Dr. Wolf has identified the upper 95% confidence limit for the quantitative rapid ELISA's negative likelihood ratio from the sensitivity analysis. The lower 95% confidence limit was 0.00, which is statistically as likely as the value for the upper 95% limit. Both of these extreme values are unlikely to occur clinically. The sensitivity analysis provided a central estimate of 0.05, which is consistent with the primary analyses. It should be noted that the value for sensitivity in the sensitivity analysis was 0.98, with a 95% confidence limit of 0.88 to 1.00, a much narrower range of values than was seen for the negative likelihood ratio. The primary analyses in our Table 1 show similar findings with narrower confidence limits. The key fact from the sensitivity analysis is that there was no shift in the observed values for sensitivity and negative likelihood ratio, although the confidence limit was broader for the latter. A recent rigorous clinical outcome study in a large number of patients supports our findings (1): Perrier and colleagues reported that the quantitative rapid ELISA was effective and safe as the first-line test for ruling out pulmonary embolism in outpatients. We agree that using the clinical probabilities adds further value to the diagnostic process and stated this in our Discussion. We also concluded from our data that a negative quantitative rapid ELISA result is as diagnostically useful as a normal or near-normal lung scan or negative duplex ultrasonography finding. As we indicated, "combining a negative rapid ELISA result with a low or moderate clinical probability for DVT or PE [pulmonary embolism] rules out these diagnoses." We agree that a high-probability clinical assessment in combination with negative results on quantitative rapid ELISA indicates the need for further testing.

We agree with the conclusion of Drs. Le Gal, Righini, and Bounameaux that "the [d-dimer] test can identify patients in whom anticoagulant therapy is not necessary, which is the true clinically relevant issue." In a commentary that accompanied our article, the Editor showed how the D-dimer test (quantitative rapid ELISA) best fits in the diagnostic process.

Regarding the comments of Drs. Philbrick, Heim, and Schectman, we stated in our Discussion that "the values for specificity and positive likelihood ratio differed among the assays, but all were within a range considered to be of little clinical value in altering probability of disease." We also said that "the clinical utility of the D-dimer assays is limited by the nonspecificity of a positive result" and that it "differs among patient samples and may be higher in outpatients."

It is unclear whether the presence of cancer interferes with the more sensitive D-dimer ELISAs. Drs. Puglisi and Federico cited non-ELISA D-dimer assays, which have a lower overall sensitivity.