Assessment of radiologic tests: control of bias and other design considerations.

摘要: The assessment of new radiologic tests can be seriously hampered by the presence systematic bias. Biases arise from incomplete verification sample population; omission uninterpretable tests; absence a definitive reference test; extraneous factors affecting interpretation; and extrapolation including variations in test efficacy among patients, hospitals, radiologists who interpret tests. authors review these biases that affect results studies provide guidelines to avoid problems.