Venous Thromboembolism Prophylaxis in Gynecologic Surgery A Systematic Review

Transcription

1 Venous Thromboembolism Prophylaxis in Gynecologic Surgery A Systematic Review David D. Rahn, MD, Mamta M. Mamik, MD, Tatiana V. D. Sanses, MD, Kristen A. Matteson, MD, MPH, Sarit O. Aschkenazi, MD, MS, Blair B. Washington, MD, Adam C. Steinberg, DO, Heidi S. Harvie, MD, MBA, MSCE, James C. Lukban, DO, Katrin Uhlig, MD, MS, Ethan M. Balk, MD, MPH, and Vivian W. Sung, MD, MPH, for the Society of Gynecologic Surgeons Systematic Review Group Reviews OBJECTIVE: To comprehensively review and critically assess the available gynecologic surgery venous thromboembolism prophylaxis literature and provide clinical practice guidelines. DATA SOURCES: MEDLINE and Cochrane databases from inception to July We included randomized controlled trials in gynecologic surgery populations. Interventions and comparators included graduated compression stockings, intermittent pneumatic compression, See related editorial on page 973 and related article on page 978. From the University of Texas Southwestern Medical Center, Dallas, Texas; the University of New Mexico Health Sciences Center, Albuquerque, New Mexico; the University Hospitals Case Medical Center, Cleveland, Ohio; Women and Infants Hospital of Rhode Island/Brown Medical School, Providence, Rhode Island; Medical College of Wisconsin, Waukesha, Wisconsin; Hartford Hospital/University of Connecticut School of Medicine, Hartford, Connecticut; the University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania; Eastern Virginia Medical School, Norfolk, Virginia; and Tufts Medical Center, Boston, Massachusetts. The systematic review methodology experts, Drs. Balk and Uhlig, both of the Tufts Clinical and Translational Science Institute, were paid by the Society of Gynecologic Surgeons for their consultation and guidance. The authors thank Moritz Bartels, MD, and Karl Tamussino, MD, for translation of German-language studies and Joseph I. Schaffer, MD, and Matthew D. Barber, MD, MHS, from the Society of Gynecologic Surgeons for guidance and support. Presented at the 37th Annual Scientific Meeting of the Society of Gynecologic Surgeons, April 11 13, 2011, San Antonio, Texas. Corresponding author: David D. Rahn, MD, Department of Obstetrics & Gynecology, Division of Female Pelvic Medicine & Reconstructive Surgery, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX ; David.Rahn@UTSouthwestern.edu. Financial Disclosure The authors did not report any potential conflicts of interest by The American College of Obstetricians and Gynecologists. Published by Lippincott Williams & Wilkins. ISSN: /11 unfractionated heparin, and low molecular weight heparin; placebo and routine postoperative care were allowed as comparators. METHODS OF STUDY SELECTION: One thousand two hundred sixty-six articles were screened, and 14 randomized controlled trials (five benign gynecologic, nine gynecologic oncology) met eligibility criteria. In addition, nine prospective or retrospective studies with at least 150 women were identified and provided data on venous thromboembolism risk stratification, gynecologic laparoscopy, and urogynecologic populations. TABULATION, INTEGRATION, AND RESULTS: Two reviewers independently screened articles with discrepancies adjudicated by a third. Eligible randomized controlled trials were extracted for these characteristics: study, participant, surgery, intervention, comparator, and outcomes data, including venous thromboembolism incidence and bleeding complications. Studies were individually and collectively assessed for methodologic quality and strength of evidence. Overall incidence of clinical venous thromboembolism was 0 2% in the benign gynecologic population. With use of intermittent pneumatic compression for benign major procedures, venous thromboembolism incidence was less than 1%. No venous thromboembolisms were identified in prospective studies of benign laparoscopic procedures. Overall quality of evidence in the benign gynecologic literature was poor. Gynecologic oncology randomized controlled trials reported venous thromboembolism incidence (including silent venous thromboembolisms) of % with prophylaxis and up to 34.6% without prophylaxis. Fair quality of evidence supports that unfractionated heparin and intermittent pneumatic compression are both superior to placebo or no intervention but insufficient to determine whether heparins are superior to intermittent pneumatic compression for venous thromboembolism prevention. Combining two of three risks VOL. 118, NO. 5, NOVEMBER 2011 OBSTETRICS & GYNECOLOGY 1111

2 (aged 60 years or older, cancer, or personal venous thromboembolism history) substantially elevated the risk of venous thromboembolism. CONCLUSION: Intermittent pneumatic compression provides sufficient prophylaxis for the majority of gynecology patients undergoing benign surgery. Additional risk factors warrant the use of combined mechanical and pharmacologic prophylaxis. (Obstet Gynecol 2011;118: ) DOI: /AOG.0b013e318232a394 Venous thromboembolism represents a major burden to the US healthcare system. Approximately 900,000 patients present annually with clinically evident venous thromboembolism resulting in 300,000 deaths from pulmonary embolism. 1 The remaining cases yield several hundred thousand primary or extended hospitalizations. In studies using objective criteria to screen for asymptomatic and symptomatic venous thromboembolism, patients undergoing major gynecologic surgery without venous thromboembolism prophylaxis have an absolute risk of deep vein thrombosis (DVT) between 17% 2 and 40%. 2,3 Accordingly, the Agency for Healthcare Research and Quality has identified venous thromboembolism prevention as a high priority with great potential to improve the safety of hospitalized patients. 4 Organizations such as the American College of Chest Physicians have provided reviews on perioperative venous thromboembolism prophylaxis 3 in a gynecologic population based, in part, on evidence from the general surgery, urology, and colorectal literature. However, conclusions on the efficacy and safety of various methods of venous thromboembolism prophylaxis drawn from studies including men and patients undergoing nongynecologic procedures may not be directly applicable to gynecologic surgery populations. Although women overall have a lower risk of venous thromboembolism compared with men (105 compared with 114 per 100,000), 1 some women have unique risk factors that may alter venous thromboembolism risk, including use of estrogen therapy and the thrombogenic postpartum period. 5 Furthermore, gynecologic procedures often require unique patient positioning supine, high-lithotomy, low-lithotomy with and without laparoscopic insufflation. Gynecologic oncology patients have an inherently higher risk of thrombotic events given the thrombogenic nature of cancer, extensive retroperitoneal and pelvic dissection, and prolonged operative times. The most recent Practice Bulletin, Prevention of Deep Vein Thrombosis and Pulmonary Embolism, No. 84, August 2007, from the American College of Obstetricians and Gynecologists, provides a detailed review but lacks information on minimally invasive or laparoscopic gynecologic surgery. 6 The aim of this review was to comprehensively evaluate available evidence regarding thromboprophylaxis exclusively in benign and malignant gynecologic surgery populations and to identify and recommend the best (ie, most effective and safest) means of prophylaxis against perioperative venous thromboembolism. SOURCES The Society of Gynecologic Surgeons Systematic Review Group, including gynecologic surgeons and systematic review methodologists, performed a systematic search to identify studies comparing interventions aimed at venous thromboembolism prevention in women undergoing gynecologic surgery following the approach outlined by Counsell. 7 MEDLINE and Cochrane Central Register of Controlled Trials were searched from their inception until July 16, Among the Medical Subject Headings searched were multiple gynecologic surgical procedures, genital neoplasms, general surgery, pelvis and pelvic floor, vagina, cervix, uterus, prolapse, urinary incontinence, rectocele, cystocele, various anticoagulants, intermittent pneumatic compression devices, compression boots and stockings, DVT, pulmonary embolism, and venous thromboembolism (further search details available on request). We did not attempt to identify unpublished articles or abstracts, and we did not contact study authors. The search was limited to human subjects. Reference lists of selected articles and review articles were screened for additional eligible references, and relevant articles subsequent to our search date were identified from high-impact obstetrics and gynecology journals and included in our review. STUDY SELECTION Based on predefined criteria, participants of interest were women undergoing any benign and oncologic gynecologic surgical procedures. Studies including women who had other types of abdominal or vaginal surgical procedures were included if the gynecologic surgery data were separately identifiable. Interventions and comparators included graduated compression stockings, intermittent pneumatic compression, unfractionated heparin or low-dose unfractionated heparin, and low molecular weight heparin. Routine postoperative care (eg, early ambulation) and placebo were also acceptable comparators. We excluded interventions not approved for venous thromboembo Rahn et al Surgical Venous Thromboembolism Prophylaxis OBSTETRICS & GYNECOLOGY

3 lism prophylaxis by the Food and Drug Administration. The primary outcome of interest was incidence of postoperative venous thromboembolism, which could be assessed through universal objective testing or clinically evident DVT or pulmonary embolism confirmed by additional testing. Other outcomes of interest were intra- and postoperative bleeding complications as suggested by need for blood transfusion, change in hemoglobin concentration, estimated blood loss, return to the operating room for bleeding, hematoma, and clotting profile laboratory values. The titles, abstracts, and full texts (when necessary) were screened for eligibility by two reviewers, and any discrepancies were resolved by a third reviewer. Data extraction was then completed by two independent reviewers, most with experience in the systematic review process. 8,9 The following data were collected: 1) study characteristics (year, design, number enrolled or analyzed, length of follow-up); 2) patient and surgery characteristics (age of the study population, indication for surgery, type of surgery, additional risk factors for venous thromboembolism, operative time, and patient positioning); 3) intervention (type of prophylaxis, medication details [dose, frequency, duration], and length of follow-up) and comparator; and 4) outcomes listed previously. The primary systematic review included randomized controlled trials in any language. In addition, the original search was rescreened for English-language, gynecology-only prospective studies (any size) and retrospective studies with at least 150 women; these studies provided additional data on venous thromboembolism risk stratification, gynecologic laparoscopy, and urogynecologic populations, which were useful in guideline development. We assessed the methodologic quality of each study using predefined criteria from a three-category system modified from the Agency for Healthcare Research and Quality. 10 Studies were graded as good (A), fair (B), or poor (C) quality based on the likelihood of biases and the completeness of reporting. The quality of individual outcomes was separately graded within each study. Data from eligible articles were extracted and then grouped by types of study (ie, benign compared with cancer with similar interventions and comparators). For each group, we generated an evidence profile by grading the quality of evidence for each outcome (eg, venous thromboembolism incidence, transfusion rate, estimated blood loss) across studies. This process considered the methodologic quality, consistency of results across studies, directness of evidence, and other factors such as imprecision or sparseness of evidence to determine an overall quality of evidence in accordance with the Grades for Recommendation, Assessment, Development and Evaluation system. This system categorizes based on four quality ratings: high (A), moderate (B), low (C), and very low (D). 11 We then developed guideline statements incorporating the balance between benefits and harms of the compared interventions and the overall quality of evidence across all outcomes of interest. Each guideline statement was assigned an overall level of strength to the recommendation (1 strong or 2 weak ) based on the quality of the supporting evidence and the size of the net medical benefit. This strength of a recommendation indicates the extent to which one can be confident that adherence to the recommendation will do more good than harm. The wording and its implications for patients, physicians, and policymakers are detailed in the Conclusions. We decided a priori that meta-analyses would only be performed for topics with at least three studies with similar interventions, study designs, and outcome definitions. No topic had a sufficient number of studies and consequently no meta-analysis was performed. The review and guidelines were presented for public comment at the 37th Society of Gynecologic Surgeons Annual Scientific Meeting April 2011 and posted on the Society of Gynecologic Surgeons web site, where public comments were solicited for 4 weeks. RESULTS The MEDLINE and Cochrane database search identified 1,266 citations. After title and abstract screening, 117 full-text articles were evaluated in detail. A total of 15 articles reporting the findings from 14 randomized controlled trials met all inclusion criteria for the primary systematic review (Table 1). Five trials included benign gynecology populations and nine (in 10 articles) included gynecologic oncology populations All 14 randomized controlled trials evaluated women undergoing major procedures. The majority of the women in these studies underwent total abdominal hysterectomy with or without lymph node staging or other laparotomies in the supine position. An exception was the report by Studer- Schar et al in which 63% of the cases were total vaginal hysterectomies. 13 Study quality in the benign gynecologic population was poor 12,15,16 to fair 13,14 ; the oncologic literature included four good (A), 19,23,24,26 four fair (B), 17,20,22,25 and one poor (C) 21 randomized controlled trial. Sample sizes ranged from 28 to 822 participants per study arm. 12,20 DVTs were diagnosed by venography, I-fibrinogen-uptake test, or Dopp- VOL. 118, NO. 5, NOVEMBER 2011 Rahn et al Surgical Venous Thromboembolism Prophylaxis 1113

4 Table 1. Description of Randomized Controlled Trials of Venous Thromboembolism Prophylaxis Treatments in Gynecologic Surgery Study First Author, Country, Year Study Quality Intervention (n) Comparator (n) Benign gynecology patients, major surgeries Welti, 12 Germany, 1981 Studer-Schar, 13 Switzerland, 1982 C B UFH 5,000 units sq 1 2 h preoperatively, every 12 h postoperatively 10 d (777) UFH 5,000 units sq preoperatively, every 12 h postoperatively 7 d (33) TED hose, foot elevation, early ambulation (822) Saline sq, same intervals (36) Docherty, 14 United Kingdom, 1983 B UFH 5,000 units sq preoperatively, every 12 h postoperatively 5 d (47) Saline sq, same intervals (45) Borstad, 15 Norway, 1988 C Dalteparin 5,000 units sq 1 h preoperatively, per day postoperatively 7 d (105) UFH 5,000 units sq preoperatively, every 12 h postoperatively 7 d (110) Borstad, 16 Norway, 1992 C Dalteparin 2,500 units sq 1 h preoperatively, per day postoperatively 7 d (71) UFH 5,000 units sq preoperatively, every 12 h postoperatively 7 d (70) Gynecologic oncology patients, major surgeries Clarke-Pearson, 17 United States, 1983 Clarke-Pearson, 18 United States, 1984 B Ancillary report to Clarke- Pearson, UFH 5,000 units sq 2 h preoperatively, every 12 h postoperatively 7 d (88) Foot elevation, early ambulation (97) Clarke-Pearson, 19 United States, 1990 A UFH 5,000 units sq 2 h preoperatively, every 8 h postoperatively 7 d (104) No intervention (103) Fricker, 20 France, 1988 B UFH 5, 000 units sq every 8 h preoperatively (2 9 doses), every 8 h postoperatively 7 d (97) UFH 5,000 units sq 2 h preoperatively, every 8 h postoperatively 7 d (104) Dalteparin 2,500 units sq 2 h preoperatively, 12 h postoperatively, then 5,000 units/d postoperatively 10 d (28) No intervention (103) UFH 5,000 units sq every 8 h preoperatively (two to nine doses), every 8 h postoperatively 7 d (97) UFH 5,000 units sq 2 h preoperatively, every 8 h postoperatively 10 d (32) 1114 Rahn et al Surgical Venous Thromboembolism Prophylaxis OBSTETRICS & GYNECOLOGY

5 Follow-Up Duration VTE Detection Method VTE Incidence Bleeding Complications 11 d Overt VTE, V/Q scan, I-fibrinogen 1.0% compared with 1.4% (NS) Ill-defined, no differences 7 d NA Not assessed No differences in transfusion rate (24% compared with 25%), hematoma, subjective bleeding 8 d Clinical PE 2% compared with 0% (NS) Greater EBL with UFH: 401 (SEM 75) ml compared with 246 (15) ml (P.025); no differences in transfusion rate, hematoma, Hgb change 7 d Overt VTE 0% Greater transfusion rate with LMWH: 26% compared with 10% (P.02) and wound hematoma rate: 25% compared with 14% (P.02) 5-d bleeding, 1-mo DVT Overt VTE, VTE questionnaire 0% No differences in transfusion rate (16% compared with 11%), hematoma, return to OR for bleeding 42 d I-fibrinogen and other methods 30 d I-fibrinogen and other methods 15% compared with 12% (NS) Fewer VTE with UFH: 8.7% compared with 18.4% (P.04) 3-arm trial Fewer VTE with UFH: 4.1% compared with 18.4% (P.001) 3-arm trial 8.7% compared with 4.1% (NS) Reported in ancillary article Greater Hemovac drainage with UFH: ml compared with ml (P.001); no differences in EBL, transfusion rate, wound hematoma rate No differences in transfusion rate (0%), median EBL ( ml), suction volume, or hematoma rate 10 d Lung scintigraphy (PE), I-fibrinogen (DVT) PE: 0% compared with 6.3% (NS); DVT: 0% No differences in EBL, transfusion volume, drain output, Hgb, Hct, or Plt changes (unable to isolate bleeding complication data for gynecologic cancers only) (continued) VOL. 118, NO. 5, NOVEMBER 2011 Rahn et al Surgical Venous Thromboembolism Prophylaxis 1115

6 Table 1. Description of Randomized Controlled Trials of Venous Thromboembolism Prophylaxis Treatments in Gynecologic Surgery (continued) Study First Author, Country, Year Study Quality Intervention (n) Comparator (n) Ward, 21 Australia, 1998 C Dalteparin 5,000 units sq 12 h preoperatively, per day postoperatively 5 d (280) UFH 5,000 units sq 12 h preoperatively, every 12 h postoperatively 5 d (286) Baykal, 22 Turkey, 2001 Clarke-Pearson, 23 United States, 1984 Clarke-Pearson, 24 United States, 1984 Clarke-Pearson, 25 United States, 1993 B Enoxaparin 2,500 units 2 h preoperatively, per day postoperatively 7 d (47) A IPC applied before induction and removed in PACU (38) or 24 h postoperatively (59) A IPC applied before induction and removed in 5 d (or at discharge) (55) B UFH 5,000 units sq 2, 10, and 18 h preoperatively, every 8 h postoperatively 7 d (107) UFH 5,000 units sq preoperatively, every 8 h postoperatively 7 d (55) Foot elevation, early ambulation (97) Foot elevation, early ambulation (52) IPC applied before induction and removed in 5 d (or at discharge) (101) Maxwell, 26 United States, 2001 A Dalteparin 2,500 units sq 1 2 h preoperatively, 12 h postoperatively, then 5,000 units per day postoperatively 5 d (106) IPC applied before induction and removed in 5 d (or at discharge) (105) VTE, venous thromboembolism; UFH, unfractionated heparin; sq, subcutaneously; TED, thrombo-embolic deterrent; V/Q, ventilation perfusion; NS, nonsignificant; NA, not applicable; PE, pulmonary embolism; EBL, estimated blood loss; SEM, standard error of mean; LMWH, low molecular weight heparin; DVT, deep vein thrombosis; OR, operating room; Hgb, hemoglobin; Hct, hematocrit; Plt, platelet; IPC, intermittent pneumatic compression; PACU, postanesthesia care unit; aptt, activated partial thromboplastin time. ler ultrasonography and pulmonary emboli by ventilation perfusion scan, pulmonary angiogram, or computed tomography angiogram. These randomized controlled trials were insufficient to describe the incidence of venous thromboembolism or indications for venous thromboembolism prophylaxis for other gynecologic procedures, including those done in the lithotomy position, laparoscopy, hysteroscopy, or other gynecologic subspecialty procedures such as cystoscopy, anti-incontinence, or prolapse procedures. Therefore, a secondary search for supportive nonrandomized controlled trial literature was completed and identified 14 studies, 2,27 39 nine of which were used to inform the practice guidelines ,39 Figure 1 illustrates the search yield. Of the five randomized controlled trials in benign gynecologic surgery populations, three compared unfractionated heparin (given subcutaneously preoperatively and every 12 hours postoperatively) to usual care or placebo Welti et al conducted a large (n 2,367) randomized controlled trial of unfractionated heparin compared with usual care 12 in women undergoing a variety of gynecologic procedures but also included cesarean deliveries and postpartum tubal ligations. In this study, venous thromboembolism occurrence was determined by varying means from clinically overt cases to objective screening with I-fibrinogen and ventilation perfusion scans; they found no statistically significant differences in venous thromboembolism incidence between study arms (11 cases [1.0%] in the unfractionated heparin arm compared with 17 cases [1.4%] in the arm receiving usual care). Of the trials comparing unfractionated heparin with placebo, Studer-Schar et al 13 did not report on venous thromboembolism incidence and Docherty 14 found no significant difference in clinical pulmonary 1116 Rahn et al Surgical Venous Thromboembolism Prophylaxis OBSTETRICS & GYNECOLOGY

7 Follow-Up Duration VTE Detection Method VTE Incidence Bleeding Complications 6 wk Clinical DVT (confirmed by Doppler); clinical PE (confirmed by V/Q) DVT: 0% compared with 0.3% (NS); PE: 1.8% compared with 0.3% (NS) No differences in transfusion rate (20% compared with 14%) or no. of units transfused (3.3 compared with 3.9) 9 d Clinical VTE 0% No differences in EBL, transfusions, drain output, or change in Hct 42 d I-fibrinogen and other methods 18.6% compared with 12.4% (NS) Not described 42 d I-fibrinogen and other methods 12.7% compared with 34.6%, P.05 No difference in EBL 30 d I-fibrinogen, clinical VTE Total VTE 6.5% compared with 4.0%, P d Bilateral lower extremity Doppler 0.9% compared with 1.9% (NS) Greater postop transfusion rate with UFH: 31.8% compared with 15.9%, P.03; greater median suction volume with UFH: 1,560 compared with 936 ml, P.02; greater no. of patients with aptt greater than 1.5 control with UFH: 23.4% compared with 4.0% (P.001) No differences in transfusion rate, EBL, wound hematoma; greater final Hct with IPC (difference 1.6%, P.004); longer aptt with LMWH (2.93 min, P.01) embolus rate between unfractionated heparin and placebo (one [2.1%] compared with zero [0%], respectively). As for bleeding complications, neither Welti nor Studer-Schar found differences between study arms, whereas Docherty et al who assessed blood loss by several methods reported a higher mean ( standard error of mean) estimated blood loss for women receiving unfractionated heparin compared with placebo ( ml compared with ml, P.025) and a greater proportion of patients with blood loss greater than 500 ml (six [12.8%] compared with zero [0%], P.05); there were no differences in other bleeding outcomes including transfusion rate, wound hematomas, suction volume, and change in hemoglobin. Synthesizing these studies, it is uncertain whether unfractionated heparin (5,000 units preoperatively and twice-daily postoperatively) is preferable to early ambulation (or placebo) for venous thromboembolism prophylaxis in a population undergoing benign major gynecologic surgery. Use of unfractionated heparin at these doses inconsistently resulted in greater estimated blood loss (approximately 150 ml) but no other differences in bleeding complication rates (low-quality evidence). Data are insufficient to compare differences in rate of venous thromboembolism occurrence. Two randomized controlled trials from the same authors compared perioperative low molecular weight heparin (5,000 units dalteparin in one study 15 or 2,500 units in the other 16 ) with 5,000 units unfractionated heparin in women undergoing abdominal hysterectomy and other laparotomies. Using plethysmography followed by venography as needed, neither study identified any venous thromboembolisms. Both studies were powered to find differences in transfusion rates and other bleeding complications VOL. 118, NO. 5, NOVEMBER 2011 Rahn et al Surgical Venous Thromboembolism Prophylaxis 1117

8 Articles published in obstetrics and gynecology journals after dates of search n=3 Gynecologic oncology population: n=10 Unfractionated heparin compared with placebo or control: 3 Unfractionated heparin compared with low molecular weight heparin: 3 Intermittent pneumatic compression device compared with control: 2 Unfractionated heparin or low molecular weight heparin compared with intermittent pneumatic compression device: 2 Articles assessed n=120 Articles (trials) included in systematic review n=15 (14) Literature search performed and citations screened as described in Sources and Study Selection N=1,266 Full-text articles assessed n=117 Benign gynecology population: n=5 Unfractionated heparin compared with placebo or control: 3 Unfractionated heparin compared with low molecular weight heparin: 2 Titles, abstracts, or both reviewed and excluded: n=1,149 Venous thromboembolism prophylaxis not studied: 959 Ineligible study design (eg, review): 130 No gynecologic surgery group or subgroup: 41 Ineligible intervention or comparator: 9 Other: 10 Full-text articles reviewed and excluded from systematic review: n=105 Venous thromboembolism prophylaxis not studied: 9 Ineligible study design for systematic review: 19* No gynecologic surgery group or subgroup: 56 Ineligible intervention or comparator: 14 Other: 7 Fig. 1. Flow diagram of study search and systematic review. *After full-text review, 19 articles were ineligible for inclusion in the primary systematic review because they were not randomized controlled trials; 14 of these studies were either prospective or large retrospective studies in gynecologic surgery populations, nine of which were helpful in guideline formation. Rahn. Surgical Venous Thromboembolism Prophylaxis. Obstet Gynecol rather than a difference in venous thromboembolism rates. The study using the higher dose of low molecular weight heparin did have a higher transfusion rate (27 [26%] compared with 11 [10%], P.02) and more wound hematomas (26 [25%] compared with 15 [14%], P.02) compared with unfractionated heparin. 15 No bleeding complication differences between study arms were identified in the trial using the lower dose of low molecular weight heparin. 16 Synthesizing these studies, comparing low molecular weight heparin (2,500 or 5,000 units dalteparin 1 hour preoperatively and daily for 7 days postoperatively) with unfractionated heparin (5,000 units preoperatively and twice-daily postoperatively) for prophylaxis against venous thromboembolism in a population undergoing benign major gynecologic surgery, it is uncertain whether low molecular weight heparin is preferable to unfractionated heparin. At 2,500 units daily, dalteparin had no difference in bleeding or wound complications; at 5,000 units, greater bleeding complications were identified compared with using unfractionated heparin (low-quality evidence). Data are insufficient to compare differences in rate of venous thromboembolism occurrence. Of the nine randomized controlled trials (in 10 articles) from gynecologic oncology patients undergoing major procedures, two compared unfractionated heparin with control and usual care, three compared unfractionated heparin with low molecular weight heparin, two compared intermittent pneumatic compression with control, 23,24 and two compared heparin (unfractionated heparin or low molecular weight heparin) with intermittent pneumatic compression. 25,26 One randomized controlled trial (in two articles) by Clarke-Pearson et al 17,18 compared 5,000 units unfractionated heparin 2 hours preoperatively and twice-daily postoperatively with usual care (elevation of foot of bed, early ambulation) in 185 women (90% underwent laparotomies) and found no differences in venous thromboembolism incidence between groups as measured by I-fibrinogen: 13 (15%) and 12 (12%), respectively. 17 Although no differences in estimated blood loss, transfusion rate, or wound hematoma rate were found, there was a greater amount of 1118 Rahn et al Surgical Venous Thromboembolism Prophylaxis OBSTETRICS & GYNECOLOGY

9 Hemovac drainage in the patients receiving the heparin compared with usual care (mean standard deviation; ml compared with ml; P.001). 18 Another randomized controlled trial (three-armed) 19 compared unfractionated heparin (5,000 units) given in varying frequencies of preoperative dosing (arms 1 and 2) and three times per day postoperatively with a no intervention control group (third arm). Statistically fewer venous thromboembolisms (as measured by I-fibrinogen) occurred in both unfractionated heparin regimens (arm 1: 8.7% and arm 2: 4.1%) compared with control (arm 3: 18.4%) but not compared with each other (Table 1). There were no differences in transfusion rate (0%), median estimated blood loss ( ml), suction volume, or hematoma rates among the three arms. Synthesizing these studies, comparing unfractionated heparin (5,000 units 2 hours preoperatively and twice or three times daily for 7 days postoperatively) with usual perioperative care with no pharmacologic venous thromboembolism prophylaxis in a gynecologic oncology population undergoing major surgery, there are net benefits for unfractionated heparin (when delivered three times per day postoperatively) with fewer venous thromboembolisms observed and no difference in bleeding complications compared with standard care. Unfractionated heparin delivered twice daily postoperatively did not result in significantly fewer venous thromboembolisms compared with standard care (moderate quality of evidence). Three randomized controlled trials compared prophylaxis with low molecular weight heparin with unfractionated heparin in patients with cancer undergoing major procedures Ward et al and Baykal et al exclusively studied gynecologic oncology patients. Although Fricker et al included urologic and digestive tract cancers, the gynecology patients venous thromboembolism incidence data could be culled separately. None of the studies identified significant differences in venous thromboembolism incidence (range 0 6.3%) or in bleeding complications, although none reported specific power analyses. Synthesizing these studies, comparing low molecular weight heparin (2,500 5,000 units of dalteparin 2 or more hours preoperatively and daily for 5 10 days postoperatively or 2,500 units of enoxaparin 2 hours preoperatively and daily for 7 days postoperatively) with unfractionated heparin (5,000 units 2 hours preoperatively and three times per day for 7 days postoperatively) for prophylaxis against venous thromboembolism in a gynecologic oncology population undergoing major surgery, it is uncertain whether low molecular weight heparin is preferable to unfractionated heparin. No differences were observed in bleeding complications such as transfusion (moderate quality of evidence), but data are insufficient or underpowered to compare differences in rate of venous thromboembolism occurrence. Two randomized controlled trials by Clarke-Pearson et al compared the use of intermittent pneumatic compression with usual care (elevation of foot of bed and early ambulation) for prophylaxis against perioperative venous thromboembolism for gynecologic oncology patients. 23,24 In both studies, intermittent pneumatic compression devices were applied at time of anesthesia induction. In one study, intermittent pneumatic compression was used until patients were in the postanesthesia care unit (or up to 24 hours postoperatively), 23 and in the other study, intermittent pneumatic compression devices were ordered until time of discharge or for a minimum of 5 days postoperatively. 24 In the study with the shorter duration of postoperative intermittent pneumatic compression use, there was no difference in venous thromboembolism incidence compared with usual care as measured by I-fibrinogen: 18 (18.6%) compared with 12 (12.4%), respectively. 23 However, in the study that compared intermittent pneumatic compression maintained until time of discharge to usual care, venous thromboembolism incidence was significantly decreased in the patients using intermittent pneumatic compression (seven [12.7%] compared with 18 [34.6%], respectively, P.05). 24 Intermittent pneumatic compression use was not associated with differences in bleeding complications for either study. Synthesizing these studies, comparing intermittent pneumatic compression use (from time of anesthesia induction until early postoperatively or until discharge) and standard perioperative care with no pharmacologic venous thromboembolism prophylaxis in a gynecologic oncology population undergoing major surgery, there are net benefits for intermittent pneumatic compression use when maintained postoperatively for 5 days or until discharge with fewer venous thromboembolisms observed and no difference in bleeding complications compared with usual care (moderate quality of evidence). Finally, two randomized controlled trials compared use of heparin (unfractionated heparin 25 or low molecular weight heparin 26 ) with intermittent pneumatic compression devices placed before induction of anesthesia and maintained until time of discharge or a minimum of 5 days postoperatively. Neither study was powered to detect a difference in incidence of venous thromboembolism. Combining clinically overt venous thromboembolism and those detected by I-fibrinogen screening, there were no differences VOL. 118, NO. 5, NOVEMBER 2011 Rahn et al Surgical Venous Thromboembolism Prophylaxis 1119

10 between the unfractionated heparin and intermittent pneumatic compression interventions (seven [6.5%] compared with four [4.0%], respectively, P.54). 25 Similarly, using bilateral lower extremity Doppler screening on postoperative days 3 5, Maxwell et al found no differences in DVT incidence between the low molecular weight heparin and intermittent pneumatic compression interventions (one [0.9%] compared with two [1.9%], respectively, P nonsignificant). 26 Comparing unfractionated heparin with intermittent pneumatic compression use, heparin was associated with a higher postoperative transfusion rate (34 [32%] compared with 17 [16%], respectively, P.03) and higher median (range) retroperitoneal suction volume (1,560 [460 3,265] ml compared with 936 [195 2,032] ml, P.02). The number of participants with a prolonged activated partial prothrombin time was significantly higher in the arm receiving unfractionated heparin (25 [23.4%] compared with four [4.0%], P.001). 25 The study of low molecular weight heparin compared with intermittent pneumatic compression did not identify any differences in transfusion rate, estimated blood loss, or wound hematoma, but the patients using the intermittent pneumatic compression had a significantly greater final hematocrit (difference of 1.64%, P.004), and the maximum activated thromboplastin time was significantly longer (2.93 minutes) in the low molecular weight heparin group (P.01). Synthesizing these studies, comparing unfractionated heparin (5,000 units 2 hours preoperatively and three times per day for 7 days postoperatively) with intermittent pneumatic compression use (from time of anesthesia induction until 5 days postoperatively or until discharge) for prophylaxis against venous thromboembolism in a gynecologic oncology population undergoing major surgery, there are tradeoffs to the use of these interventions: intermittent pneumatic compression use may preferentially reduce bleeding complications, but data are insufficient to comment on differences in venous thromboembolism occurrence rates (poor quality of evidence). Comparing low molecular weight heparin (2,500 units dalteparin 1 2 hours preoperatively, 2,500 units 12 hours after surgery, and then 5,000 units daily for 5 days postoperatively) with the same intermittent pneumatic compression intervention as previously, it is uncertain whether low molecular weight heparin is preferable to intermittent pneumatic compression use. No differences are observed in clinical bleeding complications, but data are insufficient to comment on differences in venous thromboembolism occurrence rates (moderate quality of evidence). A secondary literature screen for nonrandomized controlled trials identified five prospective 2,27,31 33 and nine large retrospective (at least 150) studies of gynecology-only procedures ,34 39 After review, nine of these were used to inform the clinical practice guidelines regarding venous thromboembolism prophylaxis for patients undergoing gynecologic laparoscopic procedures, the incidence of venous thromboembolism in urogynecologic patients, characteristics that increase postoperative venous thromboembolism risk, the effect of combining methods of venous thromboembolism prophylaxis, and the optimal duration of venous thromboembolism prophylaxis. To determine the risk of venous thromboembolism in patients undergoing benign laparoscopic gynecologic procedures, three prospective cohort studies objectively screened patients postoperatively for evidence of lower extremity DVT. The largest study by Ageno et al 32 (n 266, consecutive cases) had a mean operating time of 60 minutes (majority greater than 45 minutes) and used no pharmacologic or mechanical venous thromboembolism prophylaxis. Patients had duplex Doppler scan on postoperative days 4 and 7 and completed a venous thromboembolism phone questionnaire 30 and 90 days postoperatively. In this population, no DVTs and no clinical venous thromboembolisms were identified. Although this was largely a low-risk population, it did include women using oral contraceptives (14%) and who had varicose veins (5%), a family history of venous thromboembolism (12%), obesity (5%), and tobacco use (20%). The two other studies on patients undergoing benign gynecologic laparoscopic procedures (n and n ) similarly found no postoperative cases of venous thromboembolism by Doppler scanning 24 hours and 7 days postoperatively, although some of these patients used intraoperative intermittent pneumatic compression devices. Nick et al 35 retrospectively reviewed 849 patients undergoing laparoscopic gynecologic surgery who routinely used intermittent pneumatic compression before induction of anesthesia and early mobilization postoperatively. Four hundred thirty (51%) had gynecologic malignancies. There was a low incidence of symptomatic venous thromboembolism overall: six (0.7%). Although five of the six patients had cancer (ie, 1.2% venous thromboembolism rate in patients with cancer and 0.2% in women with benign disease), malignancy was not found to significantly increase the risk of venous thromboembolism. Rather, in this cohort, surgical complexity was associated with increasing risk of venous thromboembolism, in which high complexity cases (radical hysterectomy, pel Rahn et al Surgical Venous Thromboembolism Prophylaxis OBSTETRICS & GYNECOLOGY

11 vic or para-aortic lymphadenectomy, splenectomy, bowel resection) significantly increased the risk for venous thromboembolism (2.8%) compared with low complexity cases (diagnostic and second-look laparoscopy) (0%) or intermediate complexity cases (0.5%) (P.049). Ritch et al 39 retrospectively reviewed a mixture of benign and oncologic cases of 60,013 women undergoing laparoscopic hysterectomy and found a modestly higher rate of symptomatic venous thromboembolism: 579 (1.0%). Variable types of prophylaxis were used (49% mechanical, 12% pharmacologic, 39% none). Factors that increased the risk of venous thromboembolism development included age 60 years or older (odds ratio [OR] 1.64, 95% confidence interval [CI] ) and increasing comorbidities (OR 1.99, 95% CI for Charlson index of 1 and OR 3.07, 95% CI for Charlson index of 2 or more). Although the venous thromboembolism rate was higher for patients with cancer (2.3%) compared with benign patients (0.9%), multivariable analysis did not indicate this was statistically significant in this cohort. A single-site retrospective review (n 1,104) calculated the incidence of venous thromboembolism among urogynecologic patients over a 3-year period. 34 Patients in this study had undergone a variety of reconstructive and anti-incontinence procedures including transvaginal, abdominal, laparoscopic, and robotic-assisted approaches. Twenty-three percent had outpatient procedures; the mean operating time was 183 minutes. All patients used intermittent pneumatic compression devices before anesthesia induction, which were continued until patients were fully ambulatory at discharge; there was no pharmacologic prophylaxis in this cohort. During this time period, 40 postoperative patients were evaluated for possible venous thromboembolism secondary to symptoms, but only three were diagnosed with venous thromboembolism (0.3%; 95% CI ) by objectively confirmed Doppler, computed tomography angiogram, or both. Of these three, one had a pulmonary embolism and two had DVTs with pulmonary emboli. Risks and benefits of perioperative venous thromboembolism prophylaxis likely differ based on the patient s baseline risk of venous thromboembolism. Clarke-Pearson et al retrospectively reviewed 1,862 women undergoing major gynecologic procedures benign and oncologic cases to determine factors that increased patient risk of postoperative venous thromboembolism. 28 In the Clarke-Pearson study, the observed incidence of venous thromboembolism among women having total abdominal hysterectomy alone (ie, not cancer surgery, per se) was less than 1%. All patients had mechanical venous thromboembolism prophylaxis until discharge (or minimum 5 days postoperatively) with no pharmacologic prophylaxis. Clinically suspected venous thromboembolisms were confirmed by objective measures such as duplex Doppler screening, ascending venography, magnetic resonance imaging, or pulmonary arteriogram. In this population, the incidence of venous thromboembolism was less than 1% in the benign cases, 4.1% for abdominal hysterectomies with lymph node dissections, and 2.6% for radical hysterectomies. After multivariable regression, age 60 years or older, history of gynecologic cancer, and history of DVT were significantly associated with increased patient risk of venous thromboembolism. Importantly, having two of three of these risks significantly increased the risk of perioperative venous thromboembolism (3.2%) compared with having one (or none) of these risks (0.6%). A similar retrospective review of 1,373 patients by Martino et al 29 similarly showed that patients with cancer undergoing abdominal cases had a significantly higher incidence of venous thromboembolism (4.1%) than benign abdominal cases (0.3%) (P.001). Multivariable regression again demonstrated that age older than 60 years and cancer history substantially elevated the risk of clinically evident pulmonary embolism. A retrospective COHORT study (n 605) by Einstein et al 30 compared intermittent pneumatic compression use alone with intermittent pneumatic compression combined with unfractionated heparin in patients with gynecologic cancer undergoing major procedures. Patients aged older than 60 years or with a venous thromboembolism history also continued heparin every 8 hours for 2 weeks after discharge. The clinical venous thromboembolism rate was decreased from 6.5% to 1.9% in the dual prophylaxis group (OR 0.33; ). There was no difference in bleeding complications with the addition of unfractionated heparin. Lastly, we found no studies that exclusively reviewed the gynecologic (benign or oncologic) surgery population to address the question of optimal duration of pharmacologic venous thromboembolism prophylaxis. However, a landmark study by Bergqvist et al 40 evaluated venous thromboembolism incidence in patients undergoing major surgery for abdominal and pelvic cancers (n 332) who were randomized to varying durations of postoperative venous thromboembolism prophylaxis; 40% of participants were women. All patients received 40 mg enoxaparin hours preoperatively and daily for 6 10 days VOL. 118, NO. 5, NOVEMBER 2011 Rahn et al Surgical Venous Thromboembolism Prophylaxis 1121

12 postoperatively. Patients were then randomized to either the same dose of enoxaparin or placebo for an additional 21 days. The patients receiving prolonged low molecular weight heparin had fewer venous thromboembolisms as measured by bilateral venography on postoperative days 25 and 31 (eight [4.8%] compared with 20 [12.0%], P.02). There were no significant differences in bleeding complications. CONCLUSION Venous thromboembolism prevention is a public health priority, and proper perioperative venous thromboembolism prophylaxis could potentially decrease the incidence of venous thromboembolism without substantially increasing risk of bleeding complications. 4 Based on the evidence, the Society of Gynecologic Surgeons Systematic Review Group developed Clinical Practice Guidelines for venous thromboembolism prophylaxis in women undergoing gynecologic surgery (Table 2). These Clinical Practice Guidelines were based on the systematically reviewed randomized controlled trials and further informed by the nonrandomized controlled trials, which filled important clinical gaps in the randomized controlled trial evidence. Each Clinical Practice Guideline received a grade in two parts: 1) the strength of recommendation (1 we recommend or 2 we suggest ), and 2) the quality of evidence (A, B, C, D). The five Clinical Practice Guidelines generated for perioperative venous thromboembolism prophylaxis for the gynecologic surgery patient were each graded as level 2C, meaning low quality of evidence that supports a suggestion (as opposed to recommendation). This implies that the majority of patients would want to follow the recommended course of therapy but many would not. Furthermore, physicians must judge each patient independently and manage individually. From a policymaking standpoint, there is room for debate and need for further evidence. Of the 14 randomized controlled trials included in this systematic review, five compared venous thromboembolism incidence and bleeding complications in benign gynecologic populations and nine focused on patients with malignancy. Based on these studies, the incidence of clinical venous thromboembolism ranges from 0% to 2% in women undergoing benign gynecologic surgery. The nonrandomized controlled trial literature supports this estimate, and when intermittent pneumatic compression devices are routinely used during benign major procedures, clinical venous thromboembolism remains less than 1%. 28,29,34 In smaller series, no venous thromboembolisms (clinical or with objective screening) were identified in women undergoing benign laparoscopic cases, even without any venous thromboembolism prophylaxis On the contrary, the gynecologic oncology randomized controlled trials report a venous thromboembolism incidence (including subjective and clinical and objective screening) of % in women with some form of venous thromboembolism prophylaxis and up to 34.6% if no venous thromboembolism prophylaxis was used. 24 Unlike the 2007 American College of Obstetrics and Gynecology Practice Bulletin and guidelines by the American College of Chest Physicians, this systematic review critically assesses available evidence exclusively in patients undergoing gynecologic surgeries. Nonetheless, our conclusions and recommendations are similar and complementary. In particular, in each of these reviews, the highest risk patients are women aged 60 years or older with a history of venous thromboembolism, cancer, or both or a hypercoagulable state. Accordingly, these patients warrant more than mechanical prophylaxis and will require pharmacologic (in addition or instead of mechanical) prophylaxis, and, although the gynecologic literature is limited for guiding duration of prophylaxis, extended therapy should be considered. 40 However, the majority of benign gynecologic patients are at moderate to high venous thromboembolism risk according to the American College of Chest Physicians and American College of Obstetrics and Gynecology stratification, 3,6 which includes procedures longer than 30 minutes (or less than 30 minutes in a patient years old). The American College of Obstetricians and Gynecologists Practice Bulletin suggests using pharmacologic prophylaxis (unfractionated or low molecular weight heparin) or mechanical prophylaxis for these patients, 6 but we believe the gynecologic literature supports use of intermittent pneumatic compression alone for the majority of these patients. The addition of having two of three risks (age 60 years or older, cancer history, or history of venous thromboembolism) warrants added perioperative pharmacologic prophylaxis started preoperatively. In particular, large retrospective studies of a urogynecologic population 34 and of women undergoing laparoscopic hysterectomy (without additional risk factors of increasing age older than 60 years and multiple comorbidities or cancer) 39 and prospective cohort studies of women undergoing benign laparoscopy procedures would support the use of mechanical prophylaxis alone, thus decreasing the potential morbidities associated with heparin prophylaxis. 14,15 The strengths of this review are its robust methods and the transparent means by which we devel Rahn et al Surgical Venous Thromboembolism Prophylaxis OBSTETRICS & GYNECOLOGY