In the Discussion section of our paper, we stated that the number of outcome events was modest and that our meta-analysis therefore lacked statistical power to reliably detect clinically worthwhile differences in treatment effect between LMWH and unfractionated heparin. However, the pooled data presented in our paper reflect the totality of the currently available randomized evidence, and it is extremely unlikely that there will be further randomized trials comparing LMWH and unfractionated heparin for the initial treatment of pulmonary embolism. Therefore, our data are probably the best evidence that we will ever have concerning this question.

It is unclear why Drs. Karne and Shah were unable to reproduce our results, but this may relate to differences in the statistical methods used to pool the data or the programming of the meta-analysis software. We performed our analyses with Comprehensive Meta Analysis software (Biostat, Englewood, New Jersey) using a fixed-effects model based on the Mantel–Haenszel method and can confirm that the data reported in Figure 2 of our paper are correct (OR, 0.63 [CI, 0.33 to 1.18]). Furthermore, fixed- and random-effects models produced similar estimates of treatment effect, and in all cases the confidence intervals crossed unity.

We explored the potential for publication bias by creating an inverted funnel plot of treatment effect versus study precision for the primary outcome. Our funnel plot did not appear in print but was included when our manuscript was first submitted to the journal. The plot was relatively symmetrical, with 5 studies to the right and 6 studies to the left of the pooled OR and 1 study superimposed directly on the pooled OR. Although neither meticulous literature searches nor funnel plot analyses can eliminate the possibility of publication bias, we believe that our funnel plot results are consistent with a lack of major publication bias.

Drs. Karne and Shah caution that clinicians should not take the results reported in our paper as evidence of the superiority of LMWH. We concur wholeheartedly; this is also reflected in our conclusion, where we state that fixed-dose LMWH treatment appears to be as effective and safe as dose-adjusted intravenous unfractionated heparin for the initial treatment of nonmassive pulmonary embolism.