P134 Collaborating with the National Congenital Anomaly and Rare Disease Registration Service to estimate national incidence of Kawasaki disease in England

Abstract

Abstract Background The true incidence of Kawasaki disease (KD) in England is unknown. In KD treatment with intravenous immunoglobulin (IVIG) can be lifesaving, but it is both expensive and in limited supply. In collaboration with Public Health England’s National Congenital Anomaly and Rare Disease Registration Service (NCARDRS) we aimed to determine the incidence of KD in England. Methods We worked within NCARDRS to access patient-identifiable Hospital Episode Statistics (HES) data, enabled by their legal permissions (CAG 10-02(d)/2015). We extracted all patients receiving a primary or additional diagnostic code for KD between April 2003 -March 2017. For patients identified in Nottingham University Hospitals Trust, we reviewed the coding accuracy in their medical records. We calculated incidence with 95% Poisson confidence intervals among children in England using the total number of cases newly diagnosed in the study period. We used the ONS mid-year estimates for 2016 for the denominator. Results 37 patients were coded as having KD in our Trust. All had a clinical diagnosis of KD (positive predictive value 100%). 32 patients were paediatric cases treated for KD; the remaining 5 patients were diagnosed in adult life on the basis of pathognomonic angiogram findings. It was not possible to apply classification criteria because the required information was not always available. During the 13-year study period we identified 3345 cases in children aged <5, and 4234 cases in children <16 in England. The incidence in children aged <5 years was 7.50 (95% CI: 7.25-7.76)/100,000 person-years, and 40.7% of cases were female. Incidence in children aged <16 was 3.09 (95% CI: 3.00-3.19)/100,000 person-years, and 41% of cases were female. Conclusion Analysis of HES data within NCARDRS has provided population-based estimates of incidence of KD for the whole of England. Our estimates are higher than in previous studies. Validation of coding accuracy has shown our methods were specific, and they are likely to be sensitive given KD is always treated in hospital. Accurate incidence estimates are essential for appropriate resource allocation, given the limited availability of lifesaving IVIG treatment. Disclosures M. Rutter None. P.C. Lanyon None. M. Bythell None. S. Stevens None. J. Aston None. F.A. Pearce None.