Volume 8, Number 5—May 2002
Research
Sentinel Surveillance: A Reliable Way To Track Antibiotic Resistance in Communities?
Table 2
Number of isolates required to estimate accurately %PNSP in a given area and percentage of sentinel laboratory groups that met sample size requirements
Area | Actual %PNSP (target range) | No. of isolates needed to estimate %PNSPa | % of sentinel groups of 5 laboratories with > no. of required isolates |
---|---|---|---|
CA | 15 (10-20) | 94 | 100 |
CT | 18 (13-23) | 172 | 3 |
GA | 33 (28-38) | 243 | 40 |
MD | 22 (17-27) | 183 | 12 |
MN | 20 (15-25) | 163 | 70 |
NY | 15 (10-20) | 97 | 100 |
OR | 21 (16-26) | 120 | 100 |
TN | 35 (30-40) | 191 | 0 |
a No. of isolates, n, required to estimate the area’s actual %PNSP (P) within 5 percentage points (d=0.05) with 95% confidence (Z=1.96) is: n= (Z2 P(1-P))/d2, where d is the range of accepted variation around the actual %PNSP, and Z is the Z-score range within which values must fall. Because the total no. of isolates per area, N, was small, we corrected this estimate for finite population size: n=n/[1+(n-1)/N]. There is no power associated with this estimate (14).
%PNSP, percent of penicillin-nonsusceptible pneumocooccal isolates.