P108 Development of a novel algorithm to reliably identify patients with hereditary haemochromatosis in electronic patient records

<h3>Introduction</h3> Hereditary haemochromatosis (HH) is the commonest genetic condition among populations of Northern European ancestry. HH is characterised by excess iron deposition in various organs especially the liver. Despite its high prevalence, there is no specific International Classification of Diseases 10th edition (ICD-10) diagnoses code for HH making it challenging to identify these patients from electronic patient records (EPR). Patients with HH are usually assigned the ICD-10 code E83.1 ‘disorders of iron metabolism’. Many studies utilising large healthcare datasets have relied exclusively on the E83.1 code for identifying HH patients. However, patients with transfusion related iron overload from haematological conditions such as aplastic anaemia are also assigned the E83.1 code. HH is treated with venesection which has an OPCS Classification of Interventions and Procedures 4th edition (OPCS-4) code X36.2. We aimed to develop a novel algorithm to identify HH patients from EPR using ICD-10 and OPCS-4 codes. <h3>Methods</h3> All patients with an ICD-10 code of E83.1 within the Hospital Episodes Statistics Admitted Patient Care database from 1st April 2018 to 31st March 2023 in York and Scarborough Teaching Hospitals NHS Foundation Trust were included. Review of patients’ case notes was undertaken to assess their diagnosis, HH genotype, past medical history and previous medical procedures. A patient was considered to have HH if they were homozygous for the C282Y mutation. STATA MP 18 was used to create the algorithms using a combination of ICD-10 and OPCS-4 codes. <h3>Results</h3> A total of 9264 patient episodes were analysed corresponding to 787 unique patients identified with ICD-10 code E83.1. Only 479/787 (60.9%) of patients had a diagnosis of HH while 107 (13.6%) were C282Y/H63D compound heterozygotes and 42 (5.3%) had other HH genetic mutations. Of the remaining 159/787 (20.2%) patients without HH mutations, their diagnoses were 73 hyperferritinaemia, 49 incorrect coding, 17 transfusion related iron overload, 14 polycythaemia and 4 porphyria cutanea tarda (table 1). Among the patients with hyperferritinaemia, 24/73 (32.9%) had undergone venesection during the study period. Five algorithms were generated to improve the identification of patients with HH in EPR. The optimised version (algorithm 4) had a sensitivity of 85.8% and a positive predictive value (PPV) of 74.3% in detecting patients with HH. <h3>Conclusions</h3> Our innovative algorithm provides a more reliable method to identify HH patients from EPR compared to solely using the ICD-10 code E83.1. This is most valuable when carrying out population studies using large healthcare databases.