The spectrum of co-existing disease in children with established kidney failure using registry and linked electronic health record data

Abstract

Abstract Background Children with established kidney failure may have additional medical conditions influencing kidney care and outcomes. This cross-sectional study aimed to examine the prevalence of co-existing diseases captured in the electronic hospital record compared to UK Renal Registry (UKRR) data and differences in coding. Methods The study population comprised children aged < 18 years receiving kidney replacement therapy (KRT) in England and Wales on 31/12/2016. Comorbidity data at KRT start was examined in the hospital record and compared to UKRR data. Agreement was assessed by the kappa statistic. Associations between patient and clinical factors and likelihood of coding were examined using multivariable logistic regression. Results A total of 869 children (62.5% male) had data linkage for inclusion. UKRR records generally reported a higher prevalence of co-existing disease than electronic health records; congenital, non-kidney disease was most commonly reported across both datasets. The highest sensitivity in the hospital record was seen for congenital heart disease (odds ratio (OR) 0.65, 95% confidence interval (CI) 0.51, 0.78) and malignancy (OR 0.63, 95% CI 0.41, 0.85). At best, moderate agreement (kappa ≥ 0.41) was seen between the datasets. Factors associated with higher odds of coding in hospital records included age, while kidney disease and a higher number of comorbidities were associated with lower odds of coding. Conclusions Health records generally under-reported co-existing disease compared to registry data with fair-moderate agreement between datasets. Electronic health records offer a non-selective overview of co-existing disease facilitating audit and research, but registry processes are still required to capture paediatric-specific variables pertinent to kidney disease. Graphical Abstract