Conditional dependence between tests affects the diagnosis and surveillance of animal diseases

Assessment of sensitivity and specificity of bacterial culture and the VetMAX™ MastiType Multi Kit in detecting Streptococcus uberis and Escherichia coli in milk samples from dairy cows with clinical mastitis in subtropical Australia

Mastitis, a prevalent and economically important disease in the dairy industry, poses substantial challenges to dairy cow health, milk quality, and farm profitability worldwide. Mastitis is predominantly caused by bacterial infections. The objective of this study was to estimate the sensitivity (Se) and specificity (Sp) of bacterial culture and the VetMAX™ MastiType Multi Kit PCR in identified clinical mastitis pathogens. A total of 396 quarter-level milk samples were collected from 396 cows with clinical mastitis on 29 farms in the subtropical dairy region of Australia between March and December 2021. These samples were cultured and tested by PCR, and analysed using Bayesian latent class analysis under the assumption of one population two tests and also of three populations two tests, by dividing the population into subpopulations based on regions. Informative priors used in the analysis were calculated from published evidence. Models were compared using the Deviance Information Criterion (DIC). Sensitivity analysis was performed to evaluate the impact of changes in priors. The most common isolates cultured and detected by PCR were Streptococcus uberis (17.4 % and 27.3 %, respectively) and Escherichia coli (12.6 % and 25.0 %, respectively). Under the assumption of one population two tests, the Se of PCR (at cycle threshold (Ct) ≤ 37) was higher than that of bacterial culture for both pathogens: for E. coli, the Se was 50.2 % (95 % posterior probability interval (PPI): 37.4; 74.1) for bacterial culture, and 93.7 % (95 % PPI: 85.5; 98.4) for PCR. For S. uberis, the Se was 50.4 % (95 % PPI: 40.9; 61.3) for bacterial culture, and 81.5 % (73.0; 88.9) for PCR. Conversely, the Sp of bacterial culture was higher than that of PCR for both pathogens: for E. coli, the Sp was 99.2 % (97.8; 100) for bacterial culture, and 95.1 % (87.8; 99.4) for PCR. For S. uberis, the Sp was 99.2 % (95 % PPI: 97.6; 100) for bacterial culture, and 96.7 % (95 % PPI: 92.1; 99.2) for PCR. Bayesian latent class analysis with three populations two tests was only performed for S. uberis. For E. coli, this could not be performed because there were no PCR-positive results in one subpopulation. Under the assumption of three populations two tests, for S. uberis, the Se was 49.6 % (40.6; 59.4) for bacterial culture, and 81.1 % (72.6; 88.6) for PCR; and the Sp for bacterial culture was 99.1 % (97.7; 100), and for PCR was 96.9 % (93.0; 99.3). The DIC for the one population two tests model was lower than the DIC for the three populations two tests model. The sensitivity analysis for the one population two tests model demonstrated that a 10 % reduction in priors led to substantial changes in Se of both bacterial culture and PCR tests for E. coli and S. uberis, with overlap percentages ranging from 80.6 % to 92.2 %. In contrast, the Sp of bacterial culture and PCR tests remained relatively stable despite changes in priors, except for the Sp of PCR test for E. coli. In summary, the VetMAX™ MastiType Multi Kit demonstrated higher Se compared to bacterial culture, suggesting its potential as a routine test for identifying mastitis pathogens in milk samples from cows with clinical mastitis. While the bacterial culture method offered higher Sp in pathogen detection; results obtained following bacterial culture and subsequent susceptibility testing remain valuable, particularly in guiding antimicrobial treatment for mastitis.

Evaluation of three serological tests for the diagnosis of Brucella suis in dogs using Bayesian latent class analysis

Brucella suis infection of dogs is an emerging issue worldwide requiring specific management to address zoonotic risks and animal welfare concerns. Diagnosis in dogs is routinely based on serological testing, but to date these tests have only been validated for use in production animal species and humans. This study aimed to assess the diagnostic performance of three commonly used serological tests in dogs. Canine sera (n = 989) were tested with the Rose Bengal rapid plate agglutination test (RBRPT), the complement fixation test (CFT) and a competitive ELISA (C-ELISA). Diagnostic test performance was evaluated using a three test, two population Bayesian latent class analysis accounting for conditional dependence between the three tests. Positive and negative predictive values (PPV, NPV) were calculated for a range of expected prevalence estimates for the individual tests and test combinations interpreted in series and parallel. The RBRPT showed the highest individual Se of 0.902 (95 % posterior credible interval [PCI] 0.759-0.978) and the CFT the highest individual diagnostic specificity (Sp) of 0.914 (95 % PCI 0.886-0.946). The C-ELISA had marginally the best overall diagnostic performance (Youden's index = 0.807). The CFT and the C-ELISA interpreted in parallel returned the highest combined Se and Sp (0.988 and 0.885, respectively). All tests returned NPVs of > 0.982 in 2-8 % prevalence settings. Series interpretation of the three-test combination as well as the two-test combinations of the RBRPT and the C-ELISA and the CFT and the C-ELISA produced a PPV of 0.502 when the estimated prevalence was 8 %. While all tests are suitable for the detection of B. suis antibodies in dogs, they should not be interpreted in isolation as their diagnostic value is dependent on the pre-test probability of the disease. As such they are useful tools for the diagnosis of B. suis in dogs when exposure, history and clinical presentation indicate a risk of brucellosis.

A review of methods for the analysis of diagnostic tests performed in sequence

BACKGROUND

Many clinical pathways for the diagnosis of disease are based on diagnostic tests that are performed in sequence. The performance of the full diagnostic sequence is dictated by the diagnostic performance of each test in the sequence as well as the conditional dependence between them, given true disease status. Resulting estimates of performance, such as the sensitivity and specificity of the test sequence, are key parameters in health-economic evaluations. We conducted a methodological review of statistical methods for assessing the performance of diagnostic tests performed in sequence, with the aim of guiding data analysts towards classes of methods that may be suitable given the design and objectives of the testing sequence.

METHODS

We searched PubMed, Scopus and Web of Science for relevant papers describing methodology for analysing sequences of diagnostic tests. Papers were classified by the characteristics of the method used, and these were used to group methods into themes. We illustrate some of the methods using data from a cohort study of repeat faecal immunochemical testing for colorectal cancer in symptomatic patients, to highlight the importance of allowing for conditional dependence in test sequences and adjustment for an imperfect reference standard.

RESULTS

Five overall themes were identified, detailing methods for combining multiple tests in sequence, estimating conditional dependence, analysing sequences of diagnostic tests used for risk assessment, analysing test sequences in conjunction with an imperfect or incomplete reference standard, and meta-analysis of test sequences.

CONCLUSIONS

This methodological review can be used to help researchers identify suitable analytic methods for studies that use diagnostic tests performed in sequence.

Estimation of sensitivity and specificity of bulk tank milk PCR and 2 antibody ELISA tests for herd-level diagnosis of Mycoplasma bovis infection using Bayesian latent class analysis

Mycoplasmosis (due to infection with Mycoplasma bovis) is a serious disease of beef and dairy cattle that can adversely impact health, welfare and productivity (Maunsell et al. (2011)). Mycoplasmosis can lead to a range of often severe, clinical presentations. Mycoplasma bovis (M. bovis) infection can present either clinically or subclinically, with the potential for recrudescence of shedding in association with stressful periods. Infection can be maintained within herds because of intermittent shedding (Calcutt et al., 2018, Hazelton et al., 2018). M. bovis is recognized as poorly responsive to treatment which represents a major challenge for control in infected herds. Given this, particular focus is needed on biosecurity measures to prevent introduction into uninfected herds in the first place. A robust and reliable laboratory test for surveillance is important both for herd-level prevention and control. The objective of this study was to estimate the sensitivity and specificity of 3 diagnostic tests (one PCR and 2 ELISA tests) on bulk tank milk, for the herd-level detection of M. bovis using Bayesian latent class analysis. In autumn 2018, bulk tank milk samples from 11,807 herds, covering the majority of the main dairy regions in Ireland had been submitted to the Department of Agriculture testing laboratory for routine surveillance were made available. A stratified random sample approach was used to select a cohort of herds for testing from this larger sample set. A final study population of 728 herds had bulk tank milk samples analyzed using a Bio-X ELISA (ELISA 1), an IDvet ELISA (ELISA 2) and a PCR test. A Bayesian latent class analysis (BLCA) was conducted to estimate the sensitivity (Se) and specificity (Sp) of the 3 diagnostic tests applied to bulk tank milk (BTM) for the detection of the herd-level infection. An overall LCA was conducted on all herds within a single population (a 3-test, 1-population model). The herds were also split into 2 populations based on herd size (small herds had < 82 cattle) (a 3-test, 2-population model) and separately into 3 regions in Ireland (Leinster, Munster and Connacht/Ulster) (a 3-test, 3-population model). The latent variable of interest was the herd-level M. bovis infection status. In total, 363/728 (50%) were large herds, 7 (1.0%) were positive on PCR, 88 (12%) positive on ELISA 1, and 406 (56%) positive on ELISA 2. Based on the 2-population model, the sensitivity (95% Bayesian credible interval (BCI) was 0.03 (0.02, 0.05), 0.22 (0.18, 0.27), 0.94 (0.88, 0.98) for PCR, ELISA 1 and ELISA 2 respectively. The specificity (95% BCI) was 0.99 (0.99, 1.0), 0.97 (0.95, 0.99), and 0.92 (0.86, 0.97) for PCR, ELISA 1 and ELISA 2 respectively. The herd-level true prevalence was estimated at 0.43 (BCI 0.35, 0.5) for smaller herds. The true prevalence was estimated at 0.62 (BCI 0.55, 0.69) for larger herds. The true prevalence was estimated at 0.56 (BCI 0.49, 0.463) in the 1-population model. For the 3-population model, the sensitivity (95% BCI) was 0.03 (0.02, 0.05), 0.24 (0.18, 0.29), 0.95 (0.9, 0.98) for PCR, ELISA 1 and ELISA 2 respectively. The specificity (95% BCI) was 0.99 (0.99, 1.0), 0.98 (0.96, 0.99), and 0.88 (0.79, 0.95) for PCR, ELISA 1 and ELISA 2 respectively. The herd-level true prevalence (95% BCI) was estimated at 0.65 (0.56, 0.73), 0.38 (0.28, 0.46) and 0.53 (0.4, 0.65) for population 1, 2, 3 respectively. Across all 3 models, the range in true prevalence was 38% to 65% of Irish dairy herds infected with M. bovis. The operating characteristics vary substantially between tests. The IDvet ELISA had a relatively high Se (the highest Se of the 3 tests studied) but it was estimated at 0.95 at its highest in 3-test, 3-population model. This test may be an appropriate test for herd-level screening or prevalence estimation within the context of the endemically infected Irish dairy cattle population. Further work is required to optimize this test and its interpretation when applied at herd-level to offset concerns related to the lower than optimal test Sp.

Evaluation of sensitivity and specificity of Kato-Katz and circulating cathodic antigen in terms of Schistosoma japonicum using latent class analysis

Schistosoma japonicum is endemic in the Philippines. The Kato-Katz (KK) method was used to diagnose S. japonicum. This is impractical, particularly when the sample size is limited. Knowledge on point-of-care circulating cathodic antigen (CCA) test performance for S. japonicum is limited. Determining the sensitivity and specificity of new diagnostics is difficult when the gold standard test is less effective or absent. Latent class analysis (LCA) can address some limitations. A total of 484 children and 572 adults from the Philippines were screened for S. japonicum. We performed Bayesian LCA to estimate the infection prevalence, sensitivity and specificity of each test by stratifying them into two age groups. Observed prevalence assessed by KK was 50.2% and 31.8%, and by CCA was 89.9% and 66.8%, respectively. Using Bayesian LCA, among children, the sensitivity and specificity of CCA were 94.8% (88.7-99.4) and 21.5% (10.5-36.1) while those of KK were 66.0% (54.2-83.3) and 78.1% (61.1-91.3). Among adults, the sensitivity and specificity of CCA were 86.4% (76.6-96.9) and 62.8% (49.1-81.1) while those of KK were 43.6% (35.1-53.9) and 85.5% (75.8-94.6). Overall, CCA was more sensitive than KK, regardless of the age group at diagnosis, as KK was more specific. KK and CCA have different diagnostic performance, which should inform their use in the planning and implementation of S. japonicum control programs.

Bayesian evaluation of temporal changes in sensitivity and specificity of three serological tests for multiple circulating strains of rabbit haemorrhagic disease virus

Competition and indirect ELISAs are currently being used to monitor rabbit haemorrhagic disease viruses (RHDV1 and RHDV2) in rabbits worldwide. Temporal changes in the sensitivity (Se) and specificity (Sp) of RHDV1 competition-ELISA (cELISA1), RHDV2 competition-ELISA (cELISA2), and RHDV1 Immunoglobulin G (IgG1) ELISA, were investigated using Bayesian Latent Class models (BCLM) in the Australian wild rabbit population where both viruses circulate simultaneously and a long-term serological dataset exists. When cELISA1 was compared to IgG1 ELISA, the Se of cELISA1 improved while the Sp of IgG1 ELISA declined over the 2011-21. This corresponded with a decline in the true RHDV1 prevalence in 2018-21, suggesting that a large proportion of RHDV1 exposed rabbits survived the introduction and dominance of RHDV2 up to approximately 2017/2018, after which they died and were not replaced. The Se and Sp estimates for 2014-15 for both cELISA1 and IgG1 ELISA, and the true prevalence when analysing all three tests together were similar to those obtained from the analysis of cELISA1/IgG1 ELISA. The same was also true for the Se and Sp of cELISA2 and IgG1 ELISA estimates from 2018 onwards. This suggests that RHDV1 was the dominant infection type in 2014-15, but RHDV2 was the dominant infection type in 2018-21. Further, the increase in Se of cELISA2 and the low Sp of IgG1 ELISA in the cELISA2/IgG1 ELISA analysis, compared to the Se of cELISA2 and Sp of IgG1 ELISA when analysing all three tests together suggests that the underlying infection status was more influenced by RHDV2 and that the higher Se of IgG1 ELISA is due to cross-reaction of RHDV2 antibodies on IgG1 ELISA. The true prevalence data suggest that RHDV2 exposure peaked in 2017. Our findings show that test characteristics changed in response to the changing virus prevalences over time. IgG1 ELISA, currently having a high Se, should be used to monitor both viruses and will perform better than both cELISAs.

Accuracy of serological tests for bovine brucellosis: A systematic review and meta-analysis

The direct methods for diagnosis of bovine brucellosis have several limitations, therefore serological tests are the basis for the diagnosis of the disease. However, a meta-analysis estimating the diagnostic sensitivity (DSe) and diagnostic specificity (DSp) on the main tests used in bovine brucellosis control programs worldwide has not been performed. This systematic review and meta-analysis aimed to estimate the DSe, DSp and thereby accuracy of serological tests individually used in the diagnosis of bovine brucellosis. The databases CABI, Cochrane Library, PubMed/MEDLINE, SciELO, Scopus and Web of Science were used to select articles. The search resulted in 5308 studies, of which 71 were selected for systematic review using quality assessment tools and 65 studies were included in the meta-analysis. For the meta-analysis, 178 assays and 11 different serological tests were considered. To estimate DSe and DSp of the tests, studies were divided according to animal selection for the studies: (1) studies that carried out a random or consecutive selection of participants (noncasecontrol studies) and (2) all studies, including casecontrol studies. Considering only the non-case-control studies to estimate the DSe, the tests that exhibited the best and worst performance were the iELISA test (indirect enzyme immunoassay - bacterial suspension as antigen - BS) (96.5%, 95% CI: 94.1-97.9%) and 2ME (2- mercaptoethanol test) (85.0%, 95% CI: 79.6-89.1%), respectively; while for DSp, the FPA (fluorescence polarization assay) (99, 7%, 95% CI: 99.5-99.8%) and PCFIA tests (protein concentration fluorescence immunoassay) (78.5%, 95% CI: 70.0-85.1%) showed better and worse performance, respectively. Overall, our results showed an overestimation in the DSe and DSp of the eleven serological tests assessed when casecontrol studies were included in the meta-analysis, which is a concern considering its impacts on the time and costs associated with populational diagnosis of the diseases, since several of these tests are routinely used in the control and eradication programs of bovine brucellosis worldwide. Furthermore, the tests that exhibited the best DSe and DSp, iELISA (BS) and FPA, respectively, are relatively easy to perform and interpret and the test which showed the best overall accuracy was FPA.

Challenges in the real world use of classification accuracy metrics: From recall and precision to the Matthews correlation coefficient

The accuracy of a classification is fundamental to its interpretation, use and ultimately decision making. Unfortunately, the apparent accuracy assessed can differ greatly from the true accuracy. Mis-estimation of classification accuracy metrics and associated mis-interpretations are often due to variations in prevalence and the use of an imperfect reference standard. The fundamental issues underlying the problems associated with variations in prevalence and reference standard quality are revisited here for binary classifications with particular attention focused on the use of the Matthews correlation coefficient (MCC). A key attribute claimed of the MCC is that a high value can only be attained when the classification performed well on both classes in a binary classification. However, it is shown here that the apparent magnitude of a set of popular accuracy metrics used in fields such as computer science medicine and environmental science (Recall, Precision, Specificity, Negative Predictive Value, J, F1, likelihood ratios and MCC) and one key attribute (prevalence) were all influenced greatly by variations in prevalence and use of an imperfect reference standard. Simulations using realistic values for data quality in applications such as remote sensing showed each metric varied over the range of possible prevalence and at differing levels of reference standard quality. The direction and magnitude of accuracy metric mis-estimation were a function of prevalence and the size and nature of the imperfections in the reference standard. It was evident that the apparent MCC could be substantially under- or over-estimated. Additionally, a high apparent MCC arose from an unquestionably poor classification. As with some other metrics of accuracy, the utility of the MCC may be overstated and apparent values need to be interpreted with caution. Apparent accuracy and prevalence values can be mis-leading and calls for the issues to be recognised and addressed should be heeded.

Gamma-interferon assay for the ancillary diagnosis of bovine tuberculosis in dairy cattle in urban and adjacent areas of Dhaka city, Bangladesh

Background and Aim

Bovine tuberculosis (bTB) is an infectious disease of cattle, mainly caused by Mycobacterium bovis. This study aimed to evaluate the efficacy of the interferon-gamma (IFN-γ) assay and single-intradermal comparative tuberculin test (SICTT) in detecting bTB.

Materials and Methods

In an earlier study, 150 positive, 83 inconclusive, and 480 negative animals from 24 cattle herds were screened using SICTT. From these groups, 125 positive, 17 inconclusive, and six negative animals were subsequently verified using the IFN-γ assay. Single-intradermal comparative tuberculin test outcomes were interpreted according to standard guidelines, whereas blood samples were collected and stimulated with purified protein derivatives. Sandwich enzyme-linked immunosorbent assay was used to measure secreted IFN-γ. Concordant and Bayesian latent class analyses were performed to evaluate test performance.

Results

Results from the IFN-γ assay revealed that 83.2%, 64.7%, and 16.67% of the animals were positive in the SICTT-positive, inconclusive, and negative animal categories, respectively. Sensitivity (SE) and specificity (SP) of SICTT were 83.9% (95% confidence interval [CI]: 77.4-90.1) and 95.7% (95% CI: 86.9-99.7), respectively. Sensitivity and SP for the IFN-γ assay were 78.9% (95% CI: 71.9-85.4) and 83.9% (65.9-95.9), respectively. The use of both tests in parallel increases the SE of bTB detection (~94%), compared with SICTT alone.

Conclusion

Use of the IFN-γ assay with SICTT in parallel, predominantly on cattle demonstrating an inconclusive SICTT outcome, boosts bTB detection rate in low resource settings.

Perkinsus olseni in green-lipped mussels Perna canaliculus: diagnostic evaluation, prevalence, and distribution

Perkinsus olseni (Perkinsidae) is a molluscan parasite notifiable to the World Organisation for Animal Health that is reported in several shellfish hosts in New Zealand, including the native green-lipped mussel Perna canaliculus. Green-lipped mussels comprise over half of New Zealand's aquaculture export value and have historically been considered free of serious diseases based on extensive histology-based surveillance. The discovery of P. olseni in green-lipped mussels has raised questions about future disease threats to green-lipped mussels, particularly under changing ocean climatic conditions. Using mussels collected from farmed (n = 358) and wild (n = 236) populations, we aimed to determine the distribution and prevalence of P. olseni in green-lipped mussels around New Zealand, and assess the performance of diagnostic tests, including real-time PCR, conventional PCR, and culture using Ray's fluid thioglycolate medium (RFTM). Prevalence and diagnostic test performance was evaluated using Bayesian latent class analysis with informative priors. The prevalence of P. olseni was 0-3%, except for 1 wild population from a harbour where prevalence was 22%. Real-time PCR had the highest diagnostic sensitivity (87%) compared to 62 and 21% for conventional PCR and RFTM, respectively. Diagnostic specificity was similar among all methods (96-98%). No mortality was observed during the study. Our results suggest that real-time PCR is the diagnostic test best suited for surveillance of P. olseni in subclinically infected green-lipped mussels under New Zealand conditions.

SARS-CoV-2 Serology Across Scales: A Framework for Unbiased Estimation of Cumulative Incidence Incorporating Antibody Kinetics and Epidemic Recency

Serosurveys are a key resource for measuring severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) population exposure. A growing body of evidence suggests that asymptomatic and mild infections (together making up over 95% of all infections) are associated with lower antibody titers than severe infections. Antibody levels also peak a few weeks after infection and decay gradually. We developed a statistical approach to produce estimates of cumulative incidence from raw seroprevalence survey results that account for these sources of spectrum bias. We incorporate data on antibody responses on multiple assays from a postinfection longitudinal cohort, along with epidemic time series to account for the timing of a serosurvey relative to how recently individuals may have been infected. We applied this method to produce estimates of cumulative incidence from 5 large-scale SARS-CoV-2 serosurveys across different settings and study designs. We identified substantial differences between raw seroprevalence and cumulative incidence of over 2-fold in the results of some surveys, and we provide a tool for practitioners to generate cumulative incidence estimates with preset or custom parameter values. While unprecedented efforts have been launched to generate SARS-CoV-2 seroprevalence estimates over this past year, interpretation of results from these studies requires properly accounting for both population-level epidemiologic context and individual-level immune dynamics.

Sensitivity and specificity of human point-of-care circulating cathodic antigen (POC-CCA) test in African livestock for rapid diagnosis of schistosomiasis: A Bayesian latent class analysis

Schistosomiasis is a major neglected tropical disease (NTD) affecting both humans and animals. The morbidity and mortality inflicted upon livestock in the Afrotropical region has been largely overlooked, in part due to a lack of validated sensitive and specific tests, which do not require specialist training or equipment to deliver and interpret. As stressed within the recent WHO NTD 2021-2030 Roadmap and Revised Guideline for schistosomiasis, inexpensive, non-invasive, and sensitive diagnostic tests for livestock-use would also facilitate both prevalence mapping and appropriate intervention programmes. The aim of this study was to assess the sensitivity and specificity of the currently available point-of-care circulating cathodic antigen test (POC-CCA), designed for Schistosoma mansoni detection in humans, for the detection of intestinal livestock schistosomiasis caused by Schistosoma bovis and Schistosoma curassoni. POC-CCA, together with the circulating anodic antigen (CAA) test, miracidial hatching technique (MHT) and organ and mesentery inspection (for animals from abattoirs only), were applied to samples collected from 195 animals (56 cattle and 139 small ruminants (goats and sheep) from abattoirs and living populations) from Senegal. POC-CCA sensitivity was greater in the S. curassoni-dominated Barkedji livestock, both for cattle (median 81%; 95% credible interval (CrI): 55%-98%) and small ruminants (49%; CrI: 29%-87%), than in S. bovis-dominated Richard Toll ruminants (cattle: 62%; CrI: 41%-84%; small ruminants: 12%, CrI: 1%-37%). Overall, sensitivity was greater in cattle than in small ruminants. Small ruminants POC-CCA specificity was similar in both locations (91%; CrI: 77%-99%), whilst cattle POC-CCA specificity could not be assessed owing to the low number of uninfected cattle surveyed. Our results indicate that, whilst the current POC-CCA does represent a potential diagnostic tool for cattle and possibly for predominantly S. curassoni-infected livestock, future work is needed to develop parasite- and/or livestock-specific affordable and field-applicable diagnostic tests to enable determination of the true extent of livestock schistosomiasis.

Bayesian diagnostic test evaluation and true prevalence estimation of mycoplasma bovis in dairy herds

The true prevalence of dairy cattle herds with M. bovis infections in the Netherlands is unknown. Previous attempts to estimate prevalences were hampered by the absence of a diagnostic serological test that was validated under field conditions. This study estimated sensitivity and specificity of two commercial serum ELISAs and the true M. bovis herd prevalence using different Bayesian latent class models. A total of 7305 serum samples from 415 randomly chosen dairy herds were collected in fall/winter 2019 and investigated for presence of antibodies against M. bovis using the BIO-K-260 ELISA from Bio-X. Serum samples from 100 of these herds were also tested with a second ELISA, from IDvet. A Bayesian latent class model using the paired test results estimated a sensitivity of 14.1% (95% Bayesian probability interval (BPI): 11.6-16.7%) for the Bio-X ELISA and a specificity of 97.2% (95% BPI: 95.9-98.4%). Sensitivity and specificity for the IDvet ELISA were estimated at 92.5% (95% BPI: 88.3-96.5%) and 99.3% (95% BPI: 98.7-99.8%), respectively. Also, Bio-X ELISA sensitivity was considerably higher with data from calves only and with data from a selection of herds with a clinical outbreak, whereas the IDvet ELISA sensitivity was fairly constant under these conditions. These differences in test sensitivity is expected to be related to an effect of time since infection. A second Bayesian model, applied on test results of all 415 herds, estimated a true herd prevalence of 69.9% (95% BPI: 62.7-77.6%), suggesting M. bovis in endemic amongst dairy cattle herds in the Netherlands. To what extent seropositive herds have experienced a clinical outbreak needs further investigation.

Diagnostic accuracy of an immunomagnetic separation-PCR assay to detect pathogenic Leptospira spp. in urine from dairy cattle, using a Bayesian latent class model

Leptospirosis is a zoonotic disease that has spread worldwide and causes significant economic losses in the dairy industry. The causal agents of this infectious disease are members of the genus Leptospira, known as pathogenic Leptospira spp. Specific clinical signs of the infection are difficult to detect. Therefore, the disease is normally under-diagnosed, mostly due to the lack of a cost-effective technique for diagnosing animals with a low bacterial load in their urine. The aim of this study was to assess the diagnostic accuracy of a qPCR coupled with a previous Immunomagnetic separation (IMS) step (IMS-qPCR) against a qPCR without using IMS, using a Bayesian latent class model (2 tests, 3 populations) to determine the leptospirosis infectious status in naturally infected dairy cattle. The results revealed that IMS qPCR had a sensitivity (Se) of 95.7% (95% Probability Interval (PI) = 85.0; 99.4%) and a specificity (Sp) of 98% (95% PI = 96.1; 99.4%), indicating that it is more sensitive than conventional qPCR (Se = 69.7% (95% PI = 59.2; 79.0%); median difference = 25.2% (Monte Carlo Error = 10.2%); and the Sp = 98.8% (95% PI = 97.6; 99.5%), median difference = 0.8% (Monte Carlo Error = 2.1%). Therefore, results shows that IMS-qPCR is a more useful diagnostic tool in terms of accuracy for detecting infectious animals with pathogenic Leptospira in their urine.

A large-scale study on the seroprevalence of small ruminant lentiviral infection in the Polish goat population

A large-scale study was carried out in a Polish goat population in 2014-2022 to determine the herd-level (between-herd) and within-herd seroprevalence of small ruminant lentivirus (SRLV) infection. A total of 8354 adult goats (aged >1 year) from 165 herds located in various regions of Poland were serologically tested using a commercial ELISA. One hundred twenty eight herds were randomly selected while 37 were enrolled based on convenience non-random sampling. At least 1 seropositive result was obtained in 103 / 165 herds. For all these herds the probability that they were truly positive (herd-level positive predictive value) was calculated. It was ≥ 90% in 91 seropositive herds and 73% to < 90% in 12 herds in which only 1-4 goats were seropositive (22 goats in total). The seropositive goats in the latter herds were retested using a different commercial ELISA and 14 goats (9 males and 5 females) from 9 herds were confirmed to be seropositive (serial testing). The true herd-level seroprevalence was estimated at 61% (95% confidence interval [CI 95%]: 53%-68%). It differed significantly between herd size classes (p = 0.003): the highest prevalences were found in the medium (51 - 100 adult goats) and large herds (>100 adult goats) - 72% (CI 95%: 56-84%) and 86% (CI 95%: 67%-95%), respectively, while prevalences in very small (≤ 20 adult goats) and small herds (21 - 50 adult goats) were 46% (CI 95%: 34%-59%) and 57% (CI 95%: 43%-70%), respectively. The true herd-level seroprevalence differed significantly also between geographical regions of Poland (p = 0.003), with the highest values in the north-western and the lowest in the southern region of the country. The true within-herd seroprevalence estimated using a Bayesian approach ranged from 0.7% to 100% with the median (IQR) of 42% (17%-84%), and did not vary significantly between herd size classes (p = 0.393) or geographical regions of Poland (p = 0.570). Concluding, SRLV infection is widespread in the Polish goat population, the north-western region of Poland is most extensively infected, and herds counting > 50 adult goats are more often infected.

Development of in-house ELISA for detection of antibodies against lumpy skin disease virus in cattle and assessment of its performance using a bayesian approach

Lumpy skin disease (LSD) is a contagious disease among cattle and buffalo worldwide. Currently, an enzyme-linked immunosorbent assay (ELISA) has been recognized as an efficient diagnostic tool that is less time-consuming and easier than the viral neutralization test to measure the antibody levels. In the present study, an in-house method of indirect ELISA was developed to detect the bovine antibodies against Lumpy skin disease virus (LSDV) and its performance was assessed using field samples. This in-house method has been compared with the commercial ELISA test kit for detection of bovine antibodies against LSDV. The sensitivity (Se) and the specificity (Sp) of the test were estimated using a Bayesian latent class model. Checkerboard titration was performed using the naturally LSDV-infected bovine sera and colostrum-deprived calf sera. The LSDV antigen concentrations (1 TCID₅₀/mL), the sample serum (1:500), and goat anti-bovine immunoglobulin G (IgG) labeled with horseradish peroxidase (HRP) (1:10,000) were determined to be optimal for this assay. The calculated cut-off value was 0.067, and there were no differences in the results of tests that utilized positive and negative sera (p < 0.05). The characteristics of two diagnostic tests were evaluated using a conditional dependent and one-population Bayesian model. The Se value of an in-house indirect ELISA were almost similar to ELISA test kit. On the other hand, the Sp value of the in-house ELISA test was lower than that of the commercial ELISA test with the median values of 89% (95% PPI = 75.9-99.3%) and 91.4% (95% PPI = 85.3-95.5%), respectively. A posterior estimate for the prevalence was 66.9% (95% PPI = 60.8-83.3%) and higher than initially expected.

Modeling the accuracy of a novel PCR and antibody ELISA for African swine fever virus detection using Bayesian latent class analysis

Introduction

Diagnostic test evaluation for African swine fever (ASF) in field settings like Vietnam is critical to understanding test application in intended populations for surveillance and control strategies. Bayesian latent class analysis (BLCA) uses the results of multiple imperfect tests applied to an individual of unknown disease status to estimate the diagnostic sensitivity and specificity of each test, forgoing the need for a reference test.

Methods

Here, we estimated and compared the diagnostic sensitivity and specificity of a novel indirect ELISA (iELISA) for ASF virus p30 antibody (Innoceleris LLC.) and the VetAlert™ ASF virus DNA Test Kit (qPCR, Tetracore Inc.) in field samples from Vietnam by assuming that disease status 1) is known and 2) is unknown using a BLCA model. In this cross-sectional study, 398 paired, individual swine serum/oral fluid (OF) samples were collected from 30 acutely ASF-affected farms, 37 chronically ASF-affected farms, and 20 ASF-unaffected farms in Vietnam. Samples were tested using both diagnostic assays. Diagnostic sensitivity was calculated assuming samples from ASF-affected farms were true positives and diagnostic sensitivity by assuming samples from unaffected farms were true negatives. ROC curves were plotted and AUC calculated for each test/sample combination. For comparison, a conditionally dependent, four test/sample combination, three population BLCA model was fit.

Results

When considering all assumed ASF-affected samples, qPCR sensitivity was higher for serum (65.2%, 95% Confidence Interval [CI] 58.1-71.8) and OF (52%, 95%CI 44.8-59.2) compared to the iELISA (serum: 42.9%, 95%CI 35.9-50.1; OF: 33.3%, 95%CI 26.8-40.4). qPCR-serum had the highest AUC (0.895, 95%CI 0.863-0.928). BLCA estimates were nearly identical to those obtained when assuming disease status and were robust to changes in priors. qPCR sensitivity was considerably higher than ELISA in the acutely-affected population, while ELISA sensitivity was higher in the chronically-affected population. Specificity was nearly perfect for all test/sample types.

Discussion

The effect of disease chronicity on sensitivity and specificity could not be well characterized here due to limited data, but future studies should aim to elucidate these trends to understand the best use of virus and antibody detection methods for ASF. Results presented here will help the design of surveillance and control strategies in Vietnam and other countries affected by ASF.

Screening for Coxiella Burnetii in Dairy Cattle Herds in Poland

Introduction

The intracellular bacterium Coxiella burnetii is the aetiological agent of Q fever, a zoonosis affecting many animal species worldwide. Cattle and small ruminants are considered the major reservoirs of the bacteria and they shed it through multiple routes.

Material and Methods

A total of 2,180 sera samples from 801 cattle herds in all Polish voivodeships were tested by ELISA for the presence of specific antibodies. Milk samples were obtained from seropositive cows in 133 herds as part of a separate study. The milk samples were examined by ELISA and real-time PCR tests.

Results

Seroprevalence at the animal level was 7.06% and true positive seroprevalence was 6.0% (95% confidence interval (CI) 1.1-9.4). Seroprevalence at the herd level was estimated at 11.1% and true positive seroprevalence was 10.5% (95% CI 3.2-15.8). Shedding of the pathogen in milk was detected by real-time PCR in 33 out of 133 tested herds (24.81%, 95% CI 17.74-33.04%) and the presence of C. burnetii antibodies was confirmed in 85 of them (63.9%, 95% CI 55.13-72.05%). The highest level of conformity between ELISA and real-time PCR results was obtained for bulk tank milk samples.

Conclusion

Coxiella burnetii infections are quite common in cattle herds across the country, which emphasises the crucial roles of surveillance and adequate biosecurity measures in the prevention and limitation of Q fever spread in Poland.

Evaluation of three commercial ELISA tests for serological detection of maedi-visna virus using Bayesian latent class analysis

Early and accurate diagnosis is fundamental for successful surveillance and control of maedi-visna virus (MVV). MVV was detected in Norway in 2019, almost 14 years after the previous outbreak. Genetic analysis indicates persistence of the virus in the sheep population since 2005. The virus was not detected despite continuous serological surveillance. This emphasises the need for improved surveillance, which relies on an understanding of both diagnostic test performance, sampling strategy and the prevalence of the disease. This study therefore aims to evaluate three commercial ELISA tests for MVV antibodies. We conducted a retrospective study using 615 samples from six flocks diagnosed with MVV in 2019. We ran all samples with the following three tests: ID Screen® MVV/CAEV Indirect (IDvet, Grabels, France), IDEXX MVV/CAEV p28 Ab Verification Test (IDEXX Laboratories, Maine, USA) and Elitest MVV/CAEV (Hyphen Biomed, Neuville-sur-Oise, France), hereinafter referred to as ID Screen, IDEXXp28 and Elitest respectively. Without a perfect reference test, we used Bayesian latent class analysis, including conditional dependence between tests, to estimate diagnostic accuracy and true prevalence in the flocks. Using recommended cut-off values, we found that ID Screen and Elitest had significantly higher sensitivity (Se) estimates (99.3 % [97.4-100.0, 95 % Posterior Credible Interval] and 97.4 % [94.1-99.7 %], respectively) than IDEXXp28 (79.5 % [72.3-86.0 %]), while IDEXXp28 and ID Screen had significantly higher specificity (Sp) estimates than Elitest (99.7 % [99.1-100.0], 99.1 % [98.0-99.8 %] and 93.7 % [91.4-95.7 %], respectively). The estimated true prevalence in the six flocks ranged from a median of 0.8-93.5 %. Combining ID Screen and Elitest in serial interpretation showed the highest median Se and Sp (96.7 % [92.0-99.1] and 100.0 % [99.9-100.0], respectively), as well as the highest median positive predictive value (PPV) for the population with the lowest prevalence. Our study supports the use of ID Screen for screening. Further verification with Elitest in serial interpretation will enhance the PPV.

Estimating the sensitivity and specificity of serum ELISA and pooled and individual fecal PCR for detecting Mycobacterium avium subspecies paratuberculosis in Canadian cow-calf herds using Bayesian latent class models

While Johne's disease (JD) is less common in beef than in dairy herds, consolidation is increasing transmission risk. Estimates of Mycobacterium avium spp. paratuberculosis (MAP) prevalence and test performance in cow-calf herds are needed to inform control programs. Objectives of this study included describing the prevalence of MAP in Canadian cow-calf herds and comparing the relative performance of a serum ELISA, pooled fecal PCR and individual fecal PCR using Bayesian latent class models, and to investigate factors associated with positive MAP tests. Blood and fecal samples (n = 3,171) were collected from 159 Canadian cow-calf herds. All samples were analyzed using serum ELISA and fecal PCR (pools of five samples) and a subset of 913 fecal samples were also tested with individual PCR. Based on latent class analysis, MAP prevalence was higher in eastern compared to western Canada for both animals {East, 3% [95% Credible Interval (CrI) 1-7%]; West, 1% [95% CrI 0.2-2%]} and herds [East, 15% (95% CrI 2-35%); West, 10% (95% CrI 1-26%), based on one or more positive results]. Sensitivity (Se) and specificity (Sp) for animal level individual PCR were 96% (95% CrI 80-100%) and 98% (95% CrI 96-100%), respectively followed by pooled PCR [Se = 54% (95% CrI 36-72%), Sp > 99.9% (95% CrI 99.8-100%)] and ELISA [Se = 36% (95% CrI 22-52%), Sp = 98% (95% CrI 96-99%)]. Based on 20 samples per herd, the herd level Se of ELISA was 79% (95% CrI 47-100%) (at least one positive sample) compared to 43% (95% CrI 14-94%) for pooled PCR. Herd-level Sp was 99% (95% CrI 96-100%) for pooled PCR and 90% (95% CrI 83-100%) for ELISA. Cows from herds with dairy cattle on farm and cows with symptoms of JD in the past 3 years were more likely to be MAP positive. Herds that had animals with JD symptoms in the previous 3 years and those with more breeding females were most likely to test positive for MAP. While serum ELISA can be effective for herd screening, PCR performed better for animal testing. Pooled PCR testing could be a less costly option; however, determining the most cost-effective approach will require further economic analysis.

Bayesian latent class analysis to estimate the optimal cut-off for the MilA ELISA for the detection of Mycoplasma bovis antibodies in sera, accounting for repeated measures

The MilA ELISA has been identified as a highly effective diagnostic tool for the detection of Mycoplasma bovis specific antibodies and has been validated for serological use in previous studies. This study aimed to estimate the optimal cut-off and corresponding estimates of diagnostic sensitivity (DSe) and diagnostic specificity (DSp) of the MilA ELISA for testing bovine serum. Serum samples from 298 feedlot cattle from 14 feedlots across four Australian states were tested on entry into the feedlot and approximately 42 days later. The paired serum samples were tested with the MilA ELISA, BIO K302 (Bio-X Diagnostics, Belgium) and BIO K260 (Bio-X Diagnostics, Belgium). A cut-off of 135 AU was estimated to be optimal using Bayesian latent class analysis with three tests in multiple populations, accounting for conditional dependence between tests. At this cut-off, the DSe and DSp of the MilA ELISA were estimated to be 92.1 % (95 % highest probability density [HPD] interval: 87.4, 95.8) and 95.5 % (95 % HPD: 92.4, 97.8), respectively. The DSes of the BIO K260 and BIO K302 ELISAs were estimated to be 60.5 % (95 % HPD: 54.0, 66.9) and 44.6 % (95 % HPD: 38.7, 50.7), respectively. DSps were 95.6 % (95 % HPD: 92.9, 97.7) and 97.8 % (95 % HPD: 95.9, 99.0), respectively. Mycoplasma bovis seroprevalence was remarkably high at follow-up after 42 days on the feedlots. Overall, this study estimated a cut-off, DSe and DSp for the MilA ELISA with less dependence on prior information than previous analyses and demonstrated that the MilA ELISA has higher DSe than the BIO K260 and BIO K302 assays.

Detection of Low MAP Shedder Prevalence in Large Free-Stall Dairy Herds by Repeated Testing of Environmental Samples and Pooled Milk Samples

Simple Summary

Paratuberculosis is a disease which affects ruminants worldwide. Many countries have implemented certification and monitoring systems to control the disease, particularly in dairy herds. Monitoring herds certified as paratuberculosis non-suspect is an important component of paratuberculosis herd certification programs. The challenge is to detect the introduction or reintroduction of the infectious agent as early as possible with reasonable efforts but high certainty. In our study, we evaluated different low-cost testing schemes in herds where the share of infected animals was low, resulting in a low within-herd prevalence of animals shedding the bacteria that causes paratuberculosis in their feces. The test methods used were repeated pooled milk samples and fecal samples from the barn environment. Our study showed that numerous repetitions of different samples are necessary to monitor such herds with sufficiently high certainty. In the case of herds with a very low prevalence, our study showed that a combination of different sampling approaches is required.

Abstract

An easy-to-use and affordable surveillance system is crucial for paratuberculosis control. The use of environmental samples and milk pools has been proven to be effective for the detection of Mycobacterium avium subsp. paratuberculosis (MAP)-infected herds, but not for monitoring dairy herds certified as MAP non-suspect. We aimed to evaluate methods for the repeated testing of large dairy herds with a very low prevalence of MAP shedders, using different sets of environmental samples or pooled milk samples, collected monthly over a period of one year in 36 herds with known MAP shedder prevalence. Environmental samples were analyzed by bacterial culture and fecal PCR, and pools of 25 and 50 individual milk samples were analyzed by ELISA for MAP-specific antibodies. We estimated the cumulative sensitivity and specificity for up to twelve sampling events by adapting a Bayesian latent class model and taking into account the between- and within-test correlation. Our study revealed that at least seven repeated samplings of feces from the barn environment are necessary to achieve a sensitivity of 95% in herds with a within-herd shedder prevalence of at least 2%. The detection of herds with a prevalence of less than 2% is more challenging and, in addition to numerous repetitions, requires a combination of different samples.

Evaluation of the sensitivity and specificity of three diagnostic tests for Coxiella burnetii infection in cattle and buffaloes in Punjab (India) using Bayesian latent class analysis

Q Fever is a zoonotic disease of significant animal and public health concern, caused by Coxiella burnetii (C. burnetii), an obligate intracellular bacterium. This study was done to evaluate the diagnostic sensitivity (DSe) and diagnostic specificity (DSp) of three diagnostic methods to diagnose C. burnetii infection in cattle and buffaloes in Punjab, India: an indirect ELISA method applied in serum samples and a trans-Polymerase Chain Reaction (trans-PCR) technique applied in milk samples and genital swabs, using a Bayesian latent class analysis. Conditional independence was assumed between the tests, given (i) the different biological principle of ELISA and trans-PCR and (ii) the fact that the trans-PCR was performed on different tissues. The ELISA method in the serum samples showed the highest DSe of 0.97 (95% Probability Intervals (PIs): 0.93; 0.99) compared to the trans-PCR method applied in milk samples 0.76 (0.63; 0.87) and genital swabs 0.73 (0.58; 0.85). The DSps of all tests were high, with trans-PCR in genital swabs recording the highest DSp of 0.99 (0.98; 1), while the DSp of trans-PCR in milk samples and ELISA in serum samples were 0.97 (0.95; 0.99) and 0.95 (0.93; 0.97) respectively. The study results show that none of the applied tests are perfect, therefore, a testing regimen based on the diagnostic characteristic of the tests may be considered for diagnosis of C. burnetii.

Assessment of Real-Time Quaking-Induced Conversion (RT-QuIC) Assay, Immunohistochemistry and ELISA for Detection of Chronic Wasting Disease under Field Conditions in White-Tailed Deer: A Bayesian Approach

Chronic wasting disease (CWD) is a transmissible prion disease of the cervidae family. ELISA and IHC tests performed postmortem on the medial retropharyngeal lymph nodes (RPLN) or obex are considered diagnostic gold standards for prion detection. However, differences in CWD transmission, stage of infection, pathogenesis, and strain can limit performance. To overcome these uncertainties, we used Bayesian statistics to assess the accuracy of RT-QuIC, an increasingly used prion amplification assay, to diagnose CWD on tonsil (TLN), parotid (PLN) and submandibular lymph nodes (SMLN), and ELISA/IHC on RPLN of white-tailed deer (WTD) sampled from Minnesota. Dichotomous RT-QuIC and ELISA/IHC results from wild (n = 61) and captive (n = 46) WTD were analyzed with two-dependent-test, one-population models. RT-QuIC performed on TLN and SMLN of the wild WTD population had similar sensitivity (median range (MR): 92.2–95.1) to ELISA/IHC on RPLN (MR: 91.1–92.3). Slightly lower (4–7%) sensitivity estimates were obtained from farmed animal and PLN models. RT-QuIC specificity estimates were high (MR: 94.5–98.5%) and similar to ELISA/IHC estimates (MR: 95.7–97.6%) in all models. This study offers new insights on RT-QuIC and ELISA/IHC performance at the population level and under field conditions, an important step in CWD diagnosis and management.

Two-test algorithms for infectious disease diagnosis: Implications for COVID-19

Diagnostic assays for various infectious diseases, including COVID-19, have been challenged for their utility as standalone point-of-care diagnostic tests due to suboptimal accuracy, complexity, high cost or long turnaround times for results. It is therefore critical to optimise their use to meet the needs of users. We used a simulation approach to estimate diagnostic outcomes, number of tests required and average turnaround time of using two-test algorithms compared with singular testing; the two tests were reverse transcription polymerase chain reaction (RT-PCR) and an antigen-based rapid diagnostic test (Ag-RDT). A web-based application of the model was developed to visualise and compare diagnostic outcomes for different disease prevalence and test performance characteristics (sensitivity and specificity). We tested the model using hypothetical prevalence data for COVID-19, representing low- and high-prevalence contexts and performance characteristics of RT-PCR and Ag-RDTs. The two-test algorithm when RT-PCR was applied to samples negative by Ag-RDT predicted gains in sensitivity of 27% and 7%, respectively, compared with Ag-RDT and RT-PCR alone. Similarly, when RT-PCR was applied to samples positive by Ag-RDT, specificity gains of 2.9% and 1.9%, respectively, were predicted. The algorithm using Ag-RDT followed by RT-PCR as a confirmatory test for positive patients limited the requirement of RT-PCR testing resources to 16,400 and 3,034 tests when testing a population of 100,000 with an infection prevalence of 20% and 0.05%, respectively. A two-test algorithm comprising a rapid screening test followed by confirmatory laboratory testing can reduce false positive rate, produce rapid results and conserve laboratory resources, but can lead to large number of missed cases in high prevalence setting. The web application of the model can identify the best testing strategies, tailored to specific use cases and we also present some examples how it was used as part of the Access to Covid-19 Tools (ACT) Accelerator Diagnostics Pillar.

An intercomparison study of ELISAs for the detection of porcine reproductive and respiratory syndrome virus – evaluating six conditionally dependent tests

Latent class analysis is a widely used statistical method for evaluating diagnostic tests without any gold standard. It requires the results of at least two tests applied to the same individuals. Based on the resulting response patterns, the method estimates the test accuracy and the unknown disease status for all individuals in the sample. An important assumption is the conditional independence of the tests. If tests with the same biological principle are used, the assumption is not fulfilled, which may lead to biased results. In a recent publication, we developed a method that considers the dependencies in the latent class model and estimates all parameters using frequentist methods. Here, we evaluate the practicability of the method by applying it to the results of six ELISA tests for antibodies against the porcine reproductive and respiratory syndrome (PRRS) virus in pigs that generally follow the same biological principle. First, we present different methods of identifying suitable starting values for the algorithm and apply these to the dataset and a vaccinated subgroup. We present the calculated values of the test accuracies, the estimated proportion of antibody-positive animals and the dependency structure for both datasets. Different starting values led to matching results for the entire dataset. For the vaccinated subgroup, the results were more dependent on the selected starting values. All six ELISA tests are well suited to detect antibodies against PRRS virus, whereas none of the tests had the best values for sensitivity and specificity simultaneously. The results thus show that the method used is able to determine the parameter values of conditionally dependent tests with suitable starting values. The choice of test should be based on the general fit-for-purpose concept and the population under study.

Application of Bayesian modeling for diagnostic assays of Mycobacterium avium subsp. paratuberculosis in sheep and goats flocks

Background

This study aimed to screen the sera of goats and sheep from flocks suspected of Mycobacterium avium subsp. paratuberculosis (MAP) infection by a newly standardized Mce-truncated ELISA (Mt-ELISA) kit for the detection of antibodies against MAP. Four diagnostic applied tests were evaluated including Indigenous plate-ELISA (IP-ELISA), Mt-ELISA, fecal Polymerase Chain Reaction (f-PCR) and fecal culture (FC).

Materials and methods

Assuming the absence of a gold standard, latent-class models in a Bayesian framework were used to estimate the diagnostic accuracy of the four tests for MAP.

Results

Mt-ELISA had higher Sensitivity (Se) in sheep (posterior median: 0.68 (95% Probability Interval (PI): 0.43–0.95), while IP-ELISA recorded the highest Se in goats as 0.83 (95% PI, 0.61–0.97). The f-PCR Se estimate slightly differed between species [sheep 0.36 (0.19–0.58), goats 0.19 (0.08–0.35)], while the Se of FC was similar between species [sheep 0.29 (0.15–0.51), goats 0.27 (0.13–0.45)]. The specificity estimates for all tests were high, close to unity, and similar between species.

Conclusion

Overall, the results showed that the Mt-ELISA method can be used for MAP detection in small ruminants’ flocks.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12917-022-03141-7.

Occurrence and risk factors of brucellosis among domestic animals: an artificial neural network approach

Brucellosis is a zoonotic infectious disease with a worldwide distribution. This cross-sectional study aimed to determine the occurrence of Brucella in milk and blood samples of ruminants and the importance factors associated with animal brucellosis in Eastern Iran. A total of 200 paired samples, including blood (100) and milk (100), were obtained from the goats, sheep, and cows in Eastern Iran. Serum agglutination (SAT) and 2-mercapto ethanol (2-ME) tests were performed on the sera. A multiplex-polymerase chain reaction (m-PCR) assay was performed to identify the following species of Brucella, including B. abortus biovar 1, 2, 4, 3b, 5, 6, and 9, B. abortus S19, B. melitensis, and B. melitensis Rev.1 vaccine strain. B. abortus RB51 vaccine strain was also investigated in a second PCR assay. Risk factors for infection with Brucella spp. including the effect of abortion, contact with the wild animals, herd type, age, and previous vaccination in predicting blood contamination with B. abortus biovar 1, 2, and 4 were modeled by use of the artificial neural network. A total of 23 samples were seropositive regarding SAT and 2-ME tests. In total, B. abortus was detected in 35% and 15% of blood and milk samples, respectively, by the m-PCR assay. One sample of each of milk and blood was detected to have B. melitensis. Some samples were simultaneously contaminated with two Brucella species or two biovars of B. abortus. B. abortus S19 and B. melitensis Rev.1 vaccine strains were also detected in milk and blood samples. The sensitivity and specificity of the ANN model were 81% and 62%, respectively. In conclusion, B. abortus had higher frequency than B. melitensis in blood and milk samples. ANN determined herd type, previous vaccination, and age of the animal as the largest predictors of blood contamination with B. abortus.

Evaluation of diagnostic tests' sensitivity, specificity and predictive values in bovine carcasses showing brucellosis suggestive lesions, condemned by Brazilian Federal Meat Inspection Service in the Amazon Region of Brazil

Cervical bursitis is a suggestive lesion of bovine brucellosis. Diagnostic sensitivity and specificity of two brucellosis serological tests, Rose Bengal (RB) and serum agglutination test with 2-mercaptoethanol (SAT/2-ME), and of isolation and identification (bacteriology) were evaluated through Bayesian latent class analysis (BLCA). A total of 165 paired serum and cervical bursitis samples detected at inspection by Brazilian federal meat inspection services were analyzed. The best model fit to the data occurred when accounting for the conditional dependence between serological tests. According to this model, RB and SAT/2-ME had almost the same sensitivity, 0.960 [0.903 - 0.992] and 0.963 [0.906 - 0.994] with 95 % Credible Interval (95 %CrI), respectively. Specificities were 0.9068 [0.562 - 0.997] and 0.875 [0.546 - 0.990] for RB and SAT/2-ME, respectively, also with 95 %CrI. Bacteriology had lower sensitivity than serological tests, 0.594 (95 %CrI: [0.525 - 0.794]) and the highest specificity of all evaluated tests, 0.992 (95 %CrI: [0.961-1.00]). Prevalence of infected animals was 0.829 (95 %CrI: [0.700-0.900]). BLCA showed that both RB and SAT/2-ME fitted to the purpose of initial screening the brucellosis suspect in carcasses with cervical bursitis in a reliable way. The results of RB or SAT/2-ME can guide the sanitary actions for brucellosis control and help the implementation of a risk-based surveillance system in the meat production chain. This strategy is especially true in remote areas with large beef cattle herds, raised extensively, where in vivo tests are rarely performed due to logistic and management constraints, as in the Northern region of Brazil.

Meta-Analysis Methods of Diagnostic Test Accuracy Studies

Meta-analytic techniques are used to combine the results of different studies that have evaluated the accuracy of diagnostic tests. In this article, we present univariate and multivariate meta-analysis methods for a single test and we provide an extensive description of methods for meta-analysis and comparison of multiple diagnostic tests. We close with a practical example of a meta-analysis that aimed to determine whether Rheumatoid Factor identifies patients with Rheumatoid Arthritis.

Bayesian Estimation of Diagnostic Accuracy of Three Diagnostic Tests for Bovine Tuberculosis in Egyptian Dairy Cattle Using Latent Class Models

The aim of the present study was to calculate the sensitivity (Se) and specificity (Sp) of the single cervical tuberculin test (SCT), rapid lateral flow test (RLFT), and real-time polymerase chain reaction (RT-PCR) for the diagnosis of Mycobacterium bovis (M. bovis) infection in Egyptian dairy cattle herds within a Bayesian framework. The true M. bovis infection within-herd prevalence was assessed as a secondary objective. Data on the test results of SCT, RLFT, and RT-PCR for the detection of M. bovis were available from 245 cows in eleven herds in six major governorates in Egypt. A Bayesian latent class model was built for the estimation of the characteristics of the three tests. Our findings showed that Se of SCT (0.93 (95% Posterior credible interval (PCI): 0.89–0.93)) was higher than that of RT-PCR (0.83 (95% PCI: 0.28–0.93)) but was similar to the Se of RLFT (0.93 (95% PCI: 0.31–0.99)). On the contrary, SCT showed the lowest Sp estimate (0.60 (95% PCI: 0.59–0.65)), whereas Sp estimates of RT-PCR (0.99 (95% PCI: 0.95–1.00)) and RLFT (0.99 (95% PCI: 0.95–1.00)) were comparable. The true prevalence of M. bovis ranged between 0.07 and 0.71. In conclusion, overall, RT-PCR and RLFT registered superior performance to SCT, making them good candidates for routine use in the Egyptian bovine tuberculosis control program.

SARS-CoV-2 serology across scales: a framework for unbiased seroprevalence estimation incorporating antibody kinetics and epidemic recency

Serosurveys are a key resource for measuring SARS-CoV-2 cumulative incidence. A growing body of evidence suggests that asymptomatic and mild infections (together making up over 95% of all infections) are associated with lower antibody titers than severe infections. Antibody levels also peak a few weeks after infection and decay gradually. We developed a statistical approach to produce adjusted estimates of seroprevalence from raw serosurvey results that account for these sources of spectrum bias. We incorporate data on antibody responses on multiple assays from a post-infection longitudinal cohort, along with epidemic time series to account for the timing of a serosurvey relative to how recently individuals may have been infected. We applied this method to produce adjusted seroprevalence estimates from five large-scale SARS-CoV-2 serosurveys across different settings and study designs. We identify substantial differences between reported and adjusted estimates of over two-fold in the results of some surveys, and provide a tool for practitioners to generate adjusted estimates with pre-set or custom parameter values. While unprecedented efforts have been launched to generate SARS-CoV-2 seroprevalence estimates over this past year, interpretation of results from these studies requires properly accounting for both population-level epidemiologic context and individual-level immune dynamics.

An abattoir-based study on the prevalence of bovine tuberculosis from culled adult dairy cows in Wuhan, China

Abattoir surveillance is an integral component of a bovine tuberculosis (bTB) eradication programme. The objective of this study was to determine the prevalence of bTB among culled adult dairy cows in Wuhan, China and to further assess two diagnostic procedures as an adjunct to the confirmation of M. bovis in animals with TB-like lesions. The study was conducted in an abattoir located in Wuhan, China over a period of 41 days from June to July 2019. A total of 171 culled adult dairy cows were sampled and inspected, and blood samples collected from 134 of these. The viscera and lymph nodes of the carcasses were visually inspected and palpated for TB-like lesions. A total of 28.1 % (48/171) of the carcasses had gross TB-like lesions. 89.6 % (43/48) of the animals with TB-like lesions were positive to a PCR procedure for bTB. The sensitivity and specificity for post-mortem examination for TB-like lesions using a Bayesian latent class analysis model was estimated to be 60.8 % and 86.6 %, respectively. A seroprevalence of 20.9 % (28/134) was recorded for antibody response to M. bovis antigens MPB70 and MPB83 based on an ELISA procedure. There was a low-moderate agreement between the ELISA and PCR results in the detection of bTB (Kappa = 0.46, 95 % CI: 0.24-0.67). The study confirms a high prevalence of bTB among culled adult dairy cows in the abattoir and highlights the need to implement surveillance for bTB based on post-mortem examination and ELISA and PCR methods in association with backward tracing of infected dairy herds.

Bayesian Estimation of the Prevalence and Test Characteristics (Sensitivity and Specificity) of Two Serological Tests (RB and SAT-EDTA) for the Diagnosis of Bovine Brucellosis in Small and Medium Cattle Holders in Ecuador

In Ecuador, a national program for bovine brucellosis control has been in implementation since 2008. Given the costs, small- and medium-sized livestock holders are not completely committed to it. The objective of this study was to determine true prevalence (TP) of bovine brucellosis in small- and medium-sized herd populations, as well as the diagnostic sensitivity and specificity of the Rose Bengal (RB) test and the sero-agglutination test (SAT)-EDTA using a Bayesian approach. Between 2011 and 2016, 2733 cattle herds were visited, and 22,592 animal blood samples were taken in nineteen provinces on mainland Ecuador. Bayes-p and deviance information criterion (DIC) statistics were used to select models. Additionally, risk-factor analysis was used for herds according to their brucellosis test status. True prevalence (TP) in herds was estimated by pool testing. National seroprevalence of farms was 7.9% (95% CI: 6.79–9.03), and TP was 12.2% (95% CI: 7.8–17.9). Apparent prevalence (AP) in animals was 2.2% (95% CI: 1.82–2.67), and TP was 1.6% (95% CrI: 1.0–2.4). Similarly, the sensitivity of the RB was estimated at 64.6% (95% CrI: 42.6–85.3) and specificity at 98.9% (95% CrI: 98.6–99.0); for the SAT-EDTA test, sensitivity was 62.3% (95% CrI: 40.0–84.8) and 98.9% (95% CrI: 98.6–99.1) for specificity. Results of the two tests were highly correlated in infected and uninfected animals. Likewise, high spatial variation was observed, with the Coastal Region being the zone with the highest TP at 2.5%. (95% CrI: 1.3–3.8%) in individual animals and 28.2% (95% CI: 15.7–39.8) in herds. Risk factors include herd size, type of production (milk, beef, and mixed), abortions recorded, and vaccination. The results of this study serve to guide authorities to make decisions based on parallel testing at the beginning of a bovine brucellosis program for small livestock holders to increase sensitivity level of the screening tests in Ecuador.

Citywide serosurveillance of the initial SARS-CoV-2 outbreak in San Francisco using electronic health records

Serosurveillance provides a unique opportunity to quantify the proportion of the population that has been exposed to pathogens. Here, we developed and piloted Serosurveillance for Continuous, ActionabLe Epidemiologic Intelligence of Transmission (SCALE-IT), a platform through which we systematically tested remnant samples from routine blood draws in two major hospital networks in San Francisco for SARS-CoV-2 antibodies during the early months of the pandemic. Importantly, SCALE-IT allows for algorithmic sample selection and rich data on covariates by leveraging electronic health record data. We estimated overall seroprevalence at 4.2%, corresponding to a case ascertainment rate of only 4.9%, and identified important heterogeneities by neighborhood, homelessness status, and race/ethnicity. Neighborhood seroprevalence estimates from SCALE-IT were comparable to local community-based surveys, while providing results encompassing the entire city that have been previously unavailable. Leveraging this hybrid serosurveillance approach has strong potential for application beyond this local context and for diseases other than SARS-CoV-2.

The detection dogs test is more sensitive than real-time PCR in screening for SARS-CoV-2

In January 2020, the coronavirus disease was declared, by the World Health Organization as a global public health emergency. Recommendations from the WHO COVID Emergency Committee continue to support strengthening COVID surveillance systems, including timely access to effective diagnostics. Questions were raised about the validity of considering the RT-PCR as the gold standard in COVID-19 diagnosis. It has been suggested that a variety of methods should be used to evaluate advocated tests. Dogs had been successfully trained and employed to detect diseases in humans. Here we show that upon training explosives detection dogs on sniffing COVID-19 odor in patients’ sweat, those dogs were able to successfully screen out 3249 individuals who tested negative for the SARS-CoV-2, from a cohort of 3290 individuals. Additionally, using Bayesian analysis, the sensitivity of the K9 test was found to be superior to the RT-PCR test performed on nasal swabs from a cohort of 3134 persons. Given its high sensitivity, short turn-around-time, low cost, less invasiveness, and ease of application, the detection dogs test lends itself as a better alternative to the RT-PCR in screening for SARS-CoV-2 in asymptomatic individuals.

Hag-Ali and colleagues highlight the potential for using trained dogs for detecting COVID-19 positive patients. The dogs, originally trained for explosives detection, were able to detect COVID-19 positive sweat samples with a sensitivity rivaling the gold-standard RT-PCR test currently used.

Bayesian latent class model estimates of diagnostic accuracy for three test methods designed to detect spring viremia of carp virus

Spring viremia of carp virus (SVCV) causes a systemic hemorrhagic disease that poses a significant risk to wild and cultured fish and is listed as notifiable by the World Organization for Animal Health. Validated molecular diagnostic tools for SVCV are required to accurately describe and analyze the ecology of the virus. Here, the diagnostic specificity (DSp) and sensitivity (DSe) (i.e. accuracy) of three SVCV diagnostic tests - 2 reverse transcription quantitative polymerase chain reaction (RT-qPCR) assays Q1G and Q2N and virus isolation by cell culture (VI) - were evaluated using 2-class latent class models run in maximum likelihood (ML) and Bayesian frameworks. Virus-free or experimentally-infected koi were sorted into three populations with low, moderate or high prevalence levels of SVCV (n = 269 fish in total). Koi kidney tissues were tested using Q2N and Q1G and for the VI assay, pools of kidney, spleen and gill tissues were used. All samples were blinded and analyzed in one laboratory. The ML and Bayesian approaches successfully estimated the diagnostic accuracy of the 3 tests with the exception of 1 ML model. The estimates were consistent across the two frameworks. The DSe estimates were higher for Q1G (>98 %) and Q2N (>96 %) compared to VI (>60 %). The DSp of all three tests varied by 12-15 % (79-91 % for Q1G, 79-94 % for Q2N and 81-97 % for VI) across same-fish samples revealing the potential range in test performance for one sample. The 3 fish populations had distinct SVCV prevalence levels estimated at 0-3 % (low), 70-73 % (moderate) and 95-96 % (high). The Bayesian covariance models revealed minor DSe dependence between Q1G and Q2N. The results suggested that SVCV diagnostic tests Q2N and Q1G are suitable for use as diagnostic assays and are fit for presumptive diagnosis, surveillance, and certification of populations or individuals as SVCV free.

Evaluation of histopathology, PCR, and qPCR to detect Mikrocytos mackini in oysters Crassostrea gigas using Bayesian latent class analysis

Latent class analysis (LCA) is a common method to evaluate the diagnostic sensitivity (DSe) and specificity (DSp) for pathogen detection assays in the absence of a perfect reference standard. Here we used LCA to evaluate the diagnostic accuracy of 3 tests for the detection of Mikrocytos mackini in Pacific oysters Crassostrea gigas: conventional polymerase chain reaction (PCR), real-time quantitative PCR (qPCR), and histopathology. A total of 802 Pacific oysters collected over 12 sampling events from 9 locations were assessed. Preliminary investigations indicated that standard LCA assumptions of test independence and constant detection accuracy across locations were likely unrealistic. This was mitigated by restructuring the LCA in a Bayesian framework to include test-derived knowledge about pathogen prevalence and load for categorizing populations into 2 classes of infection severity (low or high) and assessing separate DSe and DSp estimates for each class. Median DSp estimates were high (>96%) for all 3 tests in both population classes. DSe estimates varied between tests and population classes but were consistently highest for qPCR (87-99%) and lowest for histopathology (21-51%). Acknowledging that detection of M. mackini may be fitted to multiple diagnostic and management purposes, qPCR had the highest DSe while maintaining similar DSp to both conventional PCR and histopathology and thus is generally well-suited to most applications.

An Iterative, Frequentist Approach for Latent Class Analysis to Evaluate Conditionally Dependent Diagnostic Tests

Latent class analysis is a well-established method in human and veterinary medicine for evaluating the accuracy of diagnostic tests without a gold standard. An important assumption of this procedure is the conditional independence of the tests. If tests with the same biological principle are used, this assumption is no longer met. Therefore, the model has to be adapted so that the dependencies between the tests can be considered. Our approach extends the traditional latent class model with a term for the conditional dependency of the tests. This extension increases the number of parameters to be estimated and leads to negative degrees of freedom of the model, meaning that not enough information is contained in the existing data to obtain a unique estimate. As a result, there is no clear solution. Hence, an iterative algorithm was developed to keep the number of parameters to be estimated small. Given adequate starting values, our approach first estimates the conditional dependencies and then regards the resulting values as fixed to recalculate the test accuracies and the prevalence with the same method used for independent tests. Subsequently, the new values of the test accuracy and prevalence are used to recalculate the terms for the conditional dependencies. These two steps are repeated until the model converges. We simulated five application scenarios based on diagnostic tests used in veterinary medicine. The results suggest that our method and the Bayesian approach produce similar precise results. However, while the presented approach is able to calculate more accurate results than the Bayesian approach if the test accuracies are initially misjudged, the estimates of the Bayesian method are more precise when incorrect dependencies are assumed. This finding shows that our approach is a useful addition to the existing Bayesian methods, while it has the advantage of allowing simpler and more objective estimations.

Disease Severity, Fever, Age, and Sex Correlate With SARS-CoV-2 Neutralizing Antibody Responses

Clinical trials on the use of COVID-19 convalescent plasma remain inconclusive. While data on safety is increasingly available, evidence for efficacy is still sparse. Subgroup analyses hint to a dose-response relationship between convalescent plasma neutralizing antibody levels and mortality. In particular, patients with primary and secondary antibody deficiency might benefit from this approach. However, testing of neutralizing antibodies is limited to specialized biosafety level 3 laboratories and is a time- and labor-intense procedure. In this single center study of 206 COVID-19 convalescent patients, clinical data, results of commercially available ELISA testing of SARS-CoV-2 spike-IgG and -IgA, and levels of neutralizing antibodies, determined by plaque reduction neutralization testing (PRNT), were analyzed. At a medium time point of 58 days after symptom onset, only 12.6% of potential plasma donors showed high levels of neutralizing antibodies (PRNT50 ≥ 1:320). Multivariable proportional odds logistic regression analysis revealed need for hospitalization due to COVID-19 (odds ratio 6.87; p-value 0.0004) and fever (odds ratio 3.00; p-value 0.0001) as leading factors affecting levels of SARS-CoV-2 neutralizing antibody titers in convalescent plasma donors. Using penalized estimation, a predictive proportional odds logistic regression model including the most important variables hospitalization, fever, age, sex, and anosmia or dysgeusia was developed. The predictive discrimination for PRNT50 ≥ 1:320 was reasonably good with AUC: 0.86 (with 95% CI: 0.79-0.92). Combining clinical and ELISA-based pre-screening, assessment of neutralizing antibodies could be spared in 75% of potential donors with a maximal loss of 10% of true positives (PRNT50 ≥ 1:320).

A Bayesian analysis of a Test and Vaccinate or Remove study to control bovine tuberculosis in badgers (Meles meles)

A novel five year Test and Vaccinate or Remove (TVR) wildlife research intervention project in badgers (Meles meles) commenced in 2014 in a 100km2 area of Northern Ireland. It aimed to increase the evidence base around badgers and bovine TB and help create well-informed and evidence-based strategies to address the issue of cattle-to-cattle spread and spread between cattle and badgers. It involved real-time trap-side testing of captured badgers and vaccinating those that tested negative for bTB (BadgerBCG-BCG Danish 1331) and removal of those that tested bTB positive using the Dual-Path Platform VetTB test (DPP) for cervids (Chembio Diagnostic Systems, Medford, NY USA). Four diagnostic tests were utilised within the study interferon gamma release assay (IGRA), culture (clinical samples and post mortem), DPP using both whole blood and DPP using serum. BCG Sofia (SL222) was used in the final two years because of supply issues with BadgerBCG. Objectives for this study were to evaluate the performance of the DPP in field conditions and whether any trend was apparent in infection prevalence over the study period. A Bayesian latent class model of diagnostic test evaluation in the absence of a gold standard was applied to the data. Temporal variation in the sensitivity of DPP and interferon gamma release assay (IGRA) due to the impact of control measures was investigated using logistic regression and individual variability was assessed. Bayesian latent class analysis estimated DPP with serum to have a sensitivity of 0.58 (95% CrI: 0.40-0.76) and specificity of 0.97 (95% CrI: 0.95-0.98). The DPP with whole blood showed a higher sensitivity (0.69 (95% CrI: 0.48-0.88)) but similar specificity (0.98 (95% Crl: 0.96-0.99)). The change from BCG Danish to BCG Sofia significantly impacted on DPP serum test characteristics. In addition, there was weak evidence of increasing sensitivity of IGRA over time and differences in DPP test sensitivity between adults and cubs. An exponential decline model was an appropriate representation of the infection prevalence over the 5 years, with a starting prevalence of 14% (95% CrI: 0.10-0.20), and an annual reduction of 39.1% (95% CrI: 26.5-50.9). The resulting estimate of infection prevalence in year 5 of the study was 1.9% (95% CrI: 0.8-3.8). These results provide field evidence of a statistically significant reduction in badger TB prevalence supporting a TVR approach to badger intervention. They give confidence in the reliability and reproducibility in the DPP Whole Blood as a real time trap-side diagnostic test for badgers, and describe the effect of vaccination and reduced infection prevalence on test characteristics.

Adapting a porcine reproductive and respiratory syndrome virus (PRRSV) oral fluid antibody ELISA to routine surveillance

Distinct from tests used in diagnostics, tests used in surveillance must provide for detection while avoiding false alarms, i.e., acceptable diagnostic sensitivity but high diagnostic specificity. In the case of the reproductive and respiratory syndrome virus (PRRSV), RNA detection meets these requirements during the period of viremia, but antibody detection better meets these requirements in the post-viremic stage of the infection. Using the manufacturer's recommended cut-off (S/P ≥ 0.4), the diagnostic specificity of a PRRSV oral fluid antibody ELISA (IDEXX Laboratories, Inc., Westbrook, ME, USA) evaluated in this study was previously reported as ≥ 97 %. The aim of this study was to improve its use in surveillance by identifying a cut-off that would increase diagnostic specificity yet minimally impact its diagnostic sensitivity. Three sample sets were used to achieve this goal: oral fluids (n = 596) from pigs vaccinated with a modified live PRRSV vaccine under experimental conditions, field oral fluids (n = 1574) from 94 production sites of known negative status, and field oral fluids (n = 1380) from 211 sites of unknown PRRSV status. Based on the analysis of samples of known status (experimental samples and field samples from negative sites), a cut-off of S/P ≥ 1.0 resulted in a diagnostic specificity of 99.2 (95 % CI: 98.8, 99.7) and a diagnostic sensitivity of 96.5 (95 % CI: 85.2, 99.2). Among 211 sites of unknown status, 81 sites were classified as antibody positive using the manufacturer's cut-off; 20 of which were reclassified as negative using a cut-off of S/P ≥ 1.0. Further analysis showed that these 20 sites had a small proportion of samples (18.0 %) with S/P values just exceeding the manufacturer's cut-off (x̄ = 0.5). Whereas the remainder of positive sites (n = 61) had a high proportion of samples (76.3 %) with high S/P values (x̄ = 6.6). Thus, the manufacturer's cut-off (S/P ≥ 0.4) is appropriate for diagnostic applications, but a cut-off of S/P ≥ 1.0 provided the higher specificity required for surveillance. A previously unreported finding in this study was a statistically significant association between unexpected reactors and specific production sites and animal ages or stages. While beyond the scope of this study, these data suggested that certain animal husbandry or production practices may be associated with non-specific reactions.

Estimation of the sensitivity and specificity of four serum ELISA and one fecal PCR for diagnosis of paratuberculosis in adult dairy cattle in New Zealand using Bayesian latent class analysis

In New Zealand, a new diagnostic approach for the control of paratuberculosis in mixed aged milking cows has been developed using a combination of ELISA and quantitative fecal PCR (f-qPCR). Our analysis was designed to evaluate performance of these individual tests in infected or infectious mixed aged cows across the prevalence of infection typically encountered on NZ dairy farms and calculate test accuracy when used as a screening test of serological ELISAs for four separate antigens read in parallel followed by a confirmatory quantitative f-qPCR test. Data from a cross-sectional study of 20 moderate prevalence herds was combined with existing data from 2 low and 20 high prevalence herds forming a dataset of 3845 paired serum and fecal samples. Incidence of clinical Johne's disease (JD) was used to classify herds into three prevalence categories. High (≥ 3% annual clinical JD for the last three years), moderate (<3 - 1%) and low (<1% incidence for at least the last five years). Positive tests were declared if> 50 ELISA units and f-qPCR at two cut-points (≥1 × 10⁴ genomes/mL or >1 × 10³ genomes/mL). Fixed Bayesian latent class models at both f-qPCR cut-points, accounted for conditional independence and paired conditional dependence. Mixed models at both f-qPCR cut-points, using a different mechanism to account for conditional dependencies between tests were also implemented. Models (24 in number) were constructed using OpenBUGS. The aim was to identify Mycobacterium avium subsp. paratuberculosis (MAP) infected cows that met at least one of two criteria: shedding sufficient MAP in feces to be detected by f-qPCR or mounting a detectable MAP antibody response. The best fit to the data was obtained by modelling pairwise dependencies between tests in a fixed model or by accounting for dependencies in a mixed model at a fecal cut-off of ≥1 × 10⁴ genomes/mL. Test performance differed with prevalence, but models were robust to prior assumptions. For the fixed model, at a prevalence of 0.29 (95 % probability interval (PI) = 0.25-0.33), as a screening plus confirmatory f-qPCR, post-test probability for disease in a positive animal was 0.84 (95 %PI = 0.80-0.88) and 0.16 (95 %PI = 0.15-0.18) for disease in a test negative animal. In low prevalence herds (0.01(95 %PI = 0.00-0.04)) the equivalent figures were 0.84 (95 %PI = 0.08-0.92) and 0.00 (95 %PI = 0.00-0.02). These results suggest this is a useful tool to control JD on dairy farms, particularly in herds with higher levels of infection, where the sampling and testing cost per animal is defrayed across more detected animals.

Validation of a Novel ELISA for the Diagnosis of Hemorrhagic Septicemia in Dairy Cattle from Thailand Using a Bayesian Approach

The objective of this study was to estimate sensitivity (Se) and specificity (Sp) of a novel enzyme-linked immunosorbent assay (ELISA) test (using a coating antigen from Pasteurella multocida M-1404 via heat extraction) and an indirect hemagglutination (IHA) test for detection of Hemorrhagic septicemia (HS) in dairy cows, under Thai conditions, using a Bayesian approach. Dairy cow sera with a total of 1236 samples from 44 farms were tested with the two tests to detect immune responses against the HS. Percentages of positive samples for the ELISA and IHA tests were 73% (901/1236) and 70% (860/1236), respectively. Estimated sensitivity and estimated specificity of the ELISA test were 90.5% (95% posterior probability interval (PPI) = 83.2–95.4%) and 70.8% (95% PPI = 60.8–79.8%), respectively. Additionally, estimates for the Se and Sp values of the IHA test were 77.0% (95% PPI = 70.8–84.1%) and 51.1% (PPI = 36.8–66.3%), respectively. The estimated prevalence of the disease was 71.7% (95% PPI = 62.7–82.6%). These results demonstrate that the ELISA test can be a useful tool for the detection of the presence of an antibody against the HS in dairy cows. Notably, the cows in this area indicated a high percentage of exposure to Pasteurella multocida.

Early-Life Factors and Risk of Multiple Sclerosis: An MR-EWAS

BACKGROUND

Although several risk factors are associated with multiple sclerosis (MS) in adulthood, evidence for risk factors acting from birth to adolescence is scarce.

METHODS

We conceived a 2-step study design, where signals from an Environment-Wide Association Study are prioritized for follow-up in a Mendelian Randomization study (MR-EWAS), to examine the association of early-life factors with risk of MS. The EWAS was conducted in UK Biobank, where we agnostically selected all the available risk factors acting from the perinatal period until the adolescence, including perinatal factors, anthropometric characteristics during childhood, male and female sexual factors, and skin phenotypic characteristics. We prioritized statistically significant risk factors to perform a 2-sample MR study using publicly available summary-level genetic data. We also calculated the power of the 2-step MR-EWAS approach under several scenarios and compared it against a 1-step hypothesis-free MR approach to detect risk factors of MS.

RESULTS

In the EWAS, an increase per 1 year in age at menarche was associated with a lower risk of MS (OR = 0.93; 95% CI: 0.90-0.96) and a plumper than average body size at the age of 10 was associated with a higher risk of MS (OR = 1.42; 95% CI: 1.24-1.61). Individuals getting very tanned or moderately tanned were at higher risk of MS compared with individuals that never tan or get mildly to occasionally tanned (OR = 0.86; 95% CI: 0.79-0.94). The MR analysis supported the association of age at menarche and childhood body mass index (BMI) without presence of pleiotropic effects. In the multivariable MR analysis, the association of age at menarche was not statistically significant after adjusting for childhood BMI. The MR analysis for ease of tanning did not reveal a statistically significant association. In multiple scenarios, the power of MR-EWAS approach was larger than the power of a hypothesis-free MR approach.

CONCLUSIONS

We introduced the MR-EWAS, a 2-step approach that is more powerful compared with the hypothesis-free MR approach under certain scenarios, to test potential causal signals. Our comprehensive assessment of early-life risk factors of MS highlighted a potential causal role of early menarche and elevated childhood BMI for risk of MS.