Risk of Hemorrhage in Patients with Chronic Liver

Disease and Coagulopathy Receiving Pharmacologic

Venous Thromboembolism Prophylaxis

Jacob A. Reichert,* Peter F. Hlavinka, and Jill C. Stolzfus

Pharmacy Department, St. Luke’s University Health Network, Bethlehem, Pennsylvania

STUDY OBJECTIVE To evaluate the impact of pharmacologic venous thromboembolism (VTE) prophylaxison risk of hemorrhage while hospitalized in patients with chronic liver disease (CLD) and concurrentcoagulopathy.

DESIGN Retrospective, multicenter chart review.SETTING Five hospital, university-affiliated network in eastern Pennsylvania.PATIENTS Patients admitted to the network from January 1, 2012, until December 31, 2012, with

ICD-9 code consistent with CLD and elevated international normalized ratio of 1.5 or greater notsecondary to anticoagulation.

MEASUREMENTS AND MAIN RESULTS Two hundred fifty-six patients met criteria for analysis, with 80 havingreceived pharmacologic VTE prophylaxis and 176 having received no pharmacologic VTE prophylaxis.Differences were observed in the primary outcome of overall hemorrhage (composite of major andminor hemorrhage) for patients receiving VTE prophylaxis versus no VTE prophylaxis (17.5% vs 7.4%,p=0.02). Logistic regression revealed covariates independently associated with increased hemorrhagerisk were pharmacologic VTE prophylaxis use (adjusted odds ratio [AOR] 3.64, p=0.004), increasinginternational normalized ratio (AOR 1.31, p=0.007), and decreasing platelet count (AOR 0.99, p=0.03).

CONCLUSIONS Patients with CLD and concurrent coagulopathy receiving pharmacologic VTE prophy-laxis are at an increased risk of overall hemorrhage. Clinical implications remain unclear secondaryto the difference in rate of overall hemorrhage being driven primarily by a difference in minor hem-orrhage. In addition, no difference was demonstrated in many assessed clinically relevant markers.

KEY WORDS chronic liver disease, cirrhosis, venous thromboembolism, coagulopathy, low-molecularweight heparin, fondaparinux, enoxaparin, heparin, anticoagulant, hemorrhage.(Pharmacotherapy 2014;34(10):1043–1049) doi: 10.1002/phar.1464

Chronic liver disease (CLD) is a major illnessin the United States, responsible for significantmorbidity and mortality, with 31,903 reported

deaths and 101,000 discharges reporting CLD orcirrhosis as the primary diagnosis in 2010.1, 2

During these hospitalizations, associated coagul-opathies often present complex managementconsiderations for practitioners. Specifically, bal-ancing the risks of hemorrhage and thrombosiswhile considering pharmacologic prophylaxisagainst development of venous thromboembo-lism (VTE) remains a challenge.A large body of evidence from randomized,

controlled trials demonstrates the effectivenessof VTE prophylaxis in surgical and medicalpatients. However, these trials have frequentlyexcluded patients with an increased risk of

Financial support: None.Prior presentations: Eastern States Pharmacy Conference,

Hershey, PA, May 13–16, 2013; Mid-Eastern PennsylvaniaSociety of Health System Pharmacists, Night of the Residents,Lehigh Valley Hospital and Health Network, Allentown, PA,May 30, 2013; and St. Luke’s University Health NetworkResearch Symposium, Bethlehem, PA, June 13, 2013.

*Address for correspondence: Jacob A. Reichert,2512 Bethlehem Fields Way, Bethlehem, PA 18015;e-mail: jacob.reichert@gmail.com.� 2014 Pharmacotherapy Publications, Inc.

bleeding, especially those with CLD and/or coag-ulopathy, often defined as elevated internationalnormalized ratio (INR) and thrombocytopenia.3–5

Making the picture more complex are the evi-dence and theories supporting both enhancedand attenuated response to pharmacologic anti-coagulation.Earlier trials examining the hemostatic effects

of CLD have demonstrated impaired synthesis ofmany hemostatic components, including procoag-ulants and anticoagulants, such as factors II, V,VII, IX, X, and XI, antithrombin (AT), proteins Cand S, and platelet production.6–8 Because of thesechanges, INR, activated partial thromboplastintime (aPTT), and other clinical assessment mea-sures are altered leaving practitioners unable toaccurately assess response to pharmacologic anti-coagulation.The risk for both VTE and hemorrhage must

be considered when assessing the appropriate-ness of pharmacologic VTE prophylaxis. Alteredhemostasis, supported by surrogate laboratorymarkers (prolonged aPTT and INR) may suggestprotection from thrombosis and increased riskof hemorrhaging. Epidemiologic data demon-strate serious liver disease as an independentfactor associated with a decreased risk for devel-opment of VTE.9 In addition, hepatic failure andplatelet count less than 50 9 109 have beenidentified as independent risk factors for bleed-ing.10 Therefore, these findings provide plausi-bility for the theory of “autoanticoagulation” inthis patient population.9, 11

However, other data have questioned thenotion of “autoanticoagulation,” by paradoxi-cally demonstrating these patients may be at anincreased or similar risk for VTE developmentas hospitalized patients without CLD.12 Researchspecifically addressing the question of pharma-cologic VTE prophylaxis in patients with CLDhas demonstrated VTE risk similar to the generalpatient population. Additionally, this similar riskof VTE for patients with CLD has been found tocontribute to inpatient morbidity and mortal-ity.11–17 As a result of these findings, some prac-titioners recommend that patients withcoagulopathic CLD be considered for pharmaco-logic VTE prophylaxis regardless of elevatedINR.11, 15, 16, 18

Besides the inherent increased hemorrhageand VTE risk in patients with CLD, an impor-tant consideration is the possible alteredresponse to parenteral anticoagulation. Publishedtheories and evidence support both an enhancedand an attenuated response to parenteral antico-

agulation.19, 20 Enhanced response to parenteralanticoagulants has been demonstrated throughlower-than-expected antifactor Xa levels follow-ing treatment with standard dosing of low-molec-ular-weight heparins (LMWH). Confusing thisresults of lower than expected antifactor Xalevels is decreased thrombin generation in vitrofollowing the addition of LMWH to serum sam-ples, despite decreased serum AT and antifactorXa levels, suggesting an attenuated response toparenteral anticoagulants.19, 20

Because of the paucity of clinical data sup-porting the utility of pharmacologic VTE pro-phylaxis for hospitalized patients with CLD andthe competing theories regarding its appropriate-ness, further information is necessary to deter-mine its role in therapy. The objective of thisstudy was to further explore the safety profileof pharmacologic VTE prophylaxis, includinghemorrhagic risk, in patients with CLD andconcurrent coagulopathy.

Methods and Procedures

This study was conducted at five hospitals ineastern Pennsylvania. Eligible patient admissionswere identified through use of primary or second-ary International Classification of Diseases, NinthRevision (ICD-9) codes consistent with CLD(571.X, 070.22, 070.23, 070.32, 070.23, 070.44,070.54), as well as INR of 1.5 or greater duringhospitalization not secondary to systemic antico-agulation. Data were collected from January 1,2012, to December 31, 2012. Data from patientswith multiple hospital admissions were permitted.Exclusion criteria at admission were thrombosis,hemorrhage, or systemic anticoagulation. Thestudy was exempt from institutional review boardapproval, as it was a retrospective chart review,per local institutional review board policy.

Study Design and Data Collection

The primary outcome of this exploratory studywas incidence of overall hemorrhage during hos-pitalization as defined by Thrombosis In Myocar-dial Infarction (TIMI) definitions based onreceipt of VTE prophylaxis.21 Major hemorrhagewas defined as a decrease in hemoglobin ofgreater than 5.0 g/dl within 10 days or any intra-cranial hemorrhage. Minor hemorrhage wasdefined as a decrease in hemoglobin 3–5 g/dlwith witnessed bleeding or greater than 4 g/dlwithout witnessed bleeding, both within 10 days.Overall hemorrhage was a composite endpoint

1044 PHARMACOTHERAPY Volume 34, Number 10, 2014

including both major and minor hemorrhage.Hemorrhages were identified by maximum andminimum hemoglobin values across any 10 daysduring hospitalization, as well as any witnessedhemorrhage during hospitalization not related tosurgical procedures. Secondary outcomes includeddifferences between groups in overall mortality,hemorrhage-associated mortality (mortality forwhich hemorrhage was listed as a contributingfactor in the patient’s death summary), incidenceof packed red blood cell (PRBC) transfusion,number of PRBC units transfused, and rate ofthrombosis.Baseline demographic, laboratory, and admis-

sion data were collected through automated med-ical record review and confirmed throughmanual chart review. Data collected includedhospitalization duration, admission date anddiagnoses, admission laboratory values (INR,serum bilirubin, and serum creatinine), CLDdiagnoses, thrombosis diagnoses, and inpatientmortality. Thrombosis diagnoses included pulmo-nary embolism, deep vein thrombosis, portal veinthrombosis, or other venous embolism or throm-bosis. The Model for End-Stage Liver Disease(MELD) score was calculated using admissionINR, admission serum bilirubin, and admissionserum creatinine. Admission records werereviewed for active hemorrhage, current throm-bosis, concurrent anticoagulant use, and antiplat-elet use. Hospitalization records were reviewedfor pharmacologic (including unfractionated hep-arin, enoxaparin, and fondaparinux at appropri-ate VTE prophylaxis doses) or mechanical VTEprophylaxis, minimum platelet count, maximumINR, and total number of PRBC units transfused.For outcomes assessment, patient data were strat-ified as receipt of VTE prophylaxis for at least24 hours versus all other patients.

Statistical Analysis

For the primary outcome of overall bleedingand all secondary outcomes, separate v2 orFisher exact tests for the categorical variableswere performed. In addition, a direct logisticregression model was created to determine theeffect of pharmacologic VTE prophylaxis onhemorrhage, with 28 events sufficient to supportfour model covariates.22 The model adjusted forplatelet count, maximum INR, and antiplateletmedication. The continuous covariates of plate-let count and maximum INR both met theassumption of linearity, with no serious multi-collinearity based on variance inflation factors

and condition index values. Adjusted odds ratios(AORs) and 95% confidence intervals (CIs) arereported, with model robustness assessed viathe Hosmer–Lemeshow goodness-of-fit statistic.Baseline demographic and patient characteristicsdata were compared using separate independentsamples t tests for normally distributed continu-ous variables, Mann–Whitney rank sum test forskewed continuous variables, and v2 or Fisherexact tests for categorical variables. Mean valuesare presented with standard deviations (SDs)and medians with interquartile range (IQR).For all analyses, p<0.05 denotes statistical signif-icance, with no adjustment for the multiplecomparisons.

Results

Patients

A total of 440 patients fulfilled initial criteriafor analysis. Medical records for the identifiedpatients were analyzed to further assess appropri-ateness regarding exclusion criteria. Of the 440initially identified patients, 184 were subsequentlyexcluded from analysis for either pharmacologicanticoagulation, hemorrhage, or thrombosis onadmission. The remaining 256 patients were fur-ther stratified by receipt of pharmacologic VTEprophylaxis (Figure 1).Patient demographic features and baseline

characteristics are shown in Table 1. Key differ-ences between groups were age (61.0 versus54.4 years, p<0.0001), antiplatelet use on admis-sion (17.5% versus 6.8%, p=0.009), durationof hospitalization (178 hours versus 91 hours,p<0.0001), MELD score (18.7 versus 20.3,p=0.04), and minimum platelet count (80 versus61, p=0.001).

Outcomes

The primary outcome, overall hemorrhage,and stratification by severity of hemorrhage aresummarized in Table 2. Patients treated with anypharmacologic VTE prophylaxis were more likelyto experience a hemorrhage (17.5%) than werepatients not receiving VTE prophylaxis (7.4%,p=0.02). When stratified by severity of hemor-rhage, pharmacologic VTE prophylaxis was asso-ciated with a statistically greater increase inminor hemorrhage (12.5%) than no prophylaxis(4.5%, p=0.02). In contrast, rates of major hem-orrhage were not significantly different betweengroups (5.0% vs 2.8%, respectively, p=0.47).

VTE PROPHYLAXIS IN CLD SETTING Reichert et al 1045

The logistic regression model output demon-strated good overall fit (Hosmer–Lemeshowp=0.64) (Table 3). After adjusting for antiplateletmedication, maximum INR, and platelet count,pharmacologic VTE prophylaxis independentlypredicted overall bleed risk (AOR 3.64, 95% CI1.50–8.86, p=0.004). Maximum INR indepen-dently predicted higher overall hemorrhage risk(AOR 1.31, 95% CI 1.08–1.60, p=0.007), andplatelet count independently predicted loweroverall hemorrhage risk (AOR 0.99, 95% CI

Figure 1. Flow diagram of patients.

Table 1. Baseline Demographics and Characteristics

Variable VTE Prophylaxis (n=80) No VTE Prophylaxis (n=176) p-Value

Age (yrs), mean � SD 61.00 � 11.74 54.43 � 13.82 < 0.0001Gender (male), n (%) 47 (58.8) 102 (58) 0.91Race (white), n (%) 62 (77.5) 120 (68.2) 0.13Height (inches), median (IQR) 68 (63–70) 67 (64–69) 0.77Weight (kg), median (IQR) 87.65 (70.08–100.50) 82.80 (70.10–97.65) 0.61Body mass index, median (IQR) 30.39 (25.04–33.50) 29.60 (24.42–33.66) 0.68Antiplatelet on admission, n (%) 14 (17.5) 12 (6.8) 0.009Aspirin 81 mg, n (%) 9 (11.3) 11 (6.3) –Aspirin 325 mg, n (%) 4 (5.0) 1 (0.6) –Clopidogrel 75 mg, n (%) 1 (1.3) 0 (0) –ALT admission, median (IQR) 43 (33–62) 47 (36–68.75) 0.16AST admission, median (IQR) 63 (38–106) 80.50 (48–129.50) 0.07Bilirubin admission, median (IQR) 1.98 (1.32–4.19) 2.88 (1.69–5.04) 0.01Creatinine admission, median (IQR) 1.03 (0.81–1.63) 1.10 (0.71–1.79) 0.079Hospitalization duration (hrs), median (IQR) 177.95 (94.20–306.42) 90.88 (48.88–165.39) < 0.0001Maximum INR, median (IQR) 1.83 (1.63–2.18) 1.89 (1.69–2.32) 0.13MELD, median (IQR) 18.70 (15.22–23.13) 20.30 (16.15–25.12) 0.04Minimum platelet count, median (IQR) 80 (50–148) 61 (40–97) 0.001

VTE = venous thromboembolism; IQR = interquartile range; ALT = alanine aminotransferase; AST = aspartate aminotransferase; MELD =Model for End-Stage Liver Disease.

Table 2. Rate of Hemorrhage

HemorrhageClassification

VTEProphylaxis

(n=80)

No VTEProphylaxis(n=176) p-Value

Overall hemorrhage,n (%)

14 (17.5) 13 (7.4) 0.02

Major hemorrhage,n (%)

4 (5.0) 5 (2.8) 0.47

Minor hemorrhage,n (%)

10 (12.5) 8 (4.5) 0.02

VTE = venous thromboembolism.

1046 PHARMACOTHERAPY Volume 34, Number 10, 2014

0.98–1.00, p=0.03). Antiplatelet medication onadmission was not an independent predictor ofhemorrhage risk (AOR 1.67, 95% CI 0.49–5.66,p=0.41). The AOR for minimum platelet countrepresents a 1-unit increase in the odds of bleed-ing risk, adjusted for the other covariates in themodel.No statistically significant differences were

detected for the secondary outcome measures(Table 4). Rates of PRBC transfusions were 26.3%for VTE prophylaxis compared with 18.8% for noVTE prophylaxis (p=0.17).

Discussion

The findings of our study demonstrate thatpharmacologic prophylaxis for patients withCLD and concurrent coagulopathy (INR ≥ 1.5) isassociated with an increased rate of TIMI-definedoverall hemorrhage (composite of major andminor hemorrhage).21 The difference in overallhemorrhage was 10.1% greater in patients receiv-ing pharmacologic VTE prophylaxis. This 10.1%increase represents a number needed to harm(NNH) of 10 patients to cause one additionalmajor or minor hemorrhage. The regression modeldemonstrates for patients that received pharmaco-logic VTE prophylaxis, increasing maximum INR,and decreasing minimum platelet count was asso-ciated with an increased risk of hemorrhage whilehospitalized. These regression results are consis-tent with the prior analysis establishing VTE pro-phylaxis, increased INR, and decreased plateletcount as independent risk factors associated with

hemorrhaging.10 The regression analysis failed todemonstrate that antiplatelet use on admissionwas associated with increased the risk for hemor-rhage; however, antiplatelet use on admission didnot necessarily coincide with antiplatelet use dur-ing hospitalization, a variable that was not col-lected. Notably, the clinical implications of ourstudy results are questionable, with differences inclinical outcomes remaining largely unsubstanti-ated. Our study was the first to assess a patientpopulation universally experiencing CLD withconcurrent coagulopathy, evidenced by a medianmaximum INR value 1.83 in those receiving VTEprophylaxis and 1.89 in those not receiving VTEprophylaxis, not secondary to therapeutic antico-agulation.INR is a commonly used tool to evaluate

hemostasis, and elevated values suggest decreasedVTE risk and increased hemorrhage risk. Clini-cally, discrete INR demarcations serve to guideassessment of pharmacologic VTE prophylaxisappropriateness. However, this general under-standing of INR impact on clinical outcomes maynot extrapolate to patients with coagulopathysecondary to CLD. While prior understandingsuggested the theory of “autoanticoagulation” asa protective mechanism against VTE developmentand a relative contraindication for pharmacologicVTE prophylaxis, numerous recent studies havequestioned this theory by demonstrating similarrisk of VTE in CLD patients when compared tothe general population.11–17 Subsequently, thissimilar risk of VTE has prompted recommenda-tions to consider pharmacologic VTE prophylaxisdespite elevated INR.11, 15, 16, 18 In addition tothe impact on clinical outcomes, such as riskof VTE and hemorrhage, earlier research is con-flicting and unclear with regard to underlyingpharmacodynamic changes, and demonstratesboth an enhanced and attenuated anticoagulantresponse in patients with CLD.6–8 Unfortunately,in this patient population, beyond preliminaryinconclusive results, the hemostatic and clinicalefficacy and safety of pharmacologic anticoagula-tion remain largely unexplored. Because thehemostatic and clinical effect of pharmacologic

Table 3. Logistic Regression Analysis of CovariatesAssociated with Increased Risk of Hemorrhage

Covariate

AdjustedOddsRatio

95%ConfidenceInterval p-Value

Antiplatelet onadmission

1.67 0.49–5.66 0.41

VTE prophylaxis 3.64 1.50–8.86 0.004Maximum INR 1.31 1.08–1.60 0.007Platelet count 0.99 0.98–1.00 0.03

VTE = venous thromboembolism; INR = international normalizedratio.

Table 4. Secondary Outcomes

Outcome VTE Prophylaxis (n=80) No VTE Prophylaxis (n=176) p-Value

Overall mortality, n (%) 6 (7.5) 16 (9.1) 0.67Hemorrhage-associated mortality, n (%) 1 (1.3) 2 (1.1) 1.00PRBC transfusion incidence, n (%) 21 (26.3) 33 (18.8) 0.17Units PRBC transfused, median (raw range) 2 (1–11) 3 (1–12) 0.20Thrombosis, n (%) 1 (1.3) 7 (4.0) 0.25

VTE = venous thromboembolism; PRBC = packed red blood cell.

VTE PROPHYLAXIS IN CLD SETTING Reichert et al 1047

anticoagulation is largely unknown, it is crucialto establish the safety and efficacy of these agentsin settings of altered physiologic hemostasis.Past experience with pharmacologic VTE pro-

phylaxis suggests that the rate of overall hemor-rhage in medical patients hospitalized for 14 daysis 3.2%, which may increase up to 80% if criti-cally ill patients are included in the analysis.10

While our study did demonstrate an increase inoverall hemorrhage compared with placebo, theclinical implications of the results remainunclear. Although patients receiving pharmaco-logic VTE prophylaxis were at an increased riskof overall hemorrhage, this increase was primarilydriven by the increased rate of minor hemor-rhage. Analysis of individual major and minorhemorrhage revealed a statistically significant dif-ference in only rate of minor hemorrhage. Theoutcome of major hemorrhage, a more clinicallyrelevant marker, was different between groups,but its incidence was too minimal to demonstratestatistical significance. In addition, secondaryoutcomes to confirm the clinical import of theprimary outcome, overall hemorrhage, remainednot statistically significant between groups. Bothall-cause and bleeding-associated mortality didnot differ between groups, suggesting no impacton overall mortality. In addition, outcomesassessing difference in incidence of PRBC transfu-sion and number of PRBC units required werenot statistically significant.Differences in baseline demographic and

characteristic data were noted in age, antiplateletuse on admission, MELD score, and minimumplatelet count. These differences introducepotential confounders because as increasing age,antiplatelet use, increasing severity of CLD, anddecreasing platelet count are associated withincreased risk of hemorrhage.10 However, theclinical relevance of these variables is unknown,with the values possessing statistical, butperhaps not clinical, significance.Other studies have shown a similar risk of VTE

for patients with CLD and concurrent coagulopa-thy compared with the general population.11–17

More clinically relevant, in coagulopathic CLDpatients, a decrease in VTE risk has been demon-strated, suggesting a benefit.11 However, whilebenefit has been suggested, safety remains a para-mount concern, especially due to an increase inoverall hemorrhage expected with the use ofpharmacologic anticoagulation for VTE prophy-laxis, with a possible enhanced effect in CLDpatients.20 The findings from our exploratorystudy quantify the impact of pharmacologic VTE

prophylaxis on risk of hemorrhage to allowbetter assessment of clinical appropriateness ofpharmacologic VTE prophylaxis in patients withCLD and concurrent coagulopathy and establishrationale for prospective studies assessing thisimportant question.Several key limitations of our study are worthy

of note, such as the exploratory, observational,and retrospective study design, the inability toassess primary admission diagnosis, the use ofsurrogate markers to assess hemorrhage, and thesmall sample size included in the final analysis.Being exploratory in nature, no prospectivepower analysis was performed, limiting the abil-ity to draw conclusions surrounding clinicallyrelevant endpoints, especially major hemorrhage,PRBC transfusion, and mortality. Further, theretrospective and observational nature of thestudy introduces potential selection bias duringdata collection, especially with inclusion of surgi-cal patients, inability to assess primary admissiondiagnosis, and inability to collect history ofbleeding, a known risk factor for subsequenthemorrhage. Additionally, the use of surrogatelaboratory markers to assess hemorrhage remainsa major limitation, with changes in the surrogateendpoint not necessarily providing accurate iden-tification of hemorrhage.Elevated INR secondary to CLD should there-

fore not be considered a contraindication topharmacologic VTE prophylaxis. However, clini-cians should be aware that patients with CLD areat an elevated risk of hemorrhage, particularly inthe context of deteriorating liver function.

Conclusions

This exploratory study demonstrated thathospitalized patients with CLD and concurrentcoagulopathy have an increased risk for TIMI-defined overall hemorrhage during pharmaco-logic VTE prophylaxis. Prospective studies areneeded.

Acknowledgments

The author would like to thank Aaron Burton,Patrick Ferguson, A. Janelle Hermosillo, LeslieJohnson, Daniel Longyhore, Robert Menak, TriciaPapademetrious, and Brian Waldron.

References

1. National Center for Health Statistics. Deaths: final data for2010. Hyattsville, MD: National Center for Health Statistics;2010.

1048 PHARMACOTHERAPY Volume 34, Number 10, 2014

2. National Center for Health Statistics. Number, rate, and averagelength of stay for discharges from short-stay hospitals, by age,region, and sex: United States, 2010. 2013. Available from http://www.cdc.gov/nchs/fastats/hospital.htm. Accessed February 27,2013.

3. Halkin H, Goldberg J, Modan M, Modan B. Reduction ofmortality in general medical in-patients by low-dose heparinprophylaxis. Ann Intern Med 1982;96:561–5.

4. Samama MM, Cohen AT, Darmon JY, et al. A comparison ofenoxaparin with placebo for the prevention of venous throm-boembolism in acutely ill medical patients. Prophylaxis inmedical patients with Enoxaparin Study Group. N Engl J Med1999;341:793–800.

5. Cohen AT, Davidson BL, Gallus AS, et al. Efficacy and safetyof fondaparinux for the prevention of venous thromboembo-lism in older acute medical patients: randomised placebocontrolled trial. BMJ 2006;332:325–9.

6. Lisman T, Leebeek FW, de Groot PG. Haemostatic abnormali-ties in patients with liver disease. J Hepatol 2002;37:280–7.

7. Violi F, Ferro D, Quintarelli C, Saliola M, Cordova C,Balsano F. Clotting abnormalities in chronic liver disease. DigDis 1992;10:162–72.

8. G€ursoy S, Bas�kol M, Torun E, et al. Importance of anticoagu-lant proteins in chronic liver diseases. Turk J Gastroenterol2005;16:129–33.

9. Heit JA, Silverstein MD, Mohr DN, Petterson TM, O’FallonWM, Melton LJ 3rd. Risk factors for deep vein thrombosisand pulmonary embolism: a population-based case–controlstudy. Arch Intern Med 2000;160:809–15.

10. Decousus H, Tapson VF, Bergmann JF, et al. Factors atadmission associated with bleeding risk in medical patients:findings from the IMPROVE investigators. Chest 2011;139:69–79.

11. Barclay SM, Jeffres MN, Nguyen K, Nguyen T. Evaluation ofpharmacologic prophylaxis for venous thromboembolism inpatients with chronic liver disease. Pharmacotherapy 2013;33:375–82.

12. Northup PG, Sundaram V, Fallon MB, et al. Hypercoagulationand thrombophilia in liver disease. J Thromb Haemost2008;6:2–9.

13. Aldawood A, Arabi Y, Aljumah A, et al. The incidence ofvenous thromboembolism and practice of deep venous throm-bosis prophylaxis in hospitalized cirrhotic patients. Thromb J2011;9:1.

14. Ali M, Ananthakrishnan AN, McGinley EL, Saeian K. Deepvein thrombosis and pulmonary embolism in hospitalizedpatients with cirrhosis: a nationwide analysis. Dig Dis Sci2011;56:2152–9.

15. Dabbagh O, Oza A, Prakash S, Sunna R, Saettele TM. Coagul-opathy does not protect against venous thromboembolism inhospitalized patients with chronic liver disease. Chest2010;137:1145–9.

16. Pincus KJ, Tata AL, Watson K. Risk of venous thromboembo-lism in patients with chronic liver disease and the utility ofvenous thromboembolism prophylaxis. Ann Pharmacother2012;46:873–8.

17. Søgaard KK, Horv�ath-Puh�o E, Grønbaek H, Jepsen P, VilstrupH, Sørensen HT. Risk of venous thromboembolism in patientswith liver disease: a nationwide population-based case–controlstudy. Am J Gastroenterol 2009;104:96–101.

18. Koliscak L, Maynor L. Pharmacologic prophylaxis againstvenous thromboembolism in hospitalized patients with cirrho-sis and associated coagulopathies. Am J Health Syst Pharm2012;69:658–63.

19. Bechmann LP, Sichau M, Wichert M, Gerken G, Kr€oger K,Hilgard P. Low-molecular-weight heparin in patients withadvanced cirrhosis. Liver Int 2011;31:75–82.

20. Senzolo M, Rodriguez-Castro KI, Rossetto V, et al. Increasedanticoagulant response to low-molecular-weight heparin inplasma from patients with advanced cirrhosis. J Thromb Hae-most 2012;10:1823–9.

21. Chesebro JH, Knatterud G, Roberts R, et al. Thrombolysis inmyocardial infarction (TIMI) trial, phase I: a comparisonbetween intravenous tissue plasminogen activator and intrave-nous streptokinase. Clinical findings through hospital dis-charge. Circulation 1987;76:142–54.

22. Vittinghoff E, McCulloch CE. Relaxing the rule of ten eventsper variable in logistic and Cox regression. Am J Epidemiol2007;165:710–8.

VTE PROPHYLAXIS IN CLD SETTING Reichert et al 1049