For the past 40 years, the use of Papanicolaou smears to screen women has substantially reduced the morbidity and mortality of cervical cancer. However, there remain significant concerns on the part of many women and physicians about the high false-negative rate associated with the sampling and reading of Pap smears. Even under ideal conditions, false-negative rates as high as 29 percent have been reported. Cervicography is a relatively new diagnostic technique in which a photograph of the cervix is taken after the application of acetic acid. The “cervigram” is examined carefully for evidence of dysplastic changes. If such changes are found, the patient is referred for colposcopy and directed biopsies. It has been hoped that cervicography could potentially spare many women colposcopy, which is more invasive, time-intensive and expensive to perform. Nuovo and associates performed a systematic review of the literature that looked at cervicography used as a diagnostic test in detecting cervical cancer or its precursors.
The authors conducted a MEDLINE search of published studies in English from 1966 through 1996 to retrieve articles that reported data on the use of cervicography for cervical cancer screening. A manual search of Index Medicus was also performed. The first author of all selected publications was contacted to obtain information on any further relevant studies not retrieved by the literature searches. Studies were selected only if they included cervicography as well as the standard reference test—colposcopy, with or without directed biopsy. The studies were evaluated using a quality assessment scale containing nine accepted methodologic standards for the evaluation of diagnostic tests. From each study, the percentage of unsatisfactory or technically defective cervigrams and the results of colposcopy (i.e., normal, any dysplasia, low-grade dysplasia, high-grade dysplasia, cancer) were extracted. In addition, sensitivity, specificity, positive and negative predictive values, disease prevalence and likelihood ratios were calculated. A true-positive cervigram was defined as one with histologic findings on colposcopy of any dysplasia or only high-grade dysplasia.
A total of 23 published studies were initially retrieved. However, after applying the criterion listed above (colposcopy performed on all participants), only seven studies were considered acceptable for inclusion. With regard to the quality assessment score, none of the studies received a score greater than three (the maximum possible was seven). Cervicography was found to have a high sensitivity for detecting high-grade dysplasia (89 to 100 percent). It also had a high false-positive rate, which ranged from 8.2 to 61 percent for any dysplasia and from 9.8 to 63.4 percent for high-grade dysplasia. The positive predictive value for detection of any dysplasia ranged from 22 to 45.5 percent, with a median of 41 percent. In detecting high-grade dysplasia, the positive predictive value ranged from 5.4 to 31.5 percent, with a median of only 10.9 percent.
The authors conclude from their systematic review that current available research on cervicography is generally poor. The majority of published studies to date did not use the reference standard of colposcopy. Moreover, none of the seven studies reviewed adequately met the nine methodologic standards for the evaluation of a diagnostic test. Additional, well-designed studies are needed to further evaluate cervicography before it can be recommended as either a screening test or a secondary triage test.