Objective: To determine whether very low doses of warfarin are useful in thrombosis prophylaxis in patients with central venous catheters.

Design: Patients at risk for thrombosis associated with chronic indwelling central venous catheters were prospectively and randomly assigned to receive or not to receive 1 mg of warfarin, beginning 3 days before catheter insertion and continuing for 90 days. Subclavian, innominate, and superior vena cava venograms were done at onset of thrombosis symptoms or after 90 days in the study.

Results: One hundred twenty-one patients entered the study, and 82 patients completed the study. Of 42 patients completing the study while receiving warfarin, 4 had venogram-proven thrombosis. All 4 had symptoms from thrombosis. Of 40 patients completing the study while not receiving warfarin, 15 had venogram-proven thrombosis, and 10 had symptoms from thrombosis (P < 0.001). There were no measurable changes in the coagulation values assayed due to this warfarin dose, except in occasional patients who had become anorectic because of their disease or chemotherapy.

Conclusions: Very low doses of warfarin can protect against thrombosis without inducing a hemorrhagic state. This approach may be applicable to other groups of patients.

The standard therapy of using warfarin for anticoagulation imposes a risk for hemorrhage. The concept of the ideal warfarin dose is based on the perceived ideal prothrombin time which, in turn, is a function of the thromboplastin used in the in-vitro assays ¹. Substituting more sensitive thromboplastin reagents, such as the Manchester Comparative Reagent, prolongs the prothrombin time after smaller doses of warfarin ² . Recent clinical trials ^3-10 have indicated that lower doses of warfarin will substantially reduce risk for hemorrhage without affecting therapeutic efficacy. Revised recommendations for warfarin usage have been reported. ¹¹

The indications for warfarin therapy to prevent recurrence of previously established thrombi are established; however, the dose of warfarin needed for prophylaxis for de novo thrombi is less well established. ¹² It has been suggested that smaller than standard doses could be used to prevent thrombi. Low antithrombin III levels found in morbidly obese patients improved when these patients were given low doses of warfarin ¹⁰. In a prospective open nonrandomized study, low-dose warfarin therapy was associated with a reduced incidence of thrombosis when central vein catheters were used for intravenous hyperalimentation and such therapy did not prolong the prothrombin time ⁹. Poller and colleagues ⁸ reported that minidose warfarin therapy (1 mg given before and after gynecologic surgery) reduced the incidence of postoperative deep-vein thrombosis when therapy was begun 6 or more days preoperatively.

We initiated our study to investigate the same concept using an open prospective randomized schema for patients at high risk for thrombosis ¹³. Patients receiving infusion chemotherapy via chronic indwelling central venous catheters were prospectively randomly assigned either to receive or not to receive very low doses of warfarin. Subclavian venograms were done after 90 days in the study or when symptoms or signs suggesting thrombosis developed.

Top

Methods

We conducted a pilot study to establish a warfarin dose that would not alter prothrombin time, using patients with the same clinical characteristics as indicated below. Three of three patients developed prolonged prothrombin times when taking 2.5 mg of warfarin daily. One of four patients developed prolonged prothrombin times in response to 2.0 mg of warfarin daily. When taking 1 mg of warfarin daily, none of four patients had changes; therefore, the dose of 1 mg daily was selected for our study.

Thereafter, consecutive patients who acquired central venous catheters and had projected survivals of greater than 3 months were randomly assigned to receive or not receive 1 mg of warfarin daily. Assignment to receive or not receive the drug was made according to a previously established yes or no code. Because use of sclerosing chemotherapeutic drugs was a suspected risk factor for thrombosis, randomization was stratified for patients receiving sclerosing and nonsclerosing drugs. All decisions about chemotherapy drugs were made by the attending oncologists. Patients who were to receive warfarin therapy began taking warfarin 3 days before their catheters were inserted. Warfarin therapy was continued at this dose for 90 days or until there was venogram evidence of thrombosis. Prothrombin times were measured weekly for 1 month and, thereafter, monthly or more frequently if clinically indicated. If the prothrombin time became greater than 15.0 seconds (normal reference range, 11.5 to 13.5 seconds), the patient was not given warfarin, was given vitamin K, and then was given warfarin when the prothrombin time returned to normal, thereafter maintaining the prothrombin time in the reference range.

Patients were excluded from our study if they had baseline platelet counts under 125 x 109/L, acquired or congenital coagulopathies, previous subclavian vein catheters, obstructing mediastinal tumors, previous history of deep-vein thrombophlebitis, anatomic lesions that bleed (such as duodenal ulcers), or were taking drugs that suppress platelet function. We also excluded patients with serum creatinine over 140 ,umol/L (1.6 mg/dL) because of suppressive effects of uremia on platelet function, the risks that contrast dye imposes on kidney function, and alteration of warfarin metabolism. Patients gave informed consent.

Port-a-Cath subclavian catheters (Pharmacia Deltec, Inc., St. Paul, Minnesota) were inserted only by surgeons experienced in percutaneous techniques ¹⁴. Correct placement of the catheter tip in the superior vena cava or proximal innominate vein was confirmed by chest roentgenogram. If the catheter was in the jugular vein, the patient was dropped from study. Patients received a maximum of 500 units of heparin per week for catheter flush when receiving infusion therapy or 500 units monthly when not receiving therapy.

After a 10-mL discard, blood samples for the laboratory assays were drawn into plastic syringes via the Port-a-Cath and immediately placed into capped test tubes containing 3.8% sodium citrate. Samples were transported to the laboratory within 30 minutes, and the plasma separated within 1 hour at room temperature. Samples were then either placed on ice or frozen. The prothrombin time and partial thromboplastin times were measured immediately. Antithrombin III (AT III) assays were done within 6 hours. The remaining plasma was aliquoted into plastic tubes, capped, and frozen at -80°C. The laboratory personnel were unaware of the status of patients in the study.

The prothrombin times were measured using three thromboplastin reagents: Dade thromboplastin C, Dade thromboplastin FS (American Dade, American Hospital Supply, Agunda, Puerto Rico), and the Manchester Comparative Reagent (supplied by Dr. Leon Poller, National [UK] Reference Laboratory, Withington Hospital, Manchester, England). They were also measured using dilute thromboplastin to increase the sensitivity of the assays. For these latter studies, the thromboplastin was diluted to prolong the control value to approximately twice that found using undiluted thromboplastins (see Results). The partial thromboplastin time assay was the routine laboratory one-stage study using Dade Actin thromboplastin. Antithrombin III was measured using a heparin cofactor assay with a synthetic fluorogenic substrate. The whole blood antithrombin assay was done according to von Kaulla ¹⁵. Functional assays for factors II, VII, IX, and X were done according to routine one-step substitution assays. The Xa inhibitor assay (Heptest, Haemachem, Inc., St. Louis, Missouri) was done according to Yin and colleagues ¹⁶. Warfarin levels were assayed using high pressure liquid chromotography (Dr. Robert Williams, E.I. DuPont Company, Wilmington, Delaware).

Venograms were done on patients after 90 days or sooner for evaluation of ipsilateral or contralateral arm, shoulder, or neck pain; venous distention; edema; or catheter failure (inability to withdraw blood or to infuse drug). Venograms were done via the ipsilateral and, if appropriate, contralateral antecubital arm veins using subtraction technique. When needed for clarification, a selective catheterization was done from the antecubital vein to the area of the clot. Contrast injections were also made via the Port-a-Caths to evaluate clotting beyond the proximal catheter tip. Efforts were made to identify the entire length of the thrombus from the distal arm veins to the superior vena cava. Interpretation of venograms was done independently by radiologists who were unaware of patients' status.

We projected that 40 patients would be required in each arm of the study to detect with 95% CI a drop from 30% to 15% incidence of thrombosis. Discrete variables were studied by contingency tables using the chi-square statistic. Continuous variables were examined by analysis of variance with the F-statistic.

Top

Results

Outcome of Randomization

To obtain the 80 patients required for completion of the study, 121 patients were registered in the study. Table 1 shows the distribution of diseases among patients who completed the study, randomly assigned to receive and not receive warfarin. Forty-two of sixty patients (70%) receiving warfarin and 45 of 61 patients (74%) not receiving warfarin had adenocarcinomas. Table 2 shows the outcome of the randomization for age, sex, and baseline coagulation values for patients receiving and not receiving warfarin. There were no statistically detectable differences in these values between the two groups at the time of randomization.

Table 3 shows the causes for withdrawal from our study after randomization. No patient withdrew because of catheter failure or because of symptoms or signs suggestive of thrombosis.

None of the 26 patients who died from progressive disease before completing the study had complications from their catheters or warfarin therapy. None had clinical evidence of catheter-related thrombosis. An analysis for bias introduced by withdrawal or death of patients was conducted. Deaths while participating in the study were equally distributed between the two groups. On average, there were no laboratory differences between those who completed the study and those who did not, except for one laboratory value. The AT III was 102.5 + 15% (mean + SD) for patients remaining in the study and 87.8 + 18.8% for patients who died during the study (P < 0.03).

Effect of Warfarin on Thrombosis

Of the 42 patients who received warfarin and completed the study, 4 had venogram-documented thrombosis. All 4 had symptoms suggestive of thrombosis. Of the 40 patients who received no warfarin and completed the study, 15 developed venogram-documented thrombosis (P < 0.001) (Figure 1). Of these 15, 13 had symptoms suggestive of thrombosis. Table 1 shows the distribution of diseases among those who developed clots. Neither the tumor type nor the age of the patient affected the incidence of thrombosis. Mean age for those who developed thrombi while receiving warfarin was 65 years (median, 65; range, 54 to 77); whereas, for those not receiving warfarin who developed thrombosis, the mean age was 62 years (median, 62; range, 34 to 81).

Figure 1. Outcome for patients receiving and not receiving very low doses of warfarin. Each line represents one patient from time of randomization to time of venogram-proven thrombosis (closed box) or to the end of study.

Table 1. Distribution of Diseases

Cancer Location or Type	Patients Receiving Warfarin		Patients Not Receiving Warfarin
	Total	With Thrombosis	Total	With Thrombosis
Colon	10	1	14	6
Breast	6	0	8	2
Lung	8	0	3	1
Esophagus	4	1	3	0
Rectum	1	0	3	3
Pancreas	1	1	3	1
Stomach	2	1	2	1
Lymphoma	3	0	0	0
Ovary	2	0	1	0
Melanoma	0	0	2	1
Sarcoma	1	0	0	0
Anus	1	0	0	0
Gallbladder	1	0	0	0
Hodgkin lymphoma	1	0	0	0
Head and neck	1	0	0	0
Myeloma	0	0	1	0

Among the patients receiving sclerosing drugs, 2 of the 15 warfarin-treated patients developed thrombosis, whereas 6 of the 11 patients not treated with warfarin developed clots (P < 0.025). Among the patients receiving nonsclerosing drugs, 2 of 27 developed clots during warfarin therapy in comparison with 9 of 29 in the untreated group (P < 0.029).

The four patients who developed thrombosis while receiving warfarin were compliant with therapy. On review, one patient reported having had a previous deep-vein leg thrombosis that she had not reported before entry into the study. A second patient developed a subclavian thrombosis adjacent to a previously fractured clavicle with a healed deformity of the bone and diversion of the underlying suclavian vein. The two remaining patients had no apparent antecedent contributing factor leading to their thrombosis.

There was a modest tendency for clustering of the thrombi in the first half of the 90-day study (median, 24 days; mean, 38 days). Two of the nineteen patients with thrombi were asymptomatic, giving a 10.5% false-negative rate based on the clinical criteria listed above. Of the patients without thrombosis, two had suggestive clinical signs but negative venograms, giving a 3% false-positive rate.

Effects of Warfarin on Coagulation

There was no detectable warfarin or cross-reactive material at baseline for any patient. The serum warfarin levels at 30, 60, and 90 days were 0.48 (range, 0.19 to 0.74), 0.39 (range, 0.20 to 0.57), and 0.42 (range, 0.26 to 0.60) ng/mL, respectively. No differences were detected in the prothrombin times of any of the three thromboplastins; the partial thromboplastin times; the levels of factors II, VII, IX, and X; the Xa inhibitor; the AT III; or the whole blood antithrombin assay of von Kaulla times between the treated and nontreated groups. The Heptest for Xa inhibitor remained at baseline value, indicating no heparin contamination or heparin-like effect.

Table 2. Baseline Values after Randomization

Baseline Values	Patients Receiving Warfarin	Patients Not Receiving Warfarin
Sex
Male Female	27 33	32 29
Age ± SD, y	56.0 ± 13.5 (range, 26 to 81; median, 55)	60.6 ± 10.7 (range, 34 to 81; median, 62)
Prothrombin time,       mean ± SD, s
Dade thromboplastin C	11.8 ± 1.8	11.9 ± 1.8
Dade thromboplastin FS	14.0 ± 1.5	13.5 ± 3.9
Manchester Comparative Reagent	14.9 ± 1.7	14.7 ± 3.1
Partial thromboplastin time, mean ± SD, s	26.0 ± 4.4	27.3 ± 4.0
Coagulation factors, % of normal
II	93.1 ± 21.8	96.5 ± 18.2
VII	145 ± 60	116 ± 51
IX	124 ± 40	126 ± 46
X	80 ± 18	90 ± 18
Antithrombin III, % of normal	92.5 ± 26.3	97.2 ± 22.6
Whole blood anti thrombin von Kaulla time, s' 15	49.1 ± 24.7	56.O ± 29.7

An analysis of variance was conducted to determine if differences existed among the variables measured for patients with and without thrombosis. Patients not receiving warfarin who thrombosed had partial thromboplastin times of 22.5 ± 3.0 seconds, compared with 28.0 ± 3.7 seconds for those patients without thrombosis (P < 0.01). For those patients who received warfarin and who clotted, the mean partial thromboplastin time was a 23.6 ± 2.2 seconds, compared with 27.2 ± 3.9 seconds for those receiving warfarin who did not thrombose (P < 0.01). The prothrombin time with the Dade thromboplastin FS reagent for those receiving warfarin and who thrombosed was 12.3 ± 0.9 seconds, compared with 14.7 ± 2.5 seconds for those who did not thrombose (P < 0.05). Prothrombin times using thromboplastin C and the Manchester Comparative Reagent; the AT III; the whole blood antithrombin time; the levels of factors II, VII, LX, and X; and the Heptest were equal for those without and with thrombosis.

During our study, the prothrombin times were measured on days as previously indicated in Methods. The prothrombin times were also measured at various times by hospital staff when they examined the patients. Figure 2 includes all prothrombin times measured for all study patients irrespective of the number of days they participated in the study. The prothrombin times of four severely anorectic patients transiently exceeded 15.0 seconds for 1 to 6 days. Vitamin K was given to each, their prothrombin times returned to normal ranges, and they resumed warfarin therapy.

Table 3. Causes for Withdrawal from Study after Randomization

---

Causes for Withdrawal	Patients Receiving Warfarin	Patients Not Receiving Warfarin
Death from cancer	12	14
Catheter not inserted after randomization	0	3
Allergy to contrast dye	0	2
Noncompliant with warfarin therapy	1	1
Acquired the lupus anticoagulant	1	0
Catheter in jugular vein	1	1
Acquired duodenal ulcer	1	0
Acquired aspirin requirement	1	0
Acquired mediastinal mass	1	0

Dilution of the thromboplastins to approximately double the prothrombin time values of normal plasmas did not bring out a difference between the treated and untreated patients. The control time of normal volunteers was 22.7 ± 3.8 seconds (n = 11) with Dade thromboplastin C. Using this system, the results for patients not receiving warfarin was 22.5 ± 5.2 seconds (n = 27) and 21.7 ± 3.7 seconds for patients receiving warfarin (n = 29). The dilute Dade FS thromboplastin yielded prothrombin times of 23.1 ± 4.6 seconds for normal controls (n = 10), 23.9 ± 5.5 seconds (n = 27) for patients not receiving treatment, and 23.3 ± 4.8 seconds (n = 29) for patients receiving treatment. Thus, diluting the thromboplastins failed to detect the influence of the 1-mg dose schedule.

Figure 2. Prothrombin times for all patients in our study. The upper panel represents patients not receiving warfarin; and the lower panel, those receiving warfarin.

Top

Discussion

In the early 1950s, Wright and colleagues ¹⁷ proposed that the therapeutic prothrombin time for warfarin-treated patients be between 2 and 2.5 times that of controls (untreated patients) for monitoring warfarin therapy. This recommended ratio, although based on a study of over 1000 post-myocardial infarct patients, did not rest on solid evidence; however, it remained the conventionally used target for over 2 decades. Over the past 20 years, however, clinical experience has established that the doses of warfarin required to achieve this therapeutic range are associated with bleeding complications. Furthermore, it has become generally appreciated that by lowering the therapeutic ratios to between 1.2 and 1.6, a lower incidence of untoward bleeding is produced and antithrombic protection is maintained. ^{2 11 12} Other studies ¹⁸ have shown that very low doses of warfarin reduce the levels of prothrombin fragments F_{1 + 2} in patients with histories of thrombosis.

Our trial confirmed that very low doses afford prophylactic antithrombotic effect under specific clinical circumstances. Our open randomized, prospective study showed that warfarin given at 1 mg per day reduces thrombosis associated with indwelling catheters in cancer patients receiving chemotherapy. We confirm the conclusion by Poller and colleagues ⁸ that a fixed minidose of warfarin has prophylactic effect against thrombosis. Although our treatment group was protected by this very low dose of warfarin, measurements of several laboratory variables failed to show differences between treatment and control groups. This was true for functional assays of the vitamin-K-dependent coagulation factors II, VII, IX, and X, as well as more global coagulation tests. At these doses, warfarin probably affects the in-vivo function of the vitamin-K-dependent materials in ways not detectable by these ex-vivo functional assays. For example, it will be interesting to measure the reduction of gammacarboxyglutaminic acid residues attached to the factor precursors (PIVKA molecules) as a more sensitive indicator of this warfarin effect. Studies by Molhotra and colleagues ¹⁹ showed that the loss of three of ten Gla moieties disrupts the function of prothrombin molecules. An anticoagulant effect independent of warfarin's vitamin-K antagonism by still unknown mechanisms could also explain the findings of this study. Poller and colleagues ⁸ did find a slight but significant prolongation of the prothrombin times, using the Manchester Comparative Reagent on the days after major gynecologic surgery. It was suggested that the postoperative accumulation of the fibrin degradation products possibly caused this minor change in the prothrombin time.

Our investigation's findings strongly suggest that patients should receive very-low-dose warfarin therapy when they have indwelling catheters. The likelihood that this prophylactic approach may also be effective in other patient populations remains a promising field for future study. Oral anticoagulants have been used to prevent deep venous thrombosis of legs and pulmonary embolism since 1959, when Sevitt and Gallagher ²⁰ used them successfully to reduce the incidence of pulmonary emboli after trauma. Further studies covering various surgeries and summarized by Berquist ²¹ show that oral agents are most effective if started preoperatively. When the drug is started postoperatively, clinical efficacy has not been shown ^21-26. Francis and colleagues ⁷ reported a prophylactic advantage for low-dose warfarin therapy after elective hip replacements. Drug therapy was begun 10 to 14 days preoperatively and the prothrombin time was kept between 1.5 and 3.0 seconds over that of controls. Postoperatively, the dose was increased to keep the value 1.5 times that of controls. ⁷

The same results occur in other forms of surgery. Davidson and colleagues ²⁷ reported that warfarin therapy, when begun 36 hours before surgery, was equal to dextran-70 in gynecologic patients. Taberner and colleagues ⁵ reported that oral anticoagulants started 5 days preoperatively for elective gynecologic surgery reduced the incidence of thrombosis from 23% in the reference group to 6% in the treated group, a rate equal to the incidence among the low-dose-heparin-treated group. Poller and colleagues ⁸ started the fixed minidose of 1 mg of warfarin 6 to 42 days (mean, 20) before gynecologic surgery. Among general surgery patients, van der Linde ²⁸ showed an advantage for warfarin when started preoperatively. This advantage is supported by a chest surgery series prepared by Storm. ²⁹

Whether the 3-day lead or the dose used in our study is optimal for all patients is not known. It may be necessary to increase the dose modestly to provide protection for all patients. Some patients may remain resistant to warfarin's prophylactic effect no matter the dose. Finally, because the assays we reviewed did not detect the changes created by very low doses of warfarin, finding a biologic target that would change in response to any effective dose would be preferable. Such assays to explore include monoclonal antibodies directed against des-gamma-carboxyglutamic prothrombins ^{30 31 32}, chromogenic prothrombin peptide assays ³³, differential effects of the snake venom from Echis carinatus on complete or incomplete prothrombin molecules, and urinary clearance of gammacarboxyglutamic acid concentrations ³⁴. Clearly, however, there is a protective activity associated with this very-low-dose warfarin therapy, and this advantage can exist without imposing a coagulopathy.

Acknowledgments: The authors thank P. Marshall, MD; K. Stokes, MD; D. Harris, MD; and M. Clouse, MD; for their evaluation of venograms; E.T. Yin, PhD, and R.M. Williams, PhD, for their measurements of Xa inhibitor and warfarin levels; and S. Meeks and R. Tagliamonte for measuring coagulation factors.

Grant Support: By Pharmacia-NuTech, Walpole, Massachusetts, and E.I. DuPont Company, Wilmington, Delaware.

Reprinted from Annals of Internal Medicine 1990;112:423428.

Top

References

1. Loeliger EA, van den Besselaar AM, Broedmans AW. Intensity of oral anticoagulation in patients monitored with various thromboplastins [Letter]. N Engl J Med. 1983;308:1228-9.
2. Poller L. Regulating the dosage of warfarin for anticoagulation [Letter]. N Engl J Med. 1987;316:1277-8.
3. Steinberg EP, Hughes RA. Safer use of anticoagulants in outpatients [Abstract]. Clin Res. 1982;30:240A.
4. Hull R, Hirsh J, Jay R. et al. Different intensities of oral anticoagulant therapy in the treatment of proximal-vein thrombosis. N Engl J Med. 1982;307:1676
5. Taberner DA, Poller L, Burslem RW, Jones JB. Oral anticoagulants controlled by the British comparative thromboplastin versus low-dose heparin in prophylaxis of deep vein thrombosis. Br Med J. 1978;1:272-4.
6. Loeliger EA. The optimal therapeutic range in oral anticoagulation. History and proposal. Thromb Haemost. 1979,42:1141-52.
7. Francis CW, Marder VJ, Evarts CM, Yaukoolbodi S. Two-step warfarin therapy. Prevention of postoperative venous thrombosis without excessive bleeding. JAMA. 1983;249:374-8.
8. Poller L, McKernan A, Thomson JM, Elstein M, Hirsch PJ, Jones JB. Fixed minidose warfarin: a new approach to prophylaxis against venous thrombosis after major surgery. Br Med J [Clin Res]. 1987;295:1309-12.
9. Bern MM, Bothe A Jr, Bistrian B, Champagne CD, Keane MS, Blackburn GL. Prophylaxis against central vein thrombosis with low-dose warfarin. Surgery. 1986;99:216-21.
10. Bern MM, Bothe A Jr, Bistrian BR, Batiste G, Haywood E, Blackburn GL. Effects of low-dose warfarin on antithrombin III levels in morbidly obese patients. Surgery. 1983;94:78-83.
11. Hirsh J. Is the dose of warfarin prescribed by American physicians unnecessarily high? Arch Intern Med. 1987;147:769-71.
12. Wessler S. What's the dose? Adv Exp Med Biol. 1987;214:149-52.
13. Lokich JJ, Becker B. Subclavian vein thrombosis in patients treated with infusion chemotherapy for advanced malignancy. Cancer. 1983;52:1586-9.
14. Bothe A Jr, Orr G, Bistrian BR, Blackburn GL. Home hyperalimentation. Compr Ther. 1979;5:54-61.
15. von Kaulla E, von Kaulla KN. Antithrombin 3 and diseases. Am J Clin Pathol. 1967;48:69-80.
16. Yin ET, Bygdeman S, Welin-Berger T, Tangen O. A new differential assay for plasma XaI (Antithrombin III) Biological activity. Predictor of postoperative DVT? In: Neri Serneri GG, Prentice CR, eds. Haemostasis and Thrombosis. New York: Academic Press, Inc.;1979:599.
17. Wright IS, Beck DF, Marple CD. Myocardial infarction and its treatment with anticoagulants. Lancet. 1954;1:92-7.
18. Bauer KA, Rosenberg RD. The pathophysiology of the prethrombotic state in humans: insights gained from studies using markers of hemostatic system activation. Blood. 1987;70:343-50.
19. Molhotra OP, Nesheim ME, Mann KG. The kinetics of activation of normal and gamma-carboxyglutamic acid-deficient prothrombins. J Biol Chem. 1985;260:279-87.
20. Sevitt S, Gallagher NG. Prevention of venous thrombosis and pulmonary embolism in injured patients: trial of anticoagulant prophylaxis in middle-aged and elderly patients with fractures of neck or femur. Lancet. 1959;2:981-9.
21. Bergquist D. Postoperative Thromboembolism: Frequency, Etiology, Prophylaxis. New York: Springer-Verlag: 1983.
22. Barber HM, Feil EJ, Galasko CS, et al. A comparative study of dextran-70, warfarin, and low-dose heparin for the prophylaxis of thrombo-embolism following total hip replacement. Postgrad Med J. 1977;53:130-3.
23. Hume M, Kuriakose TX, Zuch L, Turner RH. 1251 fibrinogen and the prevention of venous thrombosis. Arch Surg. 1973;107:803-6.
24. Lambie JM, Barber DC, Dhall DP, Matheson NA. Dextran 70 in prophylaxis of postoperative venous thrombosis. A controlled trial. Br Med J. 1970;2:144-5.
25. Baertschi U, Schaer A, Bader P, Huber L, Morf P. A comparison of low dose heparin and oral anticoagulants in the prevention of thrombo-phlebitis following gynaecological operations [author's transl]. Geburtshilfe Frauenheilkd. 1975;35:754-60.
26. van Vroonhoven TJ, van Zijl J. Muller H. Low-dose subcutaneous heparin versus oral anticoagulants in the prevention of postoperative deep-venous thrombosis. A controlled clinical trial. Lancet. 1974;1:375-8.
27. Davidson Al, Brunt ME, Matheson NA. A further trial comparing Dextran 70 with Warfarin in the prophylaxis of postoperative venous thrombosis. Br J Surg. 1972;59:314.
28. van der Linde DL. Postoperative Deep Vein Thrombosis [Dissertation]. The Netherlands: University of Nijenrode; 1975.
29. Storm O. Anticoagulant protection in surgery. Thromb Diath Haemorrh. 1958;2:484-91.
30. Owens J, Lewis RM, Cantor A, Furie BC, Furie B. Monoclonal antibodies against human abnormal (Des-gamma-carboxy) prothrombin specific for the calcium-free conformer of prothrombin. J Biol Chem. 1984;259:13800-5.
31. Furie B, Diuguid CF, Jacobs M, Diuguid DL, Furie BC. Randomized prospective trial comparing the native prothrombin antigen and the prothrombin time for monitoring oral anticoagulant therapy [Abstract 1373]. Blood. 1988;72(Suppl 1):366A.
32. Bovill E, Bhushan F, Mann K, Church W. Measurement of abnormal prothrombin and protein C in normal warfarin-treated individuals [Abstract 1362]. Blood 1988;72(Suppl 1):363A.
33. O'Donnell JR, Walker ID, Davidson JF. Control of oral anticoagulant treatment by chromogenic prothrombin assay. J Clin Pathol. 1987;40:5004.
34. Berger RG, Featherstone GL, Raasch RH, McCartney WH, Hadler NM. Treatment of calcinosis universalis with low-dose warfarin. Am J Med. 1987;83:72 6.