RESUMEN
A nonparametric method proposed by DeLong et al in 1988 for comparing areas under correlated receiver operating characteristic curves is used widely in practice. However, the DeLong method as implemented in popular software quietly deletes individuals with any missing values, yielding potentially invalid and/or inefficient results. We simplify the DeLong algorithm using ranks and extend it to accommodate missing data by using a mixed model approach for multivariate data. Simulation results demonstrate the validity and efficiency of our procedure for data missing at random. We illustrate our proposed procedure in SAS, Stata, and R using the original DeLong data.
Asunto(s)
Algoritmos , Área Bajo la Curva , Simulación por Computador , Curva ROC , Humanos , Modelos Estadísticos , Estadísticas no Paramétricas , Interpretación Estadística de Datos , Análisis MultivarianteRESUMEN
BACKGROUND: Crohn's disease (CD) is characterized by discontinuous inflammation. Failure to identify skipping lesions of the terminal ileum (TI) or transmural changes can lead to incorrect management. METHODS: Eligible adult patients with CD undergoing ileo-colonoscopy and computed tomography enterography or magnetic resonance enterography within 6 months. We determined the prevalence of endoscopic skipping (normal ileum on colonoscopy but proximal small bowel inflammation on cross-sectional imaging), skip lesions (discontinuous inflammation along the gastrointestinal tract identified on cross-sectional imaging), structuring, and penetrating complications. RESULTS: Among 202 patients, 45 (22.3%) had endoscopic skipping proximal to TI intubation. Fifty patients (24.5%) had small bowel skip lesions, primarily in the ileum. Strictures were identified in 34 patients (16.8%) through both imaging and ileo-colonoscopy, in 21 patients (10.4%) solely through cross-sectional imaging, and in 3 patients (1.5%) solely through ileo-colonoscopy. Approximately 36.2% of stricturing cases would be missed without cross-sectional imaging. Penetrating complications, including abscesses (2.5%) and various fistula types (4.9%), were detected in 15 (7.4%) patients. CONCLUSIONS: Ileo-colonoscopy missed detection of active CD in approximately one-fifth of cases due to more proximal disease location. Stricturing disease might be missed in more than a third of cases if cross-sectional imaging is not performed.
Crohn's disease often presents with discontinuous inflammation, with endoscopic evaluation sometimes missing proximal lesions. This study highlights the importance of cross-sectional imaging, which identifies skip lesions and strictures missed by endoscopy, crucial for accurate management.