Systematic review and meta-analysis of individual patient data to assess the sensitivity of cervical cytology for diagnosis of cervical cancer in low- And middle-income countries

Abstract

Purpose To assess the sensitivity of cervical cytology to cancer by pooling individual patient cytology results from cancers diagnosed in studies that assessed cervical screening in low- and middle-income countries. Methods Two authors reviewed studies identified through PubMed and Embase databases. We included studies that reported cervical cytology in which at least one woman was diagnosed with cervical cancer and in which abnormal cytology results were investigated at colposcopy and through a histologic sample (if appropriate). When cytology results were not reported in the manuscript, authors were contacted. Stratified analyses and meta-regression were performed to assess sources of heterogeneity between studies. Results We included 717 cancers from 23 studies. The pooled sensitivity of cytology to cancer at a cutoff of a high-grade squamous intraepithelial lesion (HSIL) or worse was 79.4% (95% CI, 67.7% to 86.0%). Results from stratified analyses did not differ significantly, except among studies that recruited symptomatic women or women referred because of abnormal cytology, when the sensitivity of cytology was much higher (95.9%; 95% CI, 86.5% to 99.9%). The cutoff of an HSIL or worse detected 85% of the cancers that would have been detected at a cutoff of atypical squamous cells of undetermined significance or worse (relative sensitivity, 85.2%; 95% CI, 80.7% to 89.7%). Conclusion Cytology at a high cutoff could be an excellent tool for targeted screening of populations at high risk of cervical cancer with a view to diagnose cancer at an earlier stage.