Bayesian Epidemiologic Screening Techniques

Graduate Group in Epidemiology | University of California, Davis
| Bayesian Epidemiologic Screening Techniques (BEST) Laboratory at University of California, Davis |
| Software Modules | Prevalence Estimation | Disease Freedom | Diagnostic Test Se and Sp Estimation | More |
| Methodological Papers | Applications |
| Presentations and Talks |
| Workshops |
| Statistics | Medicine and Epidemiology | Master of Preventive Veterinary Medicine |
| Glossary of Epidemiological Terms |
| Academic / Research Related | Scientific Journals | Software Sites | Web-based Hot Topics and Lists |
| Research Members Contact Information |

Estimation of herd-level prevalence: 1 test, multiple populations

WinBUGS 1.4 code to accompany the paper entitled "Branscum AJ, Gardner IA, Johnson WO. Bayesian modeling of animal- and herd-level prevalences. Prev Vet Med. 2004 Dec 15;66(1-4):101-112".

Code prepared by Adam Branscum, November 18, 2003
branscum@ms.uky.edu
Departments of Biostatistics and Statistics
University of Kentucky

Example from section 3.2.1.
Estimation of the herd-level prevalence (tau) and prevalence distribution of Johne's disease in California based on screening with IDEXX ELISA.



model;
{
Se ~ dbeta(58.8, 174.5) ## Mode=0.25; 95% sure < 0.30
Sp ~ dbeta(272.4, 6.5) ## Mode=0.98, 95% sure > 0.96
tau ~ dbeta(4.8, 3.6) ## Mode=0.60, 95% sure < 0.827
alpha <- mu*psi
beta <- psi*(1-mu)
mu ~ dbeta(3.283, 17.744) ## Mode=0.12, 95% sure < 0.30.
psi ~ dgamma(4.524, 0.387) ## Uses Median of 95th percentile of prevalence
## distribution=0.30 and 99% sure this number is < 0.50
for(i in 1:k)
{
prob.tpos[i] <- pi[i]*Se + (1-pi[i])*(1-Sp)
y[i] ~ dbin(prob.tpos[i], n)
inf[i] ~ dbern(tau)
pi.star[i] ~ dbeta(alpha,beta)
pi[i] <- pi.star[i] * inf[i]
}
Z30 ~ dbern(tau)
pistar30 ~ dbeta(alpha,beta)
pi30 <- Z30*pistar30
a1 <- 1-step(pi30 - 0.05)
a2 <- 1-step(pi30-0.5)
a3 <- equals(pi30,0)
b1 <- step(tau-0.50)
}
list(k=29, n=60, y=c(2,1,2,2,3,6,0,6,3,13,2,3,1,7,2,2,0,4,1,2,6,1,4,0,5,4,2,0,13))
list(Se=0.25, Sp=0.98, mu=0.12, psi=11.69, tau=0.60,
pi.star=c(0.05, 0.05,0.05,0.05,0.05,0.05, 0.05,0.05,0.05,0.05,0.05, 0.05,0.05,0.05,0.05,0.05, 0.05,0.05,0.05,0.05,0.05, 0.05,0.05,0.05,0.05,0.05, 0.05,0.05,0.05),
inf=c(1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1), Z30=0, pistar30=0.05)
 

Output

node
mean
sd
MC error
2.50%
median
97.50%
start
sample
Se
0.2642
0.02716
2.03E-04
0.2133
0.2634
0.3188
5001
50000
Sp
0.9756
0.00618
6.57E-05
0.9629
0.9758
0.987
5001
50000
a1
0.5146
0.4998
0.002515
0
1
1
5001
50000
a2
0.9702
0.1701
8.35E-04
0
1
1
5001
50000
a3
0.4201
0.4936
0.002906
0
0
1
5001
50000
alpha
1.758
1.155
0.02111
0.4283
1.47
4.756
5001
50000
b1
0.7001
0.4582
0.005842
0
1
1
5001
50000
beta
6.462
3.168
0.04273
2.081
5.863
14.25
5001
50000
mu
0.2084
0.06054
0.001253
0.1068
0.2026
0.3428
5001
50000
pi30
0.1175
0.1551
8.69E-04
0
0.04166
0.5196
5001
50000
psi
8.22
4.128
0.06068
2.638
7.415
18.43
5001
50000
tau
0.5792
0.1428
0.002234
0.3005
0.5792
0.847
5001
50000

 

©2000-2007 Department of Medicine and Epidemiology, University of California, Davis