OBJECTIVE: To compare test performance characteristics of conventional Pap tests and liquid-based cervical cytology samples.

DATA SOURCES: Eligible studies, published between 1991 and 2007, were retrieved through PubMed/EmBase searching and completed by consultation of other sources.

METHODS OF STUDY SELECTION: Studies were selected if a conventional and a liquid-based sample were prepared from the same woman or when one or the other type of sample was taken from a separate but similar cohort. The current systematic review and meta-analysis is restricted to studies where all subjects were submitted to gold standard verification, based on colposcopy and histology of colposcopy-targeted biopsies, allowing computation of absolute and relative test validity for cervical intraepithelial neoplasia grade 2 or worse. Randomized trials were selected as well if all test-positive cases were verified with the same gold standard, allowing computation of the relative sensitivity. Impact of study characteristics on accuracy was assessed by subgroup meta-analyses, meta-regression, and summary receiver operating characteristic curve regression.

TABULATION, INTEGRATION, AND RESULTS: The relative sensitivity, pooled from eight studies, with complete gold standard verification and from one randomized clinical trial, did not differ significantly from unity. Also, the specificity, considering high-grade and low-grade squamous intraepithelial lesions as cutoff, was similar in conventional and liquid cytology. However, a lower pooled specificity was found for liquid-based cytology when presence of atypical squamous cells of undetermined significance was the cutoff (ratio 0.91, 95% confidence interval 0.84-0.98). Differences in study characteristics did not explain interstudy heterogeneity.

CONCLUSION: Liquid-based cervical cytology is neither more sensitive nor more specific for detection of high-grade cervical intraepithelial neoplasia compared with the conventional Pap test.