Skip to main content

Advertisement

Table 4 Output from Bayesian LCA to estimate the sensitivity and specificity of diagnostic tests. The mean estimate for each parameter is shown, with 95% Bayesian Credible Intervals (BCI) shown in parentheses. Analysis was performed separately for CCA trace as negative (left) and CCA trace as positive (right). The covariance terms inputted into each model were selected by comparing the DIC’s of models with differing covariance terms added (see Additional file 1: Table S1). The PPV of CCA performed in Burundi was calculated using the equations described in the methods

From: Latent class analysis to evaluate performance of point-of-care CCA for low-intensity Schistosoma mansoni infections in Burundi

  CCA trace negative (%) (95% BCI) CCA trace positive (%) (95%BCI)
Sensitivity
 Kato-Katz in Burundi 15.9 (9.2–23.5) 15.0 (9.6–21.4)
 CCA in Burundi 61.1 (49.9–71.9) 91.5 (85.8–96.0)
 CCA in Leiden 79.5 (67.7–89.4) 72.0 (62.5–80.5)
 CAA in Leiden 90.3 (84.5–95.0) 91.8 (85.0–96.9)
 Covariance CCAL & CAA 0.6 (0.0–1.5)  
 Covariance CCAB & CCAL   0.3 (0.0–0.7)
Specificity
 Kato-Katz in Burundi 97.1 (94.5–99.1) 97.5 (95.2–99.3)
 CCA in Burundi 98.7 (96.6–99.9) 72.3 (65.6–78.7)
 CCA in Leiden 97.3 (94.7–99.2) 96.8 (93.9–98.8)
 CAA in Leiden 74.6 (68.3–81.2) 85.3 (79.3–91.1)
 Covariance CCAL & CAA 0.4 (0.0–1.1)  
 Covariance CCAB & CCAL   0.3 (0.0–0.8)
PPV of CCA in Burundi (95% BCI)
 Trace as negative 95.8 (89.4–99.6)
 Trace as positive 69.4 (61.7–7.1)
 Trace results only 52.2 (37.8–5.8)