Is cervicography a useful diagnostic test? A systematic overview of the literature.

BACKGROUND The appropriate approach to women with mild dyskaryotic changes on Papanicolaou smear is subject to controversy. Our aim was to assess the usefulness of cervicography as a diagnostic test in detecting cervical cancer or its precursors. METHODS We undertook an extensive literature search looking for pertinent studies of cervicography published between 1966 and 1996. Eligible studies included those in which the reference standard (colposcopy) was done on all patients. The following information was calculated: sensitivity, specificity, positive predictive value, negative predictive value, disease prevalence, and likelihood ratios. RESULTS Cervicography has a high false-positive rate. This rate ranged from 8.2 to 61.0 percent (median 42.1 percent) for any dysplasia and 9.8 to 63.4 percent (median 50.6 percent) for high-grade cervical lesions. Likelihood ratios for a positive test result ranged from 1.0 to 10.6. Likelihood ratios for a negative result ranged from 0.02 to 1.0. CONCLUSIONS The usefulness of cervicography is heavily dependent on the approach used to evaluate abnormal findings on a Papanicolaou smear. If a provider typically offers colposcopy to all patients with low-grade cytologic findings on a Papanicolaou smear, cervicography will decrease colposcopy use and allow for detection of cases of high-grade dysplasia missed by the index Papanicolaou smear. If a provider typically uses watchful waiting with repeat Papanicolaou smears for all patients who have low-grade cytologic findings, cervicography will substantially increase the use of colposcopy. Many of these colposcopies will be done as a result of false-positive cervigrams.