Weideman and Patel raise a number of salient points. We agree that our results apply only to patients who receive the forms of vitamin K that we administered and should be confirmed if other formulations (such as tablets) are to be used. Furthermore, we agree that the relative times from administration of vitamin K to subsequent INR determinations are critical. In our study, these times were available for all but two patients and were very similar (23.5 vs. 23.0 hours in the oral and subcutaneous groups, respectively [P > 0.2]). Finally, although many medications are dosed according to weight, we have been unable to demonstrate a significant correlation between body habitus and response to oral vitamin K, perhaps because much of the efficacy of oral vitamin K is related to its immediate delivery to the liver, where it has its entire therapeutic effect.

Where recorded, the indications for warfarin therapy were similar in the oral and subcutaneous groups: primary prophylaxis in the setting of atrial fibrillation (8 and 6 patients, respectively), secondary prophylaxis after deep venous thrombosis (12 and 12 patients, respectively), and primary prophylaxis after a high-risk surgical procedure (5 and 6 patients, respectively). None of the patients intentionally overdosed on warfarin, and the prolonged INR was found to be associated with excess alcohol intake in only two cases.

Ringstrom and Long inquire about a placebo group. Our group has previously published a randomized comparison of 1 mg of oral vitamin K with placebo (1). This study was methodologically similar to our more recent trial and clearly demonstrated that, when compared with placebo, 1 mg of oral vitamin K produces faster and more reliable reductions in the INR.