Table 1

Sensitivity and specificity of 100 replicates of simulations in detecting high-risk medical officer of Health areas with classical and inhibitory geostatistical design with close pairs (ICP) design

Sample sizeSensitivity
(95% CI)
Specificity
(95% CI)
Classical design
 563 (50%)85.5 (83.4 to 87.6)86.5 (84.2 to 88.7)
 372 (33%)83.2 (80.6 to 85.8)87.8 (85.3 to 90.3)
 282 (25%)80.7 (77.9 to 83.5)85.2 (82.5 to 88.0)
 225 (20%)80.7 (77.4 to 84.1)85.4 (83.0 to 87.7)
 169 (15%)78.9 (74.8 to 83.1)80.7 (77.8 to 83.6)
 113 (10%)73.5 (68.6 to 78.5)79.5 (75.8 to 83.1)
ICP design
 563 (50%)86.5 (84.8 to 88.3)81.3 (78.8 to 83.7)
 372 (33%)86.8 (84.9 to 88.6)80.9 (77.9 to 84.0)
 282 (25%)87.0 (85.1 to 88.9)80.9 (81.3 to 84.4)
 225 (20%)84.7 (82.1 to 87.2)81.7 (78.3 to 85.0)
 169 (15%)83.2 (80.3 to 86.1)77.6 (74.1 to 81.1)
 113 (10%)83.6 (80.9 to 86.3)75.0 (71.2 to 78.8)