Table 2. Predictor associations with an individual child’s probability of being infected with Schistosoma mansoni, estimated using three different diagnostic methods to account for possible discrepancies in diagnostic performance and estimates of geographical endemicity.
For each diagnostic method, a geostatistical model was built to estimate predictor effects while accounting for spatial autocorrelation. Presented are effect means and 95% credible intervals (in brackets). KK; Kato-Katz. CCA; Circulating Cathodic Antigen.
| Variable | KK | CCA with trace as positive | CCA with trace as negative |
|---|---|---|---|
| Age 10–14 y (versus 5–9 y) | 1.47 (-4.75, 10.15) | 0.01 (-0.55, 0.57) | 1.11 (-0.20, 2.73) |
| Age 15–18 y (versus 5–9 y) | 0.91 (-5.35, 9.56) | -0.16 (-0.76, 0.46) | 1.08 (-0.29, 2.72) |
| Male (versus female) | 0.54 (0.21, 0.88) | 0.08 (0.02, 0.15) | 0.30 (0.18, 0.43) |
| Land surface temperaturea | 0.91 (0.14, 1.73) | 0.08 (-0.03, 0.20) | 0.12 (-0.06, 0.30) |
| Normalized difference vegetation indexa | 0.93 (-0.08, 2.00) | 0.23 (0.08, 0.38) | 0.28 (0.03, 0.54) |
| Normalized difference water indexa | -0.36 (-1.59, 0.80) | -0.10 (-0.29, 0.08) | -0.25 (-0.56, 0.06) |
| ϕ (phi: rate of decay of spatial correlation)b | 11.90 (4.24, 19.34) | 17.36 (13.10, 19.88) | 14.81 (9.06, 19.54) |
| σ2 (sigma: variance of spatial random effect)b | 11.74 (5.93, 23.15) | 1.17 (0.95, 1.46) | 2.65 (2.00, 3.62) |
| Tau (precision) | 0.10 (0.04, 0.17) | 0.86 (0.69, 1.05) | 0.39 (0.28, 0.50) |
| Intercept | -9.57 (-18.32, -2.99) | -0.80 (-1.42, -0.19) | -4.87 (-6.54, -3.48) |
a Variables were standardized to have mean = 0 and standard deviation = 1
b Measured in decimal degrees and 3/phi determines the cluster size; one decimal degree is approximately 111km at the Equator (the size of the radii of the clusters).