Cross-Provincial Validation of a Giant Cell Arteritis Case-Identification Algorithm Using Alberta Health Data

Objectives Region-specific giant cell arteritis (GCA) epidemiology is crucial for targeted health system planning and resource allocation. In Canada, only 2 population-based studies on GCA exist to date, both of which are from Ontario.[1,2] Health system heterogeneity across Canadian provinces limits the transferability of Ontario GCA estimates. As a result, it is necessary to validate methods for GCA case ascertainment cross-provincially. Methods We conducted a chart review to validate an Ontario administrative case-identification algorithm for GCA,[1] against the ACR/EULAR GCA Classification Criteria (2022),[3] within Southern Alberta. Our study included patients aged 50 and older diagnosed with GCA from January 1, 2018, to December 31, 2023, in Alberta. We began by applying Ontario’s algorithm for GCA to a linked Alberta health administrative dataset to produce a subset of algorithm-positive cases. This pool was further sampled to generate a random selection of 100 cases within Southern Alberta for chart review. This sample size was calculated using the Wald method and assumed a 95% confidence level, a 5%-10% margin of error, and an expected positive predictive value (PPV) of 81.3% similar to Ontario findings. The 2022 ACR/EULAR GCA classification criteria served as our reference standard and reflected the current GCA diagnostic consensus.[3] As a sensitivity analysis, PPV was also computed using the treating rheumatologist diagnoses as an alternative reference standard. PPV was calculated as True Positives (TP) divided by the number of algorithm-positive charts (n = 100). Results 1,903 algorithm-positive cases were identified in Alberta using the Ontario case definition and 100 were randomly selected for chart review. When evaluating the performance of the Ontario algorithm in the Southern Alberta cohort (n = 100), the case definition yielded a PPV of 64% (95% Confidence Interval [CI] 54.6-73.4%). In the sensitivity analysis, the Ontario administrative case definition algorithm produced comparable estimates when validated against rheumatologist diagnoses as the alternative reference standard (PPV: 62%, CI 52.5-71.5%). Conclusion The Ontario administrative case definition algorithm demonstrated moderate PPV in the Southern Alberta cohort (64%; 95% CI 54.6-73.4%), which was lower than the Ontario cohort estimate (81.3%; 95% CI 73.4-89.1%).[1] The confidence intervals showed minimal overlap, with the upper limit of our CI approaching the lower limit of the Ontario estimate, suggesting possible, but limited, agreement between cohorts. Differences may reflect potential regional or methodological differences including, reference standard selection across studies and sample size. Refinement of the algorithm may improve its performance in Alberta. References [1.] Barra L. Rheumatology (Oxford) 2020;59:3250-8. [2.] Ing EB. Can J Ophthalmol 2019;54:119-24. [3.] Ponte C. Ann Rheum Dis 2022;81:1647-53.