Design-related bias in estimates of accuracy when comparing imaging tests: examples from breast imaging research
- 136 Downloads
This work highlights concepts on the potential for design-related factors to bias estimates of test accuracy in comparative imaging research. We chose two design factors, selection of eligible subjects and the reference standard, to examine the effect of design limitations on estimates of accuracy. Estimates of sensitivity in a study of the comparative accuracy of mammography and ultrasound differed according to how subjects were selected. Comparison of a new imaging test with an existing test should distinguish whether the new test is to be used as a replacement for, or as an adjunct to, the conventional test, to guide the method for subject selection. Quality of the reference standard, examined in a meta-analysis of preoperative breast MRI, varied across studies and was associated with estimates of incremental accuracy. Potential solutions to deal with the reference standard are outlined where an ideal reference standard may not be available in all subjects. These examples of breast imaging research demonstrate that design-related bias, when comparing a new imaging test with a conventional imaging test, may bias accuracy in a direction that favours the new test by overestimating the accuracy of the new test or by underestimating that of the conventional test.
KeywordsAccuracy Breast imaging Comparative design Selection bias Reference standard
- 8.Reitsma JB, Rutjes AWS, Whiting P et al (2009) Assessing methodological quality. In: Deeks JJ, Bossuyt PM, Gatsonis C (Eds.), Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy Version 1.0.0. The Cochrane Collaboration 2009. Available via The Cochrane Collaboration. http://srdta.cochrane.org. Accessed 6 February 2010
- 14.Sardanelli F, Giuseppetti GM, Panizza P et al (2004) Italian Trial for Breast MR in Multifocal/Multicentric Cancer. Sensitivity of MRI versus mammography for detecting foci of multifocal, multicentric breast cancer in fatty and dense breasts using the wholebreast pathologic examination as a gold standard. AJR Am J Roentgenol 183:1149–1157PubMedGoogle Scholar
- 18.Turnbull LW, Brown SR, Olivier C et al (2010) Multicentre randomised controlled trial examining the cost-effectiveness of contrast-enhanced high field magnetic resonance imaging in women with primary breast cancer scheduled for wide local excision (COMICE) Health Technol Assess 14. Available via http://www.hta.ac.uk/1216. Accessed 10 February 2010