ETD

• attention-deficit/hyperactivity disorder
• bipolar disorder
• comorbidity
• cyclothymia
• emotional dysregulation
• HCL-32
• hypomania
• screening
• TEMPS-M

Background: Attention-deficit/hyperactivity disorder (ADHD) and bipolar disorders (BDs) are among the most prevalent and clinically impairing psychiatric conditions in adulthood, with comorbidity rates substantially exceeding chance. The diagnostic differentiation between the two conditions is complicated by wide phenotypic overlap in the domains of emotional dysregulation, impulsivity, affective instability, and hyperactivity/hyperarousal. This intersection is particularly problematic for state-based screening instruments such as the Hypomania Checklist-32 (HCL-32), whose items capture behavioral and cognitive features of hypomania largely overlapping with ADHD symptomatology. A structurally distinct approach to examine bipolarity involves affective temperament assessment, specifically the cyclothymic subscale of the Temperament Evaluation of Memphis, Pisa, Paris and San Diego (TEMPS), which measures stable, affective dispositions rather than retrospective episodic behavioral activation. The accurate identification of cyclothymic instability and hypomanic features in ADHD patients also has direct pharmacological implications, given that stimulant treatment without adequate mood stabilizer coverage carries risk of mood destabilization in patients with underlying significative bipolarity.
Aims: The present study aimed to characterize the clinical and sociodemographic profile of adult ADHD outpatients stratified by bipolar spectrum comorbidity and to compare the discriminative performance of the HCL-32 and the TEMPS-M cyclothymic subscale (TEMPS-C), in distinguishing ADHD-only patients from those with bipolar spectrum comorbidity.
Methods: This study was conducted at the neurodevelopmental outpatient service of Psychiatry Unit 2, University Hospital of Pisa, Italy. Adults consecutively referred for ADHD evaluation between June 2020 and December 2025 were eligible if aged 18 or older with a confirmed DSM-5-TR ADHD diagnosis; patients with comorbid major depressive disorder, current psychosis, severe intellectual disability, or insufficient Italian language proficiency were excluded. ADHD diagnosis was established through the Diagnostic Interview for ADHD in Adults (DIVA 2.0 or DIVA-5). All participants completed a standardized battery comprising: the Hypomania Checklist-32 (HCL-32) and the TEMPS-M. Observer-rated ADHD severity was assessed using the Conners' Adult ADHD Rating Scales - Observer: Screening Version (CAARS-O:SV). The final descriptive analytic sample comprised 179 participants (ADHD-only n=48, ADHD+CYC n=70, ADHD+BD n=61); psychometric comparisons were conducted on 164 participants with complete scale data. Receiver operating characteristic (ROC) analyses were used to evaluate and compare the discriminative performance of the scales; binary logistic regression with stepwise backward elimination was applied to identify items with the highest independent discriminative value.
Results: The sample had a mean age of 25.7 ± 7.7 years; the ADHD+BD group was significantly older than ADHD-only (27.8 vs. 23.5 years; p=.008). Sex distribution differed markedly across groups (p<.001): ADHD-only was predominantly male (77%), ADHD+CYC showed female preponderance (63%), and ADHD+BD an intermediate distribution (39% female). Mood comorbidity groups showed consistently higher rates of family history of mood disorders, comorbid substance use disorders, eating disorders, and panic disorder, revealing a gradient of clinical complexity increasing from ADHD-only to ADHD+BD.
The TEMPS-C yielded the largest between-group effect size of the entire battery (ε²=0.183; p<.001), approximately seven times that of the HCL-32 total score (ε²=0.026; p=.045). HCL-32 Factor 1 (active/elated) failed to differentiate groups (p=.151), and no individual HCL-32 item reached significance. ROC analyses confirmed TEMPS-C superiority (AUC=0.802, 95% CI 0.736-0.869) over the HCL-32 total score (AUC=0.627) and Factor 2 (AUC=0.656; both p<.001). At the optimal cut-off of ≥20, the TEMPS-C achieved sensitivity of 80.0%, specificity of 70.9%, PPV of 88.2%, and accuracy of 77.6%. Logistic regression identified two items with independent discriminative value, but the derived two-item screener yielded an AUC of only 0.623, significantly inferior to the full scale (p=.003).
Conclusions: The TEMPS-M cyclothymic subscale substantially outperforms the HCL-32 in discriminating adult ADHD patients with bipolar spectrum comorbidity from those without. This superiority reflects a fundamental difference in the constructs measured. While the HCL-32 retrospectively captures behavioral activation extensively overlapping with ADHD features, the TEMPS-C measures a stable cyclothymic disposition that is specifically elevated in the presence of bipolar spectrum comorbidity. The two-item TEMPS-C Screener failed to replicate the discriminative performance of the full scale, suggesting that the predictive validity of the TEMPS-C arises from the aggregate contribution of all its items rather than from any single dominant component, and cannot be preserved through item reduction. Collectively, these results support the routine use of the TEMPS-C as a valid screening tool in the clinical assessment of adult ADHD outpatients and highlight the importance of systematically evaluating bipolar spectrum comorbidity in this population to guide accurate diagnosis and treatment stratification.