors’ Note
The thoughts, opinions, and recommendations presented in this article are those of the authors and do not necessarily represent the views of the Army STARRS Team, the Department of the Army and/ or the Department of Defense.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr. Wickham receives support from the Tobacco-Related Disease Research Prevention (TRDRP) Grant 24RT-0027 funded by the State of California.
