P090\u2003Validation of hospital episode statistics data to enable national registration of idiopathic inflammatory myopathies

Abstract

\n Background/Aims\u2002\n The National Congenital Anomaly and Rare Disease Registration Service (NCARDRS) aims to register all patients in England with congenital or rare diseases. Maintaining ongoing contemporaneous registers requires substantial time and resource, so novel methodologies for identification and registration of rare disorders such as the idiopathic inflammatory myopathies (IIMs) need development. Hospital Episode Statistics (HES) are routinely recorded data from every hospital admission within the National Health Service (NHS) England. This study aims to validate diagnostic ICD-10 codes within HES as a method of identifying and recording cases of IIM.\n Methods\u2002\n In collaboration with the Registration of Complex Rare Diseases - Exemplars in Rheumatology (RECORDER) project we extracted all Finished Consultant Episodes (FCE) at one hospital trust between 2010-2019 with diagnostic ICD-10 codes indicating either dermatomyositis or polymyositis under NCARDRS section 251 legal permissions (CAG 10-02(d)/2015). Co-occurring interstitial lung disease (ILD) was also considered due to ILDs association with IIMs. Hospital rheumatology and immunology databases were interrogated to identify all outpatients with IIM. Hospital electronic care records were reviewed to confirm coding accuracy. The total number of FCEs at the trust was calculated from NHS Digital reports. Sensitivity and specificity of each code and various code combination algorithms were calculated.\n Results\u2002\n HES ICD-10 codes identified 530 individuals, of whom 388 had records confirming IIM diagnosis. 146 additional patients were found through hospital records. This gives these codes a positive predictive value (PPV) of 73.1% and sensitivity of 72.7%. Due to the rarity of IIM coded FCEs compared to all FCEs, the specificity is 100.0% for every code. Dermatomyositis codes M331 and M339 had PPV 94.8% and 93.4% respectively. Juvenile DM (M330) and paraneoplastic myositis (M360) whilst rarely used had a high PPV of 91.6% and 100% respectively. Myositis unspecified (M609), was the worst performing code with PPV 61.4%. Excluding M609 from the analysis, PPV improved to 85.9%. Adding an ILD code made specificity 94.2% and sensitivity 15.5%, however it did find 51.5% of the known IIM-ILD patients. Strategies to improve PPV without impacting sensitivity were attempted. The highest PPV of 95.3% was found by selecting patients who had at least 2 different codes recorded, however sensitivity was only 30.7%. By then adding patients who also had one of the high performing codes the PPV dropped to 91.73% but sensitivity improved to 60.7%.\n Conclusion\u2002\n This is the first validation cohort to show that IIM ICD code combinations in HES have high positive predictive values, and sensitivities of up to 72.7%. The relative importance of sensitivity, specificity and PPV vary according to the research question being asked, but algorithms tested in this study could be applied across all NHS trusts to enable robust and cost-effective whole-population research into the epidemiology of IIM.\n Disclosure\u2002\n J.R. Hannah: None. P. Gordon: None. J. Galloway: None. E.J. Peach: Grants/research support; Vifor Pharma. P.C. Lanyon: Grants/research support; Vifor Pharma. M. Bythell: None. F.A. Pearce: Grants/research support; Vifor Pharma.