Thrombosis Prophylaxis and Mortality Risk Among Critically Ill Adults

Transcription

1 [ Original Research Critical Care ] Thrombosis Prophylaxis and Mortality Risk Among Critically Ill Adults Craig M. Lilly, MD, FCCP ; Xinggang Liu, MD, PhD ; Omar Badawi, PharmD, MPH ; Christine S. Franey, MPH ; and Ilene H. Zuckerman, PharmD, PhD BACKGROUND: The optimal approach for managing increased risk of VTE among critically ill adults is unknown. METHODS: An observational study of 294,896 episodes of critical illness among adults was conducted in 271 geographically dispersed US adult ICUs. The primary outcomes were all-cause ICU and in-hospital mortality after adjustment for acuity and other factors among groups of patients assigned, based on clinical judgment, to prophylactic anticoagulation, mechanical devices, both, or neither. Outcomes of those managed with prophylactic anticoagulation or mechanical devices were compared in a separate paired, propensity-matched cohort. RESULTS: After adjustment for propensity to receive VTE prophylaxis, APACHE (Acute Physiology and Chronic Health Evaluation) IV scores, and management with mechanical ventilation, the group treated with prophylactic anticoagulation was the only one with significantly lower risk of dying than those not provided VTE prophylaxis (ICU, 0.81 [95% CI, ]; hospital, 0.84 [95% CI, ; P,.0001). The mortality risk of those receiving mechanical device prophylaxis was not lower than that of patients without VTE prophylaxis. A study of 87,107 pairs of patients matched for propensity to receive VTE prophylaxis found that those managed with prophylactic anticoagulation therapy had significantly lower risk of death (ICU subhazard ratio, 0.82 [95% CI, ]; hospital subhazard ratio, 0.82 [95% CI, ]; P,.001) than those receiving only mechanical device prophylaxis. CONCLUSIONS: These findings support a recommendation for prophylactic anticoagulation therapy in preference to mechanical device prophylaxis for critically ill adult patients who do not have a contraindication to anticoagulation. CHEST 2014; 146(1):51-57 Manuscript received September 12, 2013; revision accepted March 7, 2014; originally published Online First April 10, ABBREVIATIONS: APACHE 5 Acute Physiology and Chronic Health Evaluation; JC 5 Joint Commission; NQF 5 National Quality Forum; SHR 5 subhazard ratio AFFILIATIONS: From the Department of Medicine (Dr Lilly), Department of Anesthesiology and Surgery, Clinical and Population Health Research Program, and Graduate School of Biomedical Sciences, University of Massachusetts Medical School, Worcester, MA; Department of Pharmaceutical Health Services Research (Drs Liu and Zuckerman and Ms Franey) and Department of Pharmacy Practice and Science (Dr Badawi), University of Maryland School of Pharmacy, Baltimore, MD; and Philips Healthcare (Dr Badawi), Baltimore, MD. FUNDING/SUPPORT: The database and support for analyses performed by Drs Liu and Zuckerman and Ms Franey were provided by the eicu Research Institute. CORRESPONDENCE TO: Craig M. Lilly, MD, FCCP, Department of Medicine, University of Massachusetts Medical School, UMass Memorial Medical Center, 281 Lincoln St, Worcester, MA 01605; craig.lilly@umassmed.edu 2014 AMERICAN COLLEGE OF CHEST PHYSICIANS. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details. DOI: /chest journal.publications.chestnet.org 51

2 Evidence-based strategies used for the prevention of venous thrombosis vary based on the specific risk-benefit profiles in studied populations. The ninth edition of the American College of Chest Physicians Antithrombotic Therapy and Prevention of Thrombosis guidelines recommend VTE prophylaxis tailored to VTE risk, treatment effects, and strength of evidence, which are distinct for a growing number of populations. 1 Highquality evidence, including evidence of mortality benefits, 2 supports the use of prophylactic anticoagulation (Grade 1B) or mechanical devices (Grade 1C) over no prophylaxis for high-risk surgical patients. 3 The association of VTE prophylaxis and mortality outcomes is much less clear, and the quality of the evidence for clinical practice guidelines is weaker for critically ill adults than for high-risk surgical popu lations. Current guidelines are supported by studies demonstrating that critically ill adults are at increased risk for venous thrombosis 4 and that incidence of symptomatic DVT and fatal pulmonary embolism in hospitalized medical patients can be reduced by anticoagulants. 5-7 Th e quality of the evidence supporting prophylactic anticoagulation therapy over no prophylaxis and for the use of graduated compression stockings or mechanical devices for patients who are bleeding or at high risk of bleeding is Grade 2C. Recommendations regard ing the choice of anticoagulation or mechanical devices are weak in part because a randomized trial of elastic stockings with or without enoxaparin in medical patients did not detect significant differences in all-cause mor tality. 8 Th e low quality of the evidence regarding the effectiveness of alternative forms of VTE prophylaxis for critically ill adults has led to weak recommendations and heterogeneity of practice. We measured the heterogeneity of current practice and investigated the associations among the alternative approaches to VTE prophylaxis and all-cause mortality. The overarching aim of this study is to improve the quality of the available evidence by identifying VTE prophylaxis strategies associated with lower all-cause mortality among critically ill adults. Materials and Methods The primary outcomes of this cohort study were adjusted ICU and hospital mortality among groups of critically ill adults managed with prophylactic anticoagulation therapy, thromboprophylaxis with a mechanical device, both, or neither, as assigned by clinical judgment. The study included all adult patients discharged alive or dead from participating ICUs from January 1, 2008, to September 30, Data were derived from patient information contained in the Philips eicu Research Institute data repository, 9 using abstraction, privacy protection, data aggregation, and mapping techniques as previously described. 10 Observations were entered by clinicians or transferred from a clinical information system and mapped to equivalent concepts in the eicu Research Institute database. Consistent physiologic, laboratory, diagnosis, treatment, physical examination elements, and nursing flow sheet data were included in the electronic record for all patients for the duration of their ICU stay. Acuity was measured using APACHE (Acute Physiology and Chronic Health Evaluation) IV software sublicensed from Cerner Corp. VTE Best Practice Adherence The VTE prophylaxis criteria were concordant with those endorsed by the Joint Commission (JC) and the National Quality Forum (NQF) for critically ill adult patients (NQF #0372). All adults with an ICU patient stay. 24 h were considered for inclusion. Exclusion criteria were documentation in the medical record indicating that the patient was ambulating, coagulopathic, or fully anticoagulated or received a thrombin inhibitor (argatroban, bivalirudin, lepirudin, or refludan) during the first 24 h of ICU stay. Patients were classified as being treated with a mechanical device when the medical record indicated that either an inferior vena cava filter or a compression device (boots or stockings) was in use. They were classified as receiving prophylactic anticoagulation ther apy when an order for an anticoagulant medication (dabigatran, dalteparin, enoxaparin, fondaparinux, heparin, tinzaparin, or warfarin) was present or as receiving a combination of a device and prophylactic anticoagulation when both were documented. The time of prescription was recorded as within or. 48 h after ICU admission. The medical record was reviewed by an off-site team tasked with confirming that the approach was adherent to the JC/NQF ICU thromboprophylaxis measures. All drug orders were reviewed by a pharmacist. Patients managed with mechanical devices, which are classified as JC/NQF measure-adherent for the purposes of this study, would not be classified as adherent to the ninth edition of the American College of Chest Physicians Antithrombotic Ther apy and Prevention of Thrombosis guidelines3 when they did not have a contraindication to anticoagulation. Statistical Analyses Patient characteristics were compared among groups managed with alternative approaches to preventing venous thrombosis (ie, prophylactic anticoagulation, mechanical device thromboprophylaxis, both, neither). Continuous variables were reported as means and SDs, and categorical variables were reported as counts and percentages. Group comparisons were made using the x 2 test for categorical variables and analysis of variance for continuous data, as appropriate. Because of the large sample size, statistical significance was set at the 0.01 level using two-sided distributions. Logistic regression was performed with mortality as the dependent variable. The ORs for mortality of the various treatment groups were estimated using the group that did not receive VTE prophylaxis as the reference, with and without adjustment for covariates. To control for potential confounding in the setting of not randomly assigning alternative management strategies, an inverse probabilityweighted propensity score method was used to adjust for differences in selected risk factors among patients in the four alternative management groups. The probability of receiving one of the four VTE prophylaxis management strategies (prophylactic anticoagulation and mechanical device, prophylactic anticoagulation without a mechanical device, a mechanical device only, or no prophylaxis) was estimated by multinomial logit regression. The study patients were weighted by their inverse probabilities of being in their assigned group. 11,12 Variables that were not balanced after propensity score weighting were included in a logistic regression model with ICU or hospital mortality as the dependent variable. 52 Original Research [ 146 # 1 CHEST JULY 2014 ]

3 To further elucidate the comparative effectiveness of receiving anticoagulation therapy vs mechanical device prophylaxis, separate propensity score-matched studies were performed among patients managed with a VTE prophylaxis strategy using the method developed by Fine and Gray 13 that included survival in a competing-risk cumulative incidence function model. Patients who received either anticoagulation with or without a mechanical device or mechanical device alone as VTE prophylaxis were matched one to one (no replacement with a caliper of ) on their probabilities of receiving prophylactic anticoagulation therapy. This analysis was repeated in the subgroup of patients having only prophylactic anti coagulation compared with only mechanical devices. The model estimated hazard ratios associated with the cumulative incidence function (subhazard ratio [SHR]) between patients receiving prophylactic anticoagulation and those managed only with mechanical devices. Statistical analyses were performed using SAS version 9.2 (SAS Institute Inc) and Stata 12 (StataCorp LP). Results Patient Characteristics This study included records from 294,896 hospital discharge events from 5,321 adult ICU beds in 271 ICUs located in 188 hospitals and 32 health-care systems from 31 states. Patients who failed to meet age, date, or validity criteria or were not at risk for VTE were excluded from the study ( Fig 1 ). Patient demographic characteristics by VTE prophylaxis group are presented in Table 1. Almost all patients (93%) were treated with some form of VTE prophylaxis; a combination of prophylactic anticoagulation and device prophylaxis were prescribed for 32%, anticoagulation without a device for 27%, and a device without anticoagulation for 34%. Less than 1% (1,804) of the 196,671 patients who received mechanical device prophylaxis had an inferior vena cava filter in place, and 1,159 of these patients also received prophylactic anticoagulation. Outcomes as a Function of VTE Prophylaxis Group Patients not prescribed prophylaxis were more likely to be younger, male, and of Hispanic ethnicity or unrecorded race; were less frequently managed with mechanical ventilation; were more often cared for in a cardiovascular ICU; and had lower APACHE IV acuity scores than those prescribed prophylaxis. After adjustment for propensity of treatment derived using inverse probability weighting, acuity, and treatment with mechanical ventilation, significantly lower odds of dying in the ICU and hospital were observed for patients managed with prophylactic anticoagulation than those in the other groups (Table 2 ). These analyses suggest that those managed with both prophylactic anticoagulation and mechanical devices had significantly higher mortality risk than those managed with prophylactic anticoagulation alone ( Table 2 ). Moreover, patients managed with mechanical devices alone had significantly higher odds of dying in the ICU or hospital than those not managed with VTE prophylaxis. Similar results were observed when the unadjusted analyses were repeated for all patients at risk for VTE with valid mortality outcomes but missing APACHE IV scores (e-table 1). Figure 1 Study enrollment and exclusions. APACHE 5 Acute Physiology and Chronic Health Evaluation. Outcomes for VTE Prophylaxis Propensity-Matched Groups Comparison of SHRs for ICU and hospital mortality among patients receiving prophylactic anticoagulation (with or without mechanical device therapy) or mechanical device treatment (without pharmacotherapy) was made among patients matched for their likelihood of being prescribed prophylactic anticoagulation. The population distribution of propensity scores stratified by VTE prophylaxis group is presented in Figure 2A. Propensity score matching successfully selected pairs of patients prescribed prophylactic anticoagulation or mechanical devices without pharmacotherapy that were well balanced on the following risk factors: mechanical ventilation, APACHE IV score, type of ICU, race/ethnicity, journal.publications.chestnet.org 53

4 TABLE 1 ] Baseline Patient Characteristics Among 294,896 Critically Ill Adults Discharged From the Hospital Between January 1, 2008, and September 30, 2010 Prophylaxis Group Characteristic Medication and Device (n 5 94,391) Medication Only (n 5 78,385) Device Only (n 5 100,712) Prophylaxis Not Prescribed (n 5 21,408) P Value Age, y ,.001 Female sex 45,374 (48.1) 38,184 (48.7) 45,745 (45.4) 9,457 (44.2),.001 Race,.001 African American 8,977 (9.5) 9,694 (12.4) 9,969 (9.9) 2,459 (11.5) White 72,161 (76.5) 58,280 (74.4) 74,756 (74.2) 14,534 (67.9) Hispanic 3,663 (3.9) 3,359 (4.3) 4,941 (4.9) 1,227 (5.7) Other/unknown 9,590 (10.2) 7,052 (9.0) 11,046 (11.0) 3,188 (14.9) Admission surgery 24,667 (26.1) 5,220 (6.7) 30,539 (30.3) 4,240 (19.8),.001 Sepsis a 12,231 (13.0) 8,850 (11.3) 7,133 (7.1) 1,671 (7.8),.001 Heart failure a 9,198 (9.7) 10,580 (13.5) 4,802 (4.8) 1,533 (7.2),.001 Mechanical ventilation a 41,869 (44.4) 15,273 (19.5) 29,717 (29.5) 3,127 (14.6),.001 Any cancer diagnosis 8,476 (9.0) 4,092 (5.2) 7,781 (7.7) 1,135 (5.3),.001 APACHE IV score ,.001 Pre-ICU hospital LOS, d ,.001 ICU type number,.001 Cardiovascular 24,939 (26.4) 25,323 (32.3) 26,482 (26.3) 7,675 (35.9) Medical 8,071 (8.6) 8,365 (10.7) 8,119 (8.1) 995 (4.7) Mixed or other 53,786 (57.0) 42,186 (53.8) 56,561 (56.2) 12,088 (56.5) Surgical 7,595 (8.1) 2,511 (3.2) 9,550 (9.5) 650 (3.0) Data are presented as mean SD or No. (%). APACHE 5 Acute Physiology and Chronic Health Evaluation; LOS 5 length of stay. a Within 24 h of admission to an adult ICU. cancer diagnosis, sepsis diagnosis, heart failure diagnosis, sex, age, postoperative status, and length of hospital stay before ICU admission ( Fig 2B ). An analysis of 87,107 pairs of propensity-matched patients demonstrated that critically ill adults managed with prophylactic anticoagulation were significantly less likely to die in the ICU (SHR, 0.82; 95% CI, ; P,.001) or hospital (SHR, 0.82; 95% CI, ; P,.001) than their counterparts prescribed mechanical device prophylaxis without prophylactic antico agulation. Analyses of 53,294 pairs grouped as having only prophylactic anticoagulation and only mechanical devices yielded similar results for dying in the ICU (SHR, 0.87; 95% CI, ; P,.001) or hospital (SHR, 0.82; 95% CI, ; P,.001). Discussion The main finding of this large observational study is that adult patients in the ICU who are managed with prophylactic anticoagulation have lower adjusted mortality than those managed with mechanical devices or not provided thrombosis prophylaxis. The risks of ICU and in-hospital mortality in patients managed without VTE prophylaxis were lower than in those managed with VTE prophylaxis based on analyses that did not adjust for differences in acuity ( Table 2 ). Analyses that adjusted for acuity, propensity for prophylaxis score, and management with mechanical ventilation demonstrated significantly lower ICU and hospital mortality for the group prescribed prophylactic anticoagulation without mechanical device thromboprophylaxis than for the other groups. Overall, patients who received VTE prophylaxis tended to be older; tended to have a higher prevalence of admission for surgery, sepsis, heart failure, and cancer; and were more likely to be managed with mechanical ventilation, which indicates that they were at higher risk of dying in the ICU. Adjustment for acu ity caused some of the ORs to change from harmful to protective. Comparison of the prophylactic anticoagulation group to the combined prophylaxis group suggested that some of the benefits of prophylactic anticoagulation may be reduced by mechanical devices. The odds of dying in the hospital for the mechanical device only group was significantly higher than that of the 54 Original Research [ 146 # 1 CHEST JULY 2014 ]

5 TABLE 2 ] ICU and Hospital Mortality as a Function of VTE Prophylaxis Group Mortality Unadjusted Anticoagulation and Device (n 5 94,391) VTE Prophylaxis Group Anticoagulation Only (n 5 78,385) Mechanical Device Only (n 5 100,712) Prophylaxis Not Prescribed (n 5 21,408) ICU 2.11 ( ), P, ( ), P, ( ), P,.0001 Reference Hospital 1.87 ( ), P, ( ), P, ( ), P,.0001 Reference Adjusted a ICU 0.96 ( ), P ( ), P, ( ), P Reference Hospital 1.00 ( ), P ( ), P, ( ), P,.0001 Reference Data are presented as OR (95% CI). See Table 1 legend for expansion of abbreviation. a After adjustment for propensity score, APACHE IV score, and treatment with mechanical ventilation within 24 h of ICU admission. group not managed with VTE prophylaxis, raising concern that these devices could be harmful under some circumstances. The modest fraction of patients in this study who were managed without VTE prophylaxis and the possibility that they may have been clinically selected using a factor that eluded our methods of adjustment make the issue of harm to adult critically ill patients from VTE prophylaxis with mechanical devices an objective for future study. This study also suggests that high levels of adherence to VTE prophylaxis guidelines are achievable. A form of VTE prophylaxis was prescribed for 93% of nonambulatory critically ill adults. The patients for whom clinicians chose not to prescribe a form of VTE prophylaxis were younger, had lower acuity scores and mortality risk, were more frequently men of Hispanic ethnicity or unrecorded race and admitted to a cardiovascular ICU, and were less frequently managed with mechanical ventilation than those prescribed VTE prophylaxis. The high prevalence of VTE prophylaxis and heterogeneity of practice, with a nearly equal dis tribution based on the form of VTE prophylaxis prescribed, allowed us to determine whether propensity for VTE prophylaxismatched patients managed with prophylactic anticoagulation or mechanical devices alone was associated with a lower mortality risk. We analyzed 87,107 pairs of propensity-matched patients managed with either prophylactic anticoagulation (with or without a mechanical device) or a mechanical device without prophylactic anticoagulation. The SHRs for ICU and hospital mortality were significantly lower for the group prescribed prophylactic anticoagulation ther apy than for the group prescribed a mechanical device without pharmacotherapy. The 18% lower risk of dying in the ICU or hospital demonstrates that prescription of prophylactic anticoagulation was associated with clinically important differences in mortality risk. Clinical trials in postsurgical patients demonstrated that anticoagulants and mechanical devices have equivalent effectiveness 14 ; trials in noncritically ill medical patients 8 and trials showing reduced rates of DVT among immobile patients with acute stroke treated with intermittent compression devices 15 do not appear to apply to heterogeneous clinical populations of critically ill adults. These findings are compatible with the hypotheses that anticoagulants are inherently more effective than mechanical devices for critically ill adults and that prophylactic anticoagulation is more reliably delivered in critical care practice than is mechanical device prophylaxis. Audits in postoperative 16 and trauma patients 17 prescribed intermittent pneumatic compression documented that one-half of the patients had nonfunctional devices due to lack of power or tube occlusion. In addition to a lack of benefit when devices are nonfunctional, compression stockings and boots are removed periodically in clinical practice. Taken together, these findings provide a higher level of evidence to support the ninth edition of the American College of Chest Physicians Antithrombotic Therapy and Prevention of Thrombosis guidelines recommendation 1 that nonambulatory critically ill adults who do not have contraindications receive prophylactic anticoagulation for the prevention of VTE. This large observational study of clinical practices and outcomes has several limitations. First, treatments were not randomly assigned, and there were clinically important differences among the groups. Clinician selection of treatment can introduce bias into clinical practice studies that have not been fully adjusted for. In our comparative effectiveness analysis of anticoagulation vs mechanical device alone, some treatment selection bias was mitigated by the fact that all patients included in the propensity-matched analyses had been treated with journal.publications.chestnet.org 55

6 Figure 2 A, Histogram of propensity scores among patients managed with prophylactic anticoagulation treatment and those treated with only a mechanical device. B, Standardized difference of covariates (prophylactic anticoagulation vs mechanical device only) before and after propensity matching. Note that a positive % bias indicates that a patient who received a prophylactic anticoagulant had a higher probability of receiving pharmacotherapy than a patient managed only with mechanical device thromboprophylaxis. Conversely, a negative % bias indicates that a patient who received a prophylactic anticoagulant had a lower probability of receiving pharmacotherapy than a patient managed only with mechanical device thromboprophylaxis. See Figure 1 legend for expansion of abbreviation. some form of VTE prophylaxis. This clinical practice study may provide different insights into the effectiveness of thromboprophylaxis in real-world practice settings than can be inferred from available randomized clinical trials. Generalization to clinical practice from clinical trials of thromboprophylaxis has been problematic because even the best trials did not enroll more than one-third of eligible patients. 18 In addition, interpretation of the increased mortality associated with the prescription of a mechanical device is limited by a lack of information regarding which patients had devices that were nonfunctional or removed for substantial intervals. Finally, the rates of adherence with VTE prophylaxis guidelines may be lower for ICUs that did not 56 Original Research [ 146 # 1 CHEST JULY 2014 ]

7 use real-time population management tools that were available for this study population or that did not measure and report rates of adherence to the clinical leaders. Despite these important limitations, the study provides important new information that supports the use of prophylactic anticoagulation for critically ill adults based on mortality benefits. In summary, the association of prophylactic anticoagulation with lower hospital and ICU mortality risk among nonambulatory critically ill adults supports the recommendations of the current thromboprophylaxis guidelines. 3,19 The association of lower ICU and hospital mortality risk with antithrombotic pharmacotherapy than VTE prophylaxis with mechanical devices alone supports the ninth edition of the American College of Chest Physicians Antithrombotic Therapy and Prevention of Thrombosis guidelines 3 recommendation for the preferential use of anticoagulants for VTE prevention in critically ill adult patients who do not have contraindications. Acknowledgments Author contributions: C. M. L. had full access to the data and takes responsibility for its integrity and the accuracy of the analyses. C. M. L., O. B., and I. H. Z. contributed to the study concept; C. M. L., X. L., O. B., C. S. F., and I. H. Z. contributed to the study design; C. M. L. and O. B. contributed to data acquisition; C. M. L., X. L., O. B., C. S. F., and I. H. Z. contributed to data interpretation; X. L., O. B., C. S. F., and I. H. Z. contributed to data analysis; X. L. and C. S. F. contributed to structuring the databases; C. M. L. contributed to the writing of the first and final versions; and X. L., O. B., C. S. F., and I. H. Z. contributed to the review and editing of all versions of the manuscript. Financial/nonfinancial disclosures: Th e authors have reported to CHEST the following conflicts of interest: Dr Badawi is an employee of Philips Healthcare. Dr Zuckerman is a principal investigator of a contract funded by Philips Healthcare. Drs Lilly and Liu and Ms Franey have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Other contributions: This manuscript has been reviewed and approved by the publication committee of the eicu Research Institute. Additional information: The e-table can be found in the Supplemental Materials section of the online article. References 1. Kahn SR, Lim W, Dunn AS, et al. Prevention of VTE in nonsurgical patients: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidencebased clinical practice guidelines. Chest ;141(2_suppl):e195S-e226S. 2. Collins R, Scrimgeour A, Yusuf S, Peto R. Reduction in fatal pulmonary embolism and venous thrombosis by perioperative administration of subcutaneous heparin. Overview of results of randomized trials in general, orthopedic, and urologic surgery. N Engl J Med ;318(18): Guyatt GH, Akl EA, Crowther M, Gutterman DD, Schuünemann HJ. Executive summary: antithrombotic therapy and preven tion of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest ; 141(2_suppl):7S-47S. 4. Spyropoulos AC, Anderson FA Jr, Fitzgerald G, et al ; IMPROVE Investigators. Predictive and associative models to identify hospitalized medical patients at risk for VTE. Chest ; 140 (3): Dentali F, Douketis JD, Gianni M, Lim W, Crowther MA. Meta-analysis: anticoagulant prophylaxis to prevent symptomatic venous thromboembolism in hospitalized medical patients. Ann Intern Med ;146(4): Lloyd NS, Douketis JD, Moinuddin I, Lim W, Crowther MA. Anticoagulant prophylaxis to prevent asymptomatic deep vein thrombosis in hospitalized medical patients: a systematic review and meta-analysis. J Thromb Haemost ;6(3): Alikhan R, Cohen AT. Heparin for the prevention of venous thromboembolism in general medical patients (excluding stroke and myocardial infarction). Cochrane Database Syst Rev ;(3):CD Kakkar AK, Cimminiello C, Goldhaber SZ, Parakh R, Wang C, Bergmann JF ; LIFENOX Investigators. Low-molecular-weight heparin and mortality in acutely ill medical patients. N Engl J Med ;365(26): McShea M, Holl R, Badawi O, Riker RR, Silfen E. The eicu Research Institute - a collaboration between industry, healthcare providers, and academia. IEEE Eng Med Biol Mag ; 29 (2): Lilly CM, Zuckerman IH, Badawi O, Riker RR. Benchmark data from more than 240,000 adults that reflect the current practice of critical care in the United States. Chest ;140(5): Imbens GW. The role of the propensity score in estimating dose-response functions. Biometrika ;87(3): Foster EM. Propensity score matching: an illustrative analysis of dose response. Med Care ;41(10): Fine JP, Gray RJ. A proportional hazards model for the sub distribution of a competing risk. J Am Stat Assoc ;94(446): Eppsteiner RW, Shin JJ, Johnson J, van Dam RM. Mechanical compression versus subcutaneous heparin therapy in postoperative and posttrauma patients: a systematic review and meta-analysis. World J Surg ;34(1): Dennis M, Sandercock P, Reid J, Graham C, Forbes J, Murray G ; CLOTS (Clots in Legs Or stockings after Stroke) Trials Collaboration. Effectiveness of intermittent pneumatic compression in reduction of risk of deep vein thrombosis in patients who have had a stroke (CLOTS 3): a multicentre randomised controlled trial. Lancet ;382(9891): Comerota AJ, Katz ML, White JV. Why does prophylaxis with external pneumatic compression for deep vein thrombosis fail? Am J Surg ;164(3): Cornwell EE III, Chang D, Velmahos G, et al. Compliance with sequential compression device prophylaxis in at-risk trauma patients: a prospective analysis. Am Surg ;68(5): Cook D, Meade M, Guyatt G, et al ; PROTECT Investigators for the Canadian Critical Care Trials Group and the Australian and New Zealand Intensive Care Society Clinical Trials Group. Dalteparin versus unfractionated heparin in critically ill patients. N Engl J Med ;364(14): Kearon C, Akl EA, Comerota AJ. Antithrombotic therapy for VTE disease: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest ; 141(2_suppl):e419S-e494S. journal.publications.chestnet.org 57