Dementia Identification with EMR

PH1717 / 3756

Doney ASF, Bonney W, Jefferson E, Walesby KE, Bittern R, Trucco E, Connelly P, McCrimmon RJ, Palmer CNA, Camilleri MPJ, Whiteley W

May 14, 2025

Clinical-Coded Phenotype
  1. Overview

    Phenotype Type
    Disease or syndrome
    Sex
    Both
    Valid Event Date Range
    No data
    Coding System
    BNF codesICD10 codes
    Collections
    No data
    Tags
    No data
    Ontology
    No data
  2. Definition

    The major data sources for identification of dementia cases in the UK comprise EMRhospitalisation data, death records and dispensed prescribing records. From thesesources, we categorise any dementia into the following clinicopathological subtypes:Alzheimer's dementia, Vascular dementia, Other dementia (any dementia type otherthan Alzheimer's or Vascular), Unspecified dementia (where a specific sub-type isnot recorded) and Possible dementia (where the diagnosis is not clear for it to definitivelybe dementia). For individuals with dementia, the earliest date of mention of dementia(in any of the sources) is taken as a surrogate for the incident date of dementia.

    For more information about the phenotype, please refer to linked publications.

  3. Implementation

    Implementation

    We use a combination of ICD-10 codes (for hospitalisation and death records) andBNF codes (prescribing records) to adjudicate dementia sub-type as listed in theconcept list below. With regards to prescriptions, given that in the UK, medicationsfor dementia (donepezil hydrochloride, galantamine, rivastigmine, and memantine hydrochloride)are largely indicated for individuals with Alzheimer dementia, a history of redeeminga prescription for one of these drugs can be considered indicative of a diagnosisof Alzheimer's.

    Given the longitudinal nature of our records, we make use of the entire patient historyto determine a distribution over dementia sub-types (if any). Specifically, we summariseeach patient's coding episode into a binary indication of whether each dementia subtype(Alzheimer's, Vascular, Other, Unspecified or Possible) is mentioned or not. Foreach patient, we then count the number of coding episodes for each subtype and divideit by the total number of coding episodes over all subtypes, to yield a proportion(probability) over dementia subtypes for that patient - if the patient has no evidenceof dementia, then this proportion is taken to be all 0s. Subsequently, for dementiasubjects, the earliest date of mention of dementia (in any of the sources) is takenas a surrogate for the incident date of dementia, while the last date on record fora non-dementia patient is taken to be the last date of follow-up for that subject.

    For more detail on implementation, please refer to linked publications.

  4. Clinical Codelist

  5. Publication

    Related publications

    • Doney ASF, Bonney W, Jefferson E, Walesby KE, Bittern R, Trucco E, Connelly P, McCrimmon RJ, Palmer CNA. Investigating the Relationship Between Type 2 Diabetes and Dementia Using Electronic Medical Records in the GoDARTS Bioresource. Diabetes Care. 2019 Oct;42(10):1973-1980. doi: 10.2337/dc19-0380. Epub 2019 Aug 7. PMID: 31391202.

      (DOI: 10.2337/dc19-0380)

    Citation Example

    Doney ASF, Bonney W, Jefferson E, Walesby KE, Bittern R, Trucco E, Connelly P, McCrimmon RJ, Palmer CNA, Camilleri MPJ, Whiteley W. PH1717 / 3756 - Dementia Identification with EMR. Phenotype Library [Online]. 14 May 2025. Available from: http://phenotypes.healthdatagateway.org/phenotypes/PH1717/version/3756/detail/. [Accessed 06 June 2026]

  6. API

    To Export Phenotype Details:

    FormatAPI
    JSON site_root/api/v1/phenotypes/PH1717/version/3756/detail/?format=json
    R Package

    # Download here

    library(ConceptLibraryClient)


    # Connect to API

    client = ConceptLibraryClient::Connection$new(public=TRUE)


    # Get details of Phenotype

    phenotype_details = client$phenotypes$get_detail(
     'PH1717',
     version_id=3756
    )

    Py Package

    # Download here

    from pyconceptlibraryclient import Client


    # Connect to API

    client = Client(public=True)


    # Get details of Phenotype

    phenotype_detail = client.phenotypes.get_detail(
     'PH1717',
     version_id=3756
    )

    To Export Phenotype Code List:

    FormatAPI
    JSON site_root/api/v1/phenotypes/PH1717/version/3756/export/codes/?format=json
    R Package

    # Download here

    library(ConceptLibraryClient)


    # Connect to API

    client = ConceptLibraryClient::Connection$new(public=TRUE)


    # Get codelist of Phenotype

    phenotype_codelist = client$phenotypes$get_codelist(
     'PH1717',
     version_id=3756
    )

    Py Package

    # Download here

    from pyconceptlibraryclient import Client


    # Connect to API

    client = Client(public=True)


    # Get codelist of Phenotype

    phenotype_codelist = client.phenotypes.get_codelist(
     'PH1717',
     version_id=3756
    )

  7. Version History

    Version IDNameOwnerPublish date
    3973 Dementia Identification with EMR skrueger2025-06-11 16:15
    Dementia Identification with EMR skrueger2025-05-14 17:08
    3652 Dementia Identification with EMR skrueger2025-03-03 14:58
    3642 Dementia Identification with EMR skrueger2025-02-13 14:47