Fig. 4
Sequential entomological sampling for classification. a Stop lines corresponding to a Wald’s sequence probability ratio test (SPRT)71 sampling plan for classification based on entomological infection thresholds, as measured by simple random sampling of vectors. Results for a scenario with p0 = 0.00005 ( = 95% EPT L3 prevalence threshold) and p1 = 0.0001 ( = 5% EPT threshold), α = 0.05, β = 0.10 are presented. The cumulative total number of sampling units (mosquitoes or black flies) assessed as infected (say using dissections) is denoted on the y axis. n is the cumulative total number of sampling units assessed. Sampling units are observed sequentially and the cumulative total of those assessed as infected is noted. The criteria for cessation of sampling are characterized by the two parallel straight lines (the stop lines) shown. The lower stop line corresponds to the critical infection threshold (e.g., the 95% EPT L3 threshold prevalence), whereas the upper line denotes the emergence threshold (e.g., the 5% EPT L3 breakpoint prevalence value). If the observed data fall below the lower stop line following cumulative sampling bouts, the appropriate decision is to cease sampling and accept the null hypothesis (i.e., p ≤ pCRIT( = 95EPT)). If the observed data fall between the two stop lines, sampling continues. If the data fall above the upper stop line, then the appropriate decision is to accept the alternate hypothesis (interpreted as p ≥ pCRIT). b Average Sampling number (ASN) curve for the sampling plan described above, showing that sample sizes per sampling bout will be maximal (approximately = 350 randomly sampled vectors) close to the critical prevalence adopted as a threshold value (i.e., p = 0.00005 in this case) and much smaller when p is substantially below or above pCRIT
