1
2
3
Diagnostic accuracy of PCR for Jaagsiekte sheep retrovirus using field data from 125
Scottish sheep flocks
4
5
6
7
8
9
10
11
12
13
14
15
16
17
F.I. Lewis a,*, F. Brülisauer a, C. Cousens c, I.J. McKendrick b, G.J. Gunn a
a
Epidemiology Research Unit, SAC (Scottish Agricultural College), King’s Buildings, West
Mains Road, Edinburgh, EH9 3JG, UK
b
Biomathematics and Statistics Scotland, King’s Buildings, West Mains Road, Edinburgh,
EH9 3JZ, UK
c
Moredun Research Institute, Pentlands Science Park, Bush Loan, Penicuik, EH26 0PZ, UK
* Corresponding author. Tel.: +44 1463 243030
Email address: fraser.lewis@sac.ac.uk (F.I. Lewis).
18
19
Abstract
Using a representative sample of Scottish sheep comprising 125 flocks, the sensitivity
20
and specificity of PCR for Jaagsiekte sheep retrovirus (JSRV) was estimated. By combining
21
and adapting existing methods, the characteristics of the diagnostic test were estimated (in the
22
absence of a gold standard reference) using repeated laboratory replicates. As the results of
23
replicates within the same animal cannot be considered to be independent, the performance of
24
the PCR was calculated at individual replicate level.
25
26
The median diagnostic specificity of the PCR when applied to individual animals drawn
27
from the Scottish flock was estimated to be 0.997 (95% confidence interval [CI] 0.996-0.999),
28
whereas the median sensitivity was 0.107 (95% CI 0.077-0.152). Considering the diagnostic
29
test as three replicates where a positive result on any one or more replicates results in a
30
positive test, the median sensitivity increased to 0.279. Reasons for the low observed
31
sensitivity were explored by comparing the performance of the test as a function of the
32
concentration of target DNA using spiked positive controls with known concentrations of
33
target DNA. The median sensitivity of the test when used with positive samples with a mean
34
concentration of 1.0 target DNA sequence per 25 μL was estimated to be 0.160, which
35
suggests that the PCR had a high true (analytical) sensitivity and that the low observed
36
(diagnostic) sensitivity in individual samples was due to low concentrations of target DNA in
37
the blood of clinically healthy animals.
38
39
40
Keywords: Jaagsiekte sheep retrovirus; Diagnostic test validation; PCR; Modelling
41
42
Introduction
Jaagsiekte sheep retrovirus (JSRV) is the aetiological agent of ovine pulmonary
43
adenocarcinoma (OPA), an infectious lung tumour of sheep occurring in almost all countries
44
but absent from Australia, New Zealand and Iceland. Currently, there is no treatment or
45
vaccination for JSRV infection and clinical OPA is inevitably fatal. OPA can cause
46
substantial losses in affected flocks and, in order to prevent spread of JSRV infection, a
47
reliable diagnostic test for detection of infected sheep is needed.
48
49
No cost effective serological assays are available for JSRV, since the virus does not
50
induce a specific antibody response in infected animals (Sharp and Herring, 1983; Ortin et al.,
51
1998). Current JSRV diagnostic tests are based on virus detection, e.g. from blood or
52
bronchoalveolar lavage samples, observation of clinical signs of OPA in advanced clinical
53
cases, and identification of OPA lesions at post mortem examination. However, no routine
54
assays for pre-clinical diagnosis of JSRV infection are available.
55
56
PCR for JSRV is not used routinely since there are reservations regarding its suitability
57
for diagnosis of JSRV infection outside the research environment (De las Heras et al., 2005;
58
Voigt et al., 2007). Viable implementation of any assay into routine diagnostics is dependent
59
upon the accuracy of the diagnostic test. Hence, thorough validation of the test against the
60
target population is essential.
61
62
In this study, we used blood samples from a national survey commissioned by the
63
Scottish Government for validation of a JSRV PCR assay. The diagnostic test used is similar
64
to the hemi-nested PCR described by De las Heras et al. (2005). Previous work suggested that
65
the diagnostic accuracy of this test is highly dependent upon (1) the specimen tested and (2)
66
the stage of disease in the animal being sampled. De las Heras et al. (2005) noted that the
67
sensitivity of the test when based on blood samples from infected but clinically healthy
68
animals was too low to provide a reliable result at the individual animal level, and these
69
authors recommended flock level testing. This conclusion was based on sampling from six
70
animals infected with JSRV, but with no clinical evidence of disease.
71
72
Voigt et al. (2007) suggested that the sensitivity of a similar JSRV PCR used with blood
73
samples may be as low as 10% at the individual animal level; this estimate was based on a
74
study population of 47 Grey Heath sheep with histologically confirmed OPA lesions. These
75
experimental studies used small sample sizes with repeated sampling of individual animals
76
and confirmatory tests in live and dead animals.
77
78
Although certain findings from these two experimental studies may not be applicable for
79
diagnosis of JSRV infection under field conditions, the observed association between disease
80
status and diagnostic accuracy is of relevance because in prevalence surveys it is expected
81
that the majority of animals tested will be clinically healthy, i.e. the likelihood of detecting an
82
individual infected sheep will be low. Diagnostic sensitivity and specificity are population
83
parameters that describe the test performance for a given reference population (Greiner and
84
Gardner, 2000). So it is important to question the accuracy of the JSRV PCR assay under
85
given circumstances, in our case when applied to the Scottish sheep flock. The answer to this
86
question has implications for future disease monitoring and control in the target population.
87
88
Hughes and Totten (2003) proposed that the sensitivity of PCR assays should be
89
specified as a function of the number of target DNA molecules present. However, in field
90
samples the concentration of target DNA is unknown and estimates of sensitivity and
91
specificity of the test used are not functions of concentration but rather averages over the
92
range of possible concentrations which occur biologically within the subjects being sampled.
93
94
An extensive body of literature exists on methods of validation for diagnostic tests in the
95
absence of a gold standard reference test. Hui and Walter (1980) defined the necessary
96
conditions for test sensitivities and specificities to be estimated using maximum likelihood
97
methods. Later additions include Bayesian approaches (Joseph et al., 1995) and allowance for
98
covariance between tests (Dendukuri and Joseph, 2001). The use of non-gold standard
99
methods, particularly Bayesian methods, in diagnostic testing features heavily in modern
100
veterinary epidemiology (Enoe et al., 2000).
101
102
In this study, our primary objective was to estimate the accuracy of the JSRV PCR when
103
applied to the Scottish sheep flock. A secondary objective was to present a novel statistical
104
approach for estimating sensitivity and specificity of a diagnostic test in the absence of a gold
105
standard reference test, using laboratory replicates to increase the amount of data available for
106
analysis.
107
108
Materials and methods
109
Data
110
Data were collected from a representative random sample of 125 Scottish sheep flocks.
111
Study farms were stratified by the Scottish Government Animal Health Division to take
112
account of the distribution of sheep flocks in different Scottish regions (see Supplementary
113
material, Appendix A). Only flocks with at least 50 breeding ewes were eligible to take part in
114
the study. In each flock, blood was collected from a random sample of animals, typically 27
115
sheep, and each blood sample was subsequently tested for the presence of JSRV proviral
116
DNA using a hemi-nested PCR (Palmarini et al., 1996), except that 800 ng DNA were used
117
per replicate and the second round was a Taqman PCR using the carboxyfluorescein (FAM)
118
labelled probe 5’-AGCAAACATCCGAGCCTTAAGAGCTTTC-3’ using an Applied
119
Biosystems SDS7000.
120
121
Samples from each flock were tested separately and comprised three replicate aliquots
122
from each blood sample, along with a set of three positive controls of varying JSRV DNA
123
concentrations (each with one aliquot) and typically four negative control samples (each with
124
three replicate aliquots). The negative controls were from differing sources, namely, cow
125
blood, Icelandic sheep blood, distilled water and a buffer solution. A total of 499 negative
126
control samples were available, each with three replicate aliquots; samples from one flock
127
were tested with three rather than four negative controls.
128
129
Table 1 summarises the test results of the negative controls and field samples. We
130
ignored the source of the negative control samples, as there was no evidence to suggest any
131
differences associated with source in the mean proportion of replicates falsely testing positive.
132
A total of 121 positive control samples was included in the analysis. Table 2 provides a
133
summary of the positive control data.
134
135
136
Statistical method
In the analysis of field samples three issues were relevant to statistical estimation of the
137
sensitivity and specificity of the JSRV PCR. Firstly, test results from individual blood
138
samples were not validated against a gold standard reference test to determine the true status
139
of each sample. Secondly, replicate aliquots were available from each blood sample, which
140
increased the amount of data available; however, these results could not be assumed to be
141
independent and therefore an appropriate adjustment was needed to correct for correlations
142
among replicates. Finally, the probability of a flock being free from the infectious agent (i.e.
143
the within flock prevalence can equal zero with non-zero probability) needed to be accounted
144
for. To accommodate each of these complications, we used a Bayesian non-gold standard
145
latent variable model (Enoe et al., 2000), with conditional dependence between replicates
146
from the same sample (Dendukuri and Joseph, 2001), where the latent variable denoting
147
within flock JSRV prevalence has a mixture distribution (Branscum et al., 2004).
148
149
Our study was based on a single diagnostic test with conditionally dependent replicates
150
and was considered to be a special case based on the approach of Dendukuri and Joseph
151
(2001), with the sensitivity and specificity being the same in each test. The observed data
152
within a single flock were modelled using a multinomial distribution, which defines the
153
probability of observing animals with zero, one, two or three positive replicates, given a fixed
154
total number of animals sampled (see Supplementary file).
155
156
The statistical model allowed the prevalence of JSRV to vary between flocks and we
157
estimated sensitivity and specificity across all flocks. The likelihood function for a single
158
flock is multinomial and the likelihood function for all flocks in our study is the product of
159
the likelihood functions for individual flocks, where we allowed the prevalence of JSRV in
160
each flock to vary independently (see Supplementary file). We used a Bayesian model with
161
uninformative priors for all parameters and fitted the model using JAGS, an open source
162
software package for running Markov chain Monte Carlo analyses similar to WinBUGS (see
163
Supplementary file).
164
165
In the analysis of control samples we estimated the sensitivity and specificity of the test
166
when applied to the JSRV positive and JSRV negative control samples. These control samples
167
were primarily used as quality assurance checks during the laboratory testing process;
168
however, they also provided potential bounds on the accuracy of the test when applied to
169
samples of unknown status. Analysis of the negative control samples followed the same
170
method as for the field samples, since they were distinct samples with three replicates each,
171
with the knowledge that the true status of the sample was negative.
172
173
The positive control samples required a different approach, since we had no replicates
174
but rather a single sample at three different dilutions. We adopted the parametric approach of
175
Hughes and Totten (2003), which discriminated between ‘observed’ sensitivity and ‘true’
176
sensitivity. The former includes test error due to (1) the aliquot under study contains no copies
177
of the target DNA sequence; or (2) although target DNA is present, the PCR fails to amplify
178
the DNA. It is argued that ‘true’ sensitivity only includes the error associated with (2) and that
179
sensitivity should be a function of the number of target DNA molecules. The observed
180
sensitivity may be estimated using standard methods, such as logistic regression with dilution
181
as a covariate. In contrast, estimating true sensitivity requires certain probabilistic
182
assumptions, e.g. the number of DNA molecules follows a Poisson distribution (see
183
Supplementary file).
184
185
Results
186
Field samples
187
Fitting our statistical model to the field data, we estimated that sensitivity (S) of the PCR
188
had a posterior median of 0.107 and a 95% CI of 0.077-0.152. In contrast, we found that the
189
test was highly specific, with a posterior median for specificity (C) of 0.997 (95% CI 0.996-
190
0.999). Estimates of the posterior densities for S and C are illustrated in Fig. 1. The estimated
191
covariance within sample replicates was low, with a median of 2.59 x 10-3 when JSRV was
192
present (covs) and a median of 3.63 x 10-6 when JSRV was absent (covc).
193
194
195
Control samples
Fitting our statistical model to the negative control samples (Table 1), the estimated 95%
196
CI was 0.982-0.993 for S and 1.04 x 10-6 to 1.41 x 10-4 for covc. Using the method of Hughes
197
and Totten (2003) for estimating the true S of the test on the positive control samples, median
198
S estimates for mean concentrations of 1, 6, 12.5 and 25 target DNA molecules per 25 µL
199
were 0.160, 0.648, 0.886 and 0.987, respectively. In contrast, the raw observed S of the test
200
using the data in Table 2 were 0.793, 0.884 and 0.901 for mean concentrations of 6, 12.5 and
201
25 target DNA molecules per 25 μL, respectively. The method also allows for explicit
202
estimation of C; however, given that we had median estimates for C from both the field
203
samples and negative control samples in excess of 0.99, we assumed that the probability of
204
observing a false positive is zero.
205
206
Fig. 2 shows estimates of the posterior density for S at the three observed concentrations,
207
plus extrapolation when the mean concentration is one copy of target DNA sequence in 25
208
μL. A key parameter in the mechanistic model used by Hughes and Totten (2003) is the
209
probability that each of the target molecules in the sample being tested fails to escape
210
amplification by PCR, where we estimate this to be 0.160. As we have assumed that the
211
specificity is 1.0, then this is equivalent to the S when the mean concentration is 1.0 target
212
DNA sequence per 25 μL.
213
214
Summary probabilities
215
Table 3 contains summary statistics for each of the eight independent conditional
216
probabilities estimated from our model. These summarise the probabilities of observing zero
217
or more positive replicates conditional on whether JSRV infection is truly present. The
218
probability of observing one or more positive replicates is considerably larger when JSRV is
219
truly present, as should be expected.
220
221
222
Discussion
We estimated that the median sensitivity of the JSRV PCR was 0.107 per individual
223
replicate, where this accounted for covariance between replicates from the same sample. We
224
found that the covariance between replicates was low, which was unsurprising, given the
225
observed data: out of a total 3,361 sets of triple replicates, only 106 had one positive replicate,
226
14 had two positive replicates and four had three positive replicates. Hence, there is little
227
obvious covariance between positive replicates, even from positive animals.
228
229
Considering the performance of the PCR when applied to control samples it was
230
interesting to note that estimates of C using the field samples were slightly higher than those
231
derived from the negative controls. This could in part be explained by the fact that S and C in
232
the field data were negatively correlated. This is to be expected, since S + C must be greater
233
than 1 to have a ‘legitimate’ (better than random guessing) diagnostic test. In our modelling,
234
we assumed uninformative independent priors for S and C. An alternative would be to
235
explicitly model this correlation using a joint prior, as for Chu et al. (2006).
236
237
For the positive control samples, we found that the estimates of observed S were lower
238
than estimates of true S; this was to be expected, since the former also includes the error due
239
to individual samples not containing any target DNA. Table 3 showed an alternative and
240
potentially more informative assessment of test accuracy. The probability of observing no
241
positive test replicates, given that JSRV is truly present within the blood sample, is in excess
242
of 0.70. If we define a positive test result as where any one of the available three replicates is
243
positive, then we can approximate the sensitivity of the test across all three replicates by
244
summing the individual probabilities in rows 3, 5 and 7 in Table 3, giving a median observed
245
sensitivity of 0.279. This result is in line with observations in previous studies, where PCR
246
results based on blood samples were compared with other diagnostic procedures (Gonzalez et
247
al., 2001; De las Heras et al., 2005; Voigt et al., 2007).
248
249
The primary objective of our modelling was to estimate the diagnostic accuracy of the
250
JSRV PCR when applied to the Scottish sheep flock. A necessary and interdependent part of
251
this process is estimation of the prevalence of JSRV within each flock sampled. We explicitly
252
allowed the prevalence of JSRV to vary independently within each of the study flocks. If,
253
instead, the primary goal of our study was estimation of prevalence and, in particular, at
254
regional or national flock level, then a natural extension to our model would be to incorporate
255
it into the hierarchical framework of Branscum et al. (2004).
256
257
In previous work, the study population consisted largely of animals which were likely to
258
be in early and late stages of OPA; these studies concluded that the ‘observed’ sensitivity in
259
preclinical animals is considerably lower than in animals with clinical OPA (De las Heras et
260
al., 2005; Voigt et al., 2007). Samples for the current study were taken from a random
261
selection of Scottish sheep; the vast majority of these animals were clinically healthy. The test
262
characteristics assessed in this validation based on a representative field sample would
263
therefore be applicable for prevalence studies or diagnostic screening of clinically healthy
264
sheep.
265
266
We estimated the ‘true’ sensitivity of the JSRV PCR as a function of the number of
267
target DNA molecules present using the associated spiked positive control samples to assess
268
the detection limit of the PCR. There were two key findings from this analysis. Firstly, we
269
estimated a median sensitivity of 0.160 when the mean number of target DNA molecules per
270
25 μL is 1.0; this observation is indicative of a high technical performance of the PCR assay
271
and a good ‘true’ sensitivity. Secondly, comparing the estimate of 0.160 with the median
272
sensitivity from field samples of 0.107 strongly suggests that the concentrations of target
273
DNA in the latter are generally very low. The fact that only a few samples from infected
274
animals tested positive in more than one replicate leads to the same conclusion and hence
275
potentially explains the low ‘observed’ sensitivity.
276
277
We used a non-gold standard method to validate the diagnostic test. The use of non-gold
278
standard methods in practice requires considerable care, e.g. requiring sufficient observations
279
to enable robust estimation (Toft et al., 2005). In the work presented, we required data from at
280
least four flocks, each with different levels of prevalence, in order to calculate test
281
characteristics (see Supplementary file).
282
283
The JSRV PCR assay was assessed to have a generally low ‘observed’ sensitivity when
284
used with blood samples from clinically healthy sheep. Therefore, improvements and
285
adjustments would be necessary should the test become part of routine diagnostic
286
investigations. Theoretically, the assay could be further enhanced, but, given that various
287
technical refinements have been implemented in the past, the PCR seems to have reached
288
what is currently possible with state of the art technology. On an individual animal level, the
289
‘observed’ sensitivity could be improved by testing specimens which have a higher
290
concentration of JSRV proviral DNA in infected animals, e.g. bronchoalveolar lavage
291
samples. An obvious next step is estimation of the flock level sensitivity of the test when
292
applied to the Scottish sheep flock. This is, however, a considerably more complex task, since
293
flock level sensitivity depends jointly on the accuracy of the JSRV PCR assay at individual
294
animal/replicate level, the number of animals sampled from within each flock and, crucially,
295
the distribution of within flock prevalence of JSRV in the population under study.
296
297
298
Conclusions
Using non-gold standard methods, which make use of laboratory replicates to maximise
299
available data, the JSRV PCR was assessed to have a high ‘true’ sensitivity and low
300
‘observed’ sensitivity, where the latter can be explained by the low concentration of JSRV
301
proviral DNA in the blood of infected sheep. The analytical method presented is generic and
302
applicable to diagnostic test validation when repeated measurements are available.
303
304
305
306
Conflict of interest statement
None of the authors has any financial or personal relationships that could
inappropriately influence or bias the content of the paper.
307
308
309
Acknowledgements
Fieldwork and data collection were funded as part of RERAD grant MRI843/04 awarded
310
to MRI, BioSS and SAC. Statistical analyses were undertaken by the Scottish Government
311
Centre of Excellence in epidemiology, population health and infectious disease control
312
(EPIC). Leenadevi Thonur and Joanne Crawford performed all PCR assays. Biomathematics
313
and Statistics Scotland, Moredun Research Institute and the Scottish Agricultural College all
314
receive financial support from the Scottish Government (RERAD).
315
316
Appendix A. Supplementary material
317
318
Supplementary data associated with this article can be found in the online version, at
doi:….
319
320
References
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
Branscum, A. J., Gardner, I. A., Johnson, W. O., 2004. Bayesian modeling of animal- and
herd-level prevalences. Preventive Veterinary Medicine 66, 101-112.
Chu, H., Wang, Z., Cole, S. R., Greenland, S., 2006. Sensitivity analysis of misclassification:
a graphical and a Bayesian approach. Annals of Epidemiology 16, 834-841.
De las Heras, M., Ortin, A., Salvatori, D., de Villareal, M. P., Cousens, C., Ferrer, L. M.,
Cebrian, L. M., de Jalon, J. A. G., Gonzalez, L., Sharp, J. M., 2005. A PCR technique for
the detection of Jaagsiekte sheep retrovirus in the blood suitable for the screening of
ovine pulmonary adenocarcinoma in field conditions. Research in Veterinary Science 79,
259-264.
Dendukuri, N., Joseph, L., 2001. Bayesian approaches to modeling the conditional
dependence between multiple diagnostic tests. Biometrics 57, 158-167.
Enoe, C., Georgiadis, M. P., Johnson, W. O., 2000. Estimation of sensitivity and specificity of
diagnostic tests and disease prevalence when the true disease state is unknown.
Preventive Veterinary Medicine 45, 61-81.
Gonzalez, L., Garcia-Goti, M., Cousens, C., Dewar, P., Cortabarria, N., Extramiana, A. B.,
Ortin, A., De las Heras, M., Sharp, J. M., 2001. Jaagsiekte sheep retrovirus can be
detected in the peripheral blood during the pre-clinical period of sheep pulmonary
adenomatosis. Journal of General Virology 82, 1355-1358.
Greiner, M., Gardner, I. A., 2000. Epidemiologic issues in the validation of veterinary
diagnostic tests. Preventive Veterinary Medicine 45, 3-22.
Hughes, J. P., Totten, P., 2003. Estimating the accuracy of polymerase chain reaction-based
tests using endpoint dilution. Biometrics 59, 505-511.
Hui, S. L., Walter, S. D., 1980. Estimating the error rates of diagnostic-tests. Biometrics 36,
167-171.
Joseph, L., Gyorkos, T. W., Coupal, L., 1995. Bayesian-estimation of disease prevalence and
the parameters of diagnostic-tests in the absence of a gold standard. American Journal of
Epidemiology 141, 263-272.
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
Ortin, A., Minguijon, E., Dewar, P., Garcia, M., Ferrer, L. M., Palmarini, M., Gonzalez, L.,
Sharp, J. M., De las Heras, M., 1998. Lack of a specific immune response against a
recombinant capsid protein of Jaagsiekte sheep retrovirus in sheep and goats naturally
affected by enzootic nasal tumour or sheep pulmonary adenomatosis. Veterinary
Immunology and Immunopathology 61, 229-237.
Palmarini, M., Holland, M. J., Cousens, C., Dalziel, R. G., Sharp, J. M., 1996. Jaagsiekte
retrovirus establishes a disseminated infection of the lymphoid tissues of sheep affected
by pulmonary adenomatosis. Journal of General Virology 77, 2991-2998.
Sharp, J. M., Herring, A. J., 1983. Sheep pulmonary adenomatosis - demonstration of a
protein which cross-reacts with the major core proteins of Mason-Pfizer monkey virus
and mouse mammary-tumor virus. Journal of General Virology 64, 2323-2327.
Toft, N., Jørgensen, E., Højsgaard, S., 2005. Diagnosing diagnostic tests: evaluating the
assumptions underlying the estimation of sensitivity and specificity in the absence of a
gold standard. Preventive Veterinary Medicine 68, 19-33.
Voigt, K., Brugmann, M., Huber, K., Dewar, P., Cousens, C., Hall, M., Sharp, J. M., Ganter,
M., 2007. PCR examination of bronchoalveolar lavage samples is a useful tool in preclinical diagnosis of ovine pulmonary adenocarcinoma (Jaagsiekte). Research in
Veterinary Science 83, 419-427.
381
Table 1
382
Observed test results from field and negative control samples
383
Number of positive
Number of field
Number of negative
replicates
samples
control samples
0
3237
486
1
106
11
2
14
2
3
4
0
Total
3361
499
384
385
Each observation is the number of replicates out of a total of three which tested positive for
386
JSRV.
387
Table 2
388
Observed test results from positive control samples
389
390
Mean number of
Number of
Number of samples
DNA plasmids
samples
positive for JSRV
6
121
96
12.5
121
107
25
121
109
391
Table 3
392
Posterior estimates of test accuracy applied to field samples.
393
Parameter
Median
95% Confidence interval
P (0; I+) a
7.21 x 10-1
6.24 x 10-1;7.92 x 10-1
P (0; I-)
9.92 x 10-1
9.87 x 10-1;9.96 x 10-1
P (1; I+)
2.48 x 10-1
1.92 x 10-1;3.14 x 10-1
P (1; I-)
7.61 x 10-3
3.82 x 10-3;1.25 x 10-2
P (2; I+)
2.29 x 10-2
1.14 x 10-2;4.67 x 10-2
P (2; I-)
8.59 x 10-6
3.81 x 10-7;3.61 x 10-5
P (3; I+)
8.16 x 10-3
2.61 x 10-3;2.13 x 10-2
P (3; I-)
8.39 x 10-6
3.54 x 10-7;3.58 x 10-5
394
395
a
396
infection negative for JSRV, respectively.
397
Number of positive replicates; Infection status. I+ and I- denote infection positive and
398
Figure legends
399
400
Fig. 1. Posterior density estimates for test sensitivity and specificity against field samples
401
from 125 flocks.
402
403
Fig. 2. Posterior density estimates for test sensitivity (S) using positive control samples at
404
mean concentrations of 1.0, 6.0, 12.5 and 25 target DNA molecules per 25 μL.