Unaided visual inspection or "downstaging" has been suggested as a potential alternative method for cervical cancer screening in developing countries. Our study was designed to evaluate the accuracy of downstaging to detect cervical neoplasia in a low-resource setting. A total of 6,399 women aged 30-64 years were screened with downstaging by trained nonmedical health workers. Two thresholds were used to define positive downstaging: "low threshold" when any visible abnormality on the cervix was considered positive and "high threshold" when selected abnormalities such as bleeding on touch, bleeding erosion, hypertrophied oedematous cervix, congested stippled cervix and growth or ulcer constituted the positive test. All women underwent a colposcopy examination. Biopsies were directed when colposcopy revealed abnormal lesions. True disease status was defined as histologically proven moderate dysplasia and worse lesions. Since all the participants received a diagnostic (reference) investigation (biopsy and/or colposcopy), sensitivity, specificity and predictive values were estimated directly. Low- and high-threshold downstaging were positive in 1,585 (24.8%) and 460 (7.2%) women, respectively. The sensitivities of low- and high-threshold downstaging to detect high-grade precursors and invasive cancers were 48.9% and 31.9%, respectively. The specificities were 75.8% and 93.3%, respectively. These results indicate that downstaging is not suitable as an independent primary screening modality for cervical neoplasia.
Copyright 2002 Wiley-Liss, Inc.