Article Text

PDF
PWE-053 Development of data linkage techniques to support the uk ibd registry: making the most of routine nhs datasets
  1. K Bodger1,
  2. M Shawihdi1,
  3. M Pearson2,
  4. R Driscoll3,
  5. F Cummings4,
  6. S Bloom5
  7. UK IBD Registry
  1. 1Department of Gastroenterology, Institute of Translational Medicine
  2. 2Clinical Evaluation, University of Liverpool, Liverpool
  3. 3The Riva Partnership Ltd, Barnet
  4. 4Gastroenterology, University Hospital Southamptom, Southampton
  5. 5Gastroenterology, University College London Hospital, London, UK

Abstract

Introduction The IBD Registry aims to provide the first ever UK-wide repository of anonymized, individual-level data to support service delivery, audit and research. We report the Registry’s progress with development of methodologies to analyse the routinely coded data collected by all NHS hospitals (Hospital Episode Statistics, Outpatient and A&E attendances).

Method National HES data were interrogated to identify all admissions ‘belonging’ to people coded with an IBD-specific ICD-10 code (2-yr period), exploring coding variations and devising algorithms to distinguish IBD-related from unrelated admissions. At one NHS Trust, a linked dataset containing HES, OPD and A&E events (7-yr period) was merged in order to link all IBD-related activity. In collaboration with the HSCIC, we established the feasibility of linking individual (anonymized) patients from the UK IBD Registry with their routine HES data.

Results National HES data

84,314 patients had ≥1 admission with an IBD-specific diagnostic code (any position). We extracted their all-cause admissions (271,567). Of these, 119,045 admissions had a primary diagnosis of IBD. A further 77,957 admissions had IBD codes in position 2 or 3, but 39,515 (50%) had a relevant primary diagnosis (e.g. symptom or complication) or a procedure code that would indicate an IBD-related admission. A further 25,978 admissions had no IBD diagnosis coded but a relevant primary diagnosis or procedure (e.g. a colonoscopy).

Single Trust Datasets

1,941 IBD cases were identified by screening discharges over a 7 yr period, with 6,583 admissions categorised as IBD-related using the coding algorithms developed on national data. The activity rose to 7,507 by adding admissions recorded under an IBD consultant where relevant diagnosis or procedure codes were missing. We linked cases successfully to their OPD events (22,743 for GI specialists; 16,710 non-GI) and all-cause AED attendances (6,276; 39% discharged home). Reports of activity, patient characteristics, admission types, procedures and candidate metrics have been generated for IBD, UC and CD. Linkage between UK IBD Registry and HES has been achieved at patient-level, allowing outputs of routine data to be extracted and analysed for all registered cases.

Conclusion The UK IBD Registry is developing and testing algorithms that interrogate routine NHS data in ways that account for the realities of coding quality and bring together information from inpatient, outpatient and AED sources. The linkage of routine data to registry information offers the potential to build powerful tools to support IBD care delivery, audit and research.

Disclosure of interest None Declared.

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.