Page 62 - Screening for Cervical Cancer: Systematic Evidence Review

Chapter III. Results

versus 61.0%; specificity of 96.7% versus 95.1%; negative predictive value of 98.4% versus

98.6%, positive likelihood ratios of 19.4 versus 12.5, and negative likelihood ratios of 0.43 for

both tests. In some studies, Hybrid Capture I delivers test performance comparable to SHARP

PCR and Hybrid Capture II. 61,65 However, in the Cuzick et al. analysis (which allows direct

comparison in the same cohort), the newer HPV testing modalities appear to be superior.

All of the studies that consider HPV and LSIL detection evaluated Hybrid Capture II.

We calculated summary performance measures using the aggregate numbers of true positives,

false negatives, true negatives, and false positives in these studies with a combined total

enrollment of 5,674. In aggregate, the sensitivity of Hybrid Capture II for detecting LSIL or

more severe lesion, was 67.3%; specificity, 80.1%, positive predictive value, 36.7%; and

63
negative predictive value, 93.4%. Eliminating the HIV-positive women from the estimates, the

aggregate positive predictive value falls slightly to 35.7%; negative predictive value is

essentially unchanged at 93.6%; sensitivity is 65.2%; and specificity is 81.2%. Lastly,

eliminating all groups with a 10% or greater prevalence of LSIL or higher lesions, the resulting

performance is sensitivity of 73.5%; specificity, 90.5%; positive predictive value, 29.2%; and

negative predictive value, 98.5%.

Triage Use of HPV Testing

A proposed role for HPV testing is to serve as a diagnostic tool to facilitate triage of

women with an abnormal Pap tests. To estimate its test characteristics in this role included 7

articles relevant to triage uses of HPV (Evidence Table 3B, Appendix C) that met the following

criteria for inclusion:

62