Attention deficit hyperactivity disorder (ADHD) is a chronic condition affecting an estimated 2%–7% of the adult population, and is associated with significant impairments in behavioral, cognitive, and social functioning (1–5). Limited access to specialized medical professionals for ADHD assessment and management worldwide has raised concerns about underdiagnosis and undertreatment, and screening tools have been developed to facilitate diagnosis in primary care (6–8).

One such screening tool is the Adult ADHD Self-Report Scale (ASRS) that was developed in 2005 in collaboration with the World Health Organization (8). The ASRS is now one of the most widely used ADHD screening tools for adults in primary care due to its brevity, simplicity, and comprehensiveness (9, 10). The ASRS has a high negative predictive value (+95%), meaning a negative screen strongly suggests the absence of ADHD (11). However, its positive predictive value is relatively low, estimated between 12%–22%, suggesting that a positive screen result is more likely to be a false alarm, with only a minority of individuals who screen positive actually having ADHD (11, 12).

The ASRS consists of 18 questions aligned with DSM-IV criteria, with responses considered indicative of ADHD symptoms (i.e., qualifying responses) ranging from “sometimes” to “often” depending on the question (8). To fit within the time constraints of primary care, the ASRS was revised to divide questions into Part A, comprising the six most sensitive and specific items, and Part B, containing the remaining 12 items for additional context (13). A positive screen is typically defined as at least 4 of the 6 Part A items meeting their respective thresholds (13). For ease of scoring, the standard ASRS format shades the qualifying response options and groups Part A and Part B items under separate headings (14).

Substantial research on survey design has found that the layout (e.g., question grouping) and visual design (e.g., color) of a survey can influence the validity and reliability of its results. For example, satisficing, when respondents choose an answer that is acceptable rather than optimal, tends to increase when similar questions are grouped together (15). Additionally, the use of different color hues between response options has been found to increase the selection of more extreme responses, especially when response options do not have a label (16).

The aim of this study was to evaluate whether these design features of the ASRS—shading and question grouping—influence the rate of positive screening results. Our hypothesis was that removing these features may lower the positive screen rate and potentially improve the positive predictive value of the ASRS.

This is a 2x2 factorial design randomized controlled trial. The trial was registered on clincaltrials.gov July 30, 2024 (https://clinicaltrials.gov/study/NCT06530758). It was approved by the University of Alberta Health Research Ethics Board (Pro00144202). This trial is reported using the CONSORT 2025 reporting guidelines. (17).

Figure 1 presents the participant flow. Of the 1,976 eligible individuals, 1,254 were not offered the survey due to staff availability, 40 had completed it previously, 87 declined, and 595 completed it. Although age and gender/sex differed significantly between groups, the differences were not practically meaningful: mean age ranged from 40 to 45 years, and the proportion of women ranged from 77% to 78% (Supplementary Material 3). There was variation in the percentage of patients offered the survey across days of the week, but this difference was not statistically significant (Supplementary Material 4). Additionally, in total 45% of surveys (IQR 27%-57%) were completed in the morning and the remaining in the afternoon.

Table 2 provides the participant characteristics. No statistically significant differences were found between participants of each survey version. The mean age across groups ranged from 39 to 41 years. Overall, 79% of participants identified as women; 54% identified as White and 33% as Asian. Additionally, 85% reported having completed at least some post-secondary education. Approximately 81% of participants self-reported having a family member diagnosed with ADHD, 14% reported being diagnosed with ADHD and 23% suspected they had ADHD.

Table 3 shows the percentage of screen positives for each version of the ASRS. The screen-positive rate across all participants was 32%, with the highest screening positive rates in participants who had been diagnosed or who suspected they had ADHD. Additionally with diagnosed participants excluded, the screen positive rate was 24% (120/510 participants).

Table 4 provides the logistic regression results. Variables associated with the screen positive rate were previous diagnoses of ADHD (OR 47.4, 95% CI 23.1–97.0, p < 0.0001), suspected ADHD (OR 16.2, 95% CI 9.7–27.2, p = 0.0009), and age (OR 0.97, 95% CI 0.96-0.99, p – 0.01). ASRS survey versions were not statistically significant predictors of ADHD positive screening (p ≥ 0.05). The results were similar for the other three other scoring methods except for ‘Part A and B, Dichotomous scoring,’ not shading reduced the odds of screen positive (OR 0.72, 95% CI 0.54–0.96) and for ‘Part A, Scaled scoring,’ not grouping increased the odds of screen positive (1.35, 95% CI 1.03–1.76) (Supplementary Material 5).

Table 5 provides the characteristics of participants diagnosed with ADHD. In total, 85 participants who completed the survey self-reported an ADHD diagnosis. Participants were diagnosed at a median age of 27 (IQR 20-35); had been diagnosed for a median of 3 years (IQR 0–8); and over half (52%) were currently taking ADHD medications.

To our knowledge, this is the first study published to examine whether the format of the ASRS increases the likelihood of a positive ADHD screen. Although no shading (grouping only) tended to decrease the positive rate and no grouping (shading only) tended to increase it—with both effects reaching statistical significance in other scoring methods—neither effect was statistically significant when using the standard scoring method. These findings suggest primary care providers should continue using the ASRS version with grouping and shading, especially since it is faster to score.

These results differ from our original hypothesis, which predicted that eliminating shading and grouping would reduce the screen-positive rate. The trend toward increased positive screens not grouping may reflect satisficing behavior, when all 18 items on the ASRS appear as one group, respondents may be more likely to provide quick, adequate answers rather than optimal ones (15). Not shading, meanwhile, showed a non-significant trend toward reducing positive screens. This is consistent with prior research suggesting that color cues are less influential when each response option is clearly labeled, as is the case with the ASRS (16). This may also explain why not shading with dichotomous scoring for Parts A and B was associated with a statistically significant reduction in the odds of a positive screen, it may have made it harder for respondents to associate the header labels with the individual questions further down the ASRS.

The screen-positive rate of the ASRS in this trial was substantially higher than the estimated population prevalence, of 2-7% (2–5). The rate was 32% when all participants were included and 24% when those with a prior ADHD diagnosis were excluded. Although the patient population at Shifa is not fully representative of Canadian adults (55% vs. 70% Caucasian, 79% vs. 50% women, and 85% vs. 58% with post-secondary education, respectively), the results may be reflective of the broader Canadian population as screen positive rate aligns with findings from similar studies conducted in the UK (26%) and the US (17%) (12, 20–22). Further, it is unlikely that the rate is inflated due to selection bias. Even if all 87 patients who declined participation had screened negative, the positive rate would remain approximately 28%. Furthermore, the 1,254 patients who were not offered the survey were excluded due to staff time limitations rather than the likelihood of having ADHD, and the observed differences in gender/sex and age between those offered and not offered the survey were not practically meaningful.

This high screen positive rate highlights the importance of conducting a comprehensive ADHD evaluation in addition to the ASRS. Such an assessment should involve a detailed clinical interview and collateral reports from informants to evaluate symptoms across the lifespan, assess functional impairments, and rule out alternative diagnoses (23).

However, the feasibility of conducting comprehensive assessments within primary care is questionable. First, the time required for a proper evaluation—often estimated at a minimum of two hours—far exceeds the typical 15-minute appointment slot available in Canadian family practice, where providers are already facing significant time constraints (24–26). Second, many primary care providers may lack the necessary expertise or confidence to carry out these assessments (8, 24, 27). According to the recently published Adult ADHD Assessment Quality Assurance Standard, clinicians need formal training in ADHD, supervised experience assessing at least 10–20 ADHD cases, and a minimum of one year’s experience working in a mental health setting to be considered qualified (24).

This places primary care providers in a precarious position. There is increasing demand and pressure to perform ADHD assessments, but they do not have the time, expertise, or tools to properly conduct the assessments (6). Indeed, in Canada, diagnosing and managing ADHD has increasingly shifted to family physicians over the past decade, resulting in more than half of all ADHD assessments now being conducted in primary care, and this trend was also seen in this study (7, 28). This raises the risk of overdiagnosis, which was observed in this trial (18% of patients had a diagnoses of ADHD compared to 2-7% prevalence in the adult population). This can have serious consequences, including inappropriate treatment, medication misuse, adverse drug effects, the psychological burden of an ADHD label, and the misallocation of healthcare resources (29, 30).

It would be useful to have screening tools in primary care that can better distinguish between ADHD and other related conditions and identify malingering, while still being feasible in the busy clinical setting. One potential approach is to adapt existing tools for this purpose. In the case of the ASRS, modifications such as removing shading for positive questions, not grouping Part A and Part B questions, and using different scoring methods did not appear to improve the effectiveness. However, other strategies may prove beneficial—such as revising the scoring criteria, raising the threshold for a positive result, incorporating questions to screen for comorbid conditions and malingering, or combining the ASRS with other tools that address these areas. For instance, the ADHD Symptom Infrequency Scale (ASIS) includes embedded items designed to detect malingering (18).

This study has several limitations. Although we obtained some demographic and diagnostic data from individuals who self-reported an ADHD diagnosis (Table 5), we do not have sufficient information to assess the accuracy of these diagnoses. We had attempted to gather additional details on diagnostic methods through a question asking participants how they were diagnosed (e.g., clinical interview, screening tool, or cognitive assessment); however, participants may not have accurately understood the terminology, so we have not reported this data. Lastly, our findings may not generalize to rural populations, as planned recruitment from two rural clinics was not completed due to staffing changes.

Shading screen-positive responses and grouping the key questions in Part A does not appear to impact the screen positive rate. However, the high observed screen-positive rate is a concern in the context of limited time and expertise of primary care providers. Additional screening tools would be useful to help primary care providers better identify patients who have ADHD.