A voluntary trichomonosis inter-laboratory comparison study in South Africa

Abstract

Trichomonosis is currently the most important venereal disease of cattle in South Africa with adverse economic implications to the beef production industry due to cow abortions, infertility and culling of carrier bulls. Once diagnosed in a herd, eradication is difficult due to financial and biological implications. Bulls are asymptomatic carriers and susceptibility increases with age. In infected females, clinical signs include embryonal death, abortion, pyometra, foetal maceration and uterine discharge. Diagnostic accuracy is one of the major clinical problems preventing easy eradication of trichomonosis from a herd and can be influenced by biological variance in the occurrence of the organism, sampling errors, sample degradation during sample transport and diagnostic laboratory inaccuracies. This study aimed to validate the accuracy of voluntarily enrolled private (n = 8) and state-owned (n = 5) laboratories that perform trichomonosis diagnostic tests by estimating the sensitivity (Se) and specificity (Sp) per laboratory. It was hypothesized that diagnostic laboratories in South Africa play an insignificant role in the inaccuracy of the diagnosis of trichomonosis. Laboratories performed either the culture method (n = 5), polymerase chain reaction (PCR) (n = 6) or a combination of culture and PCR (n= 2). Fresh preputial scrapings from four bulls with known negative status for trichomonosis were pooled in 200ml of phosphate buffered saline (PBS) to form the sample base for 12 subsamples of 13ml each. Duplicate subsamples were then contaminated with 2ml originating from four different laboratory cultures of Tritrichomonas foetus or 2ml of culture medium for four negative samples. Aliquots of the subsamples were transferred to an anaerobic transport medium, and the final concentration reached in these samples submitted to the laboratories, were categorised as follows: weak (<10 organisms/μl), moderate (10 – 30 organisms/μl) or strong (>30 organisms/μl). A total of 312 samples were sent by courier in two separate rounds: eight (4 duplicates) positive and four negative samples per round. Multiple logistic regression was performed on sensitivity, using sampling round, laboratory sector, diagnostic test type and sample concentration as independent variables, and removing variables in a stepwise manner based on the highest P-value. Two public laboratories only reported on one round of sampling, and one batch of 12 samples was severely delayed in reaching another public laboratory. The sample identifications of a further two batches were not recorded by the respective private laboratories. The results from these 60 unreported samples were not included in the analysis. Laboratories that performed the PCR assay (solely, or in addition to culture) were grouped for data analysis. The overall specificity (Sp) was 100% and the sensitivity was 88.7% (95% CI 83.9% - 93.5%). Laboratories using PCR recorded higher sensitivity than those using the culture method (95.5%; 95% CI 91.0% – 99.9% and 81.3%; 95% CI 72.5% - 90.0% respectively, P < 0.01), and private laboratories recorded higher Se than public laboratories (96.4%; 95% CI 92.9% - 99.9% and 73.2%; 95% CI 61.2% - 85.2%, P < 0.01). For laboratories using PCR, weak positive samples recorded a lower sensitivity than strong positive samples (86.4%; 95% CI 70.8% - 101.9% and 100%; 95% CI 100% - 100%, respectively, P < 0.01). One public and six private laboratories obtained 100% accuracy during the two sampling rounds. In the logistic regression model, private sector (compared to public), an increasing concentration of organisms in the sample and the second round of sampling (compared to the first round) were independent predictors of laboratory sensitivity for the detection of Tritrichomonas foetus. It is concluded that inaccuracies in the diagnostic laboratory contributes to the deficiencies in diagnostic sensitivity for trichomonosis in South Africa, but does not influence diagnostic specificity. It is further concluded that diagnostic sensitivity was independently influenced by the sector in which the laboratory operates (private vs public) and the concentration of Tritrichomonas foetus organisms in the sample.