How this was built

Methodology, scope & limitations

This page states plainly how the directory was assembled, what is included and excluded, how references were checked, and where the resource falls short. It is a curated, evidence-linked inventory — not a systematic review or a formal clinical guideline.

What each entry contains — and what it doesn't

Each entry provides an official source where available; a checked development, validation, review or context reference where one was located; and selected key studies and reviews for the best-characterised tools. The reference type is shown because a confirmed paper is not necessarily a validation study.

It is not a fresh systematic review of each individual tool, and it does not claim to be. What this directory adds over a static card set is fast discovery, side-by-side comparison, explicit evidence labels, and implementation context (see Real-world use).

How tools were identified

The starting list was compiled from the curator’s working knowledge of the field and then checked against the NIDUS adult delirium measurement inventory and standard reviews. It is not the output of a registered systematic search and does not claim complete coverage of a changing external inventory. Each specific citation was then checked against PubMed or the publisher DOI record.

Inclusion criteria

An entry is included if it is a named instrument for the detection, diagnosis, severity rating, monitoring, motor subtyping, cognitive testing, informant-based assessment, risk prediction, or record-based identification of delirium (or its synonym, acute confusional state) in humans, adult or paediatric, with a traceable published origin.

Inclusion does not imply clinical validation. Historical or development-only instruments are retained only when clearly labelled; they are not presented as suitable for clinical use.

Exclusion criteria

• General deterioration scores whose only link to delirium is a single “new confusion” item (e.g. NEWS2) — these are physiological warning tools, not delirium instruments.
• Generic cognitive or dementia tests not specific to delirium (e.g. MMSE, MoCA), except where a test was purpose-built for delirium.
• Untraceable entries. Where an item in the source list could not be tied to any verifiable published instrument, it was removed rather than shown (see CHANGELOG — the “Global Accessibility Rating Scale” was removed on this basis).

Reference verification

For each specific reference, authors, title, journal, year and identifier were checked. This corrected multiple errors in the source data. Of 68 instruments, 62 carry a specific PubMed-indexed reference, and one more — the 1994 Delirium Assessment Scale paper — was checked against Crossref DOI metadata, giving 63 with a specific reference. The remaining five could not be tied to one canonical record and are flagged, linking to a PubMed search — never to an invented citation. The full audit trail is in the references verification log.

Accuracy figures

Sensitivity/specificity shown on tool cards are from the original validation study (often a single, sometimes small, cohort) and are labelled with an asterisk. Pooled meta-analytic figures are used only on the Compare page, for the best-studied tools, each with its own citation. Original-study and pooled figures are never mixed without labelling.

Evidence labels

Evidence labels describe the type of evidence located, not study quality or endorsement. “Pooled diagnostic-accuracy evidence” requires a cited meta-analysis for the exact instrument; “multiple validation studies” requires at least two identified validation cohorts; “single validation study” requires one actual validation cohort; “indirect/contextual evidence” denotes a review or source paper that is not a standalone validation; “development study only” denotes a published description without diagnostic validation; and “reference not confirmed” means no canonical paper was confirmed. These labels are curator-assigned and are not a formal risk-of-bias assessment.

Known limitations & omissions

Honesty about gaps is part of the method:

• Not a systematic review. There is no registered protocol, no dual independent screening of a defined search, and no formal risk-of-bias appraisal. Treat it as a well-checked index, not a definitive evidence synthesis.
• Not exhaustive. The directory has been cross-checked against NIDUS and other source lists, but those inventories evolve. The wider literature contains additional niche and non-English instruments, and new ones continue to appear. Five entries remain linked to a search because no single canonical paper was confirmed.
• Automated detection is under-represented. Fully automated machine-learning and EHR-embedded detection is a fast-moving area not yet represented as discrete, named, reusable tools; record-based and prediction examples (CHART-DEL, AWOL) are included as the nearest current equivalents.
• Language bias. Non-English instruments and validations are under-represented.
• Single maintainer. This is maintained by one person, not a funded consortium; see Disclosures.

Updates & versioning

The dataset is versioned (currently v2.7, last reviewed 15 July 2026). Links and citations are reviewed at least twice yearly, and material changes are recorded in the CHANGELOG. The underlying data is published as a machine-readable file.

Relationship to other resources

This directory is designed to complement, not replace, the authoritative resources in the field. For deeper per-tool psychometrics, use info cards and the severity crosswalk, see the NIDUS measurement inventory; for the CAM family see the American Delirium Society; for ICU tools see icudelirium.org. This site’s contribution is fast, filterable discovery and side-by-side comparison with verified links — not new evidence.