Abstract: SA-PO390
Geospatial Clustering of Incident Glomerulonephritis Suggests Disease-Specific Risk Factors
Session Information
- Glomerular Diseases: Clinical, Outcomes, Trials - III
October 27, 2018 | Location: Exhibit Hall, San Diego Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Glomerular Diseases
- 1203 Glomerular Diseases: Clinical, Outcomes, and Trials
Authors
- Canney, Mark, University of British Columbia, Vancouver, British Columbia, Canada
- Induruwage, Dilshani, BC Provincial Renal Agency, Vancouver, British Columbia, Canada
- Reich, Heather N., Toronto General Hospital, Toronto, Ontario, Canada
- Barbour, Sean, University of British Columbia, Vancouver, British Columbia, Canada
Background
Few studies have rigorously evaluated regional differences in the incidence of glomerulonephritis (GN) subtypes, thus limiting our understanding of potential sociodemographic or environmental risk factors. As such, we used population-level data from British Columbia (BC), Canada, to investigate geospatial differences in the incidence of biopsy-proven membranous nephropathy (MN), IgA nephropathy (IgAN) and ANCA-associated vasculitis (AAV).
Methods
All native kidney biopsies in BC from 1/1/2000 to 12/31/2012 were analyzed from a central registry to identify all cases of GN. Patient-level data were captured via linkage with provincial administrative databases. We used local health authorities (n=75) to define discrete geographical regions in BC (population 4.6 million) with region-level age, sex and race distributions from census data. For each GN we employed a hierarchial Bayesian model to estimate the incidence rate ratio (IRR) for each region accounting for adjacent spatial correlation. Contiguous regions with high probability of IRR>1.0 were combined into super-regions to estimate the IRR adjusted for age, sex and race.
Results
A total of 1624 patients were included: 401 MN (mean age 56, 57% male); 824 IgAN (mean age 44, 61% male); 399 AAV (mean age 61, 45% male). For each GN we identified several regions with high probability of being a cluster (Figure). The number of clusters and magnitude of increased risk were greatest for AAV (IRR 1.1-3.5), intermediate for IgAN (IRR 1.2-2.2) and lowest for MN (IRR 1.1-1.9). Results were similar for aggregated super-regions after multivariable adjustment.
Conclusion
Using a novel and robust approach, we describe geospatial clustering of biopsy-proven GN which varied by GN subtype and was not explained by inter-regional differences in age, sex and race. Our findings suggest that health services delivery for each type of GN needs to be tailored to individual regions at higher risk, and that there are likely disease-specific environmental and/or genetic risk factors that require further study.
Areas with high probability of being a cluster for MN (a), IgAN (b) and AAV (c). The color gradient represents the magnitude of IRR.