Seed data — not verified. Approximate data from public sources, not yet confirmed by cohort leads or site reps.

Why?

Definitions

This page defines every entity, category, scoring axis, and classification used across the DIGIT Capabilities Database. It is the reference for interpreting data on the Cohort Map, Capabilities Browser, and Query Builder. All definitions are grounded in published UK health research standards including the HRA/CAG consent framework, HDRUK Data Utility Framework, DPUK metadata standards, and NIHR clinical research infrastructure classifications.

Overview

The DIGIT Capabilities Database is a metadata catalogue of UK mental health research sites, cohorts, and trial capabilities, built for the UKRI Mental Health Goals Programme. It helps Alliance Managers match industry enquiries with UK capabilities, helps researchers discover trial sites and cohort access, and supports statisticians with trial design and feasibility planning.

The database organises information around four core entities (sites, cohorts, capabilities, and diagnosis areas) and uses a three-dimensional mapping model to visualise cohorts along three independent axes: consent level, characterisation depth, and data availability.

Core Entities

Site

A UK location where a mental health clinical trial could be designed, recruited into, or run. Includes NIHR BRCs, NHS Trusts, university departments, Clinical Trials Units, and more.

Inclusion test: “Would a CRO feasibility analyst consider this location when planning a UK mental health trial?”

Cohort

A defined research cohort, data resource, biobank, or electronic health record system containing mental health data that could support trial recruitment, feasibility assessment, or observational research.

Inclusion test: “Could a researcher use this resource to identify potential participants or extract mental health data?”

Capability

A piece of research infrastructure or expertise at a site relevant to planning a mental health trial. Capabilities describe what a site can do — infrastructure and capacity.

Inclusion test: “Could this infrastructure be used to generate new data for a clinical trial?”

Diagnosis Area

A mental health condition or condition group that a site has experience researching, or that a cohort has participants diagnosed with. Maps to ICD-10 codes.

Inclusion test: “Is this a distinct clinical population with its own trial design considerations?”

The 3D Mapping Axes

The Cohort Map visualises cohorts along three independent axes forming a 4x4x4 cube. Each axis uses a 4-level ordinal scale following the precedent set by the HDRUK Data Utility Framework. The axes are designed to be orthogonal — each carries genuinely independent information:

AxisQuestion it answersHow it is scored
Consent LevelHow can participants be approached for new research?Expert-judged per HRA/CAG consent framework
Characterisation DepthHow deeply is the mental health state assessed?Expert-judged per precision psychiatry literature
Data AvailabilityHow broad is the range of data modalities available?Auto-derived from 10 boolean data type flags

Consent Level (1-4)

Grounded in the HRA/CAG consent framework, NHS Digital Section 251 guidance, Kaye et al. 2014 dynamic consent spectrum, and the SLaM Consent for Contact (C4C) model.

1
Anonymised
Anonymised / no consent
De-identified or aggregated data with no mechanism to identify or contact individuals. Legal basis: anonymisation per ICO code (not personal data under GDPR). Useful for observational and epidemiological studies only.
2
S251
Section 251 / CAG (pseudonymised, no individual consent)
Pseudonymised patient data accessible without individual consent under Section 251 of the NHS Act 2006. Subject to the national data opt-out. Individuals are identifiable within the data system but have not consented to be contacted for research.
3
Recontactable
Broad consent (recontactable for future studies)
Participants gave broad consent to future research use of their data and/or samples, and agreed to be re-contacted for further studies. Each new study may still need additional ethics approval. Legal basis: GDPR Article 6(1)(a) informed consent.
4
Active consent
Active research consent (interventions, TWiCs-eligible)
Participants have given specific informed consent for ongoing active research participation, including potential randomisation, interventions, and repeat assessments. Supports Trials Within Cohorts (TWiCs) and adaptive trial designs.

Characterisation Depth (1-4)

Grounded in NIHR BRC deep phenotyping definitions, the RDoC units of analysis (NIMH), and precision psychiatry literature. Each level is strictly cumulative — higher levels include everything from the levels below.

1
EHR
Routine EHR
Only routinely collected clinical data: ICD-10/SNOMED diagnostic codes, prescriptions, and service contacts. No research-specific mental health assessment has been applied to participants.
2
MH scales
Standardised MH instruments
Routine data plus validated mental health instruments applied systematically (e.g. PHQ-9, GAD-7, GHQ-12, WEMWBS). Provides structured symptom data beyond diagnostic codes.
3
Multi-domain
Multi-domain MH characterisation
Standardised MH instruments plus additional research domains: cognitive testing, lifestyle/environmental data, physical health measures, or structured diagnostic assessment beyond screening. Broad characterisation across multiple domains of function.
4
Deep phenotyping
Deep phenotyping, refreshed
Comprehensive multi-domain characterisation with research-grade psychiatric diagnosis (SCID/MINI/SCAN) and the cohort is actively renewed/refreshed to maintain data currency. The gold standard for precision psychiatry research.
Biosamples are deliberately excluded from the characterisation axis. Whether biosamples exist is captured by the data type flags and the data availability axis. Characterisation measures purely the depth and rigour of the clinical and behavioural assessment.

Data Availability (1-4)

This axis is automatically derived from the 10 boolean data type flags (see below). Each level is strictly cumulative.

1
Records
Clinical records only
Routine clinical or administrative data only. No research-specific data types or biological material in the archive.
2
+Research
Records + research data
Clinical records plus structured research datasets: questionnaires, cognitive test results, physical measures, and/or linked administrative data (e.g. linked HES, GP records).
3
+Biological
Adds biological specimens/molecular data
All of the above, plus biological specimens and/or molecular data: stored biosamples (blood, DNA, saliva), genotyped or sequenced data, or other -omics (proteomics, metabolomics).
4
Multi-modal
Multi-modal
All of the above, plus advanced data modalities: neuroimaging (MRI, fMRI, PET, EEG), wearable/sensor data (accelerometry, heart rate), or digital phenotyping (smartphone passive data). The broadest range of accessible data.
The data availability score is derived mechanically from the data type flags. In rare cases where the derived score does not reflect reality, a manual override can be applied with documented justification.

Cohort Data Type Flags

Each cohort has 10 boolean flags recording whether specific modalities of data exist in its archive. These are factual indicators — a flag is set to true only when there is evidence that the data type exists and is accessible to researchers.

Group A: Research Assessments

Data collected through structured research assessments administered directly to participants.

Questionnaire Datahas_questionnaire_data

Validated self-report instruments, structured interviews, or patient-reported outcomes applied to participants (e.g. PHQ-9, GAD-7, SCID, WEMWBS, GHQ-12).

Cognitive Datahas_cognitive_data

Computerised cognitive testing or formal neuropsychological assessment data (e.g. CANTAB, CogState, n-back, Stroop, TMT, WAIS).

Physical Measureshas_physical_measures

Anthropometry, vital signs, non-genomic blood biomarkers, or other objective physical measurements (e.g. height/weight/BMI, blood pressure, grip strength, blood biochemistry, ECG).

Group B: Biological

Stored biological material and molecular data derived from participant samples.

Biosampleshas_biosamples

Stored biological specimens available for analysis: blood, plasma, serum, DNA, saliva, CSF, tissue, or urine that researchers can apply to access.

Genomic Datahas_genomic_data

Genotyping array data, whole-exome sequencing (WES), whole-genome sequencing (WGS), or GWAS summary statistics for participants.

Omics Datahas_omics_data

Non-genomic molecular profiling: proteomics, metabolomics, transcriptomics, epigenomics (DNA methylation), or other -omics data.

Group C: Advanced Modalities

High-dimensional, technology-driven data modalities and administrative data linkage.

Imaging Datahas_imaging_data

Brain or body imaging data: structural MRI, functional MRI, diffusion MRI, PET, SPECT, EEG, MEG, DEXA, or retinal imaging acquired from participants.

Wearable Datahas_wearable_data

Data from body-worn devices: accelerometry, actigraphy, continuous heart rate monitoring, electrodermal activity (EDA), or sleep sensors.

Digital Phenotypinghas_digital_phenotyping

Passive smartphone or sensor data: GPS traces, screen time, call/text metadata, app usage patterns, voice/speech analysis, or similar behavioural data.

Linked Health Recordshas_linked_health_records

The cohort’s research data has been linked to NHS administrative records (HES hospital episodes, GP records, prescribing data, ONS death registry).

Cohort Groups

The group is derived by a database rule, evaluated in order — every cohort gets exactly one group, and identical inputs always produce the same group. (The groups descend from Figure 1 of the grant application, which sketched them by hand; they are now computed.)

4
1. Group 4
Population type is data resource, or consent level is 1 (anonymised, no recontact route). The records and feasibility layer.
1
2. Group 1
All three axes ≥ 3 and at least one axis at 4. The deepest consented research cohorts.
2
3. Group 2
All three axes ≥ 3. Consented, well-characterised cohorts.
3
4. Group 3
Everything else — below 3 on at least one axis. Includes registers, networks, and lighter-touch cohorts.

The rules run top to bottom; the first match wins. Because the group is computed from consent level, characterisation depth, data availability (itself derived from the data type flags), and population type, it cannot be edited directly. Rare, documented exceptions use the group_label_override column with a justification note — none are currently in use.

Population Types

Who the cohort primarily contains, from a study-planning perspective. This is deliberately separate from the 3D axes: a general-population registry and a patient cohort can carry identical consent and characterisation scores while serving completely different industry needs. Patient supply concentrates in clinical and case-enriched cohorts; general-population, occupational, and data-resource entries form the feasibility and natural-history layer.

Clinicalclinical

Recruited via clinical services or on diagnosis

Case-enrichedcase_enriched

Volunteers with a condition (registry or sign-up criteria)

General populationgeneral_population

No health-based entry criteria

Occupationaloccupational

Sampling frame is an occupation, not health status

Data resourcedata_resource

EHR, administrative, or SDE asset — not a recruited cohort

Network / umbrellanetwork_umbrella

Recruitment network or umbrella programme

Industry Engagement Status

Whether the cohort's custodian has operationalised industry or trial access. This is an organisational property of the study team, not a consent property — cohorts with identical consent levels differ completely in whether an approach from industry could actually be actioned. Values are populated from cohort-owner meetings and default to unknown until a conversation settles them.

Proven industry supplierproven

The cohort has supplied participants, samples, or data to at least one commercial trial or industry study.

Established access processestablished_process

A documented application and governance route exists that industry can use (e.g. the NIHR BioResource access process, UK Biobank AMS).

Open in principleopen_in_principle

The custodian has expressed willingness to consider industry access but no formal process exists yet.

Not yet considerednever_considered

The study team has not previously considered clinical trials or industry access — recorded directly from owner meetings, not inferred.

Declineddeclined

The custodian has decided against industry access.

Unknownunknown

Not yet established with the custodian. This is the default until an owner meeting settles it.

Cohort Relationships

Cohorts can be part of a larger programme (recorded as a parent link): a sub-study recruits within its parent (STRADL within Generation Scotland), an extension adds new participants or data layers (GLAD-omics extending GLAD), a wave is a repeat collection round, and an umbrella member sits under a programme umbrella (GLAD under the NIHR Mental Health BioResource). A sub-study only gets its own database entry when it differs from its parent on something a user filters on — axis scores, population type, recontactability, or access route; otherwise it is recorded as a note on the parent.

Sub-study (sub_study)Extension (extension)Wave (wave)Umbrella member (umbrella_member)

Site Types

Every site is classified into one institutional type. When a site could fit multiple types, the type with higher infrastructure significance takes precedence (BRC > CRF > CTU > NHS Trust > University).

BRCbrc

A site designated by NIHR as a Biomedical Research Centre with a mental health or neuroscience theme. BRC status takes precedence over other classifications.

Examples: Maudsley BRC, Oxford BRC, Cambridge BRC

NHS Trustnhs_trust

An NHS Trust or Foundation Trust whose primary function is providing mental health services, with research delivery capability.

Examples: Pennine Care NHS FT, Tees Esk and Wear Valleys NHS FT

Universityuniversity

A university department (typically psychiatry, psychology, or neuroscience) with trial delivery capability that is not an NIHR BRC.

Examples: University of Edinburgh (Psychiatry), University of Bristol

CTUctu

A UKCRC-registered Clinical Trials Unit that designs, manages, and analyses clinical trials. Provides trial methodology, statistics, and regulatory support.

Examples: King's CTU, Priment CTU (UCL), Edinburgh Clinical Trials Unit

CRFcrf

An NIHR Clinical Research Facility — a dedicated, staffed facility for conducting clinical research visits, with beds, nursing, and pharmacy.

Examples: NIHR Wellcome King's CRF, Manchester CRF

ARCarc

An NIHR Applied Research Collaboration conducting applied health research in partnership with NHS organisations, focusing on implementation science.

Examples: ARC South London, ARC North Thames

Health Boardhealth_board

A Scottish NHS Health Board or Welsh Health Board providing integrated health services with research delivery capability. The devolved equivalent of an NHS Trust.

Examples: NHS Greater Glasgow and Clyde, NHS Lothian, Betsi Cadwaladr UHB

HSCThsct

A Northern Ireland Health and Social Care Trust — the integrated health and social care provider. The NI equivalent of an NHS Trust.

Examples: Belfast HSCT, Southern HSCT

CRN Sitecrn_site

An NIHR Clinical Research Network partner site that recruits patients into trials but is not otherwise classified as a BRC, Trust, or CRF.

Examples: (Most CRN partners are better classified as another type)

Otherother

Any site that does not fit the above categories, such as charities with research arms or independent research institutes.

Examples: Research charities, independent institutes

Capability Categories

Research capabilities at sites are grouped into eight functional domains. Capabilities describe what a site can do (infrastructure), not what data a cohort already has (that is captured by the data type flags).

Imaging

Equipment and expertise for acquiring brain or body images in a research context.

Examples: MRI (structural, functional, diffusion), PET, SPECT, EEG, MEG, retinal imaging

Biomarkers

Laboratory facilities and expertise for measuring biological markers from participant samples.

Examples: Blood biomarker assays, CSF analysis, genotyping services, proteomics/metabolomics platforms

Cognitive Testing

Equipment, software, and trained staff for administering standardised cognitive assessments.

Examples: CANTAB, CogState, NIH Toolbox, paper-and-pencil neuropsychological testing (WAIS, TMT, Stroop)

Digital

Digital health technology infrastructure for research data collection.

Examples: Smartphone apps for EMA, wearable device platforms, digital phenotyping systems, remote monitoring

Specialist Facilities

Physical spaces or equipment purpose-built for research that do not fit other categories.

Examples: Sleep labs, exercise physiology labs, sensory testing rooms, mother-baby units, secure/forensic research facilities

Biobanking

Infrastructure for collecting, processing, storing, and distributing biological samples for research.

Examples: -80°C freezers, liquid nitrogen storage, sample processing facilities, LIMS, consent management

Neuromodulation

Equipment for non-invasive or invasive brain stimulation used in research or as an intervention.

Examples: TMS (repetitive, single-pulse, theta burst), tDCS, tACS, deep brain stimulation (DBS), ECT (research protocols)

Trial Methodology

Expertise and infrastructure for advanced clinical trial designs relevant to mental health.

Examples: Adaptive trial design, basket/umbrella trials, TWiCs infrastructure, decentralised trials, N-of-1 methodology

Diagnosis Area Categories

Diagnosis areas are grouped into 15 clinical categories. Each category maps to ICD-10 code ranges and contains individual conditions relevant to mental health trial design.

Psychotic DisordersICD-10: F20–F29

Conditions characterised by delusions, hallucinations, disorganised thinking, or grossly disorganised behaviour.

Includes: Schizophrenia, schizoaffective disorder, first-episode psychosis, treatment-resistant schizophrenia

Mood DisordersICD-10: F30–F39

Conditions characterised by persistent disturbance in mood (depression or mania/hypomania).

Includes: Major depressive disorder, bipolar disorder, treatment-resistant depression, dysthymia

Anxiety DisordersICD-10: F40–F41

Conditions characterised by excessive fear, anxiety, or avoidance behaviour.

Includes: Generalised anxiety disorder, social anxiety, panic disorder, specific phobias, agoraphobia

Obsessive-CompulsiveICD-10: F42, F45.2

Conditions characterised by obsessions (intrusive thoughts) and/or compulsions (repetitive behaviours).

Includes: OCD, body dysmorphic disorder, hoarding disorder, trichotillomania

Trauma-RelatedICD-10: F43

Conditions arising from exposure to traumatic or stressful events.

Includes: PTSD, complex PTSD, acute stress disorder, adjustment disorders

Eating DisordersICD-10: F50

Conditions characterised by persistent disturbance in eating behaviour.

Includes: Anorexia nervosa, bulimia nervosa, binge eating disorder, ARFID

NeurodevelopmentalICD-10: F70–F79, F84, F90, F95

Conditions with onset in the developmental period, manifesting as developmental deficits affecting functioning.

Includes: ADHD, autism spectrum disorder, Tourette syndrome, intellectual disability with co-occurring MH conditions

Substance UseICD-10: F10–F19, F63.0

Conditions arising from the use of psychoactive substances, including behavioural addictions.

Includes: Alcohol use disorder, opioid use disorder, stimulant use disorder, gambling disorder

Personality DisordersICD-10: F60–F61

Enduring patterns of inner experience and behaviour that deviate from cultural expectations and cause distress or impairment.

Includes: Borderline personality disorder (BPD/EUPD), antisocial personality disorder

NeurodegenerativeICD-10: F00–F03, G30–G31, G20

Progressive loss of neuronal structure or function, often with psychiatric manifestations.

Includes: Dementia, Alzheimer’s disease, Parkinson’s disease-related MH, frontotemporal dementia, Lewy body dementia

Sleep DisordersICD-10: F51, G47

Conditions characterised by abnormal sleep patterns causing distress or functional impairment.

Includes: Insomnia, hypersomnia, circadian rhythm disorders, sleep apnoea (in MH context)

PerinatalICD-10: F53, O99.3

Mental health conditions occurring during pregnancy or in the first year postpartum.

Includes: Perinatal depression, perinatal anxiety, postpartum psychosis, tokophobia

Self-Harm and SuicidalityICD-10: X60–X84, Z91.5

Intentional self-injury with or without suicidal intent. A critical clinical and research domain that cross-cuts diagnostic categories.

Includes: Non-suicidal self-injury (NSSI), suicidal ideation, suicide attempts

TransdiagnosticICD-10: (multiple)

Approaches that span traditional diagnostic boundaries, studying shared mechanisms or comorbidity patterns.

Includes: Anxiety-depression comorbidity, emotional dysregulation, dimensional approaches (p-factor), common mental disorders

OtherICD-10: (varies)

Conditions or research areas that do not fit the above categories.

Includes: Long COVID neuropsychiatric symptoms, chronic fatigue / ME, medically unexplained symptoms

Experience Levels

A site's experience in a specific diagnosis area is classified using four levels, based on completed and active trial counts from registries (ClinicalTrials.gov, ISRCTN, NIHR CRN portfolio).

LevelDefinitionThreshold
ExtensiveA recognised centre of excellence with sustained, high-volume trial activity.5+ completed MH trials in 10 years, or 3+ active trials, or specialist programme grant
ModerateMeaningful trial experience with multiple completed or active studies.2-4 completed MH trials in 10 years, or 1-2 active trials
SomeAt least one trial or specialist clinical expertise that could support delivery.1 completed or active trial, or specialist clinical service in the area
EmergingDeveloping capability; clinical contact with patients but no completed trials.No completed trials, but sees patients with this condition clinically

Data Access Mechanisms

How researchers access a cohort's data. This affects feasibility timelines and cost.

MechanismDefinitionExamples
Trusted Research EnvironmentData accessed within a secure, accredited TRE/SDE. Researchers submit code; results are reviewed before export.NHS SDE, OpenSAFELY, SAIL, UK Biobank RAP
Direct ApplicationResearchers apply to the data custodian and, if approved, receive a dataset extract or portal access.ALSPAC, GLAD, CPRD licence
CollaborationAccess requires a formal research collaboration with the cohort team, typically involving co-PI arrangements.Smaller cohorts, early-stage studies
Open AccessData or summary statistics freely available without formal application.Published GWAS summary stats, public-use files
Commercial LicenceIndustry or commercial access requires a paid licence or commercial data sharing agreement.CPRD commercial, IQVIA datasets
OtherAccess mechanism does not fit the above (e.g. federated analysis where no data leaves the source).Federated platforms, bespoke arrangements

Verification Statuses

Every site and cohort record carries a verification status indicating confidence in its data quality. Records progress from AI-inferred to verified as human review is completed.

StatusDefinition
VerifiedAll key fields confirmed by a human with direct knowledge (site PI, cohort lead, or programme team member) against primary sources.
Partially VerifiedSome fields confirmed by a human, but others remain unverified. Common when a PI confirms infrastructure but not trial counts.
UnverifiedData collected from published sources (websites, papers, registries) by a human researcher, but not confirmed by someone at the site or cohort.
AI-InferredData initially generated or populated with AI assistance. Requires human review before use in any formal output.

Reading the source indicators

The help icon beside each value opens its provenance, which is one of five kinds. Field-specific source: a citation pins this exact value, usually with a quote. Searched — not published: we looked for the value and could not find it published; the record lists where we checked, so a data owner can confirm or correct it (a linked document here is the place searched, not support for the value). Whole-record source: no citation pins this exact value, but documents describing the whole record are listed. Computed value: calculated from other fields (for example, the derived cohort group). AI-inferred, unverified: no provenance recorded yet — these need review first.

Conceptual Boundaries

Several concepts in the database appear similar but carry distinct meanings. Understanding these boundaries is essential for interpreting the data correctly.

Cohort Data Type Flags vs Site Capabilities

Cohort Data Type FlagSite Capability
QuestionDoes this cohort's existing archive contain this type of data?Can this site generate new data of this type for a trial?
AboutArchived data (what already exists)Infrastructure (what can be generated)
Example“ALSPAC has brain MRI data you can request”“Maudsley BRC has a 3T scanner available for trials”

Characterisation Depth vs Data Availability

These two axes are fully independent. Characterisation measures how deeply the mental health state is assessed (process quality). Data availability measures how broad the range of data modalities is (archive breadth). A cohort can have high characterisation but low data availability (e.g. detailed SCID interviews but no biosamples or imaging), or vice versa.

These definitions are maintained as part of the DIGIT Capabilities Database and are derived from the project's data dictionary (docs/data-dictionary.md). For the full technical schema, refer to that document.

Data and Digital Industry Alliance Team (DIGIT) • King's College London • IoPPN