Reference Standard: Two of the study standards considered the quality of the reference (“gold”) standard used to identify the presence of disease. The first standard required that the article included a clear definition of the reference standard. Canadian pharmacy mall A “high probability” lung scan as a reference standard to establish the presence of pulmonary emboli would be considered clearly defined; a reference standard described as a “positive” lung scan would not fulfill this standard. Subjective, nonquantitative reference standards (eg, “clinical evidence” of heart failure) needed to include explicit statements regarding factors that defined the presence of disease (eg, “crackles on chest auscultation with an S3 that cleared with diuretic therapy in the absence of an alternative diagnosis”).
The second standard related to the reference standard required that the reference standard was the best available method for assessing the presence or absence of disease considering the relative accuracy and feasibility of alternative methods. For the detection of pulmonary emboli, a lung scan alone would not be accepted as the best available reference standard, but a pulmonary angiogram or a lung scan with a long-term patient follow-up would fulfill this standard.
Spectrum Composition: Diagnostic test studies need to describe the case mix of the study population adequately to allow conclusions regarding the diagnostic accuracy of the tests and their generalizability to other patient populations. This study standard was considered fulfilled if the article provided three of the four following descriptors: (1) age distribution; (2) sex distribution; (3) summary of presenting clinical symptoms and/or disease stage in terms of early or late disease or mild, moderate, or severe disease; and (4) eligibility criteria for study subjects.
Indicated Study Sample: Diagnostic tests demonstrate higher diagnostic accuracy when examined in a “convenience” sample compared with an “indicated” sample. A convenience sample is a group comprised of healthy control subjects and patients with the target condition. An indicated sample includes patients who present with suggestive manifestations of the target disorder but may or may not have the condition. An indicated sample simulates clinical practice wherein a test is aplied to symptomatic or otherwise at-risk patients who have clinical manifestations of the target disorder. This standard required that the evaluated patients appeared to have a realistic likelihood that they had the disorder undergoing diagnostic evaluation.