Abstract
Background/Aims
Understanding real-world usage of high cost drugs and subsequent outcomes are crucial to support high quality care, adoption of innovation, and reduce unwarranted variation in treatment. Hospital Episode Statistics (HES) contain diagnoses (coded by ICD-10) and procedures/treatment (coded by OPCS) for admissions in England. However, OPCS codes are not specific for individual drugs, for example X921 (cytokine inhibitors band 1) includes rituximab and 15 other drugs. We aimed to validate the accurate identification of patients treated with rituximab for ANCA-associated vasculitis (AAV) using HES data.
Methods
We used data available to the National Congenital Anomaly and Rare Disease Registration Service (NCARDRS), part of NDRS at Public Health England and their legal permissions (CAG 10-02(d)/2015). We extracted from HES all patients treated at one hospital who ever had a coded diagnosis of either granulomatosis with polyangiitis, eosinophilic granulomatosis with polyangiitis, microscopic polyangiitis, polyarteritis nodosa or 'arteritis unspecified’, which is applied by coders to several clinical terms including “systemic vasculitis” and “ANCA associated vasculitis”. Enabled by data sharing agreements, we reviewed hospital records of those patients who had a Finished Consultant Episode (FCE) during 2018/19, to validate diagnoses and whether X921 reliably identified rituximab.
Results
Initially using codes only for granulomatosis with polyangiitis, eosinophilic granulomatosis with polyangiitis, or microscopic polyangiitis we identified 65 people, 60 of whom had AAV confirmed in their medical notes (Positive predictive value (PPV) 92.3%; 95% CI 83.0-97.5). When we expanded our search to include in addition people with codes for polyarteritis nodosa or ‘arteritis unspecified’ who had rheumatology or renal follow up, we identified an additional 13 people with AAV and 10 who had other types of systemic vasculitis (PPV for AAV 79.5, 95% CI 69.6-87.4; PPV for systemic vasculitis 90.9%, 95% CI 82.9-96.0). Among patients identified only by the specific ICD-10 codes for the 3 subtypes of AAV there were 51 episodes coded as X921, of which 50 correctly identified a rituximab infusion (and 1 was tocilizumab). No additional infusions for these patients were missed by the coding. The PPV was 98.0% (95% CI 89.6-100). Among patients identified by the more inclusive algorithm, there were 61 episodes coded as X921 in that year. Of these, 59 were for rituximab treatment for AAV and an additional 4 rituximab infusions occurred but were not coded. The PPV was 96.7% (95% CI 88.7-99.6), and the sensitivity was 93.7% (95% CI 84.5-98.2).
Conclusion
This pilot study demonstrates how the use of novel algorithms within the NDRS RECORDER project can enable the identification and registration of people with AAV and their high-cost treatment at whole-population level. Further work is now underway to study this national cohort, and to apply this methodology to other rare autoimmune diseases and high cost drugs.
Disclosure
M. Chakravorty: None. F. Pearce: Grants/research support; Recipient of research grant from Vifor Phama. Vifor Pharma had no influence on the design, interpretation or reporting of this work. M. Rutter: None. P. Lanyon: Grants/research support; Recipient of research grant from Vifor Phama. Vifor Pharma had no influence on the design, interpretation or reporting of this work. J. Aston: None. M. Bythell: None. S. Stevens: None.
Immunopathogenesis of ANCA-Associated Vasculitis
