[Skip to Content]
[Skip to Content Landing]
October 23, 1991

Diminishing Diagnostic Returns

Author Affiliations

New York, NY

New York, NY

JAMA. 1991;266(16):2224. doi:10.1001/jama.1991.03470160056028

In Reply.  —In the interest of clarity I avoided any discussion of the informational value of negative test results, referring instead to an earlier, more comprehensive article1 that demonstrates the negative information given by a negative result.The significance of negative diagnostic information is somewhat different from that suggested by Dr Fox. In fact, any test result negative with respect to a given diagnosis will decrease the posttest probability of the disease, giving a ratio of posttest to pretest probability of less than unity, the logarithm of which will be negative, yielding an amount of information that is negative but perfectly useful. It is not any more inaccurate or misleading than positive information. It does indeed increase our uncertainty about the diagnosis, but properly so.As diagnostic certainty is approached, the probability of a false-negative result approaches 1 and, since the result now carries little weight, the posttest probability