How this was built

Methodology, scope & limitations

deliriumtools.com is a curated, evidence-linked directory of delirium instrument families, versions, workflows and related measures. It is designed for discovery and comparison. It is not a completed systematic review, a clinical guideline, a diagnostic service or an endorsement of every record.

What is being counted

A single total previously obscured important differences between instruments. Content release v3.4.0 therefore reports a count breakdown: 49 independent core instrument families are represented by 82 core directory profiles, including 20 version/adaptation pages, 6 workflow pages and 7 core-emerging pages. A further 48 contextual, historical or adjacent profiles are clearly separated, giving 130 public profiles. One unresolved record is held from publication.

Profile count is not instrument-family count. A translated version, severity adaptation or workflow can deserve its own page without being misrepresented as a wholly independent instrument. Nor does inclusion mean that a measure is mature, validated for every setting, recommended, freely reusable or available as a public form.

Record types

Instrument family

A recognisable assessment family with its own intended role and structure.

Version or adaptation

A shortened, translated, setting-specific, severity or other adaptation that retains a substantial identifiable structure from a parent instrument.

Workflow

A defined sequence that combines existing instruments; it is not counted as another independent family.

Adjunctive or component test

An attention, cognitive or behavioural test studied in delirium but not, by itself, a full syndrome assessment.

Emergence-delirium or agitation measure

A peri-anaesthetic measure kept separate where arousal, pain and agitation overlap with delirium terminology.

Emerging technology

A device, software or decision-support approach with early evidence that is not presented as an established bedside instrument.

Historical or related context

A record retained for provenance or conceptual relevance but excluded from the independent clinical-family count.

A practical, author-defined classification

The public categories offer simple routes based on when a tool is used, its setting or population, its information source and what it measures. A tool may appear in more than one category.

This is a practical classification developed by the site author, not a formal consensus taxonomy. Terminology is not fully standardised, and other authors or services may group the same instruments differently. See Categories of delirium assessment tools.

How records were identified

The starting inventory was independently compiled and described on Delirium Words. Records were then cross-checked against original development and validation papers, PubMed and publisher metadata, reviews, book chapters, developer resources and other field inventories. The NIDUS Delirium Measurement Info Cards directory was one secondary cross-check among several.

The site is independently authored. It is not produced by, affiliated with or endorsed by NIDUS, Harvard Medical School, the American Delirium Society, instrument developers, the University of Edinburgh or NHS Lothian. Links to these organisations identify sources; they do not imply approval of this site.

Inclusion and classification rules

• Core family: a named clinical instrument explicitly designed to assess or detect current delirium, support serial observation, describe subtype or measure severity or course.
• Versions remain related: translations, shortened forms, severity forms and setting-specific adaptations are linked to a parent family where the relationship is known.
• Workflows are labelled: combinations of existing instruments are described as workflows rather than new independent tools.
• Evidence stage is separate: development, internal psychometrics, diagnostic accuracy, external validation, implementation and evidence synthesis are not treated as interchangeable.
• Adjacent records are visible but separated: component tests, emergence measures, risk models, chart methods and emerging technologies can be useful without belonging to the core-family count.
• Current-delirium assessment versus component test: a named multi-domain bedside method such as DelApp or LSD-4 can remain in the core directory when its publication defines delirium as the target and supplies an interpretable overall result. An isolated attention, arousal, writing or visual task such as DelBox or Lighthouse remains contextual unless it is validated and intended as a standalone delirium instrument.
• Substantive redevelopment: DRS-R-98 is linked historically to DRS but retained as a distinct family because it is a substantial redevelopment with different content, diagnostic and severity components, scoring and validation rather than a translation or minor setting adaptation.
• Uncertainty is public: a provisional date, disputed identity, conference-only source or missing public form is stated rather than silently filled.

Dated exclusions and boundary log

The following boundary decisions were rechecked for content review v3.4.0 (29 July 2026):

• General deterioration scores such as NEWS2 are excluded when delirium is represented only by a generic new-confusion item.
• Generic cognitive or dementia tests such as MMSE and MoCA are excluded unless a record is a purpose-built delirium component test, in which case it is labelled adjunctive.
• Future-risk rules such as AWOL and retrospective chart-identification methods are contextual, not current bedside assessments. Named models meeting the stated eligibility rule are grouped in the selected risk-prediction collection.
• Physiological, EEG, sensor and software methods are placed in the emerging/adjacent register until their intended use, validation and practical status justify reclassification.
• Post-anaesthetic emergence measures are separated because arousal, agitation, pain and delirium can overlap and the measures are not interchangeable with general ward delirium instruments.
• MCV-NDRS is held from the public directory because a canonical publication and instrument identity could not be verified.

These are classification decisions, not judgments that the excluded or contextual methods have no research or clinical value. Suggestions supported by a canonical source can be submitted through Contribute & corrections.

What each profile aims to contain

Profiles use independent prose and aim to state the canonical name and version, intended role, target population and setting, likely rater, broad domains, item count where verified, administration mode and timing, training needs, score direction and published interpretation, evidence stage, practical strengths and limitations, and population-specific caveats. They also separate the instrument itself from its original/source study, later validation evidence, secondary summaries and rights information.

Inapplicable fields are suppressed. “Not verified” means the present audit could not establish the fact from a reliable source; it does not prove that the information or form does not exist.

Reference and evidence checking

Authors, title, journal, year and identifier are checked against PubMed, the publisher DOI record or another primary source. A checked citation means that the record exists and relates to the instrument. It does not show that the study is high quality, independently validates diagnostic accuracy, supports a particular threshold, or makes the tool suitable for a different population.

The defined PubMed diagnostic-test-accuracy census has been screened, but wider systematic discovery, complete full-text extraction, cohort-overlap adjudication and duplicate independent review remain unfinished. The site therefore uses “DTA studies identified in this audit,” not “all validation studies.” Evidence labels report the type and amount of evidence located; they are not a formal risk-of-bias assessment.

Accuracy and evidence labels

Single-study sensitivity and specificity figures are shown only with the relevant citation and cohort context. Pooled meta-analytic values are restricted to the comparison page and clearly labelled. Development, diagnostic accuracy, external validation, implementation and systematic synthesis are kept distinct. “No eligible DTA study identified” is a transparent zero in the current audit, not proof that no study exists.

Evidence glossary

Episodic assessment

Structured tools used at first presentation or at another time when delirium is suspected. Results inform clinical judgement and are interpreted alongside history, examination and other relevant information.

Diagnostic-test accuracy (DTA)

How closely an index test agrees with an independent reference standard in a defined population. The current DTA count is an audit state, not a guarantee that every eligible study has been found.

Reference standard

The independent clinical or research assessment used to judge whether delirium was present. Its quality and independence affect how an accuracy estimate should be interpreted.

Sensitivity

The proportion of people with delirium, according to the reference standard, who have a positive result on the evaluated tool.

Specificity

The proportion of people without delirium, according to the reference standard, who have a negative result on the evaluated tool.

Overall evidence maturity

A broader description of the evidence located for a profile. It is kept separate from any single displayed sensitivity or specificity estimate.

Provisional DTA identified

At least one potentially eligible diagnostic-accuracy study is recorded, but complete full-text extraction, overlap checks or duplicate adjudication are not yet complete.

Arousal

The person’s observable level of wakefulness and responsiveness. Abnormal arousal can support assessment but is not synonymous with delirium.

Access, copyright and reuse policy

Every record has an access state and a dated rights record. “Open instrument” is used only when a reliable instrument source and compatible reuse basis have been established. A free download, open-access article or permission for clinical use does not automatically permit web republication, adaptation, translation, EHR integration or commercial use. A paper link is labelled as a paper, not as the instrument.

The editorial rule is proportionate. For the CAM family and DRS-R-98, profiles avoid reproducing protected prompts, item-level weights, worksheets or a complete executable algorithm unless an explicit licence or permission supports that use. For other tools, the site retains useful independently written detail about domains, named tasks or stimuli, item counts, score ranges, thresholds and practical scoring where that helps readers understand the instrument. This explanatory detail is not a facsimile of the source form, its visual presentation or its complete administration instructions.

There is no fixed “safe” number of words, items or percentage that guarantees lawful quotation. The editorial test is whether the commentary is independently written, properly attributed, limited to what is needed for explanation or review, and does not substitute for the instrument or its authorised form. See the UK Intellectual Property Office guidance on fair dealing and the US Copyright Office fair-use guidance. These links provide general information, not legal advice about a particular instrument.

The site’s own original editorial text may be reused under CC BY 4.0 unless a page states otherwise. That licence excludes third-party instruments, forms, article text, logos, trademarks, quotations and linked material. Instrument-specific terms always prevail. The only displayed form previews are those for which the recorded licence and attribution support embedding.

Developers, publishers and rights holders can use the rights-holder route to correct a source, request attribution changes, provide permissions or request removal. A credible request is acknowledged as promptly as practicable; a disputed preview or quotation can be temporarily removed while evidence is checked.

Editorial governance and corrections

The named curator is Professor Alasdair M J MacLullich. His authorship and other relevant relationships are disclosed on the About page. Bibliographic verification, factual clinical checking, rights review, accessibility testing and external editorial review are separate gates. A date shown for one gate must not imply that every other gate was completed on the same date.

• Material corrections record the affected record, the old and new position, the reason, the supporting source, the decision date and the editor.
• Links and rights states are scheduled for review at least every six months and when a new instrument version is identified.
• New records require a canonical identity, record type, family relationship, evidence stage, source role and truthful access/rights state before publication.
• Claims of directory completeness require a documented search, duplicate independent screening and adjudication; content release v3.4.0 does not make that claim.
• A future external editorial group should cover general/geriatric medicine, intensive care, paediatrics, perioperative care, measurement methods, rights and accessibility. No such group is named until each person has agreed and completed a declared review.

Accessibility, usability and technical review

The release includes automated checks for headings, duplicate identifiers, internal links, filter journeys and generated pages, plus desktop and mobile visual checks. The interface supports keyboard-operable navigation, visible focus, labelled controls, responsive filters and a full-shell error page. This is not a claim of WCAG certification. Independent screen-reader, 200% zoom and task-based testing with clinicians, researchers and family users remain planned and will be dated here when completed.

Release gates

• Every public record has a family, record type, evidence stage and count disposition.
• No unverified or contextual record is called a definite, validated or recommended independent tool.
• Every profile has a truthful access state, rights state and rights-check date.
• Every displayed sensitivity/specificity group names its source type, identifier, year, denominator state and reference standard before the values.
• No generic hub or article is labelled as the original instrument.
• No unlicensed CAM-family or DRS-R-98 operational reproduction is published.
• Filters, task pathways, internal links, mobile layout, accessible headings and build validation pass before deployment.
• Any release blocker found in the post-deployment red team is corrected and rechecked before the release is declared complete.

Updates, versioning and limitations

The labels are deliberately separate: the current website delivery is site release v3.7.3 dated 9 August 2026, while the directory content is clinical/content review v3.4.0 dated 29 July 2026. The machine-readable directory data, page generator and change records are maintained together. Material accepted corrections and classification changes are logged.

• Not a completed systematic review: discovery and duplicate independent adjudication remain incomplete.
• Language bias: non-English instruments and validations are likely under-represented.
• Single maintainer: the public resource is maintained by one curator, not a funded consortium.
• Access changes: developer pages, licences and hosted forms can move or change after the recorded check date.
• Clinical use remains local: a profile cannot replace the current instrument instructions, training, clinical judgment or local policy.

Relationship to other resources

For CAM-family access see the authorised CAM instrument hub; for ICU tools see ICU Delirium and Cognitive Impairment Study Group resources; and for independently produced measurement cards see the NIDUS info-card directory. NIDUS is a secondary summary source; its links and licences do not grant rights in the underlying instruments.