Multi-annual performance evaluation of laboratories in post-mortem diagnosis of animal rabies: Which techniques lead to the most reliable results in practice?

International audienceRabies diagnosis proficiency tests on animal specimens using four techniques (FAT, RTCIT, conventional RT-PCR and real-time RT-PCR) were organised over 10 years (2009–2019). Seventy-three laboratories, of which 59% were from Europe, took part. As the panels were prepared with experimentally-infected samples, the error rate of laboratories on positive and negative samples was accurately estimated. Based on fitted values produced by mixed modelling including the variable “laboratory” as a random variable to take into account the longitudinal design of our dataset, the technique that provided the most concordant results was conventional RT-PCR (99.3%; 95% CI 99.0–99.6), closely followed by FAT (99.1%; 95% CI 98.7–99.4), real-time RT-PCR (98.7%; 95% CI 98.1–99.3) and then RTCIT (96.8%; 95% CI 95.8–97.7). We also found that conventional RT-PCR provided a better diagnostic sensitivity level (99.3% ±4.4%) than FAT (98.7% ±1.6%), real-time RT-PCR (97.9% ±0.8%) and RTCIT (95.3% ±5.1%). Regarding diagnostic specificity, RTCIT was the most specific technique (96.4% ±3.9%) followed closely by FAT (95.6% ±3.8%), real-time RT-PCR (95.0% ±1.8%) and conventional RT-PCR (92.9% ±0.5%). Due to multiple testing of the samples with different techniques, the overall diagnostic conclusion was also evaluated, and found to reach an inter-laboratory concordance level of 99.3%. The concordance for diagnostic sensitivity was 99.6% ±2.0% and for diagnostic specificity, 98.0% ±8.5%. Molecular biology techniques were, however, found to be less specific than expected. The potential reasons for such findings are discussed herein. The regular organisation of performance tests has contributed to an increase in the performance of participating laboratories over time, demonstrating the benefits of such testing. Maintaining a high-quality rabies diagnosis capability on a global scale is key to achieving the goal of eliminating dog-mediated human rabies deaths. The regular organisation of exercises on each continent using selected local strains to be tested according to the local epidemiological situation is one factor that could help increase reliable diagnosis worldwide. Rabies diagnosis capabilities could indeed be enhanced by providing adequate and sustainable proficiency testing on a large scale and in the long ter