Study Finds No Benefit of Low-dose Vitamin K Administration for High INRs; But Questions Remain

Henry I. Bussey, Pharm.D.
March, 2009
Updated May, 2009

Editor's Note I contacted several of the authors of the article reviewed here and invited them to respond with comments and/or rebuttal. Comments from Mark Crowther are provided at the bottom of this page following my review below. Click here to jump to Dr. Crowther's comments.

Article Reviewed: Crowther MA, Ageno W, Garcia D, et al. Oral Vitamin K Versus Placebo to Correct Excessive Anticoagulation in Patients Receiving Warfarin, A randomized Trial. Ann Intern Med 2009; 150:293-300.

A large group of well-respected clinical investigators conducted a trial to evaluate the impact on bleeding and thromboembolism of administering a low dose of vitamin K to warfarin-treated patients with an INR between 4.5 and 10.0. Any bleeding (major, minor, and trivial), major bleeding, and thromboembolic events were no different at 90 days or at 7 days (see table).

So, should we do as the authors suggest and not feel compelled to administer vitamin K to patients with an INR between 4.5 and 10? There are a few factors that make me reluctant to adopt this conclusion:

1. Size of the study: Although this multi-center trial was relatively large, it was under-powered to identify a difference in the risk of major bleeding and, in fact, apparently no major bleeds occurred during the first 7 days.
   
   
2. Extent of INR elevation: The majority of patients included in the 7 day analysis had only modest elevations in their INRs. Specifically, 226 of 355 or 63.7% of the vitamin K group and 261 of 369 or 71% of the placebo group had INRs between 4.5 and 6.0, a level at which many clinicians might not ordinarily administer vitamin K. INRs were above 8 in only 26 patients (7%) in each group and apparently patients with an INR greater than 10 were excluded from the study.
   
   
3. Reliability of the INR: As far as I could determine, there was no evidence that the high INR readings were confirmed. In two studies conducted in our clinic many years ago, we found that 10% to 15% of INRs were seriously incorrect and misleading. Most such incorrect values were falsely elevated rather than falsely low. Therefore, one must question the degree to which lab error may have contributed to the enrollment of these patients.
   
   
4. Dosage formulation: The vitamin K was administered as a capsule reformulation compounded from the traditional 5 mg vitamin K tablet, but the dose was only 1.25 mg. Several previous studies have reported effective lowering of moderately elevated INRs with 1 mg to 2 mg of vitamin K but, as far as I know, each of those studies used the injectable liquid formulation given orally. The effectiveness of a low dose formulation as used in this study has not been established to the best of my knowledge (although a reduction in the INR was apparent in the current trial). The authors chose not to use one-half of the tablet formulation based on their anecdotal experience that such a dose would over-correct the INR in too many patients. Such over-correction was not a problem in the reports by Whitling, et al¹ and Weibert, et al² in two earlier studies (not referenced in the current manuscript). In those trials, however, the extent of INR elevation was considerably higher than the mean INRs of 6.0 and 5.8 in this study. For example, in the paper by Whitling, et al, the group mean INRs ranged from about 10 to 15. In fact, the reports by Whitling and Weibert both suggested that a 2.5 mg "half tablet" dose of vitamin K was usually effective if the INR was between 6 and 10, but that a higher dose - possibly 5 mg - might be required for INRs above 9 or 10.
   
   

Alternate conclusion: A dose of 1.25 mg as a solid dosage formulation of vitamin K failed to show a clear clinical benefit in a population of warfarin-treated patients in which the majority of patients had an INR of less than 6. The current study, however, provides little evidence with which to judge the merits of vitamin K administration for higher INR values (> 6, > 8, and/or > 10) or the effectiveness of a larger dose in tablet form for patients with higher INRs. On the other hand, I will have to admit that the results of the current study do make me a little more comfortable in not administering vitamin K when the INR is less than 6 and bleeding - or additional bleeding risks factors - are not present.

References

1. Whitling AM, Bussey HI, Lyons RM. Comparing different routes and doses of phytonadione for reversing excessive anticoagulation. Archives of Internal Medicine 1998; 158:2136-2140.
   
   
2. Weibert RT, Le DT, Kayser SR, Rapaport SL. Correction of excessive anticoagulation with low-dose oral vitamin K1. Annals of Internal Medicine 1997; 125:959-962.

Comments from Mark Crowther, MD, MSc, FRCPC

The seminal finding in the paper is the overall risk of major bleeding was very low, and that it is very unlikely (statistically) that vitamin K could have reduced this risk in a patient important way. That all having been said, it appears (to the best of our ability to discern) that the vitamin K was non-toxic and since it did drop the INR it may have utility in getting the INR down and allowing warfarin reinstitution.

Based our study, your comments and feedback I have received from many others I think the really important study (which would be difficult to do) would be to randomize patients with INR values above 6.0 to vitamin K or placebo - this is a higher bleeding risk group where the merits of vitamin K would be more apparent. However, I suspect (based on intuition and anecdote) that the vitamin K would be relatively ineffective in that group as well.

Mark Crowther, MD, MSc, FRCPC