Journal of Critical Care, Vol 15, No 4 (December), 2000: pp 127-132 127

Purpose: Acute and chronic illness, immobility, and

procedural and pharmacologic interventions may pre-

dispose patients in the intensive care unit (ICU) to ve-

nous thromboembolic (VTE) disease. The purpose of

this study was to observe potential risk factors and

diagnostic tests for VTE, and prophylaxis against VTE

in medical-surgical ICU patients.

Materials and Methods: In a prospective observational

study, 93 consecutive patients admitted to a mixed

medical-surgical ICU were followed. We recorded de-

mographics, admitting diagnoses, APACHE II score,

VTE risk factors, antithrombotic, anticoagulant and

thrombolytic agents, diagnostic tests for deep venous

thrombosis (DVT) and pulmonary embolus (PE), and

clinical outcomes.

Results: Patients were 65.5 (15.5) years old with an

APACHE II score of 21.1 (9.0); 44 (47.3%) were female.

Admission diagnoses were medical (58, 67.4%) and

surgical (35, 37.6%). The duration of ICU stay was 3

days (interquartile range: 1, 8.5 days) and the ICU mor-

tality rate was 20.4% (19 of 93). We observed 8 VTE

events among 5 of 93 patients (incidence 5.4% [0.8 to

10.0]); 2 patients had DVT and PE before admission, 1

had DVT as an admitting diagnosis, 1 had DVT on day

2 and PE on day 3, and 1 had PE on day 2. Over 804

ICU patient-days, 2 of 5 ultrasound examinations di-

agnosed DVT and 2 of 3 ventilation-perfusion lung

scans diagnosed PE. Of 64 patients in whom heparin

was not contraindicated and who were not anticoag-

ulated, subcutaneous heparin prophylaxis was pre-

scribed for 40 (62.5%) patients. ICU-acquired VTE risk

factors were mechanical ventilation (odds ratio [OR]

1.56), immobility (OR 2.14), femoral venous catheter

(OR 2.24), sedatives (OR 1.52), and paralytic drugs (OR

4.81), whereas VTE heparin prophylaxis (OR 0.08), as-

pirin (OR 0.42), and thromboembolic disease stockings

(OR 0.63) were associated with a lower risk. Only

warfarin (OR 0.07, P �.01) and intravenous heparin

(OR 0.04, P � .01) were associated with a significantly

decreased risk of VTE.

Conclusions: Several ICU-acquired risk factors for VTE

were documented in this medical-surgical ICU. VTE

prophylaxis was underprescribed, and VTE diagnostic

tests were infrequent. Further research is required to

determine the incidence, predisposing factors, attrib-

utable morbidity, mortality, and costs of VTE in med-

ical-surgical ICU patients, the optimal diagnostic test

strategies, and the most cost-effective approaches of

prophylaxis.

Copyright © 2000 by W.B. Saunders Company

Venous Thromboembolic Disease: An Observational Study in Medical-Surgical Intensive Care Unit Patients

Deborah Cook, John Attia, Bruce Weaver, Ellen McDonald, Maureen Meade, and Mark Crowther

CRITICALLY ILL patients are predisposed tovenous thromboembolism (VTE) during their

stay in the intensive care unit (ICU) due to pre-morbid medical and surgical conditions, invasivetests and treatments, prolonged immobility, vascu-lar injury from indwelling central venous catheters,and other interventions. Because features of the his-tory and physical examination for deep venousthrombosis (DVT) are insensitive and nonspecific,1

bedside assessment of DVT using a constellationof signs and symptoms with2 or without3 ultra-sonography has generated diagnostic models withstrong predictive properties in ambulatory patients.Use of a similar clinical model categorizing patientswith suspected PE into low, medium, and highpretest probability in combination with ventilation-perfusion scanning has been shown to be safe.4

However, such models to detect DVT are not avail-able in the ICU, partly due to the challenge of elic-iting symptoms from mechanically ventilated pa-tients or from those with impaired consciousness.

Cross-sectional studies at the time of admissionto medical5 and surgical6 ICUs suggest a 10%prevalence rate of deep venous thrombosis (DVT)as diagnosed by compression ultrasonography.

DVT that develops during ICU has been systemat-ically evaluated in three cohort studies of medicalor surgical ICU patients using three different diag-nostic regimens.7-9 In these studies, DVT developedwithin the first week of ICU admission in approx-imately one third of patients.

Pulmonary embolus (PE) is among the mostcommon preventable cause of death in hospitalizedpatients,10 and up to 95% of emboli arise from the

From the Departments of Medicine and Clinical Epidemiol-ogy, McMaster University Faculty of Health Sciences, Hamil-ton, Ontario, Canada; and the Centre for Clinical Epidemiol-ogy & Biostatistics, University of Newcastle, Newcastle,Australia.

This work was supported by the Father Sean O’Sullivan Research Centre of St. Joseph’s Hospital in Hamilton, Ontario.Dr. Cook is an Investigator of the Canadian Institutes of HealthResearch. Drs. Meade and Crowther are Scholars of the Med-ical Research Council of Canada.

Received August 1, 2000. Accepted August 9, 2000.Address reprint requests to Deborah J. Cook, MD, Depart-

ment of Medicine, St. Joseph’s Hospital, 50 Charlton Ave East,Hamilton, Ontario, Canada L8N 4A6.

Copyright © 2000 by W.B. Saunders Company0883-9441/00/1504-0001$10.00/0doi: 10.1053/jcrc.2000.19224

deep venous system of the lower limbs. PE maycause serious morbidity and mortality among crit-ically ill patients who have compromised car-diopulmonary reserve. Few studies exist about theincidence of PE in ICU patients. Over 1 year in arespiratory ICU in which no heparin prophylaxiswas administered, 13 of 98 patients had PE,whereas the incidence was 1 of 99 patients in thesubsequent year when prophylaxis was adminis-tered (P � .05).11 Autopsy revealed PE in 8 of 30versus 1 of 26 patients who died in years 1 and 2,respectively. In cohort studies on postmortem ex-amination, PE has been identified in 18 of 6612 and2 of 107 respiratory ICU patients. Pathology stud-ies show that PE is frequently clinically silent,13

occurring without antemortem suspicion in up to70%14 to 80%15 of hospitalized patients who die.

The incidence of reported DVT and PE is likelyto vary across populations, based on pre-ICU ad-mission characteristics, post-ICU admission eventsand exposures, clinical suspicion for VTE, thescheduling and accuracy of screening for VTE, andprophylactic interventions. The objective of thisstudy was to observe potential risk factors and di-agnostic tests for VTE, and the VTE prophylaxisprescribed in medical-surgical ICU patients in alarge urban teaching hospital. In this noninterven-tional study, our goal was not to obtain a preciseestimate of the VTE rate or to evaluate the operat-ing characteristics of diagnostic tests for VTE.

MATERIALS AND METHODS

This prospective observational study was conducted in a 15-bed medical-surgical ICU in a university-affiliated ICU inHamilton, Ontario, Canada. Cardiac surgery and neurosurgerypatients are admitted elsewhere. All consecutive patients ad-mitted from May through July 1999 were considered for thisstudy. Patients were excluded if they were admitted and eitherdied or were discharged within 12 hours of admission (from 1800hours to 0600 hours from Monday to Friday), or during week-ends (from 1800 hours on Friday to 0600 hours on Monday).

We collected data from four sources: computerized ICU flowsheets, nurses and physicians notes, electronic medical records,and a regional ICU database. We recorded premorbid VTE riskfactors (demographics, APACHE II score, admitting diagnoses,postoperative status, personal and family history of VTE, activemalignancy, thrombophilic disorder, and drugs that modify co-agulation profile) and ICU-acquired VTE risk factors (mechan-ical ventilation, immobility, femoral central venous catheters,emergency surgery, platelet count �450 � 109/L, sedatives, andparalytic agents), as well as VTE prophylaxis and treatment (un-fractionated or low-molecular-weight heparin, other anticoagu-lants, antiplatelet agents, thrombolytics, thromboembolic dis-ease [TED] stockings, pneumatic compression devices, andGreenfield filters). We also noted all diagnostic tests for VTE

(Doppler compression ultrasonography, venography, ventila-tion-perfusion scans, spiral CT chest scans, and pulmonary an-giography), and patient outcomes (DVT, PE, bleeding event,treatment complications, length of stay, and mortality status).

To avoid bias introduced by the Hawthorne effect, the ICUteam remained unaware of the conduct of this study. Data werecollected by a trained, independent ICU research nurse. All clin-ical decisions were made at the discretion of the ICU team. Thereis no formal VTE prophylaxis policy or standardized physiciansorders in our ICU. In the absence of coagulopathy or heparin-induced thrombocytopenia, all central venous and arterial linesare flushed with heparin 1000 U in 500 mL D5W at 3 mL/h.

Statistical Analysis

For continuous variables, we report means and standard de-viations and used Student’s t test to assess between group dif-ferences. For skewed distributions, we report medians and in-terquartile ranges and used the Mann-Whitney U test. Analyzingonly VTE events that were ICU-acquired, we explored poten-tial risk factors for VTE, representing both baseline character-istics and post admission, ICU-acquired, time-dependent events,and exposures. We report odds ratios with 0.5 added to eachcell16 and associated 95% confidence intervals.

RESULTS

Of 93 patients aged 65.5 (15.5) years, 44 (47.3%)were female, and the mean APACHE II score was21.1 (9.0). Admitting diagnoses and other clinicalcharacteristics are presented in Table 1. The me-dian length of ICU stay was 3 days (1, 8.5 days)and ICU mortality was 20.4% (19 of 93).

Potential factors that could influence VTE riskestablished on ICU admission were history of VTE

128 COOK ET AL

Table 1. Characteristics of 93 Medical-Surgical ICU Patients

Clinical characteristicsAge (Mean � SD) (years) 65.5 (15.5)Sex (% female) 44 (47.3%)APACHE II Score (X � SD) 21.1 (9.0%)

Primary admitting diagnosisCardiovascular medical 13 (14.0%)

surgical 12 (12.9%)Respiratory medical 13 (14.0%)

surgical 3 (3.2%)Gastrointestinal medical 8 (8.6%)

surgical 14 (15.1%)Neurological medical only 6 (6.5%)Sepsis 6 (6.5%)Miscellaneous 18 (19.4%)

Duration of ICU stay (med, IQR) (days) 3 (1, 8.5)Duration of hospital stay (med, IQR) (days) 14 (7, 29.5)ICU mortality 19 (20.4%)

NOTE: The clinical characteristics of 93 medical-surgical ICUpatients are described. Mean � SD � mean and standard de-viation.

Abbreviations: med, median; IQR, interquartile range.

(5, 5.4%), postoperative status (35, 37.6%) and ac-tive malignancy (18, 19.4%). On admission, pa-tients were receiving several drugs that modify co-agulation profile, including warfarin (3, 3.2%),dalteparin (2, 2.2%), ticlopidine (1, 1.1%), conju-gated estrogens (2, 2.2%), and vitamin K (1, 1.1%).

Potential factors occurring during the ICU staythat could influence VTE risk included mechanicalventilation (62, 66.7%), immobility (79, 84.9%),femoral venous catheters (23, 24.7%), emergencysurgery (8, 8.6%), platelet count �450 � 109/L (7,7.5%), sedatives (30, 32.3%), and paralytic agents(13, 14.0%). During their ICU stay, patients re-ceived a wide variety of drugs that modify the co-agulation profile including subcutaneous unfrac-tionated heparin 5,000 U twice a day (48, 51.6%),full anticoagulation with intravenous unfractionatedheparin (7, 7.5%), heparin flushes in central venouslines or dialysis catheters or low dose heparin in ar-terial lines (71, 76.3%), warfarin (6, 6.5%), aspirin(22, 23.7%), ticlopidine (1, 1.1%), conjugated es-trogens (3, 3.2%), desmopressin acetate (1, 1.1%),and vitamin K (16, 17.2%). No patients receivedsubcutaneous low-molecular-weight heparin,thrombolytic therapy, or pneumatic compression de-vices. TED stockings were worn by 29 (31.2%) pa-tients, all of whom were also receiving subcuta-neous unfractionated heparin 5000 U twice a day.

Of 64 patients who were not receiving anticoag-ulation for other reasons, and in whom unfraction-ated heparin was not clinically contraindicated (eg,gastrointestinal bleeding, platelet count � 50 �109/L, international normalized ratio �2.0 or par-tial thromboplastin time � 60 seconds), VTE pro-phylaxis with subcutaneous unfractionated heparinwas prescribed for 40 (62.5%) patients. Diagnostictests for VTE over 804 ICU patient-days includedfive Doppler ultrasound studies for DVT, two ofwhich were positive, and three ventilation-perfu-sion lung scans, two of which were high probabil-ity scans. Six thoracic CT scans were performed,but none were spiral CTs ordered because of a clin-ical suspicion of PE. No patients had venographyor pulmonary angiography.

The observed incidence of VTE was 5.4% (95%CI 0.8 to 10.0%); there were 8 events among 5 of93 patients. In total, 2 patients had DVT and PE be-fore admission, 1 had DVT as an ICU admitting di-agnosis, 1 had DVT on day 2 and PE on day 3, and1 had PE on day 2. Of the 2 patients with VTE be-fore ICU admission, one with septic shock had right

superficial femoral and popliteal vein thrombosisand right-sided PE, and the other had a proximal leftiliac DVT and bilateral PE following abdominal aor-tic aneurysm surgery. The latter patient also had aGreenfield filter inserted, and both patients were dis-charged from the hospital. The only patient withVTE as an admitting diagnosis also had a prior DVTand Greenfield filter. He presented with bilateral ile-ofemoral DVT extending into the inferior vena cavaand was eventually discharged from the hospital. Ofthe 2 patients with VTE following ICU admission,1 patient with pneumonia developed iliofemoralDVT and bilateral PE within 72 hours; she also hadextensive thrombus in the aortic arch and proximaldescending aorta, and died on the ward. The otherpatient had myocardial ischemia and developed aright-sided PE within 48 hours; she survived to hos-pital discharge. All VTE events were treated with in-travenous heparin. No patients had bleeding or treat-ment-related complications. There was no differencein length of stay or ICU mortality between patientswith and without VTE (P � .05).

None of the potential VTE risks factors estab-lished before or on ICU admission, such as age,sex, APACHE II score, admitting diagnosis, historyof VTE or malignancy, or drugs used for VTE pro-phylaxis or treatment were associated with ICU-ac-quired VTE (data not shown). In Table 2, we pre-sent potential ICU-acquired risk factors reflecting

VTE IN ICU 129

Table 2. ICU-Acquired Risk Factors For

ICU-Acquired Venous Thromboembolism

Odds RatioPotential Risk Factor (95% CI) P Value

Mechanical ventilation 1.56 (0.23–10.45) .64Immobility 2.14 (0.11–40.87) .61Femoral venous catheter 2.24 (0.41–12.20) .35Platelet count �450 � 109/L 0.99 (0.05–19.7) .99Sedative infusion 1.52 (0.28–8.16) .63Paralytic drug 4.81 (0.85–27.35) .08Emergency surgery in ICU 1.03 (0.19–5.50) .97VTE prophylaxis with heparin 0.08 (0.0–1.41) .05Aspirin 0.42 (0.08–2.29) .31TED stockings 0.63 (0.12–3.37) .59Central venous catheter heparin 1.05 (0.15–7.07) .96Warfarin 0.07 (0.01–0.49) .01Intravenous heparin 0.04 (0.01–0.25) �.01

NOTE: Potential risk factors are listed for VTE developing af-ter ICU admission. Odds ratios greater than one suggest anincreased risk of VTE; odds ratios less than one suggest a de-creased risk. P �.05 represents a significant association withICU-acquired VTE.

Abbreviations: VTE, venous thromboembolic; TED stock-ings, thromboembolic disease stockings.

events or exposures following ICU admission. Me-chanical ventilation (OR 1.56), immobility (OR2.14), femoral venous catheter insertion (OR 2.24),sedatives (OR 1.52), and paralytic drugs (OR 4.81)were associated with an increased risk of VTE,whereas subcutaneous unfractionated heparin 5000U twice a day (OR 0.08), aspirin (OR 0.42), andTED stockings (0.63) were associated with a lowerrisk; however, none of these predictors were statis-tically significant. Warfarin (OR 0.07,P � .01) andintravenous heparin (OR 0.04,P � .01) were as-sociated with a significantly decreased risk of VTE.

DISCUSSION

In this prospective observational study of 93medical-surgical ICU patients, 10 diagnostic testsfor VTE were conducted over 804 ICU patient-daysand the observed incidence of VTE was 5.4%. Weidentified several factors that are associated withincreased risk of VTE, including mechanical ven-tilation, immobility, femoral venous catheters,sedatives and paralytic drugs, and failure to pre-scribe VTE prophylaxis.

Mechanical ventilation, sedatives, and paralyticdrugs obviously cause or may prolong immobility.This cluster of four factors (mechanical ventilation,immobility, sedation, and paralysis) may convergeto greatly increase the risk of VTE. Sedation andparalysis also predispose to other ICU complica-tions including increased duration of mechanicalventilation,17 critical illness polyneuropathy,18 andventilator-associated pneumonia,19 which in turnmay perpetuate immobility. In other settings, suchan interaction between risk factors has been foundto markedly increase VTE rates; for example, pa-tients with factor V Leiden mutation who also takeoral contraceptives have a substantially increasedrisk of VTE compared with patients with only oneof these factors.20 Observations such as this supportthe hypothesis that ICU patients with multiple pre-disposing features may have a very high risk ofVTE. However, larger studies are needed to affirmor refute these risk factors for VTE in ICU patients.

Thrombosis and bacteremia are established com-plications of central venous catheterization, andmeta-analysis of randomized trials shows that he-parin infusion significantly decreases catheter-related thrombosis (relative risk 0.44, 95% CI 0.22to 0.87).21 However, low-dose heparin infusion andflushes do not eliminate the risk of catheter-relatedthrombosis. In a study of 76 pediatric ICU patientswho had central venous catheters infused with 100

U/mL of heparin when fluids were not administered,Doppler ultrasonography performed every 2 daysdiagnosed 15 of 17 catheter-related thromboseswithin 4 days of their insertion.22 Femoral catheterswere identified in 5 of 12 patients with DVT and18 of 64 without DVT (OR 1.8,P � .35). Hirschet al8 found that 5 out of 5 patients with upper ex-tremity DVT had prior central venous catheters.

In studies designed to measure the incidence ofVTE by conducting systematic screening and test-ing, VTE incidence rates will inevitably be higherthan those found in noninterventional studies suchas this. In 34 respiratory patients, 76% of whomwere mechanically ventilated, Moser et al7 used 125Ifibrinogen scanning for 6 days to diagnoses 3 (9%)patients with DVT. Using Doppler ultrasonographytwice weekly then at 1 week post ICU discharge in100 medical patients expected to stay in ICU � 48hours, 70% of whom were mechanically ventilated,Hirsch et al8 diagnosed DVT in 32% of 100 patientsreceiving no prophylaxis, in 40% of patients re-ceiving heparin, and in 33% of patients receivingmechanical prophylaxis. In 102 medical-surgicalICU patients admitted for � 4 days undergoingDoppler ultrasonography between days 4 and 7 andas clinically indicated, Mark et al9 found DVT in25% of patients receiving no prophylaxis, in 7% ofpatients receiving heparin, and in 19% of patientsreceiving mechanical prophylaxis.

The incidence of PE in ICU patients is very dif-ficult to ascertain due to the nonspecific nature ofrelevant signs and symptoms in mechanically ven-tilated patients. Four patients with pulmonary em-bolism were identified by Moser et al7; one had apositive ventilation-perfusion scan, one had a pos-itive angiogram, and two had autopsy diagnosis. Inthe study by Hirsch et al,8 one patient with DVTand hypoxia had a “clinically diagnosed” pul-monary embolus. Four patients with DVT and apositive ventilation-perfusion scan had pulmonaryembolism in the study by Marik et al.9 In 183 post-thoracotomy patients, Jackaman et al23 noted thatno patients developed clinical evidence of PE. Inan analysis of 119 patients who had angiographi-cally proven PE in the Prospective Investigation ofPulmonary Embolism Diagnosis (PIOPED) study,the spectrum of clinical features ranged from chestpain associated with the pulmonary infarction syn-drome, to the isolated dyspnea syndrome, to severecirculatory collapse.24 However, as with our study,these studies had no protocol for PE screening, forevaluation of suspected emboli, or routine autopsy.

130 COOK ET AL

We found that subcutaneous unfractionated he-parin was prescribed in 63% of patients in whomit was not contraindicated and who were not re-ceiving anticoagulation for other reasons. This VTEprophylaxis utilization rate is similar to 61% re-ported by Hirsch et al8 in 100 medical ICU patients,greater than 33% as reported by Keane et al25 in152 medical ICU patients,25 and less than 86% in209 medical-surgical ICU patients described byRyskamp and Trottier.26 Only one published ran-domized double-blind trial in 119 medical-surgicalICU patients has evaluated the efficacy of VTE pro-phylaxis27 comparing subcutaneous unfractionatedheparin 5000 U twice a day versus placebo injec-tions. 125I fibrinogen scanning was done for 4 to10 days and DVT rates were 13% in the heparingroup and 29% in the placebo group (P � .05).This single small trial about the benefit of unfrac-tionated heparin in medical-surgical ICU patientscontrasts with strong evidence in another seriouslyill population: trauma patients, whose proximalDVT risk is approximately 20%,28 and for whomlow-molecular-weight heparin confers a strong pre-ventive benefit over unfractionated heparin.29 In arecent randomized trial of 866 acutely ill medicalpatients, most of whom were not in ICU, 40 mgsubcutaneous enoxaparin daily was associated witha significantly lower VTE rate (5.5%) than 20 mgenoxaparin (15.0%) or placebo (14.9%).30

In this study, we prospectively enrolled patientsand systematically analyzed both admitting andICU-acquired risk factors for VTE. Blinding of theICU team to the conduct of the study allowed usto achieve the objective of an unbiased ascertain-ment of actual diagnostic and prophylactic practicepatterns which typically affect open, prospectiveobservational studies of practice patterns. Becausethis was a noninterventional study, systematic test-ing for VTE was not possible; thus, the VTE inci-dence we observed is likely an underestimate, andstudies highlighted earlier7-9 provide better evi-dence about the incidence of VTE in this popula-tion. To determine the properties of VTE diagnos-tic tests and to explore reasons why patients

received different preventive strategies would re-quire different study designs.

The 1998 American College of Chest PhysiciansAntithrombotic Consensus Conference Report con-tained no specific section on thrombosis or its pre-vention in medical-surgical ICU patients.31 Med-ical-surgical ICUs have aptly been called “the lastfrontier for prophylaxis.”32 A large prospectiveprognosis study is needed to identify more accurateand precise rates of VTE to estimate both the con-ditional probability and cumulative risk over the en-tire ICU stay. Systematic, universal diagnostic sur-veillance would be ideal in a study of heterogenouspatients to identify high-risk groups, because signsand symptoms of VTE in medical-surgical ICU pa-tients are both difficult to elicit and challenging toattribute to DVT or PE. Such a strategy will yieldkey information on the performance characteristicsof VTE diagnostic tests in medical-surgical ICU pa-tients and would help to elucidate whether all VTEevents are clinically important in ICU patients. Ahigh proportion of patients with DVT may have un-recognized PE, as shown in outpatients with veno-graphically demonstrated DVT, 50% of whom hadintermediate or high probability scans compatiblewith silent PE.33,34Though asymptomatic PE maybe well tolerated in ambulatory patients, PE maycause more serious morbidity and mortality amongICU patients who often have compromised car-diopulmonary reserve. The influence of PE on du-ration of mechanical ventilation, weaning success,and the duration of ICU stay is also unknown. Weconclude that a comprehensive research program inmedical-surgical ICU patients is warranted to un-derstand the predictors and incidence of VTE, theattributable morbidity, mortality, and costs of VTE,the optimal diagnostic test strategies, and the mostcost-effective method of prophylaxis.

ACKNOWLEDGMENTS

The authors thank Barbara Hill for help with study management;Nicole Krolicki for work on the database; and Dr. Bernard De-Jonghe for useful suggestions on this manuscript.

VTE IN ICU 131

REFERENCES1. Sandler DA, Martin JF, Duncan JS, et al: Diagnosis of

deep vein thrombosis: Comparison of clinical evaluation, ultra-sound, plethysmography, and venoscan with x-ray venogram.Lancet 29:716-719, 1984

2. Wells PS, Hirsh J, Anderson DR, et al: Accuracy of clinicalassessment of deep-vein thrombosis. Lancet 345:1326-1330, 1995

3. Anand S, Wells P, Brill-Edwards P, et al: Does this patienthave DVT? JAMA 279:1094-1099, 1998

4. Wells SP, Ginsberg JS, Anderson DR, et al: Use of a clin-ical model for safe management of patients with suspected pul-monary embolism. Ann Intern Med 129:997-1005, 1998

5. Schonhofer B, Kohler D: Prevalence of deep-vein throm-

bosis of the leg in patients with acute exacerbation of chronicobstructive pulmonary disease. Respiration 65:173-177, 1998

6. Harris LM, Curl RC, Booth FV, et al: Screening for asymp-tomatic deep vein thrombosis in surgical intensive care patients.J Vasc Surg 26:764-769, 1997

7. Moser KM, LeMoine JR, Nachtwey FJ, et al: Deep ve-nous thrombosis and pulmonary embolism: Frequency in a res-piratory intensive care unit. JAMA 246:1422-1424, 1981

8. Hirsch DR, Ingenito EP, Goldhaber SZ: Prevalence of deepvenous thrombosis among patients in medical intensive care.JAMA 274:335-337, 1995

9. Marik PE, Andrews L, Maini B: The incidence of deep ve-nous thrombosis in ICU patients. Chest 111:661-664, 1997

10. Clagett GP, Anderson FA, Heil J, et al: Prevention of ve-nous thromboembolism. Chest 108:312S-334S, 1995 (suppl)

11. Pingleton SK, Bone RC, Pingleton WW, et al: Preven-tion of pulmonary emboli in a respiratory ICU: Efficacy of lowdose heparin. Chest 79:647-650, 1981

12. Neuhaus A, Bentz RR, Weg JG: Pulmonary embolism inrespiratory failure. Chest 73:460-465, 1978

13. McKelvie PA: Autopsy evidence of pulmonary throm-boembolism. Med J Aust 160:127-128, 1994

14. Stein PD, Henry JW: Prevalence of acute pulmonary em-bolism among patients in a general hospital and at autopsy. Chest108:978-981, 1995

15. Karwinski B, Svendsen E: Comparison of clinical andpostmortem diagnosis of pulmonary embolism. J Clin Pathol42:135-139, 1989

16. Agresti A: Categorical data analysis. New York, NY, JohnWiley & Sons, 1990, pp 54

17. Kollef MH, Levy NT, Ahrens TS, et al: The use of con-tinuous intravenous sedation is associated with prolongation ofmechanical ventilation. Chest 114:541-548, 1998

18. De Jonghe BD, Cook DJ, Sharshar T, et al: Neuromus-cular disorders in critically ill patients: A systematic review. In-tensive Care Med 24:1242-1250, 1998

19. Cook DJ, Walter S, Cook RJ, et al: The incidence andrisk factors for ventilator-associated pneumonia in critically illpatients. Ann Intern Med 129:433-440, 1998

20. Vendenbroucke JP, Koster T, Briet E, et al: Increased riskof venous thrombosis in oral-contraceptive users who are carri-ers of factor V Leiden mutation. Lancet 344:1453-1457, 1994

21. Randolph AG, Cook DJ, Gonzales CA, et al: Benefit ofheparin in central venous and pulmonary artery catheters: Ameta-analysis of randomized controlled trials. Chest 113:165-171, 1998

22. Beck C, Dubois J, Grignon A, et al: Incidence and risk fac-tors of catheter-related deep venous thrombosis in a pediatric in-tensive care unit: A prospective study. J Pediatr 133:237-241, 1998

23. Jackaman FR, Perry BJ, Siddons H: Deep vein throm-bosis after thoracotomy. Thorax 33:761-763, 1998

24. Stein P, Henry JW: Clinical characteristics of patientswith acute pulmonary embolism stratified according to their pre-senting syndromes. Chest 112:974-979, 1997

25. Keane MG, Ingenito EP, Goldhaber SZ: Utilization of ve-nous thromboembolism prophylaxis in the medical intensivecare unit. Chest 106:13-22, 1995

26. Ryskamp RP, Trottier SJ: Utilization of venous throm-boembolism prophylaxis in a medical-surgical ICU. Chest113:162-164, 1998

27. Cade JF: High risk of the critically ill for venous throm-boembolism. Crit Care Med 10:448-450, 1982

28. Geerts WH, Code KI, Jay RM, et al: A prospective studyof venous thromboembolism after major trauma. N Engl J Med331:1601-1606, 1994

29. Geerts WH, Jay RM, Code KI, et al: A comparison oflow-dose heparin with low molecular-weight heparin as pro-phylaxis against venous thromboembolism after major trauma.N Engl J Med 335:701-707, 1996

30. Samama MM, Cohen AT, Darmon JY, et al: A compari-son of enoxaparin with placebo for the prevention of venousthromboembolism in acutely ill patients: Prophylaxis in medicalpatients with enoxaparin study group. N Engl J Med 341:793-800, 1999

31. Dalen JE, Hirsh J: Antithrombotic therapy: The evolvingconsensus: 1985 to 1998. Chest 114:439S-440S, 1998

32. Goldhaber SZ: Venous thromboembolism in the intensivecare unit. Chest 113:5-7, 1998

33. Huisman MV, Buller HR, ten Cate JW, et al: Unexpectedhigh prevalence of silent pulmonary embolism in patients withdeep venous thrombosis. Chest 95:498-502, 1989

34. Moser KM, Fedullo PF, LitteJohn JK, et al: Frequentasymptomatic pulmonary embolism in patients with deep ve-nous thrombosis. JAMA 271:223-225, 1994

132 COOK ET AL