Relative accuracy of point-of-care tests to rule-in heartworm infection in clinically suspected dogs using Bayesian latent class modelling.

Abstract

Our prospective cross-sectional survey of dogs from Western Fiji aimed to evaluate the fitness of common diagnostic test modalities to rule-in Dirofilaria immitis infection in patients with suggestive clinical history or signs. In the absence of a perfect reference standard, we used latent class modelling to evaluate the relative diagnostic accuracy of two point-of-care (POC) detection modalities - the modified Knott's test (MKT) to detect circulating microfilaria and four antigen lateral flow immunoassays (LFI): Anigen Rapid Canine HW Ag Test® (Bionote Co.), SNAP® Heartworm RT Test (IDEXX Laboratories), trūRapid Heartworm (Heska), WITNESS Dirofilaria® (Zoetis Inc.). The tests' fitness for ruling-in infection were compared using the likelihood ratios of a positive result (LR). The performances of the MKT and the Anigen Rapid LFI to rule-in infection on fresh blood of clinically suspected dogs were moderate to strong (LR=13.4, 95 %PCI: 6.7-114.6; LR=20.2, 95 %PCI: 5.4-138.2; respectively) but not consistently different from each other. The Anigen Rapid, SNAP and truRapid tests consistently provided the strongest evidence to rule-in infection. The LRof the WITNESS test was approximately twelve times, nine times and two times lower than the SNAP, truRapid and Anigen Rapid LFIs respectively (Bayesian p-value 0.002, 0.004 and 0.02 respectively). Overall, a positive result from MKT or LFIs is suitable to rule-in infection in dogs raising clinical suspicion and would increase the post-test probability of infection similarly. If veterinarians are choosing between LFIs, they should favour either Anigen Rapid, SNAP and truRapid over WITNESS.