Inflation of sensitivity of cervical cancer screening tests secondary to correlated error in colposcopy.

OBJECTIVE

To determine whether the sensitivity of screening tests that miss small cervical intraepithelial neoplasia (CIN) 2 or worse (e.g., acetic acid-aided visual inspection) is inflated when the criterion standard (colposcopic-directed biopsy) misses the same small CIN 2 or worse.

MATERIALS AND METHODS

One thousand nine hundred twenty-eight women were screened using acetic acid-aided visual inspection, self-tests, and direct tests for high-risk human papillomavirus, and using liquid-based cytologic screening. All women underwent colposcopy with biopsy. If a cervical quadrant had no lesion, a random biopsy at the squamocolumnar junction within that quadrant was obtained. All patients underwent endocervical curettage (ECC). Differences in sensitivity for CIN 2 or worse of screening tests were compared with criterion standards of colposcopically directed biopsy and colposcopically directed biopsy, random biopsy, plus ECC.

RESULTS

Sixty-two of 83 women with CIN 2 or worse were diagnosed by colposcopically directed biopsy, 19 by random biopsy, and 2 solely by ECC. Fifty-six of the 83 women had CIN 2 or worse involving 0 to 2 quadrants (2 with 0 quadrants were diagnosed solely by positive ECC) and 27 of 83 had CIN 2 or worse results involving 3 to 4 quadrants. Colposcopically directed biopsy detected 35 of 56 women (62.5%) with CIN 2 or worse involving 0 to 2 cervical quadrants and 27 of 27 women (100%) with CIN 2 or worse involving 3 to 4 quadrants (p < .005). Acetic acid-aided visual inspection and cytologic analysis of high-grade squamous intraepithelial lesion or cancer detected 35 of 56 (62.5%) and 36 of 56 (64.3%) lesions involving 0 to 2 quadrants and 27 of 27 (100%) (p < .005) and 24 of 27 (88.9%; p < .05) lesions involving 3 to 4 quadrants. The sensitivity of direct human papillomavirus tests (97.6%) and cytologic analysis of atypical squamous cells of undetermined significance or worse (94.0%) for CIN 2 or worse were high and unaffected by lesion size. When the criterion standard was changed from colposcopically directed biopsy, random biopsy, plus ECC to colposcopically directed biopsy, the sensitivity of acetic acid-aided visual inspection and cytologic analysis of high-grade squamous intraepithelial lesion or worse for CIN 2 or worse increased from 75.9% to 85.5% (p > .1) and from 71.1% to 79.0% (p > .25). Sensitivities of other screening tests were unaffected by changing the criterion standard.

CONCLUSIONS

Criterion standards that miss the same CIN 2 or worse as the screening tests likely cause inflation of the sensitivity of those tests.