Table 1 Model parameters
Parameter | Symbol | Values | Source/Notes | |
|---|---|---|---|---|
Village population size | N | Variable | Data | |
Screening coverage (%) | c | Variable | Data | |
Relative probability of attending screening (cases versus non-cases) | ρ | Project | Estimate (95% percentiles) | Prediction of step 1 of model. Random values for each iteration sampled from squared deviance distributions of ρ estimates. |
Kiri | 1.6 (0.7-12.8) | |||
Adjumani | 2.5 (1.2-36.6) | |||
Arua-Yumbe | 1.9 (0.9-4.0) | |||
Probability that the next person screened is S1 or S2 | pS1, pS2 | from 0 to 1 | Updated after each ith person screened. See Equations 4 and 5. | |
Ratio of observed prevalence at coverage c to observed prevalence at coverage = 100%. | βc | Computed for various values of c, and for each MSF project as a whole. | Data and model predictions. See Equation 1 and text. | |
Diagnostic accuracy | Algorithm | Mode (range) | Random values sampled from the likelihood distributions generated by Checchi et al. [13] based on a probabilistic decision model (one random value generated for each iteration). Values for the new Kiri algorithm apply to all screenings conducted since March 2005 (n = 17). | |
Diagnostic sensitivity in stage 1 (%) | σ1 | Kiri (old) | 98.0 (83.1-99.5) | |
Kiri (new) | 57.4 (41.2-78.2) | |||
Adjumani | 97.9 (74.1-99.2) | |||
Arua-Yumbe | 96.5 (74.6-98.8) | |||
Diagnostic sensitivity in stage 2 (%) | σ2 | Kiri (old) | 98.0 (83.5-99.6) | |
Kiri (new) | 67.5 (53.6-84.0) | |||
Adjumani | 97.5 (75.1-99.4) | |||
Arua-Yumbe | 97.7 (75.0-99.3) | |||
Diagnostic specificity (%) | φ | Kiri (old) | 100.0 (99.8- 100.0) | |
Kiri (new) | 100.0 (99.95-100.0) | |||
Adjumani | 100.0 (99.8-100.0) | |||
Arua-Yumbe | 100.0 (99.8-100.0) | |||
Probability of being correctly classified into stage 1 (%) | σ*1 | Kiri (old) | 67.7 (38.5-86.8) | |
Kiri (new) | 66.0 (39.0-87.2) | |||
Adjumani | 70.4 (39.1-88.6) | |||
Arua-Yumbe | 66.1 (39.2-88.5) | |||
Probability of being correctly classified into stage 2 (%) | σ*2 | Kiri (old) | 94.7 (82.1-98.6) | |
Kiri (new) | 95.1 (81.4-98.4) | |||
Adjumani | 94.0 (78.7-98.2) | |||
Arua-Yumbe | 93.1 (78.7-98.2) | |||
Probability that a false positive case will be classified into stage 1 (%) | ω | Kiri (old) | 0.0 | Based on the algorithms used in these projects, false positives can only be classified as stage 2 [13]. |
Kiri (new) | 0.0 | |||
Adjumani | 0.0 | |||
Arua-Yumbe | 0.0 | |||
Binary dummy variables | δ[…] | 0 or 1 | Denote occurrence of event in a given individual. | |