Summary receiver operating characteristic curve analysis techniques in the evaluation of diagnostic tests

Ann Thorac Surg. 2005 Jan;79(1):16-20. doi: 10.1016/j.athoracsur.2004.09.040.

Abstract

The number of studies in the literature using summary receiver operating characteristic (SROC) analysis of diagnostic accuracy is rising. The SROC is useful in many such meta-analyses, but is often poorly understood by clinicians, and its use can be inappropriate. The academic literature on this topic is not always easy to comprehend. Interpretation is therefore difficult. This report aims to explain the concept of SROC analysis, its advantages, disadvantages, indications, and interpretation for the cardiothoracic surgeon. We use a practical approach to show how SROC analysis can be applied to meta-analysis of diagnostic accuracy by using a contrived dataset of studies on virtual bronchoscopy in the diagnosis of airway lesions.

Publication types

  • Comment

MeSH terms

  • Airway Obstruction / diagnosis
  • Area Under Curve
  • Bronchoscopy / statistics & numerical data
  • Carcinoma, Non-Small-Cell Lung / diagnostic imaging
  • Carcinoma, Non-Small-Cell Lung / secondary
  • Diagnostic Tests, Routine / statistics & numerical data*
  • False Negative Reactions
  • False Positive Reactions
  • Humans
  • Lung Diseases / diagnosis
  • Lung Neoplasms / pathology
  • Lymphatic Metastasis / diagnostic imaging
  • Meta-Analysis as Topic
  • Odds Ratio
  • Positron-Emission Tomography
  • Predictive Value of Tests
  • ROC Curve*
  • Sensitivity and Specificity
  • Tomography, X-Ray Computed
  • User-Computer Interface