Original Research

Development of a Self-Assessment Instrument toDetermine Daily Intake and Variability of DietaryVitamin K

R. Rebecca Couris, PhD, RPh, Gary R. Tataronis, MS, Sarah L. Booth, PhD, Gerard E. Dallal, PhD,Jeffrey B. Blumberg, PhD, FACN, and Johanna T. Dwyer, DSc

Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University (R.R.C., S.L.B., G.E.D., J.B.B., J.T.D.),Massachusetts College of Pharmacy and Health Sciences (R.R.C., G.R.T.), Boston, Massachusetts

Key words: vitamin K, diet records, assessment tools, anticoagulation therapy

Objective: To develop and validate a brief, self-assessment instrument (K-Card) to determine dailyvariations in dietary vitamin K1 (phylloquinone) intake for use in patients receiving oral warfarin anticoagulanttherapy.

Methods: The K-Card was designed to include a checklist of selected common foods and beveragesproviding $ 5 mg vitamin K per serving in American diets and items with lower vitamin K content typicallyconsumed in quantities which contribute significantly to total vitamin K intake. The K-Card was validatedagainst records of weighed food intake from thirty-six healthy volunteers, 20 to 40 and 60 to 80 years of age,whose phylloquinone intakes and plasma concentrations had been previously measured by the MetabolicResearch Unit, Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, Boston, MAUSA. Future use of the K-Card by patients was simulated by a single investigator using 108 one-day weighedfood records to estimate phylloquinone intakes. Dietary phylloquinone calculated from the K-Card wascompared to the values of phylloquinone intake from the diet records collected on the same days, and to fastingplasma phylloquinone concentrations obtained from the same individuals on the following day.

Results:The mean dietary phylloquinone intake (6 SEM) was 138.86 15.7mg for the K-Cards comparedto 136.06 15.8mg for the diet records (p 5 0.067). Bland-Altman limits of agreement between quantities of dietaryphylloquinone calculated from the K-Card and values obtained from the weighed food records were6 38 mg.

Conclusion: In this simulation, the K-Card provided an accurate estimate of dietary phylloquinone intakeand therefore deserves further testing for use by patients receiving coumarin-based anticoagulant therapy todetermine whether variability in dietary patterns contributes to disruptions in anticoagulant drug efficacy andsafety.

INTRODUCTION

Vitamin K (phylloquinone) is involved in blood coagulationthrough its generation of the active cofactor required for thegamma-carboxylation of glutamic acid (Gla) which is a criticalpost-translational step in the synthesis of six vitamin K-depen-dent coagulation proteins: prothrombin, factors VII, IX, and X,and proteins C and S [1–5]. The synthesis of Gla residues inthese proteins permits their binding to calcium and subsequentparticipation in the clotting cascade [4–6]. Oral anticoagulants,such as warfarin, act to create a partial deficiency in the active

form of vitamin K, which is involved in the post-translationalmodification of these coagulation factors, and thus antagonizethe vitamin’s action [7]. Warfarin, a coumarin-based anticoag-ulant, has been used in human anticoagulant therapy since the1940s and today is one of the primary oral anticoagulants forthromboembolic disease and related disorders [8–10].

Warfarin’s anticoagulant effects can be disrupted by varia-tions in dietary phylloquinone (vitamin K1) intake [11, 12] thatmay result in serious, life-threatening consequences [13–18].Therefore, clinicians often recommend that patients who arereceiving warfarin therapy limit their daily variations in

Address reprint requests to: R. Rebecca Couris, Ph.D., R.Ph., Massachusetts College of Pharmacy and Health Sciences, 179 Longwood Avenue, Boston, MA 02115. E-mail:rcouris@mcp.edu

Journal of the American College of Nutrition, Vol. 19, No. 6, 801–806 (2000)Published by the American College of Nutrition

801

vitamin K consumption to no more than 250 to 500mg frombaseline intakes [16, 19]. When the intake of vitamin K remainsconstant, regardless of its absolute dietary level, no adversedrug-nutrient interaction that affects anticoagulant outcomes isevident [20]. However, when variations in vitamin K intakeexceed these levels, disruptions can occur in the new homeo-static balance that has been achieved with anticoagulant drugs.Among patients who are anticoagulated with a constant dose ofwarfarin, increases in phylloquinone intake have been associ-ated with abnormal and excessive blood clotting or warfarin-resistant anticoagulation, necessitating increases in warfarindose [13–16]. For example, warfarin resistance has been doc-umented among patients consuming large amounts of broccoli,which contains 113mg phylloquinone per 100 g serving [13].Walker [14] found that patients taking anticoagulants on weightreduction diets who consumed large amounts of green leafyvegetables, which are typically rich in phylloquinone, had agreater incidence of myocardial infarctions associated withwarfarin resistance than did their peers with lower intakes.

Although 500mg of phylloquinone has been demonstratedto overcome the disruption of vitamin K metabolism associatedwith warfarin [21], enteral products containing as little as 25 to115 mg of phylloquinone per 100 mL have also been reportedto create a dietary-induced warfarin resistance [22, 23]. Con-sistent with these observations, Oversenet al. [15] and Ped-ersenet al. [16] noted that patients taking anticoagulants whoconsumed vegetables that were poor sources of phylloquinonerarely reported adverse reactions.

In contrast, decreases in phylloquinone intake among pa-tients taking anticoagulants can result in abnormal bleeding orwarfarin-potentiated anticoagulation [17, 18]. Chowet al. [18]reported hemorrhagic complications in two patients on long-term anticoagulant therapy due to dietary modifications thatdrastically lowered their intakes of phylloquinone. Further-more, fluctuations in phylloquinone intake may prolong theinterval necessary to reach a stable dosing regimen with anti-coagulant therapy, a circumstance which may result in adversehemorrhagic and thrombotic events. Therefore, the NationalStroke Association and pharmaceutical manufacturers of anti-coagulants recommend that dietary intakes of phylloquinoneremain constant in patients receiving oral anticoagulant therapy[24, 25]. Constant dietary intakes of phylloquinone are espe-cially important for patients receiving acenocoumarol, a drugthat has an even shorter half-life than warfarin and is poten-tially affected even more dramatically by dietary phylloquinonefluctuations. A consistent phylloquinone intake that meets thecurrent Recommended Dietary Allowance of 65 to 80mg/daywould therefore be prudent patient advice [26, 27].

The recent availability of food composition data for vitaminK makes it possible to identify foods that may potentiallyinterfere with the efficacy of warfarin anticoagulant therapy.However, an efficient and inexpensive method for measuringdietary phylloquinone intake and variability in a clinical settinghas not been available. Thus, we developed a short, practical

dietary self-monitoring tool, called the K-Card. The K-Cardwas then validated to determine deviations from a standard toquantitatively assess day-to-day and week-to-week variabilityin dietary phylloquinone intake which could be used by patientsreceiving coumarin-based anticoagulant therapy.

METHODS

K-Card Development

The items selected for inclusion on the K-Card representfoods and beverages determined in previous studies to pro-vide $ 5 mg vitamin K per serving [28–30]. In addition, lowerphylloquinone content foods eaten frequently or in quantitiesthat contribute significantly to vitamin K intake were alsoincluded. Common supplements and products providing vita-min K were also incorporated into the K-Card. Fifty-five itemsare listed on one double-sided page of thick stock paper andcategorized by color coding as follows: vegetables (green),meat/poultry/fish (yellow), mixed dishes (blue), fats/oils/saladdressings (yellow), snacks (green), desserts (purple), beverages(pink), and dietary supplements (lavender). Each side of theK-Card is presented in Tables 1 and 2, respectively. Foods andfood groups that appear on the K-Card include those from theU.S. Food and Drug Administration Total Diet Study whichprovides the phylloquinone content of commonly consumedfoods and beverages in the American diet [29]. These sourceswere supplemented by provisional tables listing the phylloqui-none content of common foods and beverages and additionalitems from the vitamin K nutrient database at the Jean MayerUSDA Human Nutrition Research Center on Aging at TuftsUniversity [28–30].

K-Card Validation Utilizing Previous MetabolicBalance Study Diet Record Information

We employed the information from a previously conductedmetabolic balance study on vitamin K to validate the K-Card.Specifically, we used lists of foods documented from records ofweighed food intake of 36 volunteers whose phylloquinoneintakes and corresponding fasting plasma phylloquinone con-centrations were previously collected [30]. In this previousmetabolic balance study, subjects were stratified into fourgroups (nine individuals per group) and categorized by age (20to 40 years and 60 to 80 years) and gender. The subjects wereinstructed not to use antibiotics or supplements for at least fourweeks prior to and during the study. Each participant completedthree sets of four consecutive day-weighed diet records of allfoods, beverages and other ingredients that they consumedwhile living at home on their usual diets. Food portion sizeswere estimated with dietary scales, rulers and household mea-sures. There was a minimum of six weeks between each dietrecord period so that all months of the year were represented in

Self-Assessment of Vitamin K Intake and Variability

802 VOL. 19, NO. 6

the data for each subject. To determine vitamin K status, fastingplasma was drawn on the last day after each of the three dietrecord periods. Fasting plasma phylloquinone concentrationswere then used to determine the diet-plasma relationship for

vitamin K [30]. The results of this metabolic balance studyprovided the best available standard for phylloquinone intakeas it relates to diet records and fasting plasma phylloquinoneconcentrations.

Table 1. K-Card (Check which foods you eat at each meal everyday. If you eat more or less than one serving indicate howmuch.)

Self-Assessment of Vitamin K Intake and Variability

JOURNAL OF THE AMERICAN COLLEGE OF NUTRITION 803

Sampling of Diet Records for K-Card Validation

In this study, one of us (R.R.C.) recorded lists of foods ontothe K-Card using the diet records from the previous metabolicbalance study [30]. The investigator was blinded to all identi-fying information for each record and also to calculation ofdietary phylloquinone intake. The dietary record selected forthis simulation study was taken from each of the three sets offour consecutive day-weighed diet records and corresponded to

the day prior to the 8:00 am fasting blood sample (12 to 14hours postprandial) with plasma phylloquinone concentrationsdetermined by HPLC according to Boothet al. [30]. A total of108 one-day weighed diet records were sampled, representingthree reporting periods for each of the 36 volunteers. Validationof the K-Card was accomplished by comparing phylloquinoneintakes from the K-Cards to the completed diet records of theprevious study [30]. Recorded food intakes on the K-Card weretranslated into quantities (mg) of phylloquinone per serving and

Table 2. K-Card (Check which foods you eat at each meal everyday. If you eat more or less than one serving indicate howmuch.)

Self-Assessment of Vitamin K Intake and Variability

804 VOL. 19, NO. 6

used to determine amounts of phylloquinone consumed. Insubsequent analyses these amounts were compared to thoseobtained from calculation of phylloquinone intakes assessedfrom the previous study’s diet records on the same days usingthe Tufts University vitamin K database.

Statistical Analysis

Pearson correlation coefficients were used to compare thequantity of dietary phylloquinone from the K-Card against thatfrom weighed food records obtained from Boothet al. [30].Because strong correlations between two assessment methodsdo not account for bias between the two instruments and do notnecessarily indicate agreement, limits of agreement betweenthe quantity of dietary phylloquinone calculated from the K-Card and that obtained from the diet records were derivedaccording to the method of Bland and Altman [31]. Scatterplotswere displayed to compare plasma phylloquinone concentra-tions (nmol/L) against vitamin K intake (mg) from both theK-Card and the weighed diet records for each visit. A logarith-mic transformation of the data was used because this data wereskewed to the right. The quantity of vitamin K intake (mg) fromthe K-Card and the diet records was also compared. Repeatedmeasures ANOVA was used to compare mean differencesbetween intake as measured by diet records and K-Cards.

RESULTS

Fig. 1 depicts the differences in the quantities of dietaryphylloquinone as determined from each of the three sets of 36one-day weighed food records and the quantities of dietaryphylloquinone assessed using the K-Card. The day-to-day vari-ability was assessed by examining differences from one clinicvisit to another within patients. A repeated measures analysis ofvariance comparing differences between visits revealed no sta-tistical significance between the mean differences due to thevisits (p 5 0.578).

The mean dietary vitamin K intake (6 SEM) was 138.8615.7mg for the K-Cards compared to 136.06 15.8mg for thediet records (p 5 0.067, repeated measures ANOVA). Pearsoncorrelation coefficients were used to examine the associationbetween the quantities of phylloquinone intake assessed by theK-Card and those from the diet records for each visit asfollows: visit 1 (r 5 0.995,p , 0.05), visit 2 (r5 0.998,p ,

0.05) and visit 3 (r5 0.989,p , 0.05). As noted above, strongcorrelations between the assessment methods may not implyagreement. Therefore, the limits of agreement were determinedbetween the quantities of dietary phylloquinone calculatedfrom the K-Card and those obtained from the weighed dietrecords according to the method of Bland and Altman [31]. Inadvance of the analysis,# 50 mg of dietary phylloquinone wasdefined as an acceptable limit of variation between the twoassessment methods. This value represents a more stringent

criterion for variability compared to the amounts documentedin clinical studies that indicate variations of 250mg/day indietary phylloquinone have a significant effect on warfarinanticoagulant therapy [16, 19]. A plot of the differences indietary phylloquinone intake between the two assessment toolsagainst their mean for each individual is presented for all threevisits (Fig. 2).

DISCUSSION

In a simulation model, calculating dietary phylloquinoneintake using the K-Card provided essentially identical values tothose obtained by independent assessment from the samerecords of weighed food intake. These quantities were gener-ally within 38 mg, substantially less than the 50mg pre-definedacceptable limit of variation. The accuracy of the K-Cardderives, in part, from its inclusion of all food items contain-ing $ 5 mg vitamin K per serving. Furthermore, food itemswith lower phylloquinone content but that were typically con-sumed in quantities which contribute significantly to total phyl-loquinone intake and dietary supplements containing vitamin Kwere also included in the K-Card.

The K-Card is an easy-to-use self-assessment tool that pro-vides an accurate estimate of dietary phylloquinone intake inthis simulation exercise. Further assessment will be conductedto determine its accuracy when used by actual patients. TheK-Card food checklist permits patients to record their daily dietin fewer than 15 minutes. Quick and simple tabulations of itemscontributing to phylloquinone intake from the K-Card can beperformed by either the patient or trained personnel. In con-trast, completion of multiple diet records is time-consuming forthe patients and requires trained personnel to code and subse-quently analyze record entries. The K-Card has been tested ina prospective clinical trial to determine the effect of phylloqui-none intake on the stability of warfarin therapy (unpublisheddata).

Variability in phylloquinone intake may underlie the dis-crepancies noted in patients with fluctuating responses to cou-marin-based anticoagulant therapy from prolonged bleedingtime to clotting. Dietary intakes of phylloquinone have greaterwithin-to-between person variability than other fat-soluble vi-tamins, which may have detrimental effects on the stability oforal anticoagulant therapy [30]. Thus, when further validated,the K-Card could prove useful in monitoring serial daily phyl-loquinone intakes to establish day-to-day and week-to-weekvariability. Measuring variability in phylloquinone intake witha short and easy-to-use instrument like the K-Card may bebeneficial in education programs for patients receiving couma-rin-based anticoagulant therapy. The K-Card could also provideprimary healthcare providers with information that could im-pact on patient counseling and education to encourage compli-ance to a stable phylloquinone intake among patients receivinganticoagulant therapy.

Self-Assessment of Vitamin K Intake and Variability

JOURNAL OF THE AMERICAN COLLEGE OF NUTRITION 805

Fig. 1. Differences between dietary phylloquinone intake by patients as assessed using weighed diet records and the K-Card.

Fig. 2. Differences against means for dietary phylloquinone intake assessed from the K-Card and diet records.

Self-Assessment of Vitamin K Intake and Variability

806 VOL. 19, NO. 6

ACKNOWLEDGEMENTS

The authors would like to thank Barbara Furie for criticalevaluation of the study design and analysis, Nicola McKeownand Katherine Tucker for assessment of phylloquinone fromthe diet records, Ken Davidson for technical assistance and thevolunteers who participated in this study. This study was sup-ported by the School of Nutrition Science and Policy, FrancesStern Nutrition Center, and the Jean Mayer USDA HumanNutrition Research Center on Aging at Tufts University, Bos-ton, MA through Cooperative Agreement No. 58-1950-9-001with the U.S. Department of Agriculture.

Note: The K-Card, vitamin K intake assessment card, istrademarked and copyrighted by Tufts University. Any use orother infringement is prohibited. Contact the Associate Deanfor Administration and Finance at the Tufts University Schoolof Nutrition Science and Policy for information.

REFERENCES

1. Suttie JW, Mammah-Schendel LL, Shah DV, Lyle BJ, Greger JL:Vitamin K deficiency from dietary vitamin K restriction in hu-mans. Am J Clin Nutr 47:475–480, 1988.

2. Suttie JW: “Current Advances in Vitamin K Research.” Amster-dam: Elsevier, 1988.

3. Suttie JW: Vitamin K and human nutrition. J Am Diet Assoc92:585–590, 1992.

4. Furie BC, Furie B: Vitamin K-dependent blood coagulation pro-teins. In Arias IM, Boyer JL, Fausto N, Jakoby WB, Schacter DA,Shafritz DA (eds): “The Liver: Biology and Pathobiology,” 3rd ed.New York, NY: Raven Press, pp 1217–1225, 1994.

5. Furie BC, Furie B: Structure and mechanism of action of thevitamin K-dependent gamma-glutamyl carboxylase: recent ad-vances from mutagenesis studies. Thromb Haemost 78:595–598,1997.

6. Sadowski JA, Bovill EG, Mann KG: Warfarin and the metabolismand function of vitamin K. In Poller L (ed): “Recent Advances inBlood Coagulation.” London: Churchill, Edinburgh, Livingstone,pp 93–118, 1991.

7. Majerus PW, Broze GJ, Miletech JP, Tollefsen DM: Anticoagu-lant, thrombolytic, and antiplatelet drugs. In Gilman AG, Rall TW,Nies AS, Taylor P (eds): “The Pharmacological Basis of Thera-peutics,” 8th ed. New York: Pergaman Press, pp 1311–1331, 1990.

8. Owen Jr CA, Bowie EJW: The history of the development of oralanticoagulant drugs. In Poller L, Hirsh J (eds): “Oral Anticoagu-lants.” New York: Arnold. 1996.

9. Hirsh J, Poller L, Deykin D, Levine M, Dalen JE: Optimal thera-peutic range for oral anticoagulants. Chest 95 Suppl:5S–11S, 1989.

10. Hirsh J: Oral anticoagulants. N Engl J Med 324:1865–1875, 1991.11. Udall JA: Human sources and absorption of vitamin K in relation

to anticoagulation stability. JAMA 194:127–129, 1965.12. Udall JA: Vitamin K and coumarin drug interrelationships in man.

Curr Ther Res 8:627–631, 1966.

13. Kempin SJ: Warfarin resistance caused by broccoli. N Engl J Med39:1229–1230, 1983.

14. Walker FB: Myocardial infarction after diet-induced warfarin re-sistance. Arch Intern Med 144:2089–2090, 1984.

15. Oversen L, Lyduch S, Idorn ML: The effect of a diet rich in brusselsprouts on warfarin pharmacokinetics. Eur J Clin Pharmacol 33:521–523, 1991.

16. Pedersen FM, Hamberg O, Hess K, Ovesen L: The effect of dietaryvitamin K on warfarin-induced anticoagulation. J Int Med 229:517–520, 1991.

17. Colvin BT, Lloyd MJ: Severe coagulation defect due to a dietarydeficiency of vitamin K. J Clin Pathol 30:1147–1148, 1977.

18. Chow WH, Chow TC, Tse TM, Tai YT, Lee WT: Anticoagulationinstability with life-threatening complication after dietary modifi-cation. Postgrad Med J 66:855–857, 1990.

19. Karlson S, Leijd B, Hellstrom K: On the influence of vitaminK-rich vegetables and wine on the effectiveness of warfarin treat-ment. Acta Med Scand 220:347–350, 1986.

20. Harris JE: Interactions of dietary factors with oral anticoagulants:review and applications. J Am Diet Assoc 95:580–584, 1995.

21. Shetty HGM, Backhouse G, Bentley DP, Routledge PA: Effectivereversal of warfarin-induced excessive anticoagulation with lowdose vitamin K1. Thromb Haemost 647:13–15, 1993.

22. Howard PA, Hannaman KN: Warfarin resistance linked to enteralnutritional products. J Am Diet Assoc 85:713–715, 1985.

23. Martin JE, Lutomski DM: Warfarin resistance and enteral feed-ings. JPEN 13:206–208, 1989.

24. American Heart Association/National Stroke Association: Fightingheart disease and stroke. 70th Scientific Session Abstract StatusReport, 1997.

25. DuPont Pharma: “A Patient’s Guide to Using Coumadint atHome.” Wilmington, DE: DuPont, 1996.

26. Booth SL, Charnley JM, Sadowski JA, Saltzman E, Bovill E,Cushman M: Dietary vitamin K1 and stability of oral anticoagula-tion: proposal of a diet with constant vitamin K1 content. ThrombHaemost 77:504–509, 1997.

27. Booth SL, Tucker KL, McKeown NM, Davidson KW, Dallal GE,Sadowski JA: Relationships between dietary intakes and fastingplasma concentrations of fat-soluble vitamins in humans. J Nutr127:587–592, 1997.

28. Booth SL, Sadowski JA, Weihrauch JL, Ferland G: Vitamin K1

(phylloquinone) content of foods: a provisional table. J Food CompAnal 6:109–120, 1993.

29. Booth SL, Sadowski JA, Pennington JAT: The phylloquinone(vitamin K1) content of foods in the US-FDA Total Diet Study. JAgric Food Chem 43:1574–1579, 1995.

30. Booth SL, Pennington JAT, Sadowski JA: Food sources and di-etary intakes of vitamin K1 (phylloquinone) in the American diet:data from the FDA Total Diet Study. J Am Diet Assoc 96:149–154,1996.

31. Bland JM, Altman DG: Statistical methods for assessing agree-ment between two methods of clinical measurement. Lancet2:307–310, 1986.

Received August 4, 1999; revision accepted August 31, 2000.

Self-Assessment of Vitamin K Intake and Variability

JOURNAL OF THE AMERICAN COLLEGE OF NUTRITION 807