An update on FIV and FeLV test performance using a Bayesian statistical approach.

Area of Science:

• Veterinary Diagnostics
• Infectious Disease Epidemiology
• Statistical Modeling

Background:

• Feline retroviral screening tests (FeLV, FIV) are presumed accurate but require robust evaluation.
• Previous studies assessing diagnostic test performance have inherent limitations.
• Advanced statistical methods, like Bayesian approaches, enable more reliable sensitivity and specificity estimation.

Purpose of the Study:

• To assess the sensitivity and specificity of various feline retrovirus diagnostic tests.
• To evaluate screening tests for feline leukemia virus (FeLV) and feline immunodeficiency virus (FIV) in a potentially infected cat population.
• To apply a Bayesian statistical framework for accurate diagnostic test performance evaluation.

Main Methods:

• Utilized Bayesian statistical methods for analyzing diagnostic test results.
• Tested 490 feline blood samples for FIV using rapid immunomigration (Witness) and ELISA tests, alongside real-time PCR.
• Tested 495 feline blood samples for FeLV using rapid immunomigration (Witness) and ELISA tests, alongside virus isolation.

Main Results:

• FIV testing showed high median sensitivity (0.92-0.98) and specificity (0.93-0.99) across methods.
• FeLV testing demonstrated high median sensitivity (0.91-0.98) and specificity (0.96-0.99) across methods.
• Real-time PCR and virus isolation exhibited the highest specificity for FIV and FeLV, respectively.

Conclusions:

• Bayesian statistical methods effectively address limitations in diagnostic test evaluation, including the absence of a gold standard.
• All evaluated screening tests for FeLV and FIV exhibited high sensitivity and specificity.
• Specificity estimates for some tests were slightly lower than previously reported in literature.