Lenders and colleagues [1] accurately concluded that free plasma levels of metanephrines are better than plasma levels of catecholamines or urine levels of metanephrines in diagnosing pheochromocytoma, despite some problems with their study design (such as different centers and different assays). The important issue, however, is whether free plasma levels of metanephrines are a useful screening test. The positive and negative predictive values of 64% and 100%, respectively, reported by Lenders and colleagues are misleading. Emphasis on the negative predictive value of a test is unwarranted when the disease is rare. A clinician answering "no" to the question "Does this hypertensive patient have a pheochromocytoma?" will be correct 9999 of 10 000 times. Therefore, a test that is correct 10 000 of 10 000 times is unnecessary.

Far more important in rare illnesses is the positive predictive value of a test. This value depends on specificity, and this dependency increases with progressive decreases in prevalence. The authors, by selecting control and comparison groups, created a "universe" in which the prevalence of pheochromocytoma was 21%. In this universe, the observed specificity of 85% conferred a positive predictive power of 64% on free plasma metanephrines. It can be easily calculated that in an unselected population of hypertensive persons (prevalence of pheochromocytoma, 1 in 10 000), the positive predictive power would be 0.07%. In other words, 7 of 10 000 patients with a positive test result will actually have a pheochromocytoma; this is a negligible gain from the knowledge that 1 of 10 000 untested hypertensive persons has the disease. Selecting a subpopulation with 5% previous probability for the calculations (a task that may not be feasible, even for an expert) leads to a positive predictive power of 25%; that is, a positive test result would be a "true-positive" result in 25% of cases and a "false-positive" result in 75% of cases. Therefore, a test with a specificity of 85% is not a good screening test for pheochromocytoma.

Catecholamines and their metabolites are ubiquitous and hypersecreted in response to different physiologic and pathologic stimuli. In other words, these chemicals are not specific for chromaffin-cell tumors: Therefore, any new diagnostic test relying on their measurement will probably offer only a marginal benefit. Until a more specific marker for pheochromocytoma is found, we must continue to resort to multiple and imperfect biochemical tests, pharmacologic manipulation (for example, clonidine or glucagon), and imaging techniques for diagnosing this rare but extremely dangerous disease.