#1402 CCG-level coding for chronic kidney disease and regional kidney replacement therapy incidence: an ecological study in England

Abstract

Abstract Background and Aims With ageing of the population both prevalence of chronic kidney disease (CKD) and incidence of kidney replacement therapy (KRT) are rising. Existing research suggests that adding a formal diagnosis-code for CKD to the health record in those affected is associated with better implementation of recommended care and fewer hospitalisations for cardiovascular diseases. Furthermore, reliable estimates of the future need of KRT cannot be extrapolated from previous years after the COVID-19 pandemic. This ecological study investigates whether coding for CKD is associated with KRT incidence at clinical commissioning group (CCG) level in England. Method KRT incidence rates for CCGs in England were calculated using UK Renal Registry (UKRR) data from 01/2019 to 12/2021. Data on the percentage of uncoded CKD patients (PUCP) who had laboratory evidence of CKD but lacked a diagnostic code were obtained from the Cardiovascular Disease Prevention Audit (CVDPREVENT), a national primary care audit that extracts routinely held general practitioner (GP) data. The PUCP was modelled both as a continuous and a categorical variable using quintiles with an equally large population in each quintile to investigate potential non-linear associations. Data on potential confounders and acute kidney injury (AKI) mortality as a marker for underlying population frailty were obtained from CVDPREVENT and the UKRR, respectively. A hierarchical conceptual framework was used to assess the association between the PUCP and KRT incidence, in which covariates were grouped into sociodemographic, healthcare quality and health status covariates. Poisson models assessed the association between PUCP and KRT incidence and effect modification by AKI mortality was investigated using a simplified model for which AKI mortality was recoded to a binary variable. Results The crude overall KRT incidence rate was 141.30/1 000 000, ranging from 84.54 to 204.05/1 000 000 in the different CCGs (Fig. 1A). The mean PUCP was 15.47% (SD 5.55), ranging from 3.68% to 30.28%, with no immediately identifiable association with KRT incidence (Fig. 1B). Univariable Poisson regression showed no evidence of an association between the crude PUCP as a continuous measure and KRT incidence (p = 0.886). However, when included as a categorical variable, there was strong evidence (p < 0.001) for a difference in the crude KRT incidence rates, indicating a non-linear association. After adjusting for all covariates, there was evidence for an association between the PUCP as a continuous measure and KRT incidence (RR 1.004, 95% CI 1.000-1.008, p = 0.029). However, the PUCP was also considered as a categorical variable and there was still evidence of a difference in rates (p = 0.030) with the CCGs in the lowest PUCP quintile having a lower KRT incidence than the others. Considering the PUCP as a continuous variable, there was some evidence (p = 0.033) for effect modification: in CCGs with low AKI mortality, there was no evidence of an association between the PUCP and KRT incidence (RR 1.001, 95% CI 0.996-1.005, p = 0.823), whereas a 7% increase in KRT incidence for every 10% increase in uncoded CKD patients was found in CCGs with high AKI mortality (RR 1.007, 95% CI 1.003-1.012, p = 0.024) (Fig. 2A). Considering the PUCP as a categorical variable, there was very strong evidence (p < 0.001) for effect modification by AKI mortality: In CCGs with low AKI mortality, there was no evidence that rates were different in higher quintiles of the PUCP compared to the lowest. However, in CCGs with high AKI mortality, KRT incidence rates from the second up to the fifth quintile of the PUCP were between 10.3% and 18.5% higher than in the first quintile (Fig. 2B). Conclusion At a geographical level in England, the data suggest that the prevalence of not having formally diagnosed CKD is non-linearly associated with a higher KRT incidence rate, especially in areas with a high AKI mortality. The PUCP can be seen as a proxy for care quality. Therefore, improved care quality might be associated with lower KRT incidence rates. After the COVID-19 pandemic, PUCP follow-up data may be the first way to estimate future KRT needs and support the planning process.