Appendix CSelection Criteria for Studies Used to Estimate Sensitivity and Specificity of Cytology and HPV DNA Testing

Publication Details

The inclusion criteria for selecting studies identified by the Oregon EPC35 and used in the calculation of sensitivity and specificity are described below. The final estimates were chosen based on discussions with the AHRQ Medical Officer and Program Officer and reviewed by the Oregon EPC to confirm that they reflected the evidence report.

In the absence of the availability of summary estimates from a meta-analysis, we used the following criteria to identify estimates for use in the model.

  1. Was the study conducted in a population similar in risk to the U.S. general population? To determine applicability to the United States, we used the Oregon EPC rating of studies as poor, fair, or good applicability. Studies rated as poor, because they were conducted in countries such as India or South Africa, were eliminated from consideration for use in the model, since estimates of test performance from these populations may not reflect performance in a low risk population, taking into account disease prevalence and familiarity with using the test.
  2. Was the study graded as good quality by the Oregon EPC? We chose studies for the base-case analysis that were graded as good quality and provided estimates of absolute sensitivity and specificity from among the studies that were identified as having fair to good applicability to a U.S. population.
  3. Cytology and HPV DNA test with HC2 test performance characteristics. Since there were only a few studies that fit these criteria (Kulasingam et al and Coste et al), we used a weighted average of the two to determine the base estimates of sensitivity and specificity for cytology.14,40 However, since two additional studies showed a consistent difference between cytology and HPV DNA test accuracy (Mayrand et al and Bigras et al), we used a weighted average of these two studies to determine the incremental gain in sensitivity and decrease in specificity compared to cytology only.8,37 These estimates are presented in Table 2 in the body of the report. Since the study by Mayrand et al had the largest difference in test accuracy performance for HPV and cytology, we used these estimates as well as the Koliopoulos et al estimates (detailed below) for the HPV analyses.8,36 In total, three sets of test accuracy estimates were used for the HPV analyses. We used ranges from these studies, as well as those rated as fair quality to estimate the ranges of test sensitivity and specificity for the sensitivity analyses for Specific Aim 1 (age at which to begin screening) and Sub-Aim 1 (age at which to end screening).
    1. Estimates from Koliopoulos et al were also used in the HPV analyses, since these were from a meta-analysis of a wide variety of studies comparing HPV and cytology.36
    2. Base estimates and ranges for HC2 and cytology test performance among women with ASCUS were based on studies identified by the Oregon EPC. The suggested estimates for use in the base case were those from a single study (Manos et al) that compared both technologies that were rated as good applicability and graded as a good quality study.44 The ranges for sensitivity and specificity are from other studies that provided relevant estimates for at least one of the technologies (cytology or HC2).