- Email: [email protected]
Pulmonary Embolism Complicating Bariatric Surgery: Detailed Analysis of a Single Institution’s 24-Year Experience
Pulmonary Embolism Complicating Bariatric Surgery: Detailed Analysis of a Single Institution’s 24-Year Experience Brennan J Carmody, MD, FACS, Harvey J Sugerman, MD, FACS, John M Kellum, MD, FACS, Mohammed K Jamal, MD, Jason M Johnson, DO, Alfredo M Carbonell, DO, James W Maher, MD, FACS, Luke G Wolfe, MS, Eric J DeMaria, MD, FACS Morbidly obese patients undergoing bariatric procedures are at risk for pulmonary embolism (PE). Because large series are required to analyze low-incidence complications, factors predictive of PE have not been clearly defined. Since 1992, short-course heparin prophylaxis, beginning immediately before operation, has been used in this center. STUDY DESIGN: Prospective data on 3,861 patients undergoing bariatric procedures between 1980 and 2004 were queried. Factors analyzed included age, gender, body mass index, interval between procedure and PE, inpatient versus outpatient status, mortality, access method (open versus laparoscopic), and comorbidities. RESULTS: PE within 60 days of operation occurred in 33 patients (23 women, 10 men), for an incidence of 0.85%. No difference in incidence was noted between open (0.84%) and laparoscopic (0.88%) groups, nor did routine prophylaxis with heparin since 1992 decrease the incidence. The interval between procedure and PE was 13.2 ⫾ 2.6 (mean ⫾ SEM) days (open ⫽ 13.0 ⫾ 3.0 days, laparoscopic 14.1 ⫾ 6.49 days, p ⫽ 0.9). One-third of PEs occurred after hospital discharge. Pulmonary embolism–related mortality was 27%. A statistically greater body mass index was noted in PE patients compared with non-PE patients (57.2 ⫾ 2.4 kg.m2 versus 49.9 ⫾ 0.2 kg/m2, p ⬍ 0.01, Wilcoxon rank test). Multivariate logistic regression confirmed a primary role for preoperative weight as a predictor of PE; univariate analysis suggested an increased PE risk with obesity hypoventilation syndrome, anastomotic leak, and chronic venous insufficiency. CONCLUSIONS: Data demonstrated persistence of PE risk in the anticoagulation, laparoscopic-access era at a rate similar to that in the preanticoagulation, open-access era. Because one-third of PEs occur after hospital discharge, consideration should be given to continuing anticoagulants longer and to adopting a more aggressive policy of inferior vena cava filter prophylaxis, particularly in patients with high body mass index, obese hyperventilation syndrome, and venous insufficiency. (J Am Coll Surg 2006;203:831–837. © 2006 by the American College of Surgeons) BACKGROUND:
Because the prevalence of obesity and its complications continues to increase at an alarming rate, surgery for
morbid obesity is being used more frequently because it remains the only durable treatment for this condition. Patients undergoing surgical treatment for obesity are at increased risk for venous thromboembolism (VTE), and the incidence of perioperative pulmonary embolism (PE) has been estimated at 1%.1,2 Although uncommon, PE is a major cause of postoperative mortality, and it has recently been identified as an independent risk factor for perioperative death.3 Newer concepts in surgical care for morbidly obese patients with the potential to influence the occurrence of postoperative PE include the use of minimally invasive surgical access and new forms of anticoagulation (eg, low molecular weight heparin [LMWH]).
Competing Interests Declared: None. Presented at the American College of Surgeons 91st Annual Clinical Congress, San Francisco, CA, October 2005. Received February 8, 2006; Revised August 25, 2006; Accepted August 25, 2006. From the Division of General Surgery, Department of Surgery, Virginia Commonwealth University, Richmond, VA (Carmody, Sugerman, Kellum, Jamal, Johnson, Carbonell, Maher, Wolfe) and the Department of Surgery, Duke University, Durham, NC (DeMaria). Correspondence address: Brennan J Carmody, MD, FACS, Walter Reed Army Medical Center, 6900 Georgia Ave, NW, Washington, DC, 203075001.
© 2006 by the American College of Surgeons Published by Elsevier Inc.
831
ISSN 1072-7515/06/$32.00 doi:10.1016/j.jamcollsurg.2006.08.020
832
Carmody et al
Pulmonary Embolism Complicating Bariatric Surgery
J Am Coll Surg
Table 1. Bariatric Procedures Performed During Study Period Abbreviations and Acronyms
BMI IVC LMWH PE VTE
⫽ ⫽ ⫽ ⫽ ⫽
body mass index inferior vena cava low molecular weight heparin pulmonary embolism venous thromboembolism
It remains unclear whether or not such advances have contributed to an improvement in the incidence of postoperative PE or in PE-related mortality. Because large study cohorts are necessary to define low incidence complications like PE, factors related to incidence, prevention, time frame for occurrence, and mortality related to PE remain poorly defined. This study was undertaken in an attempt to better characterize PE after bariatric surgery by reviewing the occurrence of this complication over the 24-year history of a tertiary care hospital. Changes in surgical technique, and particularly, the shift toward laparoscopy for gastric bypass, and changes in strategy for prophylaxis, including routine anticoagulation, have occurred in this surgical center over this time frame. We hypothesized that the incidence of PE would decrease over the 24-year course of our experience as a result of the impact of these and other improvements on patient care. An additional objective of the study was to analyze risk factors in this large cohort of patients to determine patients at higher risk for PE who might benefit from more aggressive preventive measures such as prophylactic insertion of a vena caval (IVC) filter.
Procedure
VBGP PGBP DGBP Revision to PGBP Revision to DGBP HGP Revision of PGBP Revision of DGBP LLGBP Revision to LLGBP Revision of LLGBP LAGB Laparoscopic PGBP Hand assist laparoscopic PGBP Laparoscopic LLGBP Laparoscopic revision to PGBP Duodenal switch Total
n
%
101 1,369 49 191 51 51 49 21 878 58 14 37 898 25 48 14 7 3,861 (3,123 women, 738 men)
2.64 35.8 1.28 4.99 1.33 1.33 1.28 0.55 22.96 1.52 0.37 0.96 23.48 0.65 1.26 0.37 0.18 100
DGBP, distal gastric bypass; HGP, horizontal gastroplasty; LAGB, laparoscopic adjustable gastric banding; LLGBP, long limb gastric bypass; PGBP, proximal gastric bypass; VBGP, vertical banded gastroplasty.
with PE after weight reduction surgery were characterized in detail in a retrospective fashion by chart review to capture data not recorded prospectively, including interval between procedure and PE, patient status at diagnosis of PE (inpatient versus outpatient), and outcomes. The association between PE, patient comorbidities, and other complications was assessed. Venous thromboembolism prophylaxis
METHODS Patients
The database of patients undergoing bariatric procedures between 1980 and 2004 at Virginia Commonwealth University was analyzed. These data were collected prospectively and updated according to inpatient and outpatient clinic records. Institutional Review Board approval was granted for reviewing data and reporting analyses. During the study period, 3,861 patients underwent bariatric surgery (Table 1). This cohort included 2,839 open procedures and 1,022 laparoscopic cases. Prospectively collected data included detailed demographic information, comorbid conditions at the time of operation, information about surgical procedure and access technique, and outcomes including complications, weight loss, and so forth. Patients diagnosed
Between 1980 and 1992, thigh-length intermittent pneumatic compression devices were used on all patients undergoing bariatric surgery; ambulation was initiated on the evening of operation, except in patients requiring mechanical ventilation. Since 1992, routine prophylaxis included the addition of routine anticoagulation by instituting a computerized pathway for patient care. Anticoagulation consisted of either unfractionated heparin (5,000 U subcutaneously bid or every 8 hours) or LMWH (enoxaparin, Aventis Pharmaceuticals, Inc), but details about choice of anticoagulation agent, dosage, and administration schedule (such as preoperative versus intraoperative versus postoperative initiation) were not available. In 1998, standardization of the anticoagulation protocol was instituted for all patients undergoing bariatric surgery. The agent used from that time forward
Vol. 203, No. 6, December 2006
Carmody et al
Pulmonary Embolism Complicating Bariatric Surgery
833
Table 2. Demographics of Patients Experiencing Pulmonary Embolism
Table 3. Obesity-related Comorbidities in Pulmonary Embolism Patients
Demographic
Comorbidity
n
%
Hypertension Diabetes mellitus Obstructive sleep apnea Obesity hypoventilation syndrome Venous stasis disease
15 4 10 5 4
45.5 12.1 30.3 15.2 12.1
Gender 23 women 10 men Age, y Body mass index
Mean value (range)
41.4 (16–59) 57.2 kg/m2 (40.3–84)
was LMWH (enoxaparin) at a dose of 40 mg subcutaneously before operation and repeated in a daily dose until hospital discharge. Statistical analysis
All analyses were performed using SAS (SAS Institute, Inc). Means, medians (both ⫾ standard error), and ranges are shown for interval data. Univariate and multivariate analyses were used to evaluate differences between proportions. A p value ⬍ 0.05 was considered statistically significant. RESULTS Overall, 36 patients were diagnosed with PE after weight reduction operations. PE was a late event, occurring more than 4 months after operation (range 148 to 1,137 days) in three patients who were excluded from additional analysis because it was believed that the PE did not represent a perioperative event. The demographics of patients suffering PE are listed in Table 2. The mean age was 41.4 years (range 16 to 59 years), and the mean body mass index (BMI) was 57.2 ⫾ 2.4 kg/m2 (range 40.3 to 84 kg/m2). Preoperative BMI was substantially greater in patients suffering PE than it was in those in whom this complication did not develop (49.9 ⫾ 0.2 kg/m2, p ⬍ 0.01, Wilcoxon rank test). Four of the patients in whom PE developed were diagnosed with deep vein thrombosis (12%). Comorbidities are listed in Table 3. A variety of modalities were used in diagnosing PE; radiographic tests supported the diagnosis in 22 patients; in 8 patients, the diagnosis was made at autopsy (Table 4). In the remaining three patients, the diagnosis was made on clinical grounds alone by virtue of profound hypoxemia in a patient judged to be too unstable for confirmative testing. The overall incidence of perioperative PE was 0.85%. The incidence of PE did not differ between patients undergoing open surgery (0.84%) and those undergoing
laparoscopic surgery (0.88%). Pulmonary embolus incidence did not differ in patients undergoing primary bariatric procedures or revisional operations (0.9% versus 0.5%, respectively, p ⫽ 0.6, Fisher’s exact test). A notable difference in PE incidence was identified when comparing patients with BMIs ⬎ 50 kg/m2 with those with BMI ⬍ 50 kg/m2 (1.4% versus 0.5%, p ⫽ 0.005, Fisher’s exact test). The interval between procedure and PE was 13.2 ⫾ 2.6 days for the entire cohort. Strikingly, one-third of patients presented with PE after hospital discharge. No marked difference in interval between surgery and PE was found for patients undergoing open surgery (13.0 ⫾ 3.0 days) as compared with laparoscopic surgery (14.1 ⫾ 6.49 days). Nine of the 33 patients died, for a PE-associated mortality of 27% and an overall fatal PE incidence of 0.2% in the entire population of bariatric surgery patients. In eight of nine deaths, pulmonary emboli were confirmed by autopsy. The remaining patient had a large PE diagnosed by helical CT scan and died soon thereafter, but no postmortem examination was performed. The BMIs of patients experiencing fatal PE were considerably greater than the BMIs of patients experiencing nonfatal PE (65.6 ⫾ 4.4 kg/m2 versus 54.5 ⫾ 2.7 kg/m2, p ⬍ 0.01). Fatal PEs occurred sooner after operation than nonfatal PEs did. In the fatal PE group, the median interval between procedure and PE was 3 days (range 2 to 45 days); the median interval for the nonfatal PE group was Table 4. Diagnosis of Pulmonary Embolism Modality
n
CT scan Pulmonary arteriogram V/Q scan Autopsy Clinical (ABG) Total
10 1 11 8 3 33
ABG, arterial blood gas; V/Q, ventilation/perfusion scan.
834
Carmody et al
Pulmonary Embolism Complicating Bariatric Surgery
J Am Coll Surg
Table 5. Univariate Analysis: Factors Associated with Pulmonary Embolism Factor
Pulmonary embolism
No pulmonary embolism
30.3 41.4 57.2 45.5 12.1 30.3 15.2 12.1
18.9 40.5 49.9 46.7 18.4 25.1 4.2 3.3
43 7.7
2.4 4.1
Male gender, % Age, y Body mass index, kg/m2 Hypertension, % Diabetes mellitus, % Obstructive sleep apnea, % Obesity hypoventilation syndrome, % Venous stasis disease, % Gastrointestinal leak, % Pre 1992 Post 1992
p Value
0.1 0.6 0.01 0.8 0.4 0.5 0.01 0.03
95% CI*
⫺4.49, 2.63 ⫺10.73, ⫺3.79
0.0006 0.3
*Difference in the means with the 95% confidence interval.
10 days (range 1 to 51 days). This difference did not reach statistical significance. Univariate analysis (Fisher’s exact test) identified obesity hypoventilation syndrome and chronic venous insufficiency as risk factors for development of PE (Table 5). No marked difference in PE incidence was detected when the latter era of routine heparin prophylaxis (0.9%) was compared with the earlier era of inconsistent or absent heparin administration (0.6%). Multivariate analysis of all available variables confirmed preoperative weight as the only independent predictor of PE risk (361.9 ⫾ 17.1 lbs versus 307.2 ⫾ 1.1 lbs, p ⫽ 0.003). In the pre-1992 era, gastrointestinal leak complicating surgery was identified by univariate analysis as a predictor of PE, but this correlation disappeared during the more recent era of routine anticoagulation prophylaxis. IVC filters were placed prophylactically in 145 patients undergoing bariatric surgery over the 24-year period of study. Despite the IVC filter, postoperative PE developed in three patients and one died as a consequence of that PE. The median interval between procedure and PE was 5 days (range 2 to 26 days). The mean age and BMI of this subgroup were 47 years and 57.3 kg/m2, respectively, and each of the three patients abused tobacco. DISCUSSION Despite the use of routine inpatient chemoprophylaxis with heparin, the incidence of PE in bariatric surgery patients has remained relatively stable, even with the addition of standardized heparin protocols and the application of minimally invasive approaches to weight reduction surgery. Regarding access technique, the discrepancy in numbers between the open and laparoscopic
groups might mask a difference in PE incidence, which may surface as the laparoscopic group increases in size. The increased incidence of thrombosis in morbidly obese patients appears to be attributable to more than the mechanical effects of extreme obesity (decreased venous return, venous stasis).4 The endocrine activity of adipose tissue has been implicated in the dysregulation of the coagulation and fibrinolytic systems, favoring a thrombotic state. These adipocyte-secreted proteins, which are upregulated in obesity, include plasminogenactivator inhibitor-1 (PAI-1), adipsin (complement D), and adipocyte complement-related protein (Acrp30).5,6 This hypercoagulable milieu may be a partial explanation for the persistence of VTE at a consistent rate despite prophylactic measures. A wealth of data support the use of prophylactic anticoagulation in patient groups at high risk for VTE developing.7-9 The Seventh American College of Chest Physicians Conference on Antithrombotic and Thrombolytic Therapy10 endorsed routine heparin prophylaxis (unfractionated or LMWH) combined with intermittent pneumatic compression devices for high risk general surgery patients. This group also recommended extended prophylactic anticoagulation (at least 10 days) for patients undergoing elective knee or hip arthroplasty or urgent hip fracture surgery. Even within the bariatric surgery community, there was previously no consensus about the most effective prophylactic regimen for VTE. In a 2000 survey of American Society of Bariatric Surgery (ASBS) members, more than 95% of participants reported using routine prophylaxis, and 38% used two prophylactic measures simultaneously.11 Eleven percent of surveyed members routinely discharged patients with
Vol. 203, No. 6, December 2006
Carmody et al
prophylaxis. More recently, at the 2005 annual meeting of the American Society of Bariatric Surgery, an audience poll at the Master’s course revealed that no surgeon admitted to routinely withholding some form of heparin prophylaxis. So, although choice of agent and dosage likely continue to be nonstandardized, it appears that bariatric surgeons have adopted routine anticoagulation for these high risk patients. In this study, prophylaxis since 1992 included enoxaparin, 40 mg subcutaneously, daily during hospitalization. Our data collection system prevented us from comparing the various forms of chemoprophylaxis used during the study period. It is unclear whether LMWH is most efficacious using standard regimens or weight-based protocols. Several authors have compared various chemoprophylactic regimens (LMWH versus fractionated heparin, various nonweightbased dosages of LMWH) without conclusive results favoring the safest and most effective form of anticoagulation.12-15 Additionally, although a correlation has been demonstrated between antifactor Xa activity and body weight with LMWH administration, thrombosis has been reported, despite therapeutic anti-Xa levels.16 Additional data on anti-Xa activity during LMWH prophylaxis may determine whether a weightbased dosing system is more effective than standard dose regimens. The incidence of fatal PE in this study was similar to that reported by Sapala and colleagues,17 as were the associated risk factors (obesity hypoventilation syndrome, venous stasis disease, high BMI). Because multivariate analysis confirmed preoperative weight as a risk factor for PE, the BMI of patients experiencing fatal PE was substantially higher than that of those in whom nonfatal PE was developing. This study also identified a trend in the difference in median interval between procedure and PE in patients with lethal PE versus those with nonlethal PE (3 days versus 10 days). Because prophylactic measures did not differ, this finding may be explained by preexisting thrombus at the time of operation, a more virulent prothrombotic state with extreme obesity, or the fact that standardized, nonweight-based regimens are not effective in this group. These data also revealed an association between postoperative anastomotic leak and postoperative pulmonary embolism in the era before routine heparin prophylaxis. Although record review confirmed these patients received unfractionated sub-
Pulmonary Embolism Complicating Bariatric Surgery
835
cutaneous heparin (5,000 International Units subcutaneously twice or thrice daily) after leak diagnosis, they were also undergoing revisional surgery, prolonged mechanical intubation, and sedation. It is possible that the implementation of routine LMWH use in 1992, in addition to improvements in critical care, nursing, specialty beds, and a lower threshold for IVC filter insertion in critically ill patients, may be factors in the disappearance of the association between PE and leak after 1992. Autopsy studies have documented the high incidence of PE in patients who have died in the early postoperative period after gastric bypass. Melinek and associates5 reported a case series of 10 autopsies after gastric bypass; 3 of the deaths were attributable to PE (mean BMI 51.6 kg/m2, mean interval between procedure and death 11.6 days, mean age 44.6 years). In 8 of the 10 patients, however, microscopic evidence of PE was present, and in 2 patients, evidence of recent PE was documented despite previous IVC filter placement. Our data revealed a worrisome incidence of PEs occurring on an outpatient basis after hospital discharge. This finding is likely related to the decreased hospital stays associated with minimally invasive bariatric surgery. Huber and colleagues18 reported 30% more postoperative PEs when embolic events occurring after hospital discharge in postsurgical patients were included. Recently, we have considered a more aggressive approach to postdischarge chemoprophylaxis. Uniform extension of the period of prophylaxis for all bariatric surgery patients would have tremendous cost implications and would likely increase the risk of late bleeding complications beyond the risk of a PE, so this concept would clearly require more study before adoption. But an additional 2 to 4 weeks of enoxaparin for patients with BMI ⱖ 50 kg/m2 or impaired mobility is supported by this review and should also be evaluated in studies to assess the risks versus benefits of such an approach. Failure of IVC filters to prevent PE has been described by several investigators.19-22 This complication occurs at a frequency of 0.3% to 7.7% and has been reported with multiple filter types (Greenfield, SimonNitinol, Bird’s Nest). Von Bary and coworkers21 identified incorrect filter placement as a factor in three cases of reembolism, and Poletti and colleagues20 noted 20% of recurrent emboli originating distal to the filter. The eti-
836
Carmody et al
Pulmonary Embolism Complicating Bariatric Surgery
ology of most postfilter PEs is unclear. Possibilities include filter malposition, strut fracture, upper extremity source, caval thrombus distal to the filter, or hypercoagulable state (three patients used tobacco).23 Sugerman and coauthors4 advocated prophylactic IVC filter insertion in patients with respiratory insufficiency (ie, obesity hypoventilation syndrome), pulmonary hypertension (mean pulmonary artery pressure ⱖ 40 mmHg), and venous stasis disease. We now also recommend preoperative filter insertion for patients who are superobese and who have limited mobility, including those with neuromuscular disorders (eg, multiple sclerosis, muscular dystrophy) or severe degenerative joint disease. These data confirm the persistence of VTE at a stable rate despite routine heparin prophylaxis and minimally invasive techniques. Whether increased use of laparoscopic bariatric procedures and individualized heparin dosing will influence the incidence remains to be determined. When taken in context with the notable and compelling evidence that anticoagulation is effective for reducing the risk of VTE in high risk general surgery patients,7-10 we do not believe our data refute the role of routine anticoagulation in preventing VTE in bariatric surgery patients. In contrast, the appropriate conclusion from the current data is that additional strategies are needed to reduce the incidence of this life-threatening complication. Because one-third of PEs occur after hospital discharge, consideration should be given to continuing anticoagulation on an outpatient basis in selected patients. A more aggressive policy for preoperative IVC filter insertion, particularly influenced by the recent availability of temporary removable filter devices, should be considered in patients at higher risk for PE (BMI ⬎ 50 kg/m2, history of VTE, obesity hypoventilation syndrome, or venous insufficiency). Future studies to delineate superiority of various available anticoagulants, drug- and dose-related safety profiles, optimal duration of prophylaxis, and dose-related efficacy in the bariatric surgery patient are needed in this high risk population. Author Contributions
Study conception and design: Carmody, DeMaria Acquisition of data: Carmody Analysis and interpretation of data: Carmody, Sugerman, Kellum, Jamal, Johnson, Carbonell, Maher, Wolfe, DeMaria
J Am Coll Surg
Drafting of manuscript: Carmody Critical revision: Sugerman, DeMaria REFERENCES 1. Westling A, Bergqvist D, Boström A, et al. Incidence of deep venous thrombosis in patients undergoing obesity surgery. World J Surg 2002;26:470–473. 2. Eriksson S, Backman L, Ljungstrom K. The incidence of clinical postoperative thrombosis after gastric surgery for obesity during 16 years. Obes Surg 1997;7:332–335; discussion 336. 3. Fernandez A, DeMaria E, Tichansky D, et al. Multivariate analysis of risk factors for death following gastric bypass for treatment of morbid obesity. Ann Surg 2004;239:698–702; discussion 702–703. 4. Sugerman H, Sugerman E, Wolfe L, et al. Risks and benefits of gastric bypass in morbidly obese patients with severe venous stasis disease. Ann Surg 2001;234:41–46. 5. Melinek J, Livingston E, Cortina G, et al. Autopsy findings following gastric bypass surgery for morbid obesity. Arch Pathol Lab Med 2002;126:1091–1095. 6. Ahima R, Flier J. Adipose tissue as an endocrine organ. Trends Endocrinol Metab 2000;11:327–332. 7. Major K, Wilson M, Nishi G, et al. The incidence of thromboembolism in the surgical intensive care unit. Am Surg 2003; 69:857–861. 8. Tapson V, Hyers T, Waldo A, et al. Antithrombotic therapy practices in US hospitals in an era of practice guidelines. Arch Intern Med 2005;165:1458–1464. 9. Bergqvist D. Low-molecular-weight heparin for the prevention of postoperative venous thromboembolism after abdominal surgery: a review. Curr Opin Pulm Med 2005;11:392– 397. 10. Geerts W, Pineo G, Heit J, et al. Prevention of venous thromboembolism: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest 2004;126(3 Suppl): 338S–400S. 11. Wu E, Barba C. Current practices in the prophylaxis of venous thromboembolism in bariatric surgery. Obes Surg 2000;10:7–13; discussion 14. 12. Scholten D, Hoedema R, Scholten R. A comparison of two different prophylactic dose regimens of low molecular weight heparin in bariatric surgery. Obes Surg 2002;12:19–24. 13. Kalfarentzos F, Stavropoulou F, Yarmenitis S, et al. Prophylaxis of venous thromboembolism using two different doses of low-molecular-weight heparin (nadroparin) in bariatric surgery: a prospective randomized trial. Obes Surg 2001;11: 670–676. 14. Shepherd M, Rosborough T, Schwartz M. Heparin thromboprophylaxis in gastric bypass surgery. Obes Surg 2003;13:249– 253. 15. Gonzalez Q, Tishler D, Plata-Munoz J, et al. Incidence of clinically evident deep venous thrombosis after laparoscopic Roux-en-Y gastric bypass. Surg Endosc 2004;18:1082–1084. 16. Rutherford E, Schooler W, Sredzienski E, et al. Optimal dose of enoxaparin in critically ill trauma and surgical patients. J Trauma 2005;58:1167–1170. 17. Sapala J, Wood M, Schuhknecht M, et al. Fatal pulmonary embolism after bariatric operations for morbid obesity: a 24year retrospective analysis. Obes Surg 2003;13:819–825.
Vol. 203, No. 6, December 2006
Carmody et al
18. Huber O, Bounameaux H, Borst F, et al. Postoperative pulmonary embolism after hospital discharge. An underestimated risk. Arch Surg 1992;127:310–313. 19. Wolf F, Thurner S, Lammer J. Simon nitinol vena cava filters: effectiveness and complications. Fortschr Röntgenstr 2001;173: 924–930. 20. Poletti P, Becker C, Prina L, et al. Long-term results of the Simon nitinol inferior vena cava filter. Eur Radiol 1998;8: 289–294.
Pulmonary Embolism Complicating Bariatric Surgery
837
21. Von Bary S, Kuhn J, Krieger S, et al. Vena cava filter–prevention of pulmonary embolism. Report of clinical experiences. Zentralbl Chir 1999;124:27–31. 22. Becker D, Philbrick J, Selby J. Inferior vena cava filters. Indications, safety, effectiveness. Arch Intern Med 1992;152: 1985–1994. 23. Ferrell A, Byrne T, Robinson J. Placement of inferior vena cava filters in bariatric surgical patients–possible indications and technical considerations. Obes Surg 2004;14:738–743.
Because the prevalence of obesity and its complications continues to increase at an alarming rate, surgery for
morbid obesity is being used more frequently because it remains the only durable treatment for this condition. Patients undergoing surgical treatment for obesity are at increased risk for venous thromboembolism (VTE), and the incidence of perioperative pulmonary embolism (PE) has been estimated at 1%.1,2 Although uncommon, PE is a major cause of postoperative mortality, and it has recently been identified as an independent risk factor for perioperative death.3 Newer concepts in surgical care for morbidly obese patients with the potential to influence the occurrence of postoperative PE include the use of minimally invasive surgical access and new forms of anticoagulation (eg, low molecular weight heparin [LMWH]).
Competing Interests Declared: None. Presented at the American College of Surgeons 91st Annual Clinical Congress, San Francisco, CA, October 2005. Received February 8, 2006; Revised August 25, 2006; Accepted August 25, 2006. From the Division of General Surgery, Department of Surgery, Virginia Commonwealth University, Richmond, VA (Carmody, Sugerman, Kellum, Jamal, Johnson, Carbonell, Maher, Wolfe) and the Department of Surgery, Duke University, Durham, NC (DeMaria). Correspondence address: Brennan J Carmody, MD, FACS, Walter Reed Army Medical Center, 6900 Georgia Ave, NW, Washington, DC, 203075001.
© 2006 by the American College of Surgeons Published by Elsevier Inc.
831
ISSN 1072-7515/06/$32.00 doi:10.1016/j.jamcollsurg.2006.08.020
832
Carmody et al
Pulmonary Embolism Complicating Bariatric Surgery
J Am Coll Surg
Table 1. Bariatric Procedures Performed During Study Period Abbreviations and Acronyms
BMI IVC LMWH PE VTE
⫽ ⫽ ⫽ ⫽ ⫽
body mass index inferior vena cava low molecular weight heparin pulmonary embolism venous thromboembolism
It remains unclear whether or not such advances have contributed to an improvement in the incidence of postoperative PE or in PE-related mortality. Because large study cohorts are necessary to define low incidence complications like PE, factors related to incidence, prevention, time frame for occurrence, and mortality related to PE remain poorly defined. This study was undertaken in an attempt to better characterize PE after bariatric surgery by reviewing the occurrence of this complication over the 24-year history of a tertiary care hospital. Changes in surgical technique, and particularly, the shift toward laparoscopy for gastric bypass, and changes in strategy for prophylaxis, including routine anticoagulation, have occurred in this surgical center over this time frame. We hypothesized that the incidence of PE would decrease over the 24-year course of our experience as a result of the impact of these and other improvements on patient care. An additional objective of the study was to analyze risk factors in this large cohort of patients to determine patients at higher risk for PE who might benefit from more aggressive preventive measures such as prophylactic insertion of a vena caval (IVC) filter.
Procedure
VBGP PGBP DGBP Revision to PGBP Revision to DGBP HGP Revision of PGBP Revision of DGBP LLGBP Revision to LLGBP Revision of LLGBP LAGB Laparoscopic PGBP Hand assist laparoscopic PGBP Laparoscopic LLGBP Laparoscopic revision to PGBP Duodenal switch Total
n
%
101 1,369 49 191 51 51 49 21 878 58 14 37 898 25 48 14 7 3,861 (3,123 women, 738 men)
2.64 35.8 1.28 4.99 1.33 1.33 1.28 0.55 22.96 1.52 0.37 0.96 23.48 0.65 1.26 0.37 0.18 100
DGBP, distal gastric bypass; HGP, horizontal gastroplasty; LAGB, laparoscopic adjustable gastric banding; LLGBP, long limb gastric bypass; PGBP, proximal gastric bypass; VBGP, vertical banded gastroplasty.
with PE after weight reduction surgery were characterized in detail in a retrospective fashion by chart review to capture data not recorded prospectively, including interval between procedure and PE, patient status at diagnosis of PE (inpatient versus outpatient), and outcomes. The association between PE, patient comorbidities, and other complications was assessed. Venous thromboembolism prophylaxis
METHODS Patients
The database of patients undergoing bariatric procedures between 1980 and 2004 at Virginia Commonwealth University was analyzed. These data were collected prospectively and updated according to inpatient and outpatient clinic records. Institutional Review Board approval was granted for reviewing data and reporting analyses. During the study period, 3,861 patients underwent bariatric surgery (Table 1). This cohort included 2,839 open procedures and 1,022 laparoscopic cases. Prospectively collected data included detailed demographic information, comorbid conditions at the time of operation, information about surgical procedure and access technique, and outcomes including complications, weight loss, and so forth. Patients diagnosed
Between 1980 and 1992, thigh-length intermittent pneumatic compression devices were used on all patients undergoing bariatric surgery; ambulation was initiated on the evening of operation, except in patients requiring mechanical ventilation. Since 1992, routine prophylaxis included the addition of routine anticoagulation by instituting a computerized pathway for patient care. Anticoagulation consisted of either unfractionated heparin (5,000 U subcutaneously bid or every 8 hours) or LMWH (enoxaparin, Aventis Pharmaceuticals, Inc), but details about choice of anticoagulation agent, dosage, and administration schedule (such as preoperative versus intraoperative versus postoperative initiation) were not available. In 1998, standardization of the anticoagulation protocol was instituted for all patients undergoing bariatric surgery. The agent used from that time forward
Vol. 203, No. 6, December 2006
Carmody et al
Pulmonary Embolism Complicating Bariatric Surgery
833
Table 2. Demographics of Patients Experiencing Pulmonary Embolism
Table 3. Obesity-related Comorbidities in Pulmonary Embolism Patients
Demographic
Comorbidity
n
%
Hypertension Diabetes mellitus Obstructive sleep apnea Obesity hypoventilation syndrome Venous stasis disease
15 4 10 5 4
45.5 12.1 30.3 15.2 12.1
Gender 23 women 10 men Age, y Body mass index
Mean value (range)
41.4 (16–59) 57.2 kg/m2 (40.3–84)
was LMWH (enoxaparin) at a dose of 40 mg subcutaneously before operation and repeated in a daily dose until hospital discharge. Statistical analysis
All analyses were performed using SAS (SAS Institute, Inc). Means, medians (both ⫾ standard error), and ranges are shown for interval data. Univariate and multivariate analyses were used to evaluate differences between proportions. A p value ⬍ 0.05 was considered statistically significant. RESULTS Overall, 36 patients were diagnosed with PE after weight reduction operations. PE was a late event, occurring more than 4 months after operation (range 148 to 1,137 days) in three patients who were excluded from additional analysis because it was believed that the PE did not represent a perioperative event. The demographics of patients suffering PE are listed in Table 2. The mean age was 41.4 years (range 16 to 59 years), and the mean body mass index (BMI) was 57.2 ⫾ 2.4 kg/m2 (range 40.3 to 84 kg/m2). Preoperative BMI was substantially greater in patients suffering PE than it was in those in whom this complication did not develop (49.9 ⫾ 0.2 kg/m2, p ⬍ 0.01, Wilcoxon rank test). Four of the patients in whom PE developed were diagnosed with deep vein thrombosis (12%). Comorbidities are listed in Table 3. A variety of modalities were used in diagnosing PE; radiographic tests supported the diagnosis in 22 patients; in 8 patients, the diagnosis was made at autopsy (Table 4). In the remaining three patients, the diagnosis was made on clinical grounds alone by virtue of profound hypoxemia in a patient judged to be too unstable for confirmative testing. The overall incidence of perioperative PE was 0.85%. The incidence of PE did not differ between patients undergoing open surgery (0.84%) and those undergoing
laparoscopic surgery (0.88%). Pulmonary embolus incidence did not differ in patients undergoing primary bariatric procedures or revisional operations (0.9% versus 0.5%, respectively, p ⫽ 0.6, Fisher’s exact test). A notable difference in PE incidence was identified when comparing patients with BMIs ⬎ 50 kg/m2 with those with BMI ⬍ 50 kg/m2 (1.4% versus 0.5%, p ⫽ 0.005, Fisher’s exact test). The interval between procedure and PE was 13.2 ⫾ 2.6 days for the entire cohort. Strikingly, one-third of patients presented with PE after hospital discharge. No marked difference in interval between surgery and PE was found for patients undergoing open surgery (13.0 ⫾ 3.0 days) as compared with laparoscopic surgery (14.1 ⫾ 6.49 days). Nine of the 33 patients died, for a PE-associated mortality of 27% and an overall fatal PE incidence of 0.2% in the entire population of bariatric surgery patients. In eight of nine deaths, pulmonary emboli were confirmed by autopsy. The remaining patient had a large PE diagnosed by helical CT scan and died soon thereafter, but no postmortem examination was performed. The BMIs of patients experiencing fatal PE were considerably greater than the BMIs of patients experiencing nonfatal PE (65.6 ⫾ 4.4 kg/m2 versus 54.5 ⫾ 2.7 kg/m2, p ⬍ 0.01). Fatal PEs occurred sooner after operation than nonfatal PEs did. In the fatal PE group, the median interval between procedure and PE was 3 days (range 2 to 45 days); the median interval for the nonfatal PE group was Table 4. Diagnosis of Pulmonary Embolism Modality
n
CT scan Pulmonary arteriogram V/Q scan Autopsy Clinical (ABG) Total
10 1 11 8 3 33
ABG, arterial blood gas; V/Q, ventilation/perfusion scan.
834
Carmody et al
Pulmonary Embolism Complicating Bariatric Surgery
J Am Coll Surg
Table 5. Univariate Analysis: Factors Associated with Pulmonary Embolism Factor
Pulmonary embolism
No pulmonary embolism
30.3 41.4 57.2 45.5 12.1 30.3 15.2 12.1
18.9 40.5 49.9 46.7 18.4 25.1 4.2 3.3
43 7.7
2.4 4.1
Male gender, % Age, y Body mass index, kg/m2 Hypertension, % Diabetes mellitus, % Obstructive sleep apnea, % Obesity hypoventilation syndrome, % Venous stasis disease, % Gastrointestinal leak, % Pre 1992 Post 1992
p Value
0.1 0.6 0.01 0.8 0.4 0.5 0.01 0.03
95% CI*
⫺4.49, 2.63 ⫺10.73, ⫺3.79
0.0006 0.3
*Difference in the means with the 95% confidence interval.
10 days (range 1 to 51 days). This difference did not reach statistical significance. Univariate analysis (Fisher’s exact test) identified obesity hypoventilation syndrome and chronic venous insufficiency as risk factors for development of PE (Table 5). No marked difference in PE incidence was detected when the latter era of routine heparin prophylaxis (0.9%) was compared with the earlier era of inconsistent or absent heparin administration (0.6%). Multivariate analysis of all available variables confirmed preoperative weight as the only independent predictor of PE risk (361.9 ⫾ 17.1 lbs versus 307.2 ⫾ 1.1 lbs, p ⫽ 0.003). In the pre-1992 era, gastrointestinal leak complicating surgery was identified by univariate analysis as a predictor of PE, but this correlation disappeared during the more recent era of routine anticoagulation prophylaxis. IVC filters were placed prophylactically in 145 patients undergoing bariatric surgery over the 24-year period of study. Despite the IVC filter, postoperative PE developed in three patients and one died as a consequence of that PE. The median interval between procedure and PE was 5 days (range 2 to 26 days). The mean age and BMI of this subgroup were 47 years and 57.3 kg/m2, respectively, and each of the three patients abused tobacco. DISCUSSION Despite the use of routine inpatient chemoprophylaxis with heparin, the incidence of PE in bariatric surgery patients has remained relatively stable, even with the addition of standardized heparin protocols and the application of minimally invasive approaches to weight reduction surgery. Regarding access technique, the discrepancy in numbers between the open and laparoscopic
groups might mask a difference in PE incidence, which may surface as the laparoscopic group increases in size. The increased incidence of thrombosis in morbidly obese patients appears to be attributable to more than the mechanical effects of extreme obesity (decreased venous return, venous stasis).4 The endocrine activity of adipose tissue has been implicated in the dysregulation of the coagulation and fibrinolytic systems, favoring a thrombotic state. These adipocyte-secreted proteins, which are upregulated in obesity, include plasminogenactivator inhibitor-1 (PAI-1), adipsin (complement D), and adipocyte complement-related protein (Acrp30).5,6 This hypercoagulable milieu may be a partial explanation for the persistence of VTE at a consistent rate despite prophylactic measures. A wealth of data support the use of prophylactic anticoagulation in patient groups at high risk for VTE developing.7-9 The Seventh American College of Chest Physicians Conference on Antithrombotic and Thrombolytic Therapy10 endorsed routine heparin prophylaxis (unfractionated or LMWH) combined with intermittent pneumatic compression devices for high risk general surgery patients. This group also recommended extended prophylactic anticoagulation (at least 10 days) for patients undergoing elective knee or hip arthroplasty or urgent hip fracture surgery. Even within the bariatric surgery community, there was previously no consensus about the most effective prophylactic regimen for VTE. In a 2000 survey of American Society of Bariatric Surgery (ASBS) members, more than 95% of participants reported using routine prophylaxis, and 38% used two prophylactic measures simultaneously.11 Eleven percent of surveyed members routinely discharged patients with
Vol. 203, No. 6, December 2006
Carmody et al
prophylaxis. More recently, at the 2005 annual meeting of the American Society of Bariatric Surgery, an audience poll at the Master’s course revealed that no surgeon admitted to routinely withholding some form of heparin prophylaxis. So, although choice of agent and dosage likely continue to be nonstandardized, it appears that bariatric surgeons have adopted routine anticoagulation for these high risk patients. In this study, prophylaxis since 1992 included enoxaparin, 40 mg subcutaneously, daily during hospitalization. Our data collection system prevented us from comparing the various forms of chemoprophylaxis used during the study period. It is unclear whether LMWH is most efficacious using standard regimens or weight-based protocols. Several authors have compared various chemoprophylactic regimens (LMWH versus fractionated heparin, various nonweightbased dosages of LMWH) without conclusive results favoring the safest and most effective form of anticoagulation.12-15 Additionally, although a correlation has been demonstrated between antifactor Xa activity and body weight with LMWH administration, thrombosis has been reported, despite therapeutic anti-Xa levels.16 Additional data on anti-Xa activity during LMWH prophylaxis may determine whether a weightbased dosing system is more effective than standard dose regimens. The incidence of fatal PE in this study was similar to that reported by Sapala and colleagues,17 as were the associated risk factors (obesity hypoventilation syndrome, venous stasis disease, high BMI). Because multivariate analysis confirmed preoperative weight as a risk factor for PE, the BMI of patients experiencing fatal PE was substantially higher than that of those in whom nonfatal PE was developing. This study also identified a trend in the difference in median interval between procedure and PE in patients with lethal PE versus those with nonlethal PE (3 days versus 10 days). Because prophylactic measures did not differ, this finding may be explained by preexisting thrombus at the time of operation, a more virulent prothrombotic state with extreme obesity, or the fact that standardized, nonweight-based regimens are not effective in this group. These data also revealed an association between postoperative anastomotic leak and postoperative pulmonary embolism in the era before routine heparin prophylaxis. Although record review confirmed these patients received unfractionated sub-
Pulmonary Embolism Complicating Bariatric Surgery
835
cutaneous heparin (5,000 International Units subcutaneously twice or thrice daily) after leak diagnosis, they were also undergoing revisional surgery, prolonged mechanical intubation, and sedation. It is possible that the implementation of routine LMWH use in 1992, in addition to improvements in critical care, nursing, specialty beds, and a lower threshold for IVC filter insertion in critically ill patients, may be factors in the disappearance of the association between PE and leak after 1992. Autopsy studies have documented the high incidence of PE in patients who have died in the early postoperative period after gastric bypass. Melinek and associates5 reported a case series of 10 autopsies after gastric bypass; 3 of the deaths were attributable to PE (mean BMI 51.6 kg/m2, mean interval between procedure and death 11.6 days, mean age 44.6 years). In 8 of the 10 patients, however, microscopic evidence of PE was present, and in 2 patients, evidence of recent PE was documented despite previous IVC filter placement. Our data revealed a worrisome incidence of PEs occurring on an outpatient basis after hospital discharge. This finding is likely related to the decreased hospital stays associated with minimally invasive bariatric surgery. Huber and colleagues18 reported 30% more postoperative PEs when embolic events occurring after hospital discharge in postsurgical patients were included. Recently, we have considered a more aggressive approach to postdischarge chemoprophylaxis. Uniform extension of the period of prophylaxis for all bariatric surgery patients would have tremendous cost implications and would likely increase the risk of late bleeding complications beyond the risk of a PE, so this concept would clearly require more study before adoption. But an additional 2 to 4 weeks of enoxaparin for patients with BMI ⱖ 50 kg/m2 or impaired mobility is supported by this review and should also be evaluated in studies to assess the risks versus benefits of such an approach. Failure of IVC filters to prevent PE has been described by several investigators.19-22 This complication occurs at a frequency of 0.3% to 7.7% and has been reported with multiple filter types (Greenfield, SimonNitinol, Bird’s Nest). Von Bary and coworkers21 identified incorrect filter placement as a factor in three cases of reembolism, and Poletti and colleagues20 noted 20% of recurrent emboli originating distal to the filter. The eti-
836
Carmody et al
Pulmonary Embolism Complicating Bariatric Surgery
ology of most postfilter PEs is unclear. Possibilities include filter malposition, strut fracture, upper extremity source, caval thrombus distal to the filter, or hypercoagulable state (three patients used tobacco).23 Sugerman and coauthors4 advocated prophylactic IVC filter insertion in patients with respiratory insufficiency (ie, obesity hypoventilation syndrome), pulmonary hypertension (mean pulmonary artery pressure ⱖ 40 mmHg), and venous stasis disease. We now also recommend preoperative filter insertion for patients who are superobese and who have limited mobility, including those with neuromuscular disorders (eg, multiple sclerosis, muscular dystrophy) or severe degenerative joint disease. These data confirm the persistence of VTE at a stable rate despite routine heparin prophylaxis and minimally invasive techniques. Whether increased use of laparoscopic bariatric procedures and individualized heparin dosing will influence the incidence remains to be determined. When taken in context with the notable and compelling evidence that anticoagulation is effective for reducing the risk of VTE in high risk general surgery patients,7-10 we do not believe our data refute the role of routine anticoagulation in preventing VTE in bariatric surgery patients. In contrast, the appropriate conclusion from the current data is that additional strategies are needed to reduce the incidence of this life-threatening complication. Because one-third of PEs occur after hospital discharge, consideration should be given to continuing anticoagulation on an outpatient basis in selected patients. A more aggressive policy for preoperative IVC filter insertion, particularly influenced by the recent availability of temporary removable filter devices, should be considered in patients at higher risk for PE (BMI ⬎ 50 kg/m2, history of VTE, obesity hypoventilation syndrome, or venous insufficiency). Future studies to delineate superiority of various available anticoagulants, drug- and dose-related safety profiles, optimal duration of prophylaxis, and dose-related efficacy in the bariatric surgery patient are needed in this high risk population. Author Contributions
Study conception and design: Carmody, DeMaria Acquisition of data: Carmody Analysis and interpretation of data: Carmody, Sugerman, Kellum, Jamal, Johnson, Carbonell, Maher, Wolfe, DeMaria
J Am Coll Surg
Drafting of manuscript: Carmody Critical revision: Sugerman, DeMaria REFERENCES 1. Westling A, Bergqvist D, Boström A, et al. Incidence of deep venous thrombosis in patients undergoing obesity surgery. World J Surg 2002;26:470–473. 2. Eriksson S, Backman L, Ljungstrom K. The incidence of clinical postoperative thrombosis after gastric surgery for obesity during 16 years. Obes Surg 1997;7:332–335; discussion 336. 3. Fernandez A, DeMaria E, Tichansky D, et al. Multivariate analysis of risk factors for death following gastric bypass for treatment of morbid obesity. Ann Surg 2004;239:698–702; discussion 702–703. 4. Sugerman H, Sugerman E, Wolfe L, et al. Risks and benefits of gastric bypass in morbidly obese patients with severe venous stasis disease. Ann Surg 2001;234:41–46. 5. Melinek J, Livingston E, Cortina G, et al. Autopsy findings following gastric bypass surgery for morbid obesity. Arch Pathol Lab Med 2002;126:1091–1095. 6. Ahima R, Flier J. Adipose tissue as an endocrine organ. Trends Endocrinol Metab 2000;11:327–332. 7. Major K, Wilson M, Nishi G, et al. The incidence of thromboembolism in the surgical intensive care unit. Am Surg 2003; 69:857–861. 8. Tapson V, Hyers T, Waldo A, et al. Antithrombotic therapy practices in US hospitals in an era of practice guidelines. Arch Intern Med 2005;165:1458–1464. 9. Bergqvist D. Low-molecular-weight heparin for the prevention of postoperative venous thromboembolism after abdominal surgery: a review. Curr Opin Pulm Med 2005;11:392– 397. 10. Geerts W, Pineo G, Heit J, et al. Prevention of venous thromboembolism: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest 2004;126(3 Suppl): 338S–400S. 11. Wu E, Barba C. Current practices in the prophylaxis of venous thromboembolism in bariatric surgery. Obes Surg 2000;10:7–13; discussion 14. 12. Scholten D, Hoedema R, Scholten R. A comparison of two different prophylactic dose regimens of low molecular weight heparin in bariatric surgery. Obes Surg 2002;12:19–24. 13. Kalfarentzos F, Stavropoulou F, Yarmenitis S, et al. Prophylaxis of venous thromboembolism using two different doses of low-molecular-weight heparin (nadroparin) in bariatric surgery: a prospective randomized trial. Obes Surg 2001;11: 670–676. 14. Shepherd M, Rosborough T, Schwartz M. Heparin thromboprophylaxis in gastric bypass surgery. Obes Surg 2003;13:249– 253. 15. Gonzalez Q, Tishler D, Plata-Munoz J, et al. Incidence of clinically evident deep venous thrombosis after laparoscopic Roux-en-Y gastric bypass. Surg Endosc 2004;18:1082–1084. 16. Rutherford E, Schooler W, Sredzienski E, et al. Optimal dose of enoxaparin in critically ill trauma and surgical patients. J Trauma 2005;58:1167–1170. 17. Sapala J, Wood M, Schuhknecht M, et al. Fatal pulmonary embolism after bariatric operations for morbid obesity: a 24year retrospective analysis. Obes Surg 2003;13:819–825.
Vol. 203, No. 6, December 2006
Carmody et al
18. Huber O, Bounameaux H, Borst F, et al. Postoperative pulmonary embolism after hospital discharge. An underestimated risk. Arch Surg 1992;127:310–313. 19. Wolf F, Thurner S, Lammer J. Simon nitinol vena cava filters: effectiveness and complications. Fortschr Röntgenstr 2001;173: 924–930. 20. Poletti P, Becker C, Prina L, et al. Long-term results of the Simon nitinol inferior vena cava filter. Eur Radiol 1998;8: 289–294.
Pulmonary Embolism Complicating Bariatric Surgery
837
21. Von Bary S, Kuhn J, Krieger S, et al. Vena cava filter–prevention of pulmonary embolism. Report of clinical experiences. Zentralbl Chir 1999;124:27–31. 22. Becker D, Philbrick J, Selby J. Inferior vena cava filters. Indications, safety, effectiveness. Arch Intern Med 1992;152: 1985–1994. 23. Ferrell A, Byrne T, Robinson J. Placement of inferior vena cava filters in bariatric surgical patients–possible indications and technical considerations. Obes Surg 2004;14:738–743.