Unadjusted and adjusted estimates of sensitivity, specificity, LRP<sub>§</sub>, and LRN<sub>§§</sub> of HRA cytology for HSIL on HRA directed biopsy (n = 261 patients).
收藏资源简介：
*LRP: likelihood ratio positive [sensitivity/(1– specificity)]. †LRN: likelihood ratio negative [(1– sensitivity)/specificity]. 1. Adjusted assuming: (1) sensitivity and specificity of biopsy for HSIL are 0.74 and 0.91, respectively (from Byrom et al.[9]) and (2) conditional independence between cytology and biopsy given true disease status. 2. Adjusted assuming that HRA-directed biopsy has 100% specificity but unknown sensitivity. Then the sensitivity of HRA cytology is estimated without bias as the observed sensitivity (0.66) and the specificity of HRA cytology is estimated by solving equation 19 { } of Staquet et al.[11] for specificity. Here we assumed the observed HSIL prevalence of 24% by biopsy. 3. Adjusted using latent class analysis (LCA) assuming conditional independence but imposing restriction that HRA biopsy result of HSIL is measured without error (no false positives). 3′. Adjusted using latent class analysis (LCA) relaxing conditional independence assumption and imposing restriction that HRA biopsy result of HSIL is measured without error (no false positives).
