Skip to main content

Table 3 Sensitivity analysis of kernel and SaTScan analysis of PCR and serology prevalence for prediction of infection in the second year

From: Hot spot or not: a comparison of spatial statistical methods to predict prospective malaria infections

  KERNEL SaTScan
  Radius Window population size
Exposure category <1 m (Household) 100 m 500 m 1,000 m 10% 20% 50%
PCR prevalence quartiles 0.612 0.622 0.611 0.628 0.593 0.616 0.620
Area under ROC        
Proportion of total nPCR positive in the highest quartile in second year ΅ 30.1% 30.2% 33.1% 34.2% 27.1% 29.7% 29.7%
Proportion of the total study population included highest quartile 23.4% 22.7% 25.2% 24.8% 20.3% 22.4% 22.4%
AMA-1 prevalence quartiles 0.583 0.587 0.619 0.618 0.602 0.615 0.647
Area under ROC        
Proportion of total nPCR positive in the highest quartile in second year ΅ 26.6% 29.0% 31.8% 31.9 9.7% 28.9% 48.3%
Proportion of the total study population included highest quartile 22.6% 24.9% 24.8 24.9 6.72% 22.6% 38.0%
MSP-1 19 prevalence quartiles 0.559 0.533 0.602 0.622 0.595 0.612 0.591
Area under ROC        
Proportion of total nPCR positive in the highest quartile I second year 22.7% 22.8% 19.6% 17.1% 9.4% 11.8% 12.0%
Proportion of the total study population included in the highest quartile 24.9% 24.6% 24.5% 24.5% 13.7% 17.7% 17.8%
MSP-1 19 &/or AMA-1 prevalence quartiles 0.575 0.580 0.585 0.604 - - -
Area under ROC        
Proportion of total nPCR positive in the highest quartile in second year 28.7% 30.8% 31.1% 32.6%    
Proportion of the total study population included highest quartile 24.6% 24.7% 24.9% 24.5%    
  1. ΅ Proportion of total nPCR positives in the second year that are found in the highest quartile.