Not yet independently audited
This page was written from verified sources but has not yet passed the independent citation audit. Treat its claims as provisional.
Open questions — attention-deficit/hyperactivity disorder¶
Last curated: 2026-10-19 (dated sweep updates retained)
Stable IDs are retained when questions are sharpened. Tier 1 questions could change practice and are designable now; Tier 2 questions require enabling measurement, longer follow-up or mechanistic development.
Tier 1 — practice-changing and designable now¶
OQ-1 — Which treatment sequence improves function, not only acute symptoms?¶
Why open. Short-term medication ranking is supported by 133 double-blind RCTs, but 26- and 52-week comparative data were insufficient; long-term education, work, relationship and quality-of-life outcomes are sparse (Cortese 2018, PMID 30097390; Elliott 2020, PMID 33085721).
Update 2026-10-12. A new agent widens the sequencing problem. Centanafadine was FDA-approved on 24 July 2026 on placebo-only evidence (pooled g −0.37) (Hu 2026, PMID 42640319; Muneer 2026, PMID 42267239). The first registered randomized test of a personalized medication-choice tool (PETRA, NCT07812948) is a feasibility study with acceptability and recruitment endpoints only.
Update 2026-10-19. The parent-first medication trial now has posted results, but only descriptive child CGI-Severity means and no comparative analysis in its outcome module. Its two arms both include a child-treatment strategy, so it is not a test of parent medication versus no child care. Reconcile the “change” outcome title, CGI-score units and analyzed denominators against the publication before using it to answer sequencing questions (NCT04240756).
Design. Pragmatic sequential multiple-assignment trial stratified by age and comorbidity, with function and quality of life co-primary outcomes over ≥24 months.
OQ-2 — How should persistent, remitted and functionally recovered ADHD be defined?¶
Why open. Follow-up synthesis yielded about 15% full-syndrome versus 65% symptomatic persistence near age 25; the difference is definition-driven (Faraone 2006, PMID 16420712).
Design. Lifespan consensus core set validated against education, work, relationships and service need.
OQ-3 — Can missed girls and women be identified without lowering specificity?¶
Why open. Female recognition is shaped by inattention, internalizing comorbidity, compensation and referral bias, while validated sex-aware thresholds or referral algorithms are lacking (Hinshaw 2022, PMID 34231220; Young 2020, PMID 32787804).
Design. Prospective diagnostic-accuracy study in consecutive referrals with blinded expert reference assessment and long-term outcome validation.
OQ-4 — What is the safest and most effective approach to ADHD during pregnancy and postpartum?¶
Why open. Available literature must separate medication exposure from confounding by indication and untreated impairment; randomized evidence is unlikely for many outcomes (Scoten 2024, PMID 38432409).
Update 2026-10-12. Ontario data show maternal ADHD in 2.5% of pregnancies by 2023 and stimulant use in pregnancy doubling to 20.9% (Egsgaard 2026, PMID 42671284). The largest new obstetric-risk cohort lacked an unmedicated-ADHD comparator, which is exactly the confounding-by-indication control this question requires (Babinski 2026, PMID 42671280).
Update 2026-10-19. A newly retrieved pregnancy/offspring systematic review is catalogued for full-text appraisal (Komargodski 2026, PMID 42887725). Publisher retrieval supplied metadata rather than the article body; no pooled result or answer to this question is adopted from its abstract. The next appraisal should separate drug-specific estimates, exposure timing and ADHD-restricted comparators.
Design. Multinational target-trial emulation with negative controls, dose/timing, maternal function and offspring outcomes.
OQ-5 — Do cardiovascular risks emerge with long exposure or pre-existing disease?¶
Why open. Meta-analysis of 3.93 million participants found no significant overall association, but could not exclude modest arrhythmia-related risk and was imprecise in females and pre-existing cardiovascular disease (Zhang 2022, PMID 36416824).
Update 2026-10-12. Two additions. Cochrane quantifies the sustained haemodynamic shift for amphetamines (+1.93/+1.84 mmHg, +3.71 bpm, high certainty) (Chan 2025, PMID 40152309). A seven-country T2D cohort found a MACE signal confined to early-onset T2D (HR 1.58 for diagnosis <45 y) (Dong 2026, PMID 42806384). The "pre-existing disease" arm of this question now has a first subgroup estimate. The ≥10-year design is still unmet.
Design. Linked-registry active-comparator new-user cohorts with ≥10-year follow-up and validated cardiovascular endpoints.
OQ-6 — Which monitoring schedule detects meaningful growth effects without unnecessary treatment interruption?¶
Why open. Long-term methylphenidate exposure is associated with small mean height and weight z-score changes, with uncertain individual clinical importance (Carucci 2021, PMID 33080250).
Design. Prospective trajectory study using standardized growth, pubertal stage, nutrition and treatment-exposure measures.
OQ-7 — Which controls reduce stimulant misuse and diversion while preserving access?¶
Why open. Meta-analysis estimated past-year misuse at 22.6% and diversion at 18.2% among prescribed populations, but intervention evidence is thin (Forrest 2025, PMID 40698051).
Design. Cluster-randomized comparison of secure storage, smaller quantities, formulation choice and nonpunitive screening.
OQ-8 — Do digital therapeutics improve durable real-world function?¶
Why open. STARS-ADHD improved an objective attention endpoint over four weeks, but school/home transfer and durability were not established (Kollins 2020, PMID 33334505; NCT02674633).
Design. Independent active-comparator trial with blinded ratings, direct school measures, quality of life and 12-month follow-up.
OQ-9 — Which adult CBT components are active and who benefits?¶
Why open. Twenty-eight-study meta-analysis supports core and emotional symptom improvement, but protocols and comparators vary (Liu 2023, PMID 36794797).
Update 2026-10-12. Face-to-face group CBT outlasted therapist-guided internet CBT at 1 year (Pan 2026, PMID 42717581). CBT added nothing to long-term quality of life beyond psychoeducation and medication at 4–8 years (Wettstein 2026, PMID 42664236). Delivery mode and maintenance now look as important as component content.
Design. Component factorial trial with organization, cognitive restructuring, emotion regulation and therapist contact varied prospectively.
OQ-10 — Can medication-associated reductions in mortality and injury be reproduced causally?¶
Why open. Swedish target-trial emulation found 2-year all-cause mortality HR 0.79 and unnatural-cause HR 0.75, while injury meta-analysis found within-person rate ratio 0.76; both remain observational (Li 2024, PMID 38470385; Man 2017, PMID 29255995).
Design. Replicated target-trial emulations across health systems with falsification endpoints and treatment switching.
OQ-11 — What is the best assessment pathway when childhood records are absent?¶
Why open. Adult screening performs well in calibration samples, but diagnosis still depends on developmental history and clinical differentiation; retrospective tools have setting-dependent accuracy (Ustun 2017, PMID 28384801; Brevik 2020, PMID 32285644).
Design. Prospective comparison of structured interview, collateral, retrospective scales and longitudinal response-independent validation.
OQ-12 — Which school interventions produce durable academic attainment?¶
Why open. Behavioral and combined treatment can improve proximal domains, but long-term reading, grades and completion are not consistently sustained randomized outcomes (MTA Cooperative Group 1999, PMID 10591283; Swanson 2008, PMID 18573923).
Update 2026-10-12. The question is sharper. A Norwegian instrumental-variable analysis found no causal effect of medication on school grades for children on the margin of treatment (GPA −0.01; wide interval) (Lyhmann 2026, PMID 42803341). A Dutch >12-year cohort found no difference across 51 adult outcomes by childhood stimulant trajectory (van der Plas 2026, PMID 42664720). The causal case that medication improves attainment is weaker than before, but the null is imprecise, not demonstrated.
Design. School-cluster trial linking direct classroom outcomes to multi-year attainment records.
OQ-25 — Where does centanafadine belong relative to stimulants and atomoxetine?¶
Why open. Approved by the FDA on 24 July 2026 for ages ≥6 with abuse and pediatric-suicidality boxed warnings and scheduling pending (Hu 2026, PMID 42640319), on placebo-controlled trials with small-to-moderate effects (adult effect sizes −0.24 to −0.40; pooled g −0.37) (Adler 2022, PMID 35652746; Muneer 2026, PMID 42267239). Of 36 PubMed records searched on 2026-10-12, none is a head-to-head RCT. Sources also disagree on whether it should be classed as a stimulant (Ward 2026, PMID 42779129).
Update 2026-10-19. OIRA now records intended Schedule III placement in a DEA interim final rule received 14 October, still pending review. Track the final effective rule before treating scheduling as settled (OIRA/Reginfo.gov — “Schedules of Controlled Substances: Placement of Centanafadine in Schedule III”, https://www.reginfo.gov/public/do/eo/neweomeeting?rin=1117-ZA11, accessed 2026-10-19).
Design. Active-comparator RCT versus methylphenidate/lisdexamfetamine and atomoxetine with function, tolerability and misuse-liability endpoints; post-marketing abuse surveillance after scheduling.
OQ-26 — Does sustained ADHD treatment causally reduce adverse adolescent sexual and reproductive outcomes?¶
Why open. A Swedish register study associates longer medication duration with lower odds of recorded STI or unwanted-pregnancy/abortion contacts, but the untreated group also has lower odds than the short-exposure reference. Full-text Table 4 and the stated inability to adjust for ADHD severity limit a causal interpretation; differences in support, treatment persistence and healthcare contact may matter (Kopp Kallner 2026, PMID 42864120).
Design. Replicated active-comparator, new-user target-trial emulations with time-varying exposure, symptom severity, contraceptive care, family support and healthcare-contact intensity; retain separate STI and pregnancy endpoints and test transport across health systems. SciELO and LILACS could not be searched (HTTP 403 access limits), so this sweep makes no claim that regional evidence is absent.
Tier 2 — enabling science and longer-horizon questions¶
OQ-13 — Is adult-onset ADHD a distinct pathway, missed childhood disorder or mimic mixture?¶
The Dunedin cohort found 90% of study-defined adult cases lacked childhood ADHD, but adult classification omitted onset and cross-setting corroboration by design (Moffitt 2015, PMID 25998281). Update 2026-10-12: In All of Us, 71.5% of adults with ADHD were adult-diagnosed. They had fewer comorbidities and lower polygenic scores than youth-diagnosed adults, yet still differed clearly from controls (Leffa 2026, PMID 42747853). This favours "milder, later-recognized" over "mimic mixture" for many cases, but self-reported diagnosis leaves true adult onset untested.
OQ-14 — Can cognitive heterogeneity become treatment-relevant?¶
Distinct executive profiles occur in pediatric ADHD, but no reliable cognitive taxonomy guides treatment assignment (Kofler 2019, PMID 29705926).
OQ-15 — Can polygenic scores add useful prediction across ancestries?¶
GWAS identified 12 loci and a polygenic architecture, but individual classification and ancestry transport remain inadequate (Demontis 2019, PMID 30478444; Faraone 2019, PMID 29892054).
OQ-16 — Which imaging or EEG findings are causes, consequences or treatment adaptations?¶
Group-level network findings are reproducible enough to motivate mechanisms but not individual diagnosis (Posner 2020, PMID 31982036; Peterson 2024, PMID 38523599).
OQ-17 — Can any biomarker add value beyond clinical assessment?¶
A 231-study review found heterogeneous, generally low-strength diagnostic evidence across neuropsychology, EEG, imaging and biospecimens (Peterson 2024, PMID 38523599). A 2026 multicentre ocular-classifier study added adult external testing but did not evaluate prospective decision impact or outcomes from biomarker-guided care (Ranathunga 2026, PMID 42448769). Update 2026-10-12: Normative structural-connectivity deviation predicted 12-week atomoxetine but not methylphenidate response, and tracked within-person improvement (Xu 2026, PMID 42675180). This is the first differential-response signal of its kind in this knowledge base. No marker-guided randomized allocation exists yet.
OQ-18 — How do menstrual cycle and menopause alter within-person symptoms and treatment response?¶
Female lifespan consensus identifies reproductive transitions as important while direct prospective evidence remains limited (Young 2020, PMID 32787804; Hinshaw 2022, PMID 34231220). Update 2026-10-12: A registered observational pilot will collect estradiol, wearable sleep and EMA across cycles in 30 girls aged 11–15 (NCT07805382). It is feasibility work, not the within-person randomized titration that D7 requires.
OQ-19 — Which environmental associations are causal?¶
Umbrella review found multiple prenatal, perinatal and environmental candidates but variable evidence quality and confounding (Kim 2020, PMID 33069318). Update 2026-10-12: For prenatal paracetamol, an umbrella review found the ADHD association (unrestricted RR 1.29) attenuated toward null under quality restriction (Tayebi-Hillali 2026, PMID 42773841). A Danish triangulation study found null active-comparator estimates and a paternal negative-control signal consistent with familial confounding (Gram 2026, PMID 42660286). The question narrows for this exposure; other candidates are unchanged.
OQ-20 — How should emotional dysregulation enter diagnosis and outcome measurement?¶
Emotional dysregulation is common and impairing but overlaps multiple disorders and is not a core criterion (Faraone 2019, PMID 29624671).
OQ-21 — What changes the course in adults older than 60?¶
Symptom stability is reported in older adults, but diagnostic validity, cognitive differential and treatment safety evidence remain sparse (Semeijn 2016, PMID 26068984). Update 2026-10-12: EHR-recorded ADHD was associated with incident MCI/dementia at age ≥50 (HR 4.60), with a concordant polygenic-score association (Bauer-Negrini 2026, PMID 42842281). Whether this is causal, shared liability or diagnostic overlap is now the central sub-question.
OQ-22 — Which aspects of stigma causally delay care or worsen outcome?¶
Community-attitude and adult-stigma reviews document the problem, but tested anti-stigma mechanisms are limited (Bisset 2022, PMID 33769111; Krishnamoorthy 2026, PMID 42137527).
OQ-23 — Can diagnostic prevalence and service demand be reconciled?¶
High-quality post-2020 evidence did not show a clear prevalence rise despite sharp service-demand pressure; only 4 of 40 studies were low risk of bias (Martin 2025, PMID 40393551). Update 2026-10-12: US parent-reported diagnosis rose to 12.0% (NHIS 2023–24) and 11.6% (NSCH 2024) while severe ADHD stayed at 1.3–1.8% (Zablotsky 2026, PMID 42767652; Li 2026, PMID 42805290). In Denmark, prediagnostic risk-factor contrasts attenuated toward null for ADHD from 2012 to 2022 (Tarp 2026, PMID 42714883). This is converging evidence that rising diagnosis reflects ascertainment more than liability.
OQ-24 — Which accommodations change participation across education and employment?¶
School and college evidence supports structured support, but long-term participation outcomes and adult workplace trials are limited (Moore 2019, PMID 25755258; Lefler 2016, PMID 26825556).
Dots not yet connected¶
Fourteen junction-specific PubMed searches rerun on 2026-08-30 found the component literatures shown below but no published record with the exact joining design. Each row is therefore a dated, database-bounded evidence gap—not a claim that no unpublished, unindexed or differently described study exists.
| # | Dot A | Dot B | The missing junction | Powers |
|---|---|---|---|---|
| D1 | Female under-recognition (PMID 34231220) | Diagnostic-tool heterogeneity (PMID 38523599) | Sex-aware tool validated in consecutive referred patients | OQ-3 |
| D2 | Polygenic liability (PMID 30478444) | Cognitive heterogeneity (PMID 29705926) | Ancestry-robust genetic prediction of replicable cognitive profiles | OQ-14, OQ-15 |
| D3 | Acute medication efficacy (PMID 30097390) | Mortality association (PMID 38470385) | Mediation from symptom change through behavior to unnatural-cause mortality | OQ-1, OQ-10 |
| D4 | Digital attention target engagement (PMID 33334505) | School accommodations (PMID 25755258) | Trial of digital training embedded in classroom support with attainment outcomes | OQ-8, OQ-12 |
| D5 | Growth signal (PMID 33080250) | Individual treatment response | Prospective model balancing symptom gain against growth trajectory | OQ-6 |
| D6 | Cardiovascular parameter change (PMID 40203844) | Rare cardiovascular events (PMID 36416824) | Longitudinal link between early BP/pulse response and later events | OQ-5 |
| D7 | Menstrual/reproductive context (PMID 32787804) | Medication comparative efficacy (PMID 30097390) | Within-person cycle-stage randomized titration study | OQ-18 |
| D8 | Adult-onset controversy (PMID 25998281) | Screening accuracy (PMID 28384801) | Prospective screened cohort with childhood-record linkage | OQ-11, OQ-13 |
| D9 | Stimulant diversion prevalence (PMID 40698051) | Access disparities (PMID 35959536) | Equity analysis of diversion controls and treatment access | OQ-7 |
| D10 | CBT emotional benefit (PMID 36794797) | Emotional dysregulation phenotype (PMID 29624671) | Component trial using emotional dysregulation as stratifier and mediator | OQ-9, OQ-20 |
| D11 | Injury reduction association (PMID 29255995) | Cognitive timing/reaction variability | Mechanistic study linking within-person cognitive change to real-world injury | OQ-10, OQ-14 |
| D12 | Public stigma (PMID 33769111) | Diagnostic delay (PMID 23692803) | Natural experiment testing whether stigma reduction shortens pathways | OQ-22 |
| D13 | Symptomatic adult prevalence 6.76% (PMID 33692893) | Persistent adult prevalence 2.58% (same source) | Outcome validation of onset-defined versus symptom-defined adult groups | OQ-2, OQ-13 |
| D14 | Patient-priority function (PMID 30124180) | Guideline recommendations (PMID 37254562) | Guideline implementation trial with patient-priority co-primary outcomes | OQ-1 |
2026-10-19 regional-coverage limit. Europe PMC and the legacy EU Clinical Trials Register were queried. WHO ICTRP returned HTTP 403 and ISRCTN served a JavaScript/cookie challenge rather than results. CENTRAL, SciELO and LILACS could not be searched (HTTP 403 access limits documented in SOURCES.md). No cross-region evidence-absence claim is made, including for polygenic-score transport in OQ-15.