Comparison of conventional Papanicolaou smears and a fluid-based, thin-layer system for the detection of abnormal...

Objective: The primary objective of this project was to determine the diagnostic accuracy of a fluid-based, thin-layer system (ThinPrep) compared to conventional Papanicolaou smears for the detection of abnormal (cancerous and pre-cancerous) cervical lesions.Methods: Five comparative studies including 12 study sites, screening centers, and high-risk hospital centers and 10,377 patients were included in a systematic review and meta-analysis examining the cytological diagnoses of ThinPrep with conventional smears using odds ratios (ORs) and risk differences (RDs). All abnormal samples (squamous intraepithelial neoplasia [SIL] or more severe abnormalities) classified by an independent pathologist reviewer were analyzed. Sensitivity analyses using jackknife techniques examined the robustness of the results. Logistic regression models were fit to the data using study level information and the individual patient data to examine the effect of baseline covariates on the results.Results: ThinPrep results were significantly better than conventional smears. For all abnormal diagnoses, the OR = 1.95 (95% CI: 1.40-2.70, P <.0001). For each SIL diagnostic subgroup, the results also strongly favored ThinPrep: LGSIL (OR = 2.20, 95% CI: 1.50-3.22, P =.001) and HGSIL (OR = 1.33, 95% CI: 0.85-2.07, P =.091). Analyses with RDs were similar and also statistically significant. Sensitivity analyses showed the studies to be homogeneous (using Cochran's Q statistic = NS). Logistic regression models examining the meta-analysis results controlling for individual patient level and study level covariates increased the significance of the ThinPrep results.Conclusion: In a meta-analysis of the results of five comparative studies, the ThinPrep system showed significantly higher diagnostic accuracy for the detection of abnormal cervical lesions overall and by SIL diagnostic subgroups.