Risk factors for venous thromboembolism in critically ill nontrauma surgical patients who cannot receive chemical prophylaxis

The American Journal of Surgery (2013) 206, 300-306

Clinical Science

Risk factors for venous thromboembolism in critically ill nontrauma surgical patients who cannot receive chemical prophylaxis Madhukar S. Patel, M.D., M.B.A., Sc.M.a, Tyler Ewingb, Allen Kong, M.D.b, David Nguyenb, Cecilia Laub, Cristobal Barrios, M.D.b, Marianne Cinat, M.D.b,†, Matthew Dolich, M.D.b, Michael Lekawa, M.D.b, Darren Malinoski, M.D.c,* a

Department of Surgery, Massachusetts General Hospital, Boston, MA, USA; bDivision of Trauma and Critical Care, University of California, Irvine, Irvine, CA, USA; cSurgical Critical Care Section, Portland Veterans Affairs Medical Center, Portland, OR, USA

KEYWORDS: Venous thromboembolism; Anticoagulation; Prophylaxis

Abstract BACKGROUND: We sought to identify independent predictors of venous thromboembolism in critically ill general surgery patients who cannot receive chemical prophylaxis in order to identify those who may benefit from aggressive screening and/or prophylactic inferior vena cava filter placement. METHODS: Nontrauma patients in the surgical intensive care unit were prospectively followed for 2 years. Patients who had contraindications to prophylactic anticoagulation and received routine screening duplex examinations were included. Data regarding lower-extremity deep venous thrombosis or pulmonary embolism (PE) rates, past medical history (PMH), surgeries, and transfusions were collected. Logistic regression was used to identify independent predictors of lower-extremity deep venous thrombosis or PE (venous thromboembolism) with a P , .05. RESULTS: Data were complete for 204 patients. Twenty (9.8%) patients developed venous thromboembolism. Independent predictors of venous thromboembolism included postoperative blood product requirements (odds ratio 5 1.04 per unit), a PMH of PE (OR 5 10.1), and a PMH of renal insufficiency (odds ratio 5 5.1). CONCLUSIONS: Aggressive screening and/or prophylactic inferior vena cava filter may be considered when prophylactic anticoagulation is prohibited in patients with increased postoperative transfusion requirements or a PMH of either PE or renal insufficiency. Ó 2013 Elsevier Inc. All rights reserved.

The authors declare no conflict of interest. Presented as a Podium Presentation at the 2011 Annual Scientific Meeting of the Southern California Chapter of the American College of Surgeons, January, 22, 2011, Santa Barbara, CA. * Corresponding author. Tel.: 11-503-220-8262; fax: 11-503-220-3415. E-mail address: [email protected] Manuscript received June 11, 2012; revised manuscript September 20, 2012 † Deceased. 0002-9610/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjsurg.2012.10.040

M.S. Patel et al.

Risk factors for venous thromboembolism

Venous thromboembolism (VTE) continues to impact the morbidity and mortality of hospitalized patients. Patients in the intensive care unit (ICU) are not only thought to be at an increased risk of developing VTE but also may be at a higher risk for poor outcomes after VTE because of their impaired physiologic reserve.1 In addition to having all of the same risk factors as non–critically ill surgery patients, ICU patients have an increased risk of VTE because of sedation, intubation, and catheterization.1 In the ICU, deep venous thrombosis (DVT) rates can range from approximately 10% to 60% in mixed patient cohorts.1,2 Interestingly, upon autopsy of ICU patients, VTE has been reported to go clinically unrecognized, with discrepancies between premortem and postmortem diagnoses being noted.3 This suggests that some estimates of DVT rates may in fact be underestimates. Historically, although not all patients with pulmonary embolism (PE) have DVT identified, PE occurs in 12% to 38% of ICU patients4,5 with confirmed DVT and 1% to 3% are fatal.6,7 Routine VTE prophylaxis is part of standard critical care because critically ill surgical patients often fulfill all 3 criteria of Virchow’s triad: hypercoagulability, stasis, and vessel injury. Without prophylaxis, VTE rates have been documented to range from 15% to 60% in the ICU.1 Both low-dose unfractionated heparin (LDH) and low–molecularweight heparin (LMWH) are recognized as effective methods of chemical prophylaxis and anticoagulation.8 Mechanical prophylaxis, in the form of sequential compression devices (SCDs), is frequently used in lieu of or in addition to pharmacologic prevention. The American College of Chest Physicians suggests the early use of LMWH or LDH combined with SCDs for DVT prophylaxis in general surgery patients with multiple risk factors for VTE as long as contraindications are not present.8 Contraindications for chemical prophylaxis include the following: significant bleeding risk, recent or imminent surgery, renal insufficiency, anemia, recent history of gastrointestinal hemorrhage, active peptic ulcer disease, or liver disease.8,9 Surgical patients frequently present with 1 or more of the aforementioned contraindications to anticoagulation in a perioperative setting, and, when this is the case, it is recommended that mechanical prophylaxis (eg, SCDs) be used until the contraindication resolves and chemical prophylaxis can be started.8 PE prophylaxis becomes necessary when DVT prophylaxis is contraindicated or insufficient. Inferior vena cava filters (IVCFs) are used by some clinicians when faced with these circumstances. Reports analyzing the safety and efficacy of IVCFs have produced mixed results.10–16 Despite IVCFs having been associated with up to a 7-fold decrease in the occurrence of PEs,17 several studies have shown no benefit.13,14,17 Additionally, the retrieval of temporary IVCFs is also a concern. Furthermore, IVCFs may be associated with complications such as fracture, migration, and the formation of VTE within the filter itself. VTE is recognized as the most common complication associated with vena cava filters.9,11,13

301 Given the uncertainty regarding the efficacy, safety, and retrievability of IVCFs, it is important that vena cava filtration be reserved for use only in high-risk patients. According to guidelines from the Institute for Clinical Systems Improvement, IVCFs are not appropriate for routine use but rather are indicated in patients with VTE and contraindications to anticoagulation, failure of adequate anticoagulation to treat a progressive VTE, or a history of pulmonary hypertension.9 At present, there are no level 1 or level 2 recommendations for prophylactic vena cava filtration before the development of VTE in critically ill patients. However, the ability to predict which patients will fall into the first group mentioned previously (ie, those who develop VTE and cannot receive anticoagulation) could assist in the development of such recommendations. Patients who are predisposed to develop VTE may also benefit from aggressive DVT screening, a less invasive alternative to prophylactic IVCF placement, and the American College of Chest Physicians recommends reserving screening duplex ultrasound examinations for patients at an increased risk for VTE who cannot receive optimal prophylaxis.8 Although not all centers are able to screen all surgical patients, identifying risk factors for VTE could help focus resources for a screening protocol. Having recently published a report on this important topic looking only at the trauma population, the goal of this study was to expand our understanding by evaluating critically ill nontrauma patients who inherently have different risk factors for VTE formation.18 This study was designed to find independent predictors of lower-extremity DVT (LEDVT) or PE in critically ill nontrauma surgery patients who cannot receive chemical prophylaxis in order to identify patients who may benefit from aggressive screening and/or prophylactic IVCF placement.

Methods All adult nontrauma patients in the surgical ICU (SICU) service at an academic urban medical center were prospectively followed from January 2008 to December 2009. DVT prophylaxis was guided by a pre-established protocol that included SCDs (applied as early as possible to the lower extremities) and 30 mg enoxaparin (LMWH; SanofiAventis, Bridgewater, NJ) administered subcutaneously twice daily. LMWH was held because of the following contraindications: bleeding risk, renal insufficiency, or thrombocytopenia (,100,000/mL). Bleeding risk was determined at the discretion of the attending surgeon and intensivist after balancing the risk of bleeding with the risk of VTE. LDH (5,000 U administered 3 times a day) was substituted for LMWH in patients with renal insufficiency. Routine screening ultrasound duplexes were obtained for the bilateral upper and lower extremities within 48 hours of admission to the SICU and weekly thereafter. After discharge from the SICU, weekly screening continued to ensure that all inpatient VTEs were documented. PE was

302

The American Journal of Surgery, Vol 206, No 3, September 2013

diagnosed via computed tomographic angiography when clinically suspected. IVCF insertion at our institute did not follow strict objective criteria but rather was done at the discretion of the attending physician in patients with any of the following indications: high risk for VTE with contraindication for chemical prophylaxis, failure of anticoagulation to prevent progression of a VTE, or contraindication for anticoagulation in the presence of an LEDVT or PE. Data regarding age, sex, body mass index, the amount of blood products received in the first 24 hours after admission (including packed red blood cells, fresh frozen plasma, platelets, and cryoprecipitate), intravenous fluids in the first 48 hours, comorbidities, the reason for admission to the SICU, and complications were recorded. Documented comorbidities included outpatient warfarin therapy and a past medical history (PMH) of cancer, intravenous drug abuse, renal insufficiency, DVT, and PE. A PMH of DVT and a PMH of PE were additionally grouped as a new variable ‘‘VTE PMH.’’ Chronic renal insufficiency was based on an elevated serum creatinine (.1.5 mg/dL) and/or decreased creatinine clearance (,50) on admission or preadmission to the SICU in the setting of an appropriate chronic medical condition (eg, chronic renal insufficiency [CRI], cardiovascular disease, diabetes, and so on) or advanced age. Reasons for SICU admission included in the statistical analyses were abdominal surgery, craniotomy, and intracranial hemorrhage; other reasons for admission were too infrequent and did not allow for grouping. Complications recorded included systemic inflammatory response syndrome and sepsis within the first 5 days of ICU admission, and consensus guideline definitions were used.19 The primary outcome measure was the occurrence of a confirmed LEDVT or PE. LEDVT included DVT in the popliteal or more proximal veins. Upper-extremity DVTs were not included in the analyses because an IVCF does not prevent PE originating from an upper extremity. All adult patients cared for by the SICU service were enrolled, and selection criteria were applied (Fig. 1). Patients were excluded if they were admitted because of a traumatic injury, if they had a SICU LOS ,2 days, or if they did not receive an ultrasound duplex. Patients with an SICU LOS ,2 days were excluded because they were believed to be at low risk for VTE due to the potential for early mobilization and often the use of SCDs and ambulation as prophylactic measures in these patients. Of the remaining patients, only those with a contraindication to chemical prophylaxis who had no anticoagulation for at least the first 5 days of ICU admission were included. The 5-day cutoff was determined by attending consensus at our institution because it was felt that if a patient could not receive prophylactic anticoagulation for 5 days, IVCF placement would be considered depending on the patient’s risk factors for VTE. Patient characteristics, VTE risk factors, and the proportions of patients with LEDVT or PE were analyzed and compared to find variables that may contribute to a higher rate of LEDVT or PE. The Pearson chi-square or Fisher exact

Figure 1

Patient selection criteria.

tests were used to analyze categorical variables. The independent t test or Mann-Whitney U test was used to analyze continuous variables. Clinically significant variables with P , .2 in univariate analyses were subsequently included in a binomial regression model to find independent predictors of LEDVT or PE with P , .05. Separate models were used to evaluate related variables. The concordance index (c statistic) was used to determine the predictive capacity of each model, and model fit was tested by calculating a Hosmer-Lemeshow goodness-of-fit test for each model. Statistical Package for the Social Sciences software version 17.0 (SPSS Inc, Chicago, IL) was used for the analyses, and the study was approved by the Institutional Review Board of the University of California, Irvine (Irvine, CA).

Results The SICU service followed 1,269 adult patients over the 2-year study period from January 2008 to December 2009. Three hundred fifty-one of these were nontrauma patients, and they had an LEDVT rate of 8% and a PE rate of 1.1%. Of these 351 patients, 204 (58%) met the criteria for the study by having an ICU length of stay R2 days, by having a duplex before leaving the SICU, and by not receiving chemical prophylaxis within the first 5 days of admission or before an LEDVT that occurred within those first 5 days (Fig. 1). The 204 included patients had an average age of 59 years, and 57% were men. The nontrauma general surgery patients in the SICU predominantly included those

M.S. Patel et al.

Risk factors for venous thromboembolism

receiving abdominal/pelvic surgery, cranial/neurosurgery, cardiac surgery, thoracic surgery, vascular surgery, orthopedic surgery, urologic surgery, plastic surgery, and otolaryngology surgery. Specifically, the most common breakdown in our study was abdominal/pelvic surgery (36%) followed by cranial/neurologic surgery (10.8%). With the data available, further categorization becomes limited in that patients often received procedures that crossed over disciplines. Additionally, the average body mass index was 27, and patients had an SICU length of stay of 13 days and 7 ventilator days. Of the 204 patients, 20 (10%) had VTE. More specifically, 19 (9%) had LEDVT, and 2 (1.0%) had PE, with 1 patient having both. In the patient with both LEDVT and PE, LEDVT was noted 1 day before PE, and an IVCF was inserted after PE because of a persistent contraindication to anticoagulation. The overall mortality was 15%, and no patients died from PE. In the study population, 26 patients (13%) received an IVCF, with data being complete for 24 of these patients. Of these 24 patients, 11 IVCFs were placed after LEDVT or PE, and 13 were prophylactic. Of the prophylactic filters, only 1 patient (7.7%) went on to develop LEDVT. It should be noted that in the 11 patients who had an IVCF placed after the development of LEDVT, none of the patients developed PE.

303 Table 1 compares patient characteristics and VTE risk factors in patients without LEDVT or PE with those with LEDVT or PE. Six variables with P , .2 were chosen for binomial regression analysis: blood products received in the first 24 hours, a PMH of intravenous drug abuse, a PMH of renal insufficiency, a PMH of VTE, a PMH of PE, and craniotomy. Of note, a PMH of VTE and a PMH of PE are inherently related; thus, they could not be included in the same model, so 2 separate logistic regression models were used (Table 2). Model 1 was chosen because it had a slightly higher c statistic of .767 and a nonsignificant Hosmer-Lemeshow goodness-of-fit test of .254, both of which indicated an acceptable degree of discrimination20 (Table 2). Of the variables in model 1, blood products in the first 24 hours (odds ratio [OR] 5 1.04 per unit blood; confidence interval [95% CI], 1.0 to 1.07; P 5 .037), a PMH of renal insufficiency (OR 5 5.1; CI, 1.1 to 23; P 5 .036), and a PMH of PE (OR 5 10.1; CI, 1.5 to 66.6; P 5 .016) remained significant (P , .05) after multivariate analysis (Table 3).

Comments The prevention of VTE in surgical patients is a unique problem facing the intensivist. Although chemical prophylaxis

Table 1 Comparison of admission characteristics, comorbidities, reasons for admission, and complications between patients without LEDVT or PE and with LEDVT or PE (N 5 204) Variable Admission characteristics Age* Male (%)† BMI* Blood products in first 24 h (U)‡ IVF in first 48 h (mL)‡ Comorbidities (%) Outpatient warfarin PMH of cancer† PMH of IVDAx PMH of renal insufficiencyx PMH of VTEx PMH of DVTx PMH of PEx Common reasons for SICU admission (%) Abdominal surgery† ICHx Craniotomyx Complications SIRS in first 5 days† Sepsis in first 5 days†

Without LE DVT or PE (n 5 184)

With LE DVT or PE (n 5 20)

59.9 6 16.6 58.2 27.1 6 7.5 1.3 6 7.4 2,669 6 3,387

56.2 6 10.7 50 28.5 6 6.7 6.2 6 19.4 2,465 6 4,214

17.9 27.7 .5 3.8 6.5 6.0 1.6

15 15 5.0 15 15 10 10

1 .221 .187 .062 .170 .371 .076

35.3 9.2 2.2

45.0 10.0 10

.393 1 .108

43.5 30.4

45 30

P value .332 .484 .402 .081 .282

.896 .968

The bold values indicates Variables with P , .2 were chosen for binomial regression analysis and are bolded. BMI 5 body mass index; ICH 5 intracranial hemorrhage; IVDA 5 intravenous drug abuse; SIRS 5 system inflammatory response syndrome. *t test. † Chi-square test. ‡ Mann-Whitney U test. x Fisher exact test.

304 Table 2

The American Journal of Surgery, Vol 206, No 3, September 2013 Models for multivariate analysis

Model Number

Variables

C-statistic

Hosmer-Lemeshow goodness-of-fit test

Model 1

Blood products in first 24 h PMH of IVDA PMH of renal insufficiency PMH of PE Craniotomy Blood products in first 24 h PMH of IVDA PMH of renal insufficiency PMH of VTE Craniotomy

.767

.254

.754

.833

Model 2

The c-statistic has a minimum value of .5 and a maximum value of 1.0 with discriminative power categorized into the following ranges: %.6 for unacceptable discrimination, .6 to .7 for limited discrimination, .7 to .8 for acceptable discrimination, .8 to .9 for excellent discrimination, and R.9 for outstanding discrimination.20,28 A significant Hosmer-Lemeshow goodness-of-fit test (,.05) implied lack of model fit, whereas a nonsignificant test (..05) implied that the model’s estimates fit the data at an acceptable level. IVDA 5 intravenous drug abuse.

is recommended in high-risk surgery patients, many of them have contraindications to this form of anticoagulation, and, thus, their only option is mechanical prophylaxis. In such cases, the use of prophylactic IVCF placement to prevent PE may be used, but specific indications are still lacking. The aim of this study was to identify risk factors for LEDVT or PE in nontrauma SICU patients who cannot receive chemical prophylaxis so that patients who may benefit from aggressive screening and/or prophylactic IVCF placement can be better identified. In this report, LEDVT or PE was present in 10% of patients who met the inclusion criteria, and independent risk factors in a predictive model were the volume of blood products transfused within the first 24 hours of ICU admission (OR 5 1.04 per unit), a PMH of renal insufficiency (OR 5 5.1), and a PMH of PE (OR 5 10.1). In a previous systematic review of the incidence of DVTs, Attia et al21 noted the incidence of DVTs among general medical and SICU patients with no prophylaxis to be between 9% and 32%. In a 1-year prospective observational study of patients in a mixed medical SICU, Cook et al2 found the incidence of DVT to be 9.6%. This latter study may have yielded a similar incidence to the 9% Table 3 Multivariate analysis of risk factors for LEDVT or PE in surgery patients with contraindications to chemical prophylaxis (N 5 204) Variable Blood products in first 24 h (per unit) PMH of IVDA PMH of renal insufficiency PMH of PE Craniotomy

OR 1.04 15.2 5.1 10.1 5.6

95% CI

P value

1.00–1.07

.037

.9–260 1.1–23.0 1.5–66.6 0.9–36.5

.060 .036 .016 .070

The bold values indicates Variables with P , .05 were significant on binomial regression analysis and are bolded. IVDA 5 intravenous drug abuse.

(19/204) found in the current study because, similar to the current study, trauma patients were excluded and all patients received routine DVT screening. In addition to noting the incidence of VTEs, the current study specifically analyzed risk factors for VTE in critically ill nontrauma surgical patients who had contraindications to chemical prophylaxis. Comparisons to reports in the literature are limited, and conclusions should be taken with the consideration that most studies evaluating risk factors for VTE neither excluded patients based on contraindications to chemical prophylaxis nor did they specifically study critically ill, nontrauma surgical patients. In the aforementioned report by Cook et al,2 platelet transfusion was found to be an independent risk factor for DVT in a mixed medical and surgical cohort of intensive care unit patients. In a study of surgical patients, not necessarily in the ICU, Gangireddy et al22 noted postoperative transfusion to be a significant positive predictor for postoperative symptomatic VTE. In relative agreement with these previous reports, the current report found that an increased amount of blood products transfused in the first 24 hours was significant for an increase in LEDVT or PE (OR 5 1.04 per unit). For example, given this OR, if a patient had a 10-U transfusion of blood products within the first 24 hours of ICU admission, they would be at a 40% increased chance of LEDVT or PE. A possible explanation for this noted increase is that it is caused by the activation of platelets that may occur in stored units of blood products, and this activation may have an effect on patients who are transfused larger quantities of blood product.2 In addition, the requirement for blood products to be transfused postoperatively may be a marker for a larger physiologic insult from the surgery that was performed, a longer recovery period, and/or less mobility, all of which could increase the occurrence of VTE. It should be noted that blood product usage has a relatively close OR, and given that increased blood transfusions may more likely be a marker of

M.S. Patel et al.

Risk factors for venous thromboembolism

severity of illness than a cause of VTE, the relationship of this risk factor may be more associational than causal. In regards to renal insufficiency, Cook et al2 identified end-stage renal failure and Gangireddy et al22 identified acute renal insufficiency to be positive independent predictors for VTE. Interestingly, in the same study by Gangireddy et al, the authors noted hemodialysis-dependent patients to be protected from the development of VTE. The authors reasoned that both the inherent platelet dysfunction present in patients who have chronic renal failure along with the heparin anticoagulation often used in adjunct with hemodialysis to be possible explanations for their finding.22 Conversely, in order to support the finding that end-stage renal failure is a risk factor for VTE in their study of mixed medical and surgical ICU patients, Cook et al2 suggested that the associated systemic inflammation in patients with end-stage renal failure and the subsequent increase in procoagulant proteins may contribute to the increased risk of VTE formation in this subset of patients.23–25 The current study, having found a PMH of renal insufficiency to be statistically significant for an increase in VTE, corroborates the findings of Cook et al and may be similarly explained. It is important to note that a PMH of renal insufficiency in the current study was defined as previous renal insufficiency documented in a patient’s medical chart that was suggestive of a chronically decreased creatinine clearance. Unfortunately, dialysis dependence was not documented, and, thus, it could not be analyzed as a risk factor in the current study. Lastly, a personal or family history of VTE has been previously found to be an independent risk factor for DVT in a mixed medical and surgical cohort of ICU patients.2 In a different patient population, Liem et al26 found that 24% of patients with a history of LEDVT who underwent surgical procedures developed perioperative DVT extension, new-site DVT, or PE.26 Of note, this study consisted of all types of surgical patients, not specifically those who were in the SICU.26 In the current study, a PMH of PE was specifically found to increase the odds of LEDVT or PE by 10 times in critically ill surgical patients with a contraindication to chemical prophylaxis. The limitations of the study include the reliability of duplex examinations compared with a more sensitive test such as contrast venography.27 Additionally, the strict inclusion criteria may have limited the application of our data to a very specific patient cohort. Other shortcomings of the study can be attributed to limitations of the data available for analysis. For example, knowing the durations of procedures performed and the time since previous DVT or PE in patients with a PMH of VTE could have potentially added insight to the conclusions. Furthermore, a preoperative prophylaxis that may have been administered before a procedure was not documented in our dataset. Also, inclusion criteria such as bleeding risk were limited in that they were determined at the discretion of the attending surgeon and intensivist rather than by objective criteria. However, given that different physicians have different thresholds, this subjectivity likely reflects the practice in

305 multiple SICUs. Lastly, although the current report has a relatively large sample size, the possibility of type I limitations should be taken into consideration given that only 20 patients developed VTE. The strength of this investigation was that it followed a formal, prospective algorithm that was implemented for all adult critically ill surgical patients under the care of a single service. Surveillance ultrasound duplexes were routinely obtained according to a pre-established protocol, and patients were excluded from the study based on stringent criteria.

Conclusions Surgical patients are often particularly vulnerable to VTE formation. Unfortunately, chemical anticoagulation is frequently contraindicated in these patients. In this report, 10% of nontrauma SICU patients who had a contraindication and could not receive chemical prophylaxis developed LEDVT or PE. Although level 1 recommendations regarding the use of prophylactic IVCFs in this patient group are still lacking, vena cava filtration is a common last resort when anticoagulation fails to prevent VTE or when chemical prophylaxis is contraindicated. In the current investigation, only 1 patient (7.7%) who received a prophylactic filter went on to develop either LEDVT or PE, suggesting that we may be placing more IVCFs than necessary. Thus, it is critical that high-risk patients who may specifically benefit from a prophylactic IVCF be identified. Results from this study indicate that increased blood products transfused in the first 24 hours of SICU admission, a PMH of renal insufficiency, or a PMH of PE are independent predictors of VTE, and patients with these risk factors may be considered for aggressive screening and/or prophylactic IVCF placement.

References 1. Chan CM, Shorr AF. Venous thromboembolic disease in the intensive care unit. Semin Respir Crit Care Med 2010;31:39–46. 2. Cook D, Crowther M, Meade M, et al. Deep venous thrombosis in medical-surgical critically ill patients: prevalence, incidence, and risk factors. Crit Care Med 2005;33:1565–71. 3. Twigg SJ, McCrirrick A, Sanderson PM. A comparison of post mortem findings with post hoc estimated clinical diagnoses of patients who die in a United Kingdom intensive care unit. Intensive Care Med 2001;27: 706–10. 4. Cade JF. High risk of the critically ill for venous thromboembolism. Crit Care Med 1982;10:448–50. 5. Fraisse F, Holzapfel L, Couland JM, et al. Nadroparin in the prevention of deep vein thrombosis in acute decompensated COPD. The Association of Non-University Affiliated Intensive Care Specialist Physicians of France. Am J Respir Crit Care Med 2000;161:1109–14. 6. Neuhaus A, Bentz RR, Weg JG. Pulmonary embolism in respiratory failure. Chest 1978;73:460–5. 7. Cullen DJ, Nemeskal AR. The autopsy incidence of acute pulmonary embolism in critically ill surgical patients. Intensive Care Med 1986; 12:399–403. 8. Geerts WH, Bergqvist D, Pineo GF, et al. Prevention of venous thromboembolism: American College of Chest Physicians Evidence-Based

306

9. 10.

11.

12.

13. 14.

15.

16.

17.

18.

The American Journal of Surgery, Vol 206, No 3, September 2013

Clinical Practice Guidelines (8th Edition). Chest 2008;133(6 Suppl): 381S–453S. ICSI. Health Care Guidelines: VTE Thromboembolism. Available at: http://www.icsi.org. Accessed October 6, 2010. Morris CS, Rogers FB, Najarian KE, et al. Current trends in vena caval filtration with the introduction of a retrievable filter at a level I trauma center. J Trauma 2004;57:32–6. Smoot RL, Koch CA, Heller SF, et al. Inferior vena cava filters in trauma patients: efficacy, morbidity, and retrievability. J Trauma 2010;68:899–903. Imberti D, Bianchi M, Farina A, et al. Clinical experience with retrievable vena cava filters: results of a prospective observational multicenter study. J Thromb Haemost 2005;3:1370–5. Young T, Tang H, Aukes J, et al. Vena caval filters for the prevention of pulmonary embolism. Cochrane Database Syst Rev; 2007. CD006212. Gorman PH, Qadri SF, Rao-Patel A. Prophylactic inferior vena cava (IVC) filter placement may increase the relative risk of deep venous thrombosis after acute spinal cord injury. J Trauma 2009;66:707–12. Phelan HA, Gonzalez RP, Scott WC, et al. Long-term follow-up of trauma patients with permanent prophylactic vena cava filters. J Trauma 2009;67:485–9. Offner PJ, Hawkes A, Madayag R, et al. The role of temporary inferior vena cava filters in critically ill surgical patients. Arch Surg 2003;138: 591–4; discussion 594–5. Rodriguez JL, Lopez JM, Proctor MC, et al. Early placement of prophylactic vena caval filters in injured patients at high risk for pulmonary embolism. J Trauma 1996;40:797–802; discussion 802–4. Malinoski D, Ewing T, Patel MS, et al. Risk factors for venous thromboembolism in critically ill trauma patients who cannot receive chemical prophylaxis. Injury 2013;44:80–5.

19. Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/ ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003;31:1250–6. 20. LaValley MP. Logistic regression. Circulation 2008;117:2395–9. 21. Attia J, Ray JG, Cook DJ, et al. Deep vein thrombosis and its prevention in critically ill adults. Arch Intern Med 2001;161: 1268–79. 22. Gangireddy C, Rectenwald JR, Upchurch GR, et al. Risk factors and clinical impact of postoperative symptomatic venous thromboembolism. J Vasc Surg 2007;45:335–41; discussion 341–2. 23. Ikizler TA, Wingard RL, Harvell J, et al. Association of morbidity with markers of nutrition and inflammation in chronic hemodialysis patients: a prospective study. Kidney Int 1999;55:1945–51. 24. Irish A. Cardiovascular disease, fibrinogen and the acute phase response: associations with lipids and blood pressure in patients with chronic renal disease. Atherosclerosis 1998;137:133–9. 25. Irish AB, Green FR. Factor VII coagulant activity (VIIc) and hypercoagulability in chronic renal disease and dialysis: relationship with dyslipidaemia, inflammation, and factor VII genotype. Nephrol Dial Transplant 1998;13:679–84. 26. Liem TK, Huynh TM, Moseley SE, et al. Symptomatic perioperative venous thromboembolism is a frequent complication in patients with a history of deep vein thrombosis. J Vasc Surg 2010;52:651–7. 27. Davidson BL, Elliott CG, Lensing AW. Low accuracy of color Doppler ultrasound in the detection of proximal leg vein thrombosis in asymptomatic high-risk patients. The RD Heparin Arthroplasty Group. Ann Intern Med 1992;117:735–8. 28. Ohman EM, Granger CB, Harrington RA, et al. Risk stratification and therapeutic decision making in acute coronary syndromes. JAMA 2000;284:876–8.