Afina S. Glas

The diagnostic odds ratio: a single indicator of test performance

Diagnostic testing can be used to discriminate subjects with a target disorder from subjects without it. Several indicators of diagnostic performance have been proposed, such as sensitivity and specificity. Using paired indicators can be a disadvantage in comparing the performance of competing tests, especially if one test does not outperform the other on both indicators. Here we propose the use of the odds ratio as a single indicator of diagnostic performance. The diagnostic odds ratio is closely linked to existing indicators, it facilitates formal meta-analysis of studies on diagnostic test performance, and it is derived from logistic models, which allow for the inclusion of additional variables to correct for heterogeneity. A disadvantage is the impossibility of weighing the true positive and false positive rate separately. In this article the application of the diagnostic odds ratio in test evaluation is illustrated.

Computed tomography and magnetic resonance imaging in staging of uterine cervical carcinoma: a systematic review

OBJECTIVE The goal of this article is to systematically review the available evidence on the diagnostic performance of computed tomography (CT) and magnetic resonance imaging (MRI) in staging of cervical carcinoma. METHODS A comprehensive computer literature search was performed in MEDLINE and EMBASE databases from January 1985 to May 2002. Two reviewers independently scored methodological quality of included studies and extracted relevant data for data analysis. A bivariate random effect approach was used to summarize estimates of sensitivity and specificity values. Covariates were added to this model to study the influence of sample size, publication year, methodological criteria, and MRI techniques on summary estimates. RESULTS Fifty-seven articles were included. In 49 articles one imaging modality was evaluated (MRI, 38; CT, 11), and in 8 articles, both. Inclusion criteria were: minimum of 10 patients included, histopathology as reference standard, sufficient data presented to construct 2(x) 2 tables. The exclusion criterion was: data reported elsewhere in more detail. Sensitivity estimates for parametrial invasion were 74% (95% C: 68-79%) for MRI and 55% (95% CI: 44-66%) for CT, and for lymph node involvement, 60% (95% CI 52%-68%) and 43% (95% CI: 37-57%), respectively. MRI and CT had comparable specificities for parametrial invasion and lymph node involvement. For bladder invasion and rectum invasion the sensitivities for MRI were respectively 75% (95% CI: 66-83%) and 71% (95% CI: 53-83%), higher compared with CT. The specificity in evaluating bladder invasion for MRI was significantly higher compared with CT: 91% (95% CI: 83-95%) for MRI and 73% (95% CI: 52-87%) for CT. The specificities for rectum invasion were comparable. Differences in patient sample size, publication year, methodological criteria, and MRI techniques had no effect on the summary estimates. CONCLUSIONS For overall staging of cervical carcinoma, MRI is more accurate than CT.

Statistical models for quantifying diagnostic accuracy with multiple lesions per patient

We propose random-effects models to summarize and quantify the accuracy of the diagnosis of multiple lesions on a single image without assuming independence between lesions. The number of false-positive lesions was assumed to be distributed as a Poisson mixture, and the proportion of true-positive lesions was assumed to be distributed as a binomial mixture. We considered univariate and bivariate, both parametric and nonparametric mixture models. We applied our tools to simulated data and data of a study assessing diagnostic accuracy of virtual colonography with computed tomography in 200 patients suspected of having one or more polyps.