Submitted to: Parasitology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/13/2015
Publication Date: 12/2/2015
Citation: Dubey, J.P., Kwok, O.C., Gardner, I. 2015. Bayesian estimation of sensitivity and specificity of the modified agglutination test and bioassay for detection of Toxoplasma gondii in free-range chickens. Parasitology. 143:314-319.
Interpretive Summary: Toxoplasmosis continues to be a public health problem worldwide. It causes mental retardation and loss of vision in children, and abortion in livestock. Cats are the main reservoir of T. gondii because they are the only hosts that can excrete the resistant stage (oocyst) of the parasite in the feces. Humans become infected by eating under cooked meat from infected animals and food and water contaminated with oocysts. Tests to detect antibodies against the parasite are used to diagnose Toxoplasma infection.. However, it is not known if a seropositive host has viable organisms. In the present study, the authors verified that in the modified agglutination test seropositivity equated with the presence of viable parasites. The results will be of interest to biologists, parasitologists, and veterinarians.
Technical Abstract: Toxoplasma gondii infects virtually all warm-blooded animals worldwide. Serological tests, including the modified agglutination test (MAT), are often used to determine exposure to the parasite. The MAT can be used for all hosts because it does not need species-specific reagents and has been shown to be 83% sensitive and 90% specific in adult sows. The objective of the present study was to evaluate the diagnostic sensitivity and specificity of the MAT and bioassay in free-range/backyard (FR) chickens (Gallus domesticus). Previously-published T. gondii test results from 2066 chickens from 19 countries were compiled for the present study. The frequency of isolation of T. gondii increased for MAT titers between 1:5 and 1:160, and ranged from 61% to 75% for titers of 1:160, 1:320, and = 1:640. The sensitivity and specificity of the MAT and mouse bioassay were estimated using a 2-test 4-population Bayesian latent-class model that assumed conditional independence of the MAT and mouse bioassay. Median estimates of MAT sensitivity and specificity (at a cutoff of 1:10 as positive and using non-informative priors) were 99.0% and 73.5%, respectively, and median estimates of mouse bioassay sensitivity and specificity were 97.7% and > 99.1%, respectively. When a MAT cutoff of 1:20 was used as the positive threshold, MAT sensitivity decreased by about 2% and specificity increased by about 4% indicating little practical change in test accuracy. Cats fed pooled hearts from 802 FR seronegative (MAT, <1:5) chickens from several countries did not shed oocysts, indicating high test specificity because FR chickens would have been exposed to many microbes.