of the test to accurately exclude the diagnosis in those who do not have the disease, or the true negative rate It is the proportion of subjects who have a negative test result (the numerator) among all those who do not have the disease (the denominator) If the subject did not have the disease but had a positive test result, he or she is referred to as a false positive and would have the implications of needing further workup or of being labeled as having something that he or she does not have For tests that give a result that is a continuous measure, one can calculate the sensitivity and specificity using cut points from all values of the variable and plot a receiver-operating characteristic curve This curve plots sensitivity on the y-axis and the proportion of false positives on the x-axis A plot that results in a diagonal line running from the bottom left to the upper right corner represents a test that has no discriminatory ability (i.e., there is one false positive for every true positive measure of the test) Plots that are curves that bend toward the upper left corner show greater discrimination the closer they get to that corner (i.e., the number of true positives greatly exceeds that of the false positives) Alternatively, one can plot the area under this curve as a proportion The closer that the proportion of area under the curve is to 1, the more perfect the discrimination of the test The test value with the sensitivity and proportion of false positives that comes closest to the upper left corner is the value that has the best discriminatory performance and may be used as a cut point Additionally, it is possible to calculate an entity called the likelihood ratio associated with a particular result The likelihood ratio associated with a positive test result would be the ratio of the proportion of subjects with the disease who would have a positive test result, the numerator and its sensitivity, divided by the proportion of subjects without the disease who would have a positive test result, the denominator and the false positives The likelihood ratio associated with a negative test result would be the ratio of the proportion of subjects with the disease who would have a negative test result, the numerator or missed disease, divided by the proportion of subjects without the disease who would have a negative test result, this representing the denominator and specificity Likelihood ratios of 1 represent tests with no ability to differentiate those with versus without the disease Increasingly larger or smaller likelihood ratios indicate increasing performance of the test Sensitivity, specificity, and likelihood ratio are fixed aspects of a test and do not change when the test is applied to different populations with a different prevalence of the disease We are also interested in the performance of the test under different clinical situations when applied to cohorts of subjects, particularly where the prevalence of the disease in the cohort or the probability of disease in the individual before application of the test can be estimated Two performance characteristics are highly influenced by the prevalence of the disease in the population being tested Positive predictive value is the proportion of all subjects who test positive for a given test (the denominator) who actually have the disease (the numerator) The remaining proportion would represent false positives Negative predictive value is the proportion of subjects who have a negative test result (the denominator) who truly do not have the disease (the numerator) The remaining proportion would represent false negatives or missed disease Prevalence of the disease in the population to which a test is applied can highly affect the positive or negative predictive value As prevalence of the disease increases in the population, the positive predictive value increases, and for very rare diseases, the positive predictive value of a test is usually lower The critical appraisal of studies of diagnosis proceeds as follows: Was the study conducted in manner that minimized bias? ■ Were a criterion and the test uniformly both applied to all subjects in an independent and blinded manner? ■ If the design is a cohort study, does the cohort include sufficient subjects with the disease at appropriate stages for which the test might be applied for screening or diagnosis in clinical practice? ■ If the design is a case-control study, are the cases and controls representative of the cohort to whom the test might be applied in clinical practice? ■ Are the methods by which the test was performed described in sufficient detail that they can be replicated in clinical practice? Are the results reported and interpreted appropriately? ■ Are performance characteristics, such as sensitivity and specificity reported, with appropriate confidence limits? ■ Are likelihood ratios reported, or are sufficient data provided that they might be calculated? Are the results of sufficient magnitude and relevant? ■ Are the performance characteristics of the test— such as sensitivity, specificity, and likelihood ratios— sufficient that both missed disease and false-positive diagnoses will be minimized? ■ If the design is a cohort study, what is the magnitude of missed disease and false-positive diagnoses for this cohort as well as the potential implications of such? Are the findings and implications of the report applicable to my own clinical practice? ■ Is it feasible and achievable to incorporate the testing procedure into my own clinical practice? ■ Would the results of the test be applicable to the diagnostic dilemmas and patients in my clinical practice? ■ If applicable, would the results be sufficient or