Skip to main content

Table 6 Posterior medians and 95 per cent posterior credible intervals for the application of the models to simulated data, with low prevalence (π = 0.01) and large sample size (n = 20000)

From: A two-stage Bayesian method for estimating accuracy and disease prevalence for two dependent dichotomous screening tests when the status of individuals who are negative on both tests is unverified

True value

Informative priors for model

Informative priors for Se alone

Two-step informative priors for Se alone

Informative priors for Sp alone

π = 0.01

0.0094 (0.0079,0.0120)

0.0095 (0.0080,0.0119)

0.0093 (0.0075,0.0118)

0.0115 (0.0072,0.0590)

Se1 = 0.6

0.6013 (0.5035,0.6944)

0.5971 (0.5063,0.6936)

0.5972 (0.5045,0.8695)

0.4181 (0.0195,0.9603)

Se2 = 0.7

0.6998 (0.6047,0.7949)

0.6904 (0.6036,0.7940)

0.6816 (0.6040,0.9019)

0.3656 (0.0177,0.9636)

Sp1 = 0.7

0.7000 (0.6941,0.7052)

0.6998 (0.6947,0.7054)

0.700 (0.6945,0.7054)

0.6982 (0.6824,0.7057)

Sp2 = 0.6

0.5996 (0.5951,0.6045)

0.5999 (0.5948,0.6049)

0.5998 (0.5948,0.6048)

0.5964 (0.5762,0.6044)

pD

3.99

3.93

3.87

-14.72

DIC

53.84

53.74

53.63

35.09

π = 0.01

0.0099 (0.0089, 0.0110)

0.0099 (0.0089, 0.0110)

0.0099 (0.0090, 0.0111)

0.0133 (0.0093, 0.0377)

Se1 = 0.90

0.8996 (0.8526, 0.9479)

0.9019 (0.8545, 0.9479)

0.8940 (0.8528, 0.9678)

0.5017 (0.0321, 0.9563)

Se2 = 0.95

0.9445 (0.9020, 0.9876)

0.9429 (0.9031, 0.9875)

0.9341 (0.9015, 0.9759)

0.5130 (0.0720, 0.9736)

Sp1 = 0.95

0.9500 (0.9472, 0.9528)

0.9499 (0.9469, 0.9528)

0.9499 (0.9469, 0.9527)

0.9480 (0.9404, 0.9522)

Sp2 = 0.90

0.8998 (0.8969, 0.9027)

0.8998 (0.8969, 0.9028)

0.8997 (0.8967, 0.9027)

0.8975 (0.8906, 0.9018)

pD

3.82

4.01

4.17

-2.41

DIC

65.59

65.95

66.28

59.47