ORIGINAL INVESTIGATION
J Frequency of Major Hemorrhage in Patients Treated
l With Unfractionated Intravenous Heparin for Deep
Venous Thrombosis or Pulmonary Embolism
«2j A Study in Routine Clinical Practice
*« Majida Zidane; Miranda T. Schräm; Erwin W. Planken, MD, PhD; Wim H. Molendijk, MD, PhD; '6' Frits R. Rosendaal, MD, PhD; Felix ]. M. van der Meer, MD, PhD; Menno V. Huisman, MD, PhD
Background: The rate of major hemorrhage during the initial treatment with unfractionated heparin (UFH) in patients with deep venous thrombosis (DVT) and pul-monary embolism (PE) in routine clinical practice is un-derstuched. In recent clinical trials an Overall average of 3.8% was reported. However, the incidence of this com-plication in routine patient care might be higher owmg to less strict patient selection and lack of standardiza-tion in the administrastandardiza-tion of heparin. We have deter-mined major bleeding rates during heparin treatment for DVT or PE in routine practice and compared these rates with data from clinical trials.
Methode: Data on the occurrence of major hemor-rhage were retrieved according to strict critena from the records of patients who had received continuous intra-venous UFH therapy to treat objectively documented DVT or PE in 3 hospitals.
Results: After exclusion of 29 patients because of lack of objective diagnosis of DVT or PE and 25 patients be-cause of initial treatment with low-molecular-weight
hep-arin, 424 consecutive patients were available for de-tailed analysis. Among them, 17 patients (4.0%; 95% confidence interval, 2.1%-5.9%) experienced major hem-orrhage during UFH treatment, which in most patients occurred at the end of planned heparin therapy; one of the hemorrhages was fatal. Six patients (1.4%; 95% con-fidence interval, 0.3%-2.5%) developed clinically sus-pected recurrent venous thromboembolism (fatal in l case) during UFH treatment or within 7 days' cessation. Conclusions: Administration of continuous intrave-nous UFH in patients with DVT or PE in routine clini-cal practice leads to a major bleeding rate of 4.0%. This rate is comparable to the rate of major bleeding in pa-tients who received UFH in clinical trials. Our Undings are relevant to the discussion of major bleeding rates in patients with DVT and PE treated in daily clinical prac-tice with subcutaneous low-molecular-weight heparin and newer antithrombotic drugs.
Arch Intern Med. 2 000; 160:23 69-23 73
From the Department of General Internal Mediane, Leiden University Medical Center (Mss Zidane, Schräm, and Dr Huisman), the Departments of Internal Mediane, Diaconessenhuis Leiden (Dr Planken) and Rijnland Hospital, Leiderdorp (Dr Molendijk), and the Departments of Clinical Epidemiology (Dr Rosendaal) and Haematology
(Drs Rosendaal and van der Meer), Leiden University Medical Center, Leiden, the Netherlands.
T
HE MOST importantad-verse effect of initial intra-venous unfractionated hep-arin (UFH) therapy of deep venous thrombosis (DVT) and pulmonary embolism (PE) is major hemorrhage. In a large number of ran-domized clinical trials, the efficacy and safety of intravenous continuous UFH treatment has been compared with intra-venous intermittent UFH treatment,1"6 treatment with subcutaneously adminis-tered UFH,7"14 and low-molecular-weight-heparin (LMWH).15 The Overall average of major hemorrhage rates in the continu-ous intravencontinu-ous UFH groups in these stud-ies was 3.8% (95% confidence interval [CI] 3.0%-4.6%) (widely varying between the studies from 0% to 20%).
However, in spite of the widespread ap-plication of UFH therapy, surprisingly little Information is available on the incidence
of bleeding complications in routine daily practice during the period that patients re-ceive heparin administered in therapeutic doses. Moreover, it is unknown whether the bleeding rates found in clinical trials are ap-plicable in routine clinical practice. The con-ditions of routine practice could affect the complication rate of heparin therapy cause of selection criteria in trials and be-cause the administration of heparin is of-ten not standardized in routine practices. The current study was performed to accu-rately estimate the rate of major bleeding complications in patients treated with UFH in routine clinical practice for a thrombo-embolic event and to compare these data with those derived from clinical trials.
RESULTS
hos-PATIENTS AND METHODS
STUDY POPULATIONAll patients with a diagnosis of DVT or PE who had been hospitalized in 1995 and 1996 in the wards of internal medi-cine at the Academic University Medical Center in Leiden, at the Diaconessenhuis Leiden, or at the Rijnland Hospital in Leiderdorp, the Netherlands, were eligible. These hos-pitals provide immediate care for the whole region. The in-clusion criterion was the use of unfractionated heparin for a least 2 days. Patients who had had surgery were in-cluded only if treatment for thrombosis had started at least 14 days after the day of surgery. The only exclusion crite-rion was initial treatment with LMWH. The medical ethi-cal committees of all 3 participating hospitals approved the study.
TREATMENT REGIMENS
At the day of diagnosis, all patients received an intrave-nous bolus of 70 U/kg of body weight of UFH followed by a continuous infusion of UFH. The initial dose was 400 to 450 U/kg every 24 hours, resulting in a bolus of 5000 U and an initial dose of 35000 U per 24 hours. The first ac-tivated partial thromboplastin time (APTT) was per-formed 4 to 6 hours after the bolus injection was given. The heparin dose was then adjusted to maintain the APTT at the therapeutic ränge of 1.5 to 2.0 times the control APTT in 2 of the 3 hospitals. In the third hospital, a ränge of 60 to 90 seconds was targeted, with a normal ränge of 25 to 35 seconds. Nomograms for heparin treatment were not used in any of the 3 hospitals. Oral coumarin treatment was begun on the first day of heparin treatment. The coumarin dose was adjusted to maintain the international normal-ized ratio (INR) between 2.5 and 3.5, according to na-tional consensus of the Dutch Trombosis Services, The Hague, the Netherlands.
DATA COLLECTION
Data on baseline comorbidity, bleeding events that oc-curred during heparin treatment (includmg severity and treatment and recurrent thromboembolic events), and cause of death where applicable were obtained by review of medi-cal records and by means of a Standard questionnaire. In addition, the records of the blood transfusion Services of all 3 hospitals were checked for every patient to see if trans-fusion had occurred during hospitalization.
The APTT values, INR, and administered doses of hep-arin were registered for all patients who met the criteria for major hemorrhage. A random sample (n = 50) of
patients in whom major hemorrhage did not occur was taken, and the administered doses of heparin were regis-tered for comparison with those of patients with a major hemorrhage.
In addition, APTT values were reviewed in the first 24 hours of heparin treatment äs well äs prior to the fatal event for those patients who experienced fatal hemor-rhage or fatal pulmonary embolism. All patients had rou-tine follow-up visits at the outpatient department after 3 months, during which a decision with respect to continu-ing anticoagulant treatment was made.
OUTCOME EVENTS
Prior to the start of the study, bleeding was defined äs ma-jor if (1) it was associated with either a decrease in the he-moglobin level of at least 20 g/L or a transfusion was done of 2 or more units of blood, (2) the bleeding was intraperi-toneal intracranial, or gastrointestinal, or (3) cessation of treatment was warranted, or it had led to death. Only ma-jor hemorrhagic events during initial UFH treatment or within 24 hours of stopping heparin treatment were con-sidered a complication of the heparin treatment. Objec-tively demonstrated thromboembolism was defined by ab-normal findings in compression ultrasonography or contrast venography for DVT and high probability perfusion ven-tilation lung scanning or by abnormal findings on pulmo-nary angiography.
Recurrent thromboembolic events were defined äs ob-jective evidence of recurrent DVT äs demonstrated by find-ings of repeated contrast venography or compression ul-trasonography; (recurrent) pulmonary embolism was considered to have occurred if demonstrated by results of repeated perfusion lung scanning or repeated pulmonary angiography during heparin therapy or within l week af-ter stopping treatment. Any APTT values prolonged 1.5 to 2.0 times the upper limit of the control value were consid-ered adequate prolongations. Data on causes of death in patients who died during hospitalization were retrieved by reviewing the patients' medical records and, when avail-able, autopsy reports.
STATISTICAL ANALYSIS
As effect measure, the frequency of the outcome events among the patients was calculated, ie, the proportion of pa-tients with an outcome event. This may be viewed äs a cu-mulative incidence, ie, the probability of an outcome event. Since the duration of treatment was short and fairly ho-mogeneously distributed, the duration of treatment was not taken into account. The 95% CIs were calculated with the normal approximation to binomial distribution.
pitals of the region. Fifty-four patients were not eligible for the analysis: 29 who did not receive UFH treatment because the diagnosis of DVT or PE could not be con-firmed with objective methods, and 25 who initially received LMWH therapy. Thus 424 consecutive patients (Leiden University Medical Center, n=117, Diaconessenhuis, n=113, Rijnland Hospital, n=194) were included. The patients' baseline characteristics are given in Table l. These patient characteristics
compare well with those observed in recent clinical trials.16'19
BLEEDING COMPLICATIONS
Major hemorrhage during initial heparin treatment or within 24 hours of stopping the treatment occurred in 17 patients (4.0%; 95% CI, 2.1%-5.9%); in l patient it was fatal. Seven major bleeding complications occurred
In in th tu re di th U Pl
ARCH INTERN MED/VOL 160, ÄUG 14/28, 2000
2370
in patients with the diagnosis of DVT (3.1%) whüe 10 major hemorrhages occurred in patients with PE (6.2%). Major bleeding complications are given in Table 2.
The mean (SD) age of the patients who experi-enced a major bleeding complication (63.1 [21.1] years) did not significantly differ from the mean age of pa-tients without a major bleeding complication (61.5 [ 19.3] years). In 5 of the 76 patients with disseminated malig-nancy, major hemorrhage occurred (relative risk, 1.91; 95% CI, 0.65-5.61).
Most major hemorrhages occurred near the end of the planned initial heparin treatment, when the INR was close to or in the desired ränge. In l patient (listed fifth in Table 2), the results of compression ultrasonography showed DVT, and anticoagulant treatment was planned. However, this patient was first operated on to treat a para-vertebral abcess. Heparin treatment was started 2 days after the uneventful Operation.
Five patients (1.2%) experienced a major bleeding complication during hospitalization but after UFH treat-ment had been stopped. In 3 of these patients, the bleed-ing complication was fatal. Minor hemorrhage occurred in 43 patients (10%).
RECURRENT VENOUS THROMBOEMBOLISM Recurrent venous thromboembolism (VTE) within 12 weeks of UFH treatment occurred in 14 patients (3.3%; 95% CI, 1.6%-5.0). During heparin therapy or within l week of stopping therapy, 6 patients (1.4%; 95% CI, 0.3%-2.5%) had clinically suspected recurrent VTE, which was fatal in l patient. Abnormal compression ultrasonogra-phy findings confirmed the diagnosis of DVT in l pa-tient who had been treated for DVT and in l papa-tient who had been originally treated because of PE. In the other 4 patients, the diagnosis was not confirmed by objective testing.
After l week but within 12 weeks of cessation of ini-tial heparin treatment, 8 patients (1.9%; 95% CI, 0.6%-3.2) experienced recurrent VTE. In l patient fatal PE oc-curred. Objective tests confirmed recurrent VTE in 6 of the 8 patients.
DEATHS
Deaths during hospitalization occurred in 19 patients (4.5%; 95% CI, 2.5%-6.5%). Causes of death and the time of occurrence are given in Table 3.
APTT PROLONGATION AND HEPARIN DOSES
In 8 of the 17 patients who had a major hemorrhage dur-ing UFH treatment, the APTT Prolongation was supra-therapeutic within 24 hours of the bleeding complica-tion (Table 2). None of the patients who experienced a recurrent VTE during initial UFH treatment or within 7 days after this treatment had been stopped had a sub-therapeutic APTT Prolongation in the first 24 hours of UFH treatment. Of the 3 patients who died, possibly of PE, during heparin treatment, only l had a subtherapeu-tic APTT Prolongation in the first 48 hours of initial UFH
Table 1 . Baseline Characteristics of 424 Patients With Deep Venous Thrombosis or Pulmonary Embolism* Baseline Characteristic
Mean (SD) age, y M/F
Diagnosis at presentation Deep venous thrombosis Pulmonary embolism Venous thromboembohsmt IWean (SD) days of initial
unfractionated heparin treatment Mean (ränge) days of hospitalization Surgery (within <1 mo)
Previous venous thromboembolism Stroke
Chromc heartfailure
Chronic obstructive pulmonary disease Diabetes meliitus Malignancy Value 61.6(19.3) 187(44)/237(56) 227 (54) 162(38) 35 (8.3) 6.3 (2.4) 16(1-270) 19(4.5) 86 (20) 28 (6.6) 25 (5.9) 23 (5.4) 35 (8.3) 76(18)
^Unless otherwise mdicated, data are number (percentage). "[Deep venous thrombosis and pulmonary embolism
treatment (Table 4). Finally, the mean amount of UFH given did not differ significantly between patients with major bleeding complications (28764 U/d; n=17) and the randomly selected patients (29819 U/d; n=50).
COMMENT
In this large retrospective study of 424 patients treated for VTE with UFH in routme daily practice, 17 patients (4.0%; 95% CI, 2.1%-5.9%) experienced a major bleed-ing event durbleed-ing initial treatment. One patient experi-enced fatal major hemorrhage. The Overall average re-ported in clinical trials was 3.8%.1"15 This rate is
comparable to our observed rate of 4.0%. Our prior ex-pectation was that the rate of major hemorrhage in daily practice would be higher than in controlled studies. In all the trials evaluating treatment with LMWH, patients with a high risk of bleeding (eg, those with active ma-lignancies or kidney or liver function diseases) were ex-cluded.15 Specific nomograms for UFH treatment
regi-mens were often implemented in these patients.20
By assessing the APTT Prolongation the day before a major bleeding complication or the suspected throm-boembolic event, we checked whether patients received adequate heparin doses. Eight of the 17 patients (47%) who experienced a major hemorrhage during UFH treat-ment had a supratherapeutic APTT Prolongation within 24 hours of their hemorrhagic complication, which is in agreement with other studies.21 None of the 6 patients
who experienced a recurrent VTE during UFH treat-ment or within 7 days of treattreat-ment cessation had sub-therapeutic APTT Prolongation, (<1.5 X control), which is in disagreement with studies that have found a clear association between subtherapeutic APTT results at 24 to 48 hours and recurrent VTE.8·22'24
Tabie 2. Major Hemorrhage During Initial Unfractionated Heparin (UFH) Treatment of 17 Patients* Site of Bleedmg Intracranial Retroperitaneal Retropentoneal Retropentoneal hematoma Retropentoneal wound hematoma Retropentoneal and gastrointestmal Gastromtestinal Gastrointestmal Gastrointestmal Gastrointestmal Genitounnary Genitounnary Hematoma left hip
to left knee Hematoma left hip Abdominal wall
hematoma Hematoma left thigh Muscle hematoma
Time After Start of Predisposing Factor Initial Heparin Therapy, d Clippmg of aneurysm artery cerebn
antenor 47 days before UFH Pacemaker Implantation
25 days before UFH Orchidectomy 113 days
before UFH Gynecological Operation
20 days before UFH Abdominal Operation
2 days before tfFH History of peptic ulcer Dilatation therapy of the esophagus Ouodenal ulcer None Catheter Transurethral catheter Pyelonephntis
Trauma duririg hospiMh/ation' dc-criidscd hver fupclion Tumor ilf'bulkmg 18ria<'s
hefore UFH lactor V annbodies None Catheter None 6 7 7 6 7 2 3 4 7 5 2 4 g 5 5 4 4
APTT Prolongation Criteria for Ratiof Major Hemorrhage
203 190 207 340 1 70 275 177 1 23 185 150 600 165 180 388 400 189 563 Neurosurgical Operation 2 U of erythrocytes admmistered 10 U of erythrocytes admimstered 4 U of erythrocytes admimsteredj 7 U of erythrocytes admmistered Treatment cessation and 6 U of erythrocytes admmistered Oral anticoagulant treatment cessation Hb drop >20 g/L Hb drop >20 g/L 2 U of erythrocytes admmistered 3 U of erythrocytes admmistered Hb drop >20 g/L Hb drop >20 g/L 3 U of erythocytes admmistered Hb drop >20 g/L *APTTmdicates activated partial thrombob/ast/n t/me ellipses not applicable andHb hemog/obm
^APTT on the day before ma/or hemorrhage occurred divided by the upper limit of the control APTT, APTT ratlos greater than 2 0 indicate APTTprolongations exceeömg the therapeut/c ränge
$Fatal b/eeding compl/cat/on
Table 3. Causes of Death and Times of Occurrence in 19 Patients*
No. of Patients Cause of Death Who Died Durmg UFH treatment
Major hemorrhage 1 Pulmonaryembolism 3 Dissemmated malignancy 1 Liver abcess 1 Unknown 1 After UFH treatment
Major hemorrhage 2 Pulmonary embolism 2 Dissemmated malignancy 3 Othert 2 Unknown 3 Oay of Occurrence Durmg Hospitahzation 6 5,t6,7 3 10 6 8,36 10,32 6,16,29 6,19 27 32, 34
Table 4. APTT Prolongations in 3 Patients With Fatal Bleeding Complications and 5 With Pulmonary Embolism*
Cause of Death Major hemorrhage Major hemorrhage Major hemorrhage Pulmonary embolism Pulmonary embolism Pulmonary embolism Pulmonary embolism Pulmonary embolism
No. of Days APTT After Stopping Prolongation Initial Heparin Ratio Durmg Treatment First 24 hf 18 10 2 18 4 35 24 >6 4 >6 0 21 0 58 0 11φ APTT Prolongation Ratio 24 h Before Death ND 28 18 ND ND 14 27 20
*UFH mdicates unfractionated heparm ^Autopsy-proven pulmonary embolism
$0ther severe comorbid/ty pneumoma (day 6) and progressive renal insuffic/ency (day 19)
* APTTmdicates activated partial thrombob/astm time ND, not determmed because unfractionated heparm treatment had a/ready been stopped
•\APTT divided by the upper limit of the control APTT APPTprolongations greater than 15are adequate therapeutic prolongations
$APTT ratio withm the first 48 hours of unfractionated heparm treatment
records of all anticoagulated patients m the Leiden re-gion with attached mdications for anticoagulation
A study design such äs ours may result in
underre-porting or mconsistent reunderre-porting of bleeding events We
have tried to obviate this by lookmg only at major orrhages and by predefining the cntena for major hem-orrhage before studymg the patients' records We also checked patient charts and records of the transfusion
ser-ARCH INTERN MED/VOL 160 ÄUG 14/28 2000
2372
vices of all 3 hospitals. It seems unlikely, using this method, that we would have missed any rnajor hemor-rhage. The consistency of reporting major bleeding com-plications was assessed in a recent study.25
Another potential weakness of our study is related to the reporting of recurrent VTEs. To obtain a true estimate of the rate of recurrent VTE, we predefined the criteria for objective recurrent VTE, and we evaluated the records of every patient for a clinical suspicion of recurrence äs well äs whether any objective test had been performed to as-certain the diagnosis. All of these precautions strengthen the generalizability of our results to other hospitals.
Our data on bleeding rates during UFH therapy may form the background against which the complication of major bleeding during subculaneous LMWH therapy in patients with DVT and PE can be evaluated in general Community hospitals in daily clinical practice. Such stud-ies of daily practice are needed before the use of LMWH is routinely applied in an out-of-hospital setting.
Acceptedfor publication March 30, 2000.
We acknowledge M. H. Prins, MD, PhD, for his assis-tance in the preparation of this article.
Reprints: Menno V. Huisman, MD, PhD, Department of General Internal Mediane, Room B3-Q84, Leiden Uni-versity Medical Center, PO Box 9600,2300 RC, Leiden, the Netherlands (e-mail: m.v.huisman@lumc.nl).
REFERENCES
1. Salzman EW, Deykin D, Shapiro RM, et al. Management of heparm therapy. NEngl
J Med. 1975:292:1046-1050.
2. Glazier RL, Corwell EB. Randomized prospective trial of continuous versus in-termittent heparin therapy. JAMA. 1976:236:1365-1367.
3. Mant MJ, O'Brien BD, Thong KL, et al. Haemorrhagic complications of heparin therapy. Lancet. 1977;1:1133-1135.
4. Wilson JR, Lampman J. Heparin therapy: a randomized prospective study. Am HeartJ. 1979;97:155-158.
5. Fagher B, Lundh B. Heparin treatment of deep vein thrombosis. Acta MedScand. 1981,210:357-361.
6. Wilson JE, Bynum LJ, Parkey RW. Heparin therapy in venous thromboembo-lism. Am J Med. 1981 ;70:808-816.
7. Bentley PG, Kakkar W, Scully MF, et al. An objective study of alternative meth-ods of heparin administration. Thromb fies. 1980:18:177-187
8. Hüll RD, Raskob GE, Hirsh J, et al. Continuous intravenous heparin compared with intermittent subcutaneous heparin in the initial treatment of proximal-vein thrombosis. N Engl J Med. 1986:315:1109-1114.
9. Doyle DJ, Turpie AG, Hirsh J, et al. Adjusted subcutaneous heparin or continu-ous intravencontinu-ous heparin in patients with acute deep vein thrombosis: a random-ized trial. Ann Intern Med. 1987;107:441-445.
10. Walker MG, Shaw JW, Thomson GJ, Cummmg JG, Thomas ML. Subcutaneous calcium heparin versus intravenous sodium heparin in treatment of established acute deep vein thrombosis of the legs: a multicentre prospective randomised trial. BMJ. 1987:294:1189-1192.
11. Lopaciuk S, Meissner AJ, Ciesielski L, Korzycki J. Subcutaneous sodium hepa-rin versus intravenous sodium hepahepa-rin in the treatment of deep vein thrombo-sis. In: Proceedingsof the 6th International Meeting of the Danubian League Against Thrombosis and Haemorrhagic Disorders; May 31 to June 3,1989: Vienna, Aus-tria
12. Pini M, Pattacmi C, Quintavalla R, et al. Subcutaneous versus intravenous hep-arin in the treatment of deep venous thrombosis· a randomized controlled trial.
Thromb Haemost 1990:64:222-226.
13. Hommes DW, Bura A, Mazzaloi L, et al. Subcutaneous heparin compared with continuous intravenous heparin administration m the initial treatment of deep venous thrombosis: a meta-analysis. Ann Intern Med. 1992:116:279-284. 14. Levme MIM, Raskob GE, Landefeld S, Hirsh J. Hemorrhagic complications of
an-ticoagulant treatment. Cftesf. 1995,108:277-290.
15. Leizorovicz A. Comparison of the efficacy and safety of Iow molecular weight heparms and unfractionated heparin m the initial treatment of deep venous throm-bosis: an updated meta-analysis. Drugs. 1996;52(suppl 7):31-37.
16. Levine M, Gent M, Hirsh J, et al. A companson of Iow-molecular-weight hepa-rin administered primanly at hörne with unfractionated hepahepa-rin admimstered m the hospital for proximal deep-vem thrombosis. W Engl J Med. 1996:334: 677-681.
17. Koopman MMW, Prandoni P, Piovella F, et al. Treatment of venous thrombosis with intravenous unfractionated heparin administered m the hospital äs com-pared with subcutaneous Iow-molecular-weight heparin administered at home
NEnglJMed 1996:334:682-687.
18. Simonneau G, Sors H, Charbonnier B, et al. A comparison of Iow-molecular-weight heparin with unfractionated heparin for acute pulmonary embolism N Engl
JMed. 1997:337:663-669.
19. The Columbus Investigators. Low-molecular-weight heparin m the treatment of patients with venous thromboembolism. W Engl J Med. 1997:337:657-662. 20. de Groot MR, Böller HR, ten Cate JW, van Marwijk Kooy M. Use of a heparin
nomogram for treatment of patients with venous thromboembolism in a com-munity hospital. Thromb Haemost. 1998:80:70-73.
21. Levine MN, Raskob, GE, Landefeld S, Kearon C. Hemorraghic complications of anticoagulant treatment. Cftesf. 1998;114:5115-5235.
22. Hüll RD, Raskob GE, Brant FR. Relation between the time to achieve the Iower limit of the APTT therapeutic ränge and recurrent venous thromboembolism dur-ing heparin treatment for deep-vein thrombosis. Arch Intern Med. 1997:157: 2562-2568.
23. Hüll RD, Raskob GE, Brant RF, et al. The importance of initial heparin treatment on long-term clinical outcomes of antithrombotic therapy: the emerging theme of delayed recurrence. Arch Intern Med. 1997:157:2317-2321.
24. Brandts DPM, Hei|boer H, Büller HR, et al. Acenocoumarol and heparin com-pared with acenocoumarol alone in the initial treatment of proximal vein throm-bosis. N Engl J Med. 1992:327:1485-1489.
25. Graafsma YP, Prins MH, Lensmg AWA, de Haan RJ, Huisman MV, Büller HR. Bleeding classification in clinical trials: observer vanability and clinical rel-evance. Thromb Haemost. 1997:78:1189-1192.
