Sensory regulation

Which sensory approaches can be identified, what outcomes have actually been tested, and what may be said without merging ordinary access changes, sensory-based strategies, equipment and formal Ayres Sensory Integration® (ASI)?

The September 2026 review checks the material claims against 12 publications: six original trials, five original reviews and one UK professional document. Reviews overlap the trials and each other; their sample counts cannot be added as independent confirmations. Searches were targeted around the inherited claims, component identity, null results and burdens, not exhaustive or preregistered.

The complete working article and structured records (internal editorial record) retain the precise source, claim, comparison and access limits. Public-source citations below use the same numbered sequence as A007-S01–S12.

Sensory regulation is a practice area with several branches. Royal College of Occupational Therapists (RCOT)'s 2026 terminology separates participation-focused approaches from sensory-based interventions and formal ASI. Professional guidance supports individual priorities, environmental fit and competent practice; it does not establish the effect of every strategy.¹

A defensible account is therefore component- and outcome-specific. Evidence for a clinic-based sensory-motor intervention is not evidence for an equipment box, and a negative trial of one blanket does not settle every sensory adaptation. Ordinary access, voluntary movement and comfort should not be made conditional on demonstrating a medical treatment effect. That is this article's editorial boundary, not a trial finding.

Formal sensory integration: positive goals and null broad outcomes

Schaaf's small trial reported positive individual-goal and caregiver-assistance outcomes following manualised occupational therapy using sensory integration (OT/SI). Its selected population, parent contribution to outcome scoring and short endpoint limit broader interpretation. Pfeiffer's pilot also reported goal differences, with a time-matched fine-motor comparator; small size, completer selection and multiple measures constrain inference. Neither trial establishes that any activity labelled sensory produces the same effect.³⁴

SenITA supplies a larger UK pragmatic comparison. Its adjusted primary model included 106 children: the six-month carer-rated irritability difference was +0.40 points (95% confidence interval (CI) −2.33 to 3.14), with no treatment advantage. The goal ratings were collected in only an intervention subset, without control goal data. Positive within-arm goal change cannot answer whether therapy outperformed usual care on goals. The Vineland administration error and isolated negative communication result also make “every outcome was null” inaccurate. The report's economic conclusion applies to its evaluated delivery and population.²

These studies use different endpoints and delivery contexts. A goal-based positive result is not a broad treatment-effect guarantee; a null irritability result does not prove that every possible functional goal is unaffected. Neither justifies universal referral or dismissal.

Sensory strategies, equipment and movement

The 2025 sensory-based intervention (SBI) review includes diverse components, designs and outcomes. Its favourable deep-pressure and caregiver categories contain null as well as positive component findings. Alternative-seating studies did not demonstrate the proposed attention-related advantages, while the review found inconsistent delivery descriptions across the wider corpus. Its author-assigned grades are not Atlas ratings.⁷

The weighted-blanket trials illustrate why population and outcome matter. Gringras found no sleep advantage in autistic children despite preference for the heavier blanket. Lönn found small average differences in some actigraphic sleep measures in a selected attention-deficit/hyperactivity disorder (ADHD) sample, but not sleep-onset latency; reports of panic, anxiety and pain matter alongside adherence. This is not evidence that weighted items broadly calm children, improve classroom attention, or replace medicines.⁸¹²

Movement can be an available, chosen activity without a claim that it is ASI or a specific treatment. No universal movement-break, sensory-circuit or sensory-diet effect or dose is established in this reconstruction. If a proposed movement programme becomes a treatment claim, it needs its own precise protocol, comparison and evidence record within the existing A007 scope.

Environmental access

The dental crossover study tested a combined clinical environment, including optional pressure input. It found lower skin conductance and selected coded behaviours; nonspecific skin response frequency, several global behaviour scales and child-rated pain/discomfort did not all improve. The package cannot identify which ingredient mattered or certify transfer to classrooms. Requested discontinuation of components was part of its protocol, not evidence that every child enjoyed them.¹¹

The SBI review's single eligible environment study reflects its search and inclusion rules, not proof that the dental randomised controlled trial (RCT) or all other environmental research does not exist.⁷ Guidance and bounded experimental findings have distinct roles: an access rationale does not need to be inflated into evidence of neurological remediation.¹

Acuña's RCT review supports individual goals while rejecting a broad behaviour-of-concern claim; only its abstract was opened, so its full study-level membership and grading have not been certified here.⁶

Piller's 2026 ASI review combines four randomised controlled trials (RCTs) with nonrandomised and single-subject designs. Its occupational-performance synthesis includes SenITA's within-intervention change on the Canadian Occupational Performance Measure (COPM), which rates performance and satisfaction in chosen activities. Calling the parent publication an RCT does not make this particular outcome comparative. Some included programmes also add education or DIRFloortime-related components (a developmental, individual-differences and relationship-based approach), limiting attribution to ASI alone. The review itself acknowledges measurement problems with Goal Attainment Scaling (GAS), which rates progress towards individual goals.⁹

Kishida's 2026 review uses revised fidelity eligibility and explicitly excludes Schaaf 2014 and Pfeiffer 2011 for the earlier measure. Its main search was December 2022, with a 2023 citation update. Its critical conclusion is useful, but it is not a fresh same-corpus overturning of every positive review. Its concern about unverified mechanisms is kept separate from whether a particular functional outcome was validly measured.⁵

Park and Kim's 23-RCT meta-analysis reports positive pooled motor, daily-function and goal outcomes, with other pooled outcomes not significant. The opened abstract does not allow full checking of the intervention/fidelity matrix. Indexed publisher methods suggest a wider sensory integration therapy (SIT) scope; that remains an explicit transfer warning, not a verified claim that all 23 trials tested the same ASI protocol. No dosage recommendation or blanket efficacy statement is adopted.¹⁰

Review publication year, professional authority and the word “strong” cannot replace inspection of design, outcome and comparison. This reconstruction deliberately assigns no formal certainty grade or inherited headline rating.

Study doses below describe research exposure, not instructions for treatment. “Not established” marks a reporting/access limit, not evidence of absence. Further locator details appear in the structured register.

A007-S01 — Royal College of Occupational Therapists

• Identity/version: ASI; sensory-based interventions; participation-focused sensory approaches
• Design: Informed View; not a systematic review
• Population/setting: UK children/young people; no recruited sample
• Intended and actual delivery: Individualised occupation-focused planning; no universal dose
• Comparison/other support: Not comparative research
• Outcome/reporter/time: Meaningful occupational participation is the recommended planning focus, not a demonstrated effect estimate
• Fidelity/adaptation: Competence and appropriate training; generic sensory language does not establish ASI
• Burden/unwanted effects: Discuss time away from school, family effort, costs and alternatives
• Access and role: Full official four-page document; Professional guidance; identity and practice context.
• Limits: Professional position cannot establish treatment efficacy; evidence coverage January 2015–December 2025 excludes pilots.¹

A007-S02 — Randell et al.

• Identity/version: Manualised SIT delivered against ASI fidelity requirements
• Design: Pragmatic individually randomised trial; 69/69 allocated
• Population/setting: 138 children aged 4–11; autism and sensory difficulties; mainstream primary education; South Wales/South England
• Intended and actual delivery: 26 planned contacts over 26 weeks: 20 intensive face-to-face hours, four further face-to-face hours, two telephone follow-ups; median 20 contacts
• Comparison/other support: Usual care, including variable advice/other sensory strategies; little SIT contamination
• Outcome/reporter/time: Six-month carer Aberrant Behavior Checklist irritability subscale (ABC-I): adjusted difference +0.40 points, 95% CI −2.33 to 3.14, p=.77; primary model n106, outcome available n107. No overall ABC-I difference at 12 months. COPM improvement in intervention subset only, without control goal data. Vineland communication negative at 12 months, with administration error limiting interpretation
• Fidelity/adaptation: Structural fidelity met; 10/12 rated therapists consistent; 4/16 lacked recordings. Timing/receipt sensitivity analyses did not change primary conclusion
• Burden/unwanted effects: Travel and staff workload; no adverse effects reported; economic evaluation did not support cost-effectiveness in this setting
• Access and role: Full public report: methods, quantitative results and discussion opened; Original comparative trial and embedded process/economic evaluation.
• Limits: Attrition; Vineland ceiling error; exploratory subgroups; proxy outcomes; no controlled goal inference; limited transfer to other diagnoses/settings.²

A007-S03 — Schaaf et al.

• Identity/version: Manualised occupational therapy using sensory integration; 2007 fidelity measure
• Design: Randomised trial
• Population/setting: 32 autistic children aged 4–7 years, New Jersey; cognitive eligibility generally >65; one noted exception
• Intended and actual delivery: 30 one-hour sessions across 10 weeks, delivered by trained occupational therapists
• Comparison/other support: Usual care; existing services continued and logged
• Outcome/reporter/time: Post-treatment GAS favoured treatment; goal analysis 17/14. Caregiver-assistance self-care/social results also positive; do not infer general or sustained benefit
• Fidelity/adaptation: Blind independent evaluators interviewed parents; parents knew treatment but not scaled goal descriptors. Sampled sessions mean fidelity 90.1/100; low scores triggered consultation
• Burden/unwanted effects: Thirty clinic visits plus continuing support; systematic harms findings not established here
• Access and role: Full public article; Original comparative trial.
• Limits: Small convenience sample; parent-report contribution; restricted population; no longer-term endpoint; mechanism not demonstrated.³

A007-S04 — Pfeiffer et al.

• Identity/version: sensory integration (SI) using the 2007 fidelity measure
• Design: Randomised pilot
• Population/setting: 37 completers aged 6–12 in a US therapeutic summer programme; 20 SI, 17 fine-motor; four additional children withdrew before intervention
• Intended and actual delivery: 18 sessions of 45 minutes over six weeks; one child received 17
• Comparison/other support: Time-matched fine-motor intervention; other services present
• Outcome/reporter/time: Post-treatment goal ratings favoured SI; one Social Responsiveness Scale (SRS) mannerisms subscale favoured SI; other tested differences were not significant
• Fidelity/adaptation: At least one recorded session/child assessed; observed SI sessions met threshold. Parents/research evaluators described as blinded; treating-therapist goal input also recorded
• Burden/unwanted effects: Withdrawals included two children asked to leave programme before interventions; treatment harms not systematically established
• Access and role: Full original article PDF on public professional host; PMC challenge not bypassed; Original comparative pilot trial.
• Limits: Small selected completer sample; brief follow-up; multiple outcomes; goals partly proxy/therapist reported. Historical behaviour-normalisation endpoint is not adopted as an Atlas goal.⁴

A007-S05 — Kishida, Stephenson, Kemp and Salisbury

• Identity/version: Interventions adhering to revised 2011 ASI fidelity criteria
• Design: Systematic review; main search December 2022, citation update August 2023
• Population/setting: 10 studies; children under 10; four studies rated low risk of bias
• Intended and actual delivery: Varied; not one dose or programme
• Comparison/other support: Varied; not a new independent trial
• Outcome/reporter/time: Authors judged positive effects limited; low-risk studies mixed, positive GAS studies had measurement-quality concerns
• Fidelity/adaptation: Explicitly excludes Schaaf 2014/Pfeiffer 2011 for older fidelity measure; appraisal includes revised fidelity and GAS administration
• Burden/unwanted effects: Incomplete primary reporting; not a safety synthesis
• Access and role: Full public article; Original systematic review; methodological challenge.
• Limits: Older search than publication date; English scope and specific inclusion rules. Lack of established mechanism is separate from validity of a functional effect.⁵

A007-S06 — Acuña et al.

• Identity/version: ASI meeting fidelity criteria
• Design: Review of nine RCTs
• Population/setting: 344 children across nine trials; six autism studies and three other populations
• Intended and actual delivery: Not recoverable at study level from abstract
• Comparison/other support: Usual care, other treatment or no treatment within review eligibility
• Outcome/reporter/time: Authors support individual goals in autistic children; do not support behaviour-of-concern outcomes
• Fidelity/adaptation: Abstract specifies fidelity eligibility; individual adherence and risk-of-bias details not accessible
• Burden/unwanted effects: Not reported in abstract
• Access and role: Opened abstract only; publisher route failed; Original systematic review, limited-access corroboration.
• Limits: No independent verification of full search dates, study table or detailed grading; abstract conclusion not an Atlas rating.⁶

A007-S07 — Piller et al.

• Identity/version: Techniques, caregiver-focused interventions and environmental modifications; absence of ASI fidelity moves studies into SBI group
• Design: 21 studies; five Hopkins Level I and 16 Level II; English, 2015–January 2024
• Population/setting: 675 participants aged 1–21; autism, ADHD and other sensory concerns; mixed settings
• Intended and actual delivery: Heterogeneous activities, equipment and coaching; no common dose/manual
• Comparison/other support: Mixed controlled, nonrandomised and single-case comparisons
• Outcome/reporter/time: Authors give favourable deep-pressure/coaching grades, but weighted-vest and some caregiver functional results null; three alternative-seating studies found no significant attention-related gains. One eligible environmental-modification study
• Fidelity/adaptation: Implementation/measurement inconsistent; no shared SBI fidelity standard
• Burden/unwanted effects: Child perspective incompletely represented; no reliable pooled harms finding
• Access and role: Full public article; main methods/results/discussion checked; Original systematic review.
• Limits: Component pooling and source-specific grades conceal differences; one eligible study does not mean one exists in all literature; no universal movement claim.⁷

A007-S08 — Gringras et al.

• Identity/version: Weighted blanket for sleep
• Design: Randomised crossover trial
• Population/setting: 73 randomised, 67 completed; autistic children aged 5–16 years 10 months with severe sleep problems; UK
• Intended and actual delivery: Weighted and otherwise similar lighter blanket, two weeks each
• Comparison/other support: Lighter blanket; prior sleep difficulties persisted despite community intervention
• Outcome/reporter/time: No advantage on actigraphic total sleep time or other reported sleep/behaviour outcomes; children and parents preferred weighted blanket
• Fidelity/adaptation: Detailed adherence and masking not verified from full methods
• Burden/unwanted effects: Reported well tolerated over short exposure; preference is separate from measured sleep change
• Access and role: Opened abstract only; publisher full-text route failed; Original comparative equipment trial.
• Limits: Abstract-only methods; short trial; no inference to other products, daytime attention or long-term safety.⁸

A007-S09 — Piller et al.

• Identity/version: ASI plus some multicomponent programmes
• Design: 12 studies: four RCTs, four nonrandomised group and four single-subject designs; English May 2015–January 2024
• Population/setting: Children/youth; mixed diagnoses and clinic/school contexts; no independent pooled n extracted
• Intended and actual delivery: Eight ASI alone; others add education or the STAR Frame of Reference, combining sensory integration, DIRFloortime and parent education/support
• Comparison/other support: Mixed designs/comparators
• Outcome/reporter/time: Favourable goal/performance grades in authors’ framework; occupational-performance synthesis includes SenITA’s uncontrolled within-arm COPM change
• Fidelity/adaptation: ASI fidelity required; supplementary components impede attribution
• Burden/unwanted effects: Review does not establish safety or child acceptability
• Access and role: Publisher abstract and full public article PDF; PDF internal title is AJOT051130_proof.pdf, final-format equivalence not authenticated; Original systematic review.
• Limits: Study-level RCT label does not make every reported outcome controlled; heterogeneous components; GAS bias acknowledged; no independent Atlas grade.⁹

A007-S10 — Park and Kim

• Identity/version: Sensory integration therapy; do not equate all included interventions with fidelity-defined ASI
• Design: 23 RCT meta-analysis
• Population/setting: Abstract does not provide study-level populations or total n
• Intended and actual delivery: Full study-level delivery not verified
• Comparison/other support: Not independently checked at study level
• Outcome/reporter/time: Abstract reports positive pooled motor, daily-function and individual-goal effects; balance, visual construction and sensory-processing effects not significant
• Fidelity/adaptation: Full fidelity coding not verified
• Burden/unwanted effects: Not reported in opened abstract
• Access and role: Opened abstract; extended publisher methods indexed in search, full publisher page failed; Original meta-analysis, limited-access lead reconciliation.
• Limits: Positive pooled results cannot establish equivalence of broad SIT, ASI and classroom sensory options; no protocol/dose recommendation imported.¹⁰

A007-S11 — Stein Duker et al.

• Identity/version: Sensory-adapted dental environment; not a stand-alone ASI treatment
• Design: Randomised crossover
• Population/setting: Autistic children aged 6–12, Los Angeles dental hospital; 163 eligible received a cleaning, one protocol deviation excluded: 162 first/138 second cleanings
• Intended and actual delivery: Two dental visits; combined lighting/projection/music and optional pressure wrap
• Comparison/other support: Regular dental environment; same cleaning context
• Outcome/reporter/time: Skin conductance lower; selected coded distress behaviours lower. Nonspecific skin response frequency, global behaviour scales and child-reported pain/discomfort not significantly different
• Fidelity/adaptation: Checklist; light/audio components used by >90%, wrap by 74.7%; observable condition prevents behavioural assessor masking
• Burden/unwanted effects: Child/parent could request stopping a modification; participation required visits/assessments; pressure-wrap package is not a school recipe
• Access and role: Full public article; Original comparative environmental-package trial.
• Limits: Attrition; setting-specific bundled effects; child-report subsamples; no attribution to any single adaptation.¹¹

A007-S12 — Lönn et al.

• Identity/version: Individually selected fibre weighted blanket for sleep
• Design: Randomised crossover
• Population/setting: Swedish ADHD service; 95 enrolled at baseline, 94 participated, 91 available after crossover; actigraphy n=85; ages 6–14, generally uncomplicated ADHD
• Intended and actual delivery: Four weeks per blanket; occupational-therapist-selected product/weight; adherence reported
• Comparison/other support: Matching lighter fibre blanket; stable ADHD medication intended, seven continued melatonin; sensitivity analysis did not alter findings
• Outcome/reporter/time: Actigraphic total sleep time +7.72 minutes; sleep efficiency +0.82%; wake after sleep onset −2.79 minutes; sleep-onset latency not significant. Exploratory subgroups not prescribing rules
• Fidelity/adaptation: Imperfect blinding; home measurements at weeks 4/8; missing actigraphy; outcomes not universal
• Burden/unwanted effects: Reported weighted-blanket effects included panic (n=1), anxiety (n=1) and pain (n=2); some unwillingness to use blanket/actigraph. These are report counts, not necessarily unique children
• Access and role: Full public article; Original comparative equipment trial.
• Limits: Small average sleep effects; selected population; short periods; not a substitute-for-medication trial; no transfer to autism or classroom attention.¹²

Adult report, direct observation, physiological measurement and a child's account answer different questions. Goal attainment is not automatically a direct report of lived experience. Conversely, preference is a meaningful outcome even when another measure does not change. A child who cannot use a questionnaire still requires accessible ways to decline or stop; this editorial standard is not established by a proxy score.

Evidence of no reported adverse effects in one trial is not a general safety finding. Therapy can carry travel, scheduling and opportunity costs. Equipment can be unwanted or uncomfortable. Comfort cannot be inferred from quietness or less movement, and access should not be exchanged for compliance. No pressure, touch, sensory exposure or restrictive arrangement is prescribed by this profile.

Transfer is especially uncertain for older adolescents, young people with substantially different intellectual or communication profiles, physical disability, severe co-occurring health needs, multilingual families, home/school implementation without trial resources, and untested equipment. Several sources include populations beyond autism, but that does not establish every branch for every group.

• Identity and version: Name the component or programme, its edition or protocol where relevant, and the purpose being evaluated.
• Source role: Separate direct comparative evidence, descriptive implementation findings, guidance, theoretical support and firsthand experience.
• Population and setting: Retain the studied age, communication/access characteristics and setting. State when an intended application lies outside them.
• Intended and actual delivery: Record the planned components, frequency and duration (dose) separately from what was received, by whom, and with what support or interruptions.
• Comparison and other support: Name the actual comparison and concurrent changes. If there is no comparison group or condition (comparator) or it is unreported, say so rather than supply an assumed usual-care condition.
• Outcome and reporter: Record the outcome, time point and reporter. Keep the young person’s account separate from someone else’s judgement on their behalf (proxy report), and access or participation separate from task performance.
• Delivery as intended (fidelity) and adaptation: Record whether the core components were delivered and describe adaptations; distinguish an adapted application from the protocol tested.
• Burden and unwanted effects: Record time, effort, practical access, cost or workload where known, alongside refusal, fatigue, distress, restriction and other unwanted consequences. Unreported burden is not evidence of no burden.

This common key applies to the retained source summaries below. Shorter headings such as design, dose or findings are read in that context. The key supplies no missing source data, converts no proxy report into recipient testimony and assigns no new certainty grade.

1. 1.

   Royal College of Occupational Therapists. (March 2026). Using sensory integration therapy, sensory-based interventions and sensory approaches with children and young people. Accessed 12 September 2026. Full official four-page document. Locator: pp.1–3: definitions, RCOT view and practice recommendations.

   ↩1 ↩2 ↩3
2. 2.

   Randell et al.; NIHR Health Technology Assessment 26(29). (June 2022). Sensory integration therapy for children with autism and sensory processing difficulties: the SenITA RCT. Additional checked text 1. Additional checked text 2. Additional checked text 3. Accessed 12 September 2026. Full public report: methods, quantitative results and discussion opened. Locator: Ch.3 methods; Ch.6 primary analysis, adaptive behaviour and COPM; Ch.10 limitations.

   ↩1 ↩2
3. 3.

   Schaaf et al.; Journal of Autism and Developmental Disorders. (Online 10 November 2013; issue 2014). An Intervention for Sensory Difficulties in Children with Autism: A Randomized Trial. Accessed 12 September 2026. Full public article. Locator: Methods, Table1 note, Intervention/Fidelity, GAS, Results and Discussion.

   ↩1 ↩2
4. 4.

   Pfeiffer et al.; American Journal of Occupational Therapy. (2011). Effectiveness of Sensory Integration Interventions in Children With Autism Spectrum Disorders: A Pilot Study. Accessed 12 September 2026. Full original article PDF on public professional host; PMC challenge not bypassed. Locator: pp.78–83: participants, intervention, measures, results, limitations.

   ↩1 ↩2
5. 5.

   Kishida, Stephenson, Kemp and Salisbury; Journal of Autism and Developmental Disorders. (28 March 2026). Efficacy of Interventions Adhering to the Revised Ayres Sensory Integration® Fidelity Measure: A Systematic Review. Accessed 12 September 2026. Full public article. Locator: Methods/search strategy; exclusions; Results; Discussion/limitations.

   ↩1 ↩2
6. 6.

   Acuña et al.; American Journal of Occupational Therapy. (May 2025). Ayres Sensory Integration® With Children Ages 0 to 12: A Systematic Review of Randomized Controlled Trials. Accessed 12 September 2026. Opened abstract only; publisher route failed. Locator: Abstract methods/results/conclusions.

   ↩1 ↩2
7. 7.

   Piller et al.; Frontiers in Pediatrics. (2025). Systematic review of sensory-based interventions for children and youth (2015–2024). Accessed 12 September 2026. Full public article; main methods/results/discussion checked. Locator: Methods/eligibility; Results sensory techniques and caregiver interventions; Limitations.

   ↩1 ↩2 ↩3
8. 8.

   Gringras et al.; Pediatrics. (August 2014). Weighted blankets and sleep in autistic children—a randomized controlled trial. Accessed 12 September 2026. Opened abstract only; publisher full-text route failed. Locator: Abstract methods/results/conclusions.

   ↩1 ↩2
9. 9.

   Piller et al.; American Journal of Occupational Therapy. (Online 1 December 2025; issue January/February 2026). Occupational Therapy Interventions Using Ayres Sensory Integration® for Children and Youth (2015–2024): A Systematic Review. Additional checked text 1. Accessed 12 September 2026. Publisher abstract and full public article PDF; PDF internal title is AJOT051130_proof.pdf, final-format equivalence not authenticated. Locator: PDF pp.5–8, especially occupational performance and added components.

   ↩1 ↩2
10. 10.

    Park and Kim; Research in Developmental Disabilities. (Online 3 April 2026; May 2026 issue). Effects of sensory integration therapy in children: A systematic review and meta-analysis of randomized controlled trials. Accessed 12 September 2026. Opened abstract; extended publisher methods indexed in search, full publisher page failed. Locator: Abstract; indexed publisher eligibility used only as scope-warning lead.

    ↩1 ↩2
11. 11.

    Stein Duker et al.; JAMA Network Open. (1 June 2023). Sensory Adaptations to Improve Physiological and Behavioral Distress During Dental Visits in Autistic Children: A Randomized Crossover Trial. Accessed 12 September 2026. Full public article. Locator: Methods intervention; Results fidelity; Tables 2–3; Limitations.

    ↩1 ↩2
12. 12.

    Lönn et al.; Journal of Sleep Research. (Online 15 July 2023; issue April 2024). The efficacy of weighted blankets for sleep in children with attention-deficit/hyperactivity disorder—A randomized controlled crossover trial. Accessed 12 September 2026. Full public article. Locator: Methods 2.1–2.8; Results 3.1–3.3; limitations.

    ↩1 ↩2