| Objectives. (1) To identify the optimal OHIP code for identifying women infected with HCV using ROC plot, Kappa statistic, and logistic regression. (2) To test combinations requiring more than one submission of a diagnostic code. (3) To test the combination of different codes using serial and parallel methods.; Study design. Cross-sectional comparative. Study subjects. 2372 women. Primary outcome measure. ROC plot. Method. OHIP code combinations (experimental tests) of hepatitis C (reference standard). The ability of codes to identify infected women was evaluated using ROC plot, Kappa, and logistic regression. Results. “Viral hepatitis” was the optimal code on ROC plot and predicted HCV infection in logistic regression. “Drug dependence/addiction” had the highest Kappa. Multiple code submissions and parallel codes improved specificity but lost sensitivity. Serial codes increased specificity but lost sensitivity. Conclusion. “Viral hepatitis” best identified women infected with HCV. |
