Table 2

Single and selecteda two-test strategies for detection of ≥CIN 3

StrategybSensitivitycSpecificitydYouden’s indexe95% CI
HPV (+)85.3%88.2%0.740.67–0.80
Liquid-based (≥ASCUS)85.7%87.8%0.740.67–0.80
Smear (≥ASCUS)63.0%93.7%0.570.48–0.66
Cervigram (≥A)61.7%84.8%0.460.37–0.56
Smear (≥HSIL) or HPV (+)90.7%87.8%0.790.73–0.84
Liquid-based (≥HSIL) or HPV(+)90.5%88.0%0.780.73–0.84
Cervigram (≥P2) or HPV (+)89.7%88.1%0.780.72–0.84
Liquid-based (≥ASCUS) or cervigram (≥P0)93.2%83.9%0.770.72–0.82
Smear (≥HSIL) or liquid based (≥ASCUS)86.5%87.6%0.740.68–0.81
Smear (≥LSIL) or cervigram (≥P0)74.5%90.9%0.650.57–0.74
  • a For each of the six possible two-technique combinations, the table shows the performance for the cut-points with the highest accuracy as measured by Youden’s index.

  • b There were three possible thresholds for conventional and liquid-based cytology (≥ASCUS, ≥LSIL, ≥HSIL), five possible thresholds for cervicography [≥Atypical, ≥Positive(0), ≥Positive(1), ≥Positive(2), ≥Positive(3)], and a single threshold for HPV DNA testing (positive versus negative). Techniques were considered singly and in pairs at all thresholds. Two kinds of combinations were evaluated, either requiring both techniques to be positive or at least one. Overall, there were 112 strategies considered, which were ranked in order of decreasing Youden’s index.

  • c Sensitivity calculated as the percentage of cases of ≥CIN 3 detected by the screening strategy.

  • d Specificity calculated as the percentage of women without CIN 3 or cancer who tested negative by the screening strategy.

  • e Youden’s index calculated as sensitivity plus specificity (expressed as proportions) minus 1.00, with 95% CI.