Abstract
Lyme disease is emerging in eastern and central Canada, and most cases are diagnosed using the two-tier serological test (Enzyme Immuno Assay [EIA] followed by Western blot [WB]). Simplification of this algorithm would be advantageous unless it impacts test performance. In this study, accuracy of individual proteins of the IgG WB algorithm in predicting the overall test result in samples from Canadians was assessed. Because Borrelia burgdorferi strains vary geographically in Canada, geographic variations in serological responses were also explored. Metrics of relative sensitivity, specificity and the kappa statistic measure of concordance were used to assess the capacity of responses to individual proteins to predict the overall IgG WB result of 2524 EIA (C6)-positive samples from across Canada. Geographic and interannual variations in proportions of samples testing positive were explored by logistic regression. No one protein was highly concordant with the IgG WB result. Significant variations were found amongst years and geographic regions in the prevalence of samples testing positive using the overall IgG WB algorithm, and for individual proteins of the algorithm. In most cases the prevalence of samples testing positive were highest in Nova Scotia, and lower in samples from Manitoba westwards. These findings suggest that the current two tier test may not be simplified and continued use of the current two-tier test method and interpretation is recommended. Geographic and interannual variations in the prevalence of samples testing positive may be consistent with B. burgdorferi strain variation in Canada, and further studies are needed to explore this.