Surgery for Obesity and Related Diseases ] (2015) 00–00
Original article
Pulmonary embolism and gastrointestinal leak following bariatric surgery: when do major complications occur? Konstantinos Spaniolas, M.D.a,*, Kevin R. Kasten, M.D.a, Megan E. Sippey, M.D.a, John R. Pender, M.D.b, William H. Chapman, M.D.a, Walter J. Pories, M.D.a a
Department of Surgery, Brody School of Medicine, East Carolina University, Greenville, North Carolina b Department of Surgery, Billings Clinic, Billings, Montana Received March 24, 2015; accepted May 3, 2015
Abstract
Background: Complications following bariatric surgery are uncommon but potentially life threatening. Objectives: The aim of this study was to assess the timing of gastrointestinal leaks (GIL) and pulmonary embolism (PE) in patients undergoing bariatric surgery. Setting: Retrospective analysis of the nationwide American College of Surgeons National Surgical Quality Improvement Program database from 2006 to 2011. Methods: Data on patient demographic characteristics, baseline co-morbidities, procedural events, and postoperative occurrences were analyzed. Thirty-day morbidity was assessed. Median (interquartile range) and frequencies are reported. Results: We identified 71,694 bariatric surgery patients; median age was 45 years (range 36–54 yr), and median body mass index was 44.8 kg/m2 (range 40.8–50.3 kg/m2). Laparoscopic Roux-en-Y gastric bypass was performed in 39,480 patients, laparoscopic adjustable band in 21,104, laparoscopic sleeve gastrectomy in 3225, open Roux-en-Y gastric bypass in 4243, duodenal switch in 1064, revisional surgery in 1182, and other procedures in 1396 patients. Of these patients, 95.2% had no complications. GIL was found in 441 (.6%), deep vein thrombosis in 184 (.3%), and PE in 134 (.2%). These complications occurred 10 (5–15), 13 (7–20), and 11 (4–19) days after surgery, respectively. GIL and PE developed after discharge in 275 (62.4%) and 96 (71.6%), respectively. Only 35 (26.1%) of the patients who developed PE had deep vein thrombosis. There were no differences in patient characteristics between the groups of early PE versus postdischarge PE. Patients diagnosed with in-hospital GIL were more obese with more severe systemic disease compared with patients with postdischarge diagnosis. Conclusions: The majority of GILs and PEs after bariatric surgery occur after discharge. This finding goes against the routine use of contrast studies to rule out GIL. The risk of PE remains after discharge from bariatric surgery. (Surg Obes Relat Dis 2015;]:00–00.) r 2015 American Society for Metabolic and Bariatric Surgery. All rights reserved.
Keywords:
Complications; Leak; Pulmonary embolism; Organ space infection; Bariatric; NSQIP
Presented in part at the 2014 Obesity Week, November 2014, Boston, MA. * Correspondence: Konstantinos Spaniolas, M.D., Assistant Professor of Surgery, Brody School of Medicine at ECU, 600 Moye Boulevard, Greenville, NC, 27834, USA. E-mail: [email protected]
Bariatric surgery has been proven to be the most effective therapy for severe obesity and associated diseases [1]. Despite the significant improvement in morbidity and mortality following bariatric surgery over the past 2 decades, complications still occur in 4.3–14.9% of patients [2,3]. Two major areas of morbidity involve venous
http://dx.doi.org/10.1016/j.soard.2015.05.003 1550-7289/r 2015 American Society for Metabolic and Bariatric Surgery. All rights reserved.
2
K. Spaniolas et al. / Surgery for Obesity and Related Diseases ] (2015) 00–00
thromboembolism (VTE) and gastrointestinal leaks (GIL), which occur in .33–1.3 % and .07–1% of bariatric surgery patients, respectively [2–6]. Pulmonary embolism (PE) is reported in .9% of patients following bariatric surgery [5]. Additionally, PE is the cause of death in 17–33% of all postbariatric mortalities, despite routine perioperative chemoprophylaxis [7,8]. In an attempt to minimize these 2 complications, early imaging and anticoagulation therapy have been advocated. Routine contrast radiologic evaluation is associated with sensitivity less than 50% and positive predictive value of 30–70% in detecting GIL following bariatric surgery [9– 11]. Nevertheless, routine postoperative radiographic assessment for GIL remains common practice for many surgeons [12]. The value of this practice is dependent on the timing of GIL occurrence, which remains poorly defined in the literature. Single-institution studies have suggested that the majority of GIL occurs within the first postoperative week, but large studies are not available [13,14]. Extended anticoagulation is advocated as a way to decrease VTE following major surgical procedures [15]. However, there is limited literature on the role of prolonged anticoagulation for bariatric surgery patients. The value of such practice is also dependent on the timing of PE after bariatric procedures. The aim of this study was to assess the timing of GIL and PE in patients undergoing bariatric surgery and determine the differences in timing between various bariatric procedures. Methods This is a retrospective study using previously collected data from the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database. The ACS-NSQIP is an audited, validated clinical database that contains data obtained by specialized clinical abstractors on demographic characteristics, pre-existing comorbidities, intraoperative details, and postoperative outcomes. Deidentified participant use files were used for this study. The study population consisted of patients undergoing bariatric procedures, identified using Current Procedural Terminology (CPT) codes (43644, 43645, 43770, 43771, 43771, 43772, 43773, 43774, 43775, 43845, 43842, 43843, 43845, 43846, 43847, and 43848), with morbid obesity as the primary diagnosis. The presence of organ space infection was used as a surrogate for GIL, similar to previous literature [14,16,17]. Patient co-morbidities were grouped on the basis of organ systems. Congestive heart failure within 30 days, myocardial infarction within 6 months, previous percutaneous coronary intervention, previous cardiac surgery, and history of angina within 30 days before the index procedures were considered cardiac comorbidities. A history of hypertension, amputation, revascularization procedure, rest pain, and gangrene were
considered vascular co-morbidities. Impaired sensorium, coma, transient ischemic attack, cerebrovascular accident, hemiplegia, paraplegia, quadriplegia, and tumor involving the central nervous system were considered neurologic comorbidities. Pulmonary co-morbidities included a history of chronic obstructive pulmonary disease, current pneumonia, and preoperative ventilatory dependence. The timing of VTE and GIL after bariatric surgery was assessed using descriptive statistical methodology. Given that the ACS-NSQIP was the source of patient data for the present study, all outcomes were limited to 30 days after surgery. Comparisons were made between patients who developed these complications while hospitalized and those who developed them after discharge. Additional comparisons were made on the basis of the 2 most common bariatric procedures performed: laparoscopic Roux-en-Y gastric bypass (RYGB) and laparoscopic sleeve gastrectomy (SG). Approval for this study was obtained from the East Carolina University and Medical Center Institutional Review Board. The ACS-NSQIP and the hospitals participating in the ACS-NSQIP are the source of the data used in this study; the authors of this study have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived in this study. Data analysis was performed using SPSS for Windows version 20 (IBM, Somers, NY). Nominal and ordinal variables were compared using the Chi-square or Fisher’s exact test, and continuous variables were compared using the MannWhitney U-test. Results were reported as median (25th to 75th percentile) for continuous variables, and frequency for nominal and ordinal variables. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported when applicable. A P value o .05 was considered statistically significant. Results There were 71,694 bariatric patients included in the ACSNSQIP database during the time period examined; median age was 45 years (range 36–54 yr) and body mass index (BMI) 44.8 kg/m2 (range 40.7–50.3 kg/m). The majority of patients were female (78.6%). Laparoscopic RYGB was performed in 39,480 patients, laparoscopic adjustable band in 21,104, laparoscopic SG in 3225, open RYGB in 4243, duodenal switch in 1064, revisional surgery in 1182, and other procedures in 1396 patients. GIL was found in 441 patients (.6%), PE in 134 (.2%), and deep venous thrombosis (DVT) in 184 (.3%). Of the 134 patients who developed postoperative PE, 96 (71.6%) were diagnosed after hospital discharge from the index operation. The median postoperative day of diagnosis was day 11 (range, 4–19) (Fig. 1). The difference in the postoperative day of PE diagnosis between laparoscopic RYGB and SG did not reach statistical significance (11 [range, 5–19] versus 18.5 [13.25–21.25]; P ¼ .086).
Timing of Major Complications after Bariatric Surgery / Surgery for Obesity and Related Diseases ] (2015) 00–00
3
Fig. 1. Distribution of postoperative day for diagnosis of PE and GIL after bariatric surgery. Normal curve is superimposed on the histogram. (GIL¼gastrointestinal leak, PE¼pulmonary embolism)
Concurrent DVT was reported in 35 patients (26.12%). There were 149 additional patients with DVT who were not diagnosed with PE. Patients who developed PE before discharge had no significant differences in baseline characteristics compared with patients who presented with postdischarge PE (Table 1). GILs occurred 10 (5–15) days postoperatively (see Fig. 1); diagnosis was made after hospital discharge from the index operation in 275 (62.4%) patients. There was no difference in the postoperative day of GIL occurrence between laparoscopic RYGB and SG (10 [5–16] versus 11 [7–14.75]; P = .72). Patients who developed early GIL had significantly higher incidence of pre-existing vascular disease, higher BMI, advanced age, and, more commonly, severe chronic systemic illness (Table 2). Discussion Our study evaluated the prevalence and timing of 2 serious postoperative occurrences after bariatric surgery
using a large sample size. GIL occurs mostly after hospital discharge from the index bariatric procedure. This is in contrast to previous single-institution studies reporting that the majority of GIL occurs within the first postoperative week [10,18]. In a study of 50 GILs from 904 RYGB patients, 77% were diagnosed within a week from surgery [13]. Another study on 3828 RYGB patients with 150 GILs reported that median time to diagnosis of the leak ranged between 2 and 4 days, depending on the type of GIL [19]. Similar findings have been reported in a large recent retrospective study of 6030 RYGB patients with 64 GILs [14]. The median day for leak diagnosis was postoperative day 5, and 75% of leaks were diagnosed by day 12. Interestingly, greater variation is reported in the literature for the timing of GIL after SG. In a retrospective study of 15 GILs in 539 SG patients, the mean time from index procedure to diagnosis was 27 days, with a broad range (1– 102 days) [20]. A smaller study of 5 GILs after SG reported diagnosis at mean postoperative day 3 (range, 1–7) [21]. A recent systematic review of 316 GILs from 15,772 patients
Table 1 Comparison in baseline characteristics in patients with postoperative PE, based on timing Variable
PE in hospital (n ¼ 35)
PE after discharge (n ¼ 96)
P value
Odds ratio (95% confidence interval)
Vascular co-morbidities Cardiac co-morbidities Pulmonary co-morbidities Neurologic co-morbidities Smoking Diabetes Male gender Age BMI ASA 3 or 4 RYGB
18 2 1 1 5 12 8 51 47.6 27 23
67 3 7 4 8 32 32 50.5 49.7 73 57
.051 .611 .682 1.0 .331 .919 .249 .574 .544 .896 .51
2.18 0.54 2.71 1.5 0.55 0.96 1.69 1.0 1.01 .94 .76
(51.4%) (5.7%) (2.9%) (2.9%) (14.3%) (34.3%) (22.9%) (38–59) (43–58.8) (77.1%) (65.7%)
(69.8%) (3.2%) (7.4%) (4.2%) (8.3%) (33.3%) (33.3%) (42–58.75) (41.9–56.8) (76%) (59.4%)
(.99–4.82) (.09–3.36) (.32–22.81) (.16–13.85) (.17–1.8) (.43–2.17) (.69–4.13) (.98–1.04) (.98–1.05) (.38–2.35) (.34–1.71)
PE ¼ pulmonary embolism; BMI ¼ body mass index; ASA ¼ American Society of Anesthesiology class; RYGB ¼ laparoscopic gastric bypass.
K. Spaniolas et al. / Surgery for Obesity and Related Diseases ] (2015) 00–00
4
Table 2 Comparison in baseline characteristics in patients with postoperative GIL, based on timing Variable
GIL in hospital (n ¼ 160)
GIL after discharge (n ¼ 275)
P value
Odds ratio (95% confidence interval)
Vascular co-morbidities Cardiac co-morbidities Pulmonary co-morbidities Neurologic co-morbidities Smoking Diabetes Male gender Age BMI ASA 3 or 4 RYGB
110 9 6 3 25 59 40 47 47.1 130 83
156 6 9 6 48 82 59 45 45.5 189 160
.015 .057 .789 1.0 .129 .129 .409 .008 .002 .004 .201
.6 .37 .87 1.16 1.14 .73 .82 1.03 1.03 .51 1.29
(68.8%) (5.7%) (3.8%) (1.9%) (15.6%) (36.9%) (25.2%) (39–57.75) (41.8–55.1) (81.2%) (51.9%)
(56.9%) (2.2%) (3.3%) (2.2%) (17.5%) (29.8%) (21.7%) (38–52) (41.2–51.1) (68.7%) (58.2%)
(.4–.91) (.13–1.07) (.3–2.48) (.29–4.72) (.67–1.94) (.48–1.1) (.52–1.31) (1.01–1.04) (1.01–1.06) (.32–0.81) (.87–1.91)
GIL ¼ gastrointestinal leak; BMI ¼ body mass index; ASA ¼ American Society of Anesthesiology class; RYGB ¼ laparoscopic gastric bypass.
reported that the clinical presentation ranged from 1– 12 days for RYGB and 1–35 days for SG [22]. Although this difference was not statistically assessed, it is in contrast to the present study, which shows no significant difference in the timing of GIL diagnoses between the 2 procedures. Although the use of routine postoperative contrast studies to rule out GIL has been debated, multiple retrospective studies suggest low sensitivity and low positive predictive value for this approach [9–11,18,23]. There is no consensus at present, and the American Society for Metabolic and Bariatric Surgery guidelines recognize the common use of such practice [12]. Although our study was not designed to evaluate the role of routine contrast imaging, it revealed that most GIL occurs after hospital discharge and, as such, indicated the low diagnostic yield of routine contrast imaging to detect GILs. The present analysis found that most PEs occur after discharge. This is in accordance with the findings of previous studies that examined the effect of different anticoagulation regimens on the VTE rate. In a multicenter retrospective study of 668 patients undergoing open bariatric surgery, 6 patients developed symptomatic PE, and all of them presented after discharge [24]. This is similar to a retrospective single-institution study on 1125 open RYGB patients, with mean PE diagnosis at postoperative day 10 (range 4–30) [25]. The delayed presentation of PE and the decrease in the VTE rate with prolonged anticoagulation have led to the practice of extended prophylaxis, commonly for 10 days [26,27]. In a prospective study comparing 132 and 176 bariatric patients who underwent in-hospital-only or 10-day prophylaxis, respectively, the VTE rate was significantly higher in the short prophylaxis group (4.5% versus 0; P = .006) [27]. Practice patterns in terms of prolonged anticoagulation are unknown. A survey of bariatric surgeons suggested that 48.3% continue anticoagulation after discharge, but the response rate to the survey was not provided [28]. Furthermore, 2 recent prospective randomized trials of anticoagulation modalities for bariatric patients limited the duration of
prophylaxis to in-hospital or o10 days, respectively [29,30]. This is in contrast to a systematic review that recommended 1-month prophylaxis in patients undergoing major open or laparoscopic abdominal or pelvic surgery [15]. These studies collectively have suggested that the landscape of prophylaxis for bariatric surgery patients is unclear and that more information is needed. Our study reports the delayed nature of PE presentation and therefore implies that anticoagulation for 410 days may be beneficial, similar to what is suggested for other major abdominal surgery. Our study is limited by the retrospective design and selection bias. Specifically, the ACS-NSQIP database is not bariatric specific, and therefore, it is possible that postoperative occurrences may be missed or miscoded. Our definition of a GIL was based on the presence of organ space infection. Although an organ space infection may not require the presence of GIL, this association has been previously used to assess for GIL in bariatric and other gastrointestinal surgery studies [14,16,17]. This study is further limited by sample size. Specifically, the difference seen in the timing of PE after RYGB and SG was not statistically significant, but we are unable to draw definitive conclusions on this finding because our study was underpowered to adequately assess this observed difference. Furthermore, although patient follow-up in ACS-NSQIP is consistent, patients are only followed up for 30 days, and thus we were unable to assess for complications after this time point. There were minimal differences in baseline characteristics between patients who developed any of the complications examined (PE and GIL) before or after hospital discharge. This study suggests that older and chronically ill patients are more likely to develop GIL before hospital discharge, but our ability to predict which patients will develop delayed complications was limited on the basis of the data provided by this study. Conclusions Our study reveals that GIL and PE occur mostly after hospital discharge following bariatric surgery, with no
Timing of Major Complications after Bariatric Surgery / Surgery for Obesity and Related Diseases ] (2015) 00–00
significant differences between laparoscopic RYGB and SG in terms of timing. These findings do not support the routine use of contrast studies to evaluate for GIL after bariatric surgery. Furthermore, prolonged anticoagulation beyond 10 days may be beneficial in providing prophylaxis against PE for this patient population. Future directions would need to involve the cost-effectiveness of such changes in postoperative care after bariatric surgery. Disclosures The authors have no commercial associations that might be a conflict of interest in relation to this article.
[14]
[15]
[16] [17]
[18]
References [19] [1] Courcoulas AP, Yanovski SZ, Bonds D, et al. Long-term outcomes of bariatric surgery: a National Institutes of Health symposium. JAMA Surg 2014;149(12):1323–9. [2] Birkmeyer NJ, Dimick JB, Share D, et al. Michigan Bariatric Surgery Collaborative. Hospital complication rates with bariatric surgery in Michigan. JAMA 2010;304(4):435–42. [3] DeMaria EJ, Pate V, Warthen M, Winegar DA. Baseline data from American Society for Metabolic and Bariatric Surgery-designated Bariatric Surgery Centers of Excellence using the Bariatric Outcomes Longitudinal Database. Surg Obes Relat Dis 2010;6(4):347–55. [4] Flum DR, Belle SH, King WC, et al. Longitudinal Assessment of Bariatric Surgery (LABS) Consortium. Perioperative safety in the longitudinal assessment of bariatric surgery. N Engl J Med 2009;361 (5):445–54. [5] Jamal MH, Corcelles R, Shimizu H, et al. Thromboembolic events in bariatric surgery: a large multi-institutional referral center experience. Surg Endosc 2015;29(2):376–80. [6] Finks JF, English WJ, Carlin AM, et al. Michigan Bariatric Surgery Collaborative; Center for Healthcare Outcomes and Policy. Predicting risk for venous thromboembolism with bariatric surgery: results from the Michigan Bariatric Surgery Collaborative. Ann Surg 2012;255 (6):1100–4. [7] Smith MD, Patterson E, Wahed AS, et al. Thirty-day mortality after bariatric surgery: independently adjudicated causes of death in the longitudinal assessment of bariatric surgery. Obes Surg 2011;21 (11):1687–92. [8] Omalu BI, Ives DG, Buhari AM, et al. Death rates and causes of death after bariatric surgery for Pennsylvania residents, 1995 to 2004. Arch Surg 2007;142(10):923–8. [9] Hamilton EC, Sims TL, Hamilton TT, Mullican MA, Jones DB, Provost DA. Clinical predictors of leak after laparoscopic Roux-en-Y gastric bypass for morbid obesity. Surg Endosc 2003;17(5):679–84. [10] Carter JT, Tafreshian S, Campos GM, et al. Routine upper GI series after gastric bypass does not reliably identify anastomotic leaks or predict stricture formation. Surg Endosc 2007;21(12):2172–7. [11] Madan AK, Stoecklein HH, Ternovits CA, Tichansky DS, Phillips JC. Predictive value of upper gastrointestinal studies versus clinical signs for gastrointestinal leaks after laparoscopic gastric bypass. Surg Endosc 2007;21(2):194–6. [12] ASMBS Clinical Issues Committee. ASMBS guideline on the prevention and detection of gastrointestinal leak after gastric bypass including the role of imaging and surgical exploration. Surg Obes Relat Dis 2009;5(3):293–6. [13] Carucci LR, Turner MA, Conklin RC, DeMaria EJ, Kellum JM, Sugerman HJ. Roux-en-Y gastric bypass surgery for morbid obesity:
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
5
evaluation of postoperative extraluminal leaks with upper gastrointestinal series. Radiology 2006;238(1):119–27. Jacobsen HJ, Nergard BJ, Leifsson BG, et al. Management of suspected anastomotic leak after bariatric laparoscopic Roux-en-y gastric bypass. Br J Surg 2014;101(4):417–23. Rasmussen MS, Jørgensen LN, Wille-Jørgensen P. Prolonged thromboprophylaxis with low molecular weight heparin for abdominal or pelvic surgery. Cochrane Database Syst Rev 2009;(1):CD004318. Weller WE, Rosati C. Comparing outcomes of laparoscopic versus open bariatric surgery. Ann Surg 2008;248(1):10–5. Kang CY, Halabi WJ, Chaudhry OO, et al. Risk factors for anastomotic leakage after anterior resection for rectal cancer. JAMA Surg 2013;148(1):65–71. Leslie DB, Dorman RB, Anderson J, et al. Routine upper gastrointestinal imaging is superior to clinical signs for detecting gastrojejunal leak after laparoscopic Roux-en-Y gastric bypass. J Am Coll Surg 2012;214(2):208–13. Lee S, Carmody B, Wolfe L, et al. Effect of location and speed of diagnosis on anastomotic leak outcomes in 3828 gastric bypass cases. J Gastrointest Surg 2007;11(6):708–13. Moon RC, Shah N, Teixeira AF, Jawad MA. Management of staple line leaks following sleeve gastrectomy. Surg Obes Relat Dis 2015;11 (1):54–9. Juza RM, Haluck RS, Pauli EM, Rogers AM, Won EJ, LynSue JR. Gastric sleeve leak: a single institution’s experience with early combined laparoendoscopic management. Surg Obes Relat Dis 2015;11(1):60–4. Zellmer JD, Mathiason MA, Kallies KJ, Kothari SN. Is laparoscopic sleeve gastrectomy a lower risk bariatric procedure compared with laparoscopic Roux-en-Y gastric bypass? A meta-analysis. Am J Surg 2014;208(6):903–10. Doraiswamy A, Rasmussen JJ, Pierce J, Fuller W, Ali MR. The utility of routine postoperative upper GI series following laparoscopic gastric bypass. Surg Endosc 2007;21(12):2159–62. Hamad GG, Choban PS. Enoxaparin for thromboprophylaxis in morbidly obese patients undergoing bariatric surgery: findings of the prophylaxis against VTE outcomes in bariatric surgery patients receiving enoxaparin (PROBE) study. Obes Surg 2005;15 (10):1368–74. Abou-Nukta F, Alkhoury F, Arroyo K, et al. Clinical pulmonary embolus after gastric bypass surgery. Surg Obes Relat Dis 2006;2 (1):24–8. Borkgren-Okonek MJ, Hart RW, Pantano JE, et al. Enoxaparin thromboprophylaxis in gastric bypass patients: extended duration, dose stratification, and antifactor Xa activity. Surg Obes Relat Dis 2008;4(5):625–31. Raftopoulos I, Martindale C, Cronin A, Steinberg J. The effect of extended post-discharge chemical thromboprophylaxis on venous thromboembolism rates after bariatric surgery: a prospective comparison trial. Surg Endosc 2008;22(11):2384–91. Pryor HI 2nd, Singleton A, Lin E, Lin P, Vaziri K. Practice patterns in high-risk bariatric venous thromboembolism prophylaxis. Surg Endosc 2013;27(3):843–8. Imberti D, Baldini E, Pierfranceschi MG, et al. Prophylaxis of venous thromboembolism with low molecular weight heparin in bariatric surgery: a prospective, randomized pilot study evaluating two doses of parnaparin (BAFLUX Study). Obes Surg 2014;24(2):284–91. Steele KE, Canner J, Prokopowicz G, et al. The EFFORT trial: preoperative enoxaparin versus postoperative fondaparinux for thromboprophylaxis in bariatric surgical patients: a randomized doubleblind pilot trial. Surg Obes Relat Dis 2014;pii:S1550-S7289(14) 00391–8.
Original article
Pulmonary embolism and gastrointestinal leak following bariatric surgery: when do major complications occur? Konstantinos Spaniolas, M.D.a,*, Kevin R. Kasten, M.D.a, Megan E. Sippey, M.D.a, John R. Pender, M.D.b, William H. Chapman, M.D.a, Walter J. Pories, M.D.a a
Department of Surgery, Brody School of Medicine, East Carolina University, Greenville, North Carolina b Department of Surgery, Billings Clinic, Billings, Montana Received March 24, 2015; accepted May 3, 2015
Abstract
Background: Complications following bariatric surgery are uncommon but potentially life threatening. Objectives: The aim of this study was to assess the timing of gastrointestinal leaks (GIL) and pulmonary embolism (PE) in patients undergoing bariatric surgery. Setting: Retrospective analysis of the nationwide American College of Surgeons National Surgical Quality Improvement Program database from 2006 to 2011. Methods: Data on patient demographic characteristics, baseline co-morbidities, procedural events, and postoperative occurrences were analyzed. Thirty-day morbidity was assessed. Median (interquartile range) and frequencies are reported. Results: We identified 71,694 bariatric surgery patients; median age was 45 years (range 36–54 yr), and median body mass index was 44.8 kg/m2 (range 40.8–50.3 kg/m2). Laparoscopic Roux-en-Y gastric bypass was performed in 39,480 patients, laparoscopic adjustable band in 21,104, laparoscopic sleeve gastrectomy in 3225, open Roux-en-Y gastric bypass in 4243, duodenal switch in 1064, revisional surgery in 1182, and other procedures in 1396 patients. Of these patients, 95.2% had no complications. GIL was found in 441 (.6%), deep vein thrombosis in 184 (.3%), and PE in 134 (.2%). These complications occurred 10 (5–15), 13 (7–20), and 11 (4–19) days after surgery, respectively. GIL and PE developed after discharge in 275 (62.4%) and 96 (71.6%), respectively. Only 35 (26.1%) of the patients who developed PE had deep vein thrombosis. There were no differences in patient characteristics between the groups of early PE versus postdischarge PE. Patients diagnosed with in-hospital GIL were more obese with more severe systemic disease compared with patients with postdischarge diagnosis. Conclusions: The majority of GILs and PEs after bariatric surgery occur after discharge. This finding goes against the routine use of contrast studies to rule out GIL. The risk of PE remains after discharge from bariatric surgery. (Surg Obes Relat Dis 2015;]:00–00.) r 2015 American Society for Metabolic and Bariatric Surgery. All rights reserved.
Keywords:
Complications; Leak; Pulmonary embolism; Organ space infection; Bariatric; NSQIP
Presented in part at the 2014 Obesity Week, November 2014, Boston, MA. * Correspondence: Konstantinos Spaniolas, M.D., Assistant Professor of Surgery, Brody School of Medicine at ECU, 600 Moye Boulevard, Greenville, NC, 27834, USA. E-mail: [email protected]
Bariatric surgery has been proven to be the most effective therapy for severe obesity and associated diseases [1]. Despite the significant improvement in morbidity and mortality following bariatric surgery over the past 2 decades, complications still occur in 4.3–14.9% of patients [2,3]. Two major areas of morbidity involve venous
http://dx.doi.org/10.1016/j.soard.2015.05.003 1550-7289/r 2015 American Society for Metabolic and Bariatric Surgery. All rights reserved.
2
K. Spaniolas et al. / Surgery for Obesity and Related Diseases ] (2015) 00–00
thromboembolism (VTE) and gastrointestinal leaks (GIL), which occur in .33–1.3 % and .07–1% of bariatric surgery patients, respectively [2–6]. Pulmonary embolism (PE) is reported in .9% of patients following bariatric surgery [5]. Additionally, PE is the cause of death in 17–33% of all postbariatric mortalities, despite routine perioperative chemoprophylaxis [7,8]. In an attempt to minimize these 2 complications, early imaging and anticoagulation therapy have been advocated. Routine contrast radiologic evaluation is associated with sensitivity less than 50% and positive predictive value of 30–70% in detecting GIL following bariatric surgery [9– 11]. Nevertheless, routine postoperative radiographic assessment for GIL remains common practice for many surgeons [12]. The value of this practice is dependent on the timing of GIL occurrence, which remains poorly defined in the literature. Single-institution studies have suggested that the majority of GIL occurs within the first postoperative week, but large studies are not available [13,14]. Extended anticoagulation is advocated as a way to decrease VTE following major surgical procedures [15]. However, there is limited literature on the role of prolonged anticoagulation for bariatric surgery patients. The value of such practice is also dependent on the timing of PE after bariatric procedures. The aim of this study was to assess the timing of GIL and PE in patients undergoing bariatric surgery and determine the differences in timing between various bariatric procedures. Methods This is a retrospective study using previously collected data from the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database. The ACS-NSQIP is an audited, validated clinical database that contains data obtained by specialized clinical abstractors on demographic characteristics, pre-existing comorbidities, intraoperative details, and postoperative outcomes. Deidentified participant use files were used for this study. The study population consisted of patients undergoing bariatric procedures, identified using Current Procedural Terminology (CPT) codes (43644, 43645, 43770, 43771, 43771, 43772, 43773, 43774, 43775, 43845, 43842, 43843, 43845, 43846, 43847, and 43848), with morbid obesity as the primary diagnosis. The presence of organ space infection was used as a surrogate for GIL, similar to previous literature [14,16,17]. Patient co-morbidities were grouped on the basis of organ systems. Congestive heart failure within 30 days, myocardial infarction within 6 months, previous percutaneous coronary intervention, previous cardiac surgery, and history of angina within 30 days before the index procedures were considered cardiac comorbidities. A history of hypertension, amputation, revascularization procedure, rest pain, and gangrene were
considered vascular co-morbidities. Impaired sensorium, coma, transient ischemic attack, cerebrovascular accident, hemiplegia, paraplegia, quadriplegia, and tumor involving the central nervous system were considered neurologic comorbidities. Pulmonary co-morbidities included a history of chronic obstructive pulmonary disease, current pneumonia, and preoperative ventilatory dependence. The timing of VTE and GIL after bariatric surgery was assessed using descriptive statistical methodology. Given that the ACS-NSQIP was the source of patient data for the present study, all outcomes were limited to 30 days after surgery. Comparisons were made between patients who developed these complications while hospitalized and those who developed them after discharge. Additional comparisons were made on the basis of the 2 most common bariatric procedures performed: laparoscopic Roux-en-Y gastric bypass (RYGB) and laparoscopic sleeve gastrectomy (SG). Approval for this study was obtained from the East Carolina University and Medical Center Institutional Review Board. The ACS-NSQIP and the hospitals participating in the ACS-NSQIP are the source of the data used in this study; the authors of this study have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived in this study. Data analysis was performed using SPSS for Windows version 20 (IBM, Somers, NY). Nominal and ordinal variables were compared using the Chi-square or Fisher’s exact test, and continuous variables were compared using the MannWhitney U-test. Results were reported as median (25th to 75th percentile) for continuous variables, and frequency for nominal and ordinal variables. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported when applicable. A P value o .05 was considered statistically significant. Results There were 71,694 bariatric patients included in the ACSNSQIP database during the time period examined; median age was 45 years (range 36–54 yr) and body mass index (BMI) 44.8 kg/m2 (range 40.7–50.3 kg/m). The majority of patients were female (78.6%). Laparoscopic RYGB was performed in 39,480 patients, laparoscopic adjustable band in 21,104, laparoscopic SG in 3225, open RYGB in 4243, duodenal switch in 1064, revisional surgery in 1182, and other procedures in 1396 patients. GIL was found in 441 patients (.6%), PE in 134 (.2%), and deep venous thrombosis (DVT) in 184 (.3%). Of the 134 patients who developed postoperative PE, 96 (71.6%) were diagnosed after hospital discharge from the index operation. The median postoperative day of diagnosis was day 11 (range, 4–19) (Fig. 1). The difference in the postoperative day of PE diagnosis between laparoscopic RYGB and SG did not reach statistical significance (11 [range, 5–19] versus 18.5 [13.25–21.25]; P ¼ .086).
Timing of Major Complications after Bariatric Surgery / Surgery for Obesity and Related Diseases ] (2015) 00–00
3
Fig. 1. Distribution of postoperative day for diagnosis of PE and GIL after bariatric surgery. Normal curve is superimposed on the histogram. (GIL¼gastrointestinal leak, PE¼pulmonary embolism)
Concurrent DVT was reported in 35 patients (26.12%). There were 149 additional patients with DVT who were not diagnosed with PE. Patients who developed PE before discharge had no significant differences in baseline characteristics compared with patients who presented with postdischarge PE (Table 1). GILs occurred 10 (5–15) days postoperatively (see Fig. 1); diagnosis was made after hospital discharge from the index operation in 275 (62.4%) patients. There was no difference in the postoperative day of GIL occurrence between laparoscopic RYGB and SG (10 [5–16] versus 11 [7–14.75]; P = .72). Patients who developed early GIL had significantly higher incidence of pre-existing vascular disease, higher BMI, advanced age, and, more commonly, severe chronic systemic illness (Table 2). Discussion Our study evaluated the prevalence and timing of 2 serious postoperative occurrences after bariatric surgery
using a large sample size. GIL occurs mostly after hospital discharge from the index bariatric procedure. This is in contrast to previous single-institution studies reporting that the majority of GIL occurs within the first postoperative week [10,18]. In a study of 50 GILs from 904 RYGB patients, 77% were diagnosed within a week from surgery [13]. Another study on 3828 RYGB patients with 150 GILs reported that median time to diagnosis of the leak ranged between 2 and 4 days, depending on the type of GIL [19]. Similar findings have been reported in a large recent retrospective study of 6030 RYGB patients with 64 GILs [14]. The median day for leak diagnosis was postoperative day 5, and 75% of leaks were diagnosed by day 12. Interestingly, greater variation is reported in the literature for the timing of GIL after SG. In a retrospective study of 15 GILs in 539 SG patients, the mean time from index procedure to diagnosis was 27 days, with a broad range (1– 102 days) [20]. A smaller study of 5 GILs after SG reported diagnosis at mean postoperative day 3 (range, 1–7) [21]. A recent systematic review of 316 GILs from 15,772 patients
Table 1 Comparison in baseline characteristics in patients with postoperative PE, based on timing Variable
PE in hospital (n ¼ 35)
PE after discharge (n ¼ 96)
P value
Odds ratio (95% confidence interval)
Vascular co-morbidities Cardiac co-morbidities Pulmonary co-morbidities Neurologic co-morbidities Smoking Diabetes Male gender Age BMI ASA 3 or 4 RYGB
18 2 1 1 5 12 8 51 47.6 27 23
67 3 7 4 8 32 32 50.5 49.7 73 57
.051 .611 .682 1.0 .331 .919 .249 .574 .544 .896 .51
2.18 0.54 2.71 1.5 0.55 0.96 1.69 1.0 1.01 .94 .76
(51.4%) (5.7%) (2.9%) (2.9%) (14.3%) (34.3%) (22.9%) (38–59) (43–58.8) (77.1%) (65.7%)
(69.8%) (3.2%) (7.4%) (4.2%) (8.3%) (33.3%) (33.3%) (42–58.75) (41.9–56.8) (76%) (59.4%)
(.99–4.82) (.09–3.36) (.32–22.81) (.16–13.85) (.17–1.8) (.43–2.17) (.69–4.13) (.98–1.04) (.98–1.05) (.38–2.35) (.34–1.71)
PE ¼ pulmonary embolism; BMI ¼ body mass index; ASA ¼ American Society of Anesthesiology class; RYGB ¼ laparoscopic gastric bypass.
K. Spaniolas et al. / Surgery for Obesity and Related Diseases ] (2015) 00–00
4
Table 2 Comparison in baseline characteristics in patients with postoperative GIL, based on timing Variable
GIL in hospital (n ¼ 160)
GIL after discharge (n ¼ 275)
P value
Odds ratio (95% confidence interval)
Vascular co-morbidities Cardiac co-morbidities Pulmonary co-morbidities Neurologic co-morbidities Smoking Diabetes Male gender Age BMI ASA 3 or 4 RYGB
110 9 6 3 25 59 40 47 47.1 130 83
156 6 9 6 48 82 59 45 45.5 189 160
.015 .057 .789 1.0 .129 .129 .409 .008 .002 .004 .201
.6 .37 .87 1.16 1.14 .73 .82 1.03 1.03 .51 1.29
(68.8%) (5.7%) (3.8%) (1.9%) (15.6%) (36.9%) (25.2%) (39–57.75) (41.8–55.1) (81.2%) (51.9%)
(56.9%) (2.2%) (3.3%) (2.2%) (17.5%) (29.8%) (21.7%) (38–52) (41.2–51.1) (68.7%) (58.2%)
(.4–.91) (.13–1.07) (.3–2.48) (.29–4.72) (.67–1.94) (.48–1.1) (.52–1.31) (1.01–1.04) (1.01–1.06) (.32–0.81) (.87–1.91)
GIL ¼ gastrointestinal leak; BMI ¼ body mass index; ASA ¼ American Society of Anesthesiology class; RYGB ¼ laparoscopic gastric bypass.
reported that the clinical presentation ranged from 1– 12 days for RYGB and 1–35 days for SG [22]. Although this difference was not statistically assessed, it is in contrast to the present study, which shows no significant difference in the timing of GIL diagnoses between the 2 procedures. Although the use of routine postoperative contrast studies to rule out GIL has been debated, multiple retrospective studies suggest low sensitivity and low positive predictive value for this approach [9–11,18,23]. There is no consensus at present, and the American Society for Metabolic and Bariatric Surgery guidelines recognize the common use of such practice [12]. Although our study was not designed to evaluate the role of routine contrast imaging, it revealed that most GIL occurs after hospital discharge and, as such, indicated the low diagnostic yield of routine contrast imaging to detect GILs. The present analysis found that most PEs occur after discharge. This is in accordance with the findings of previous studies that examined the effect of different anticoagulation regimens on the VTE rate. In a multicenter retrospective study of 668 patients undergoing open bariatric surgery, 6 patients developed symptomatic PE, and all of them presented after discharge [24]. This is similar to a retrospective single-institution study on 1125 open RYGB patients, with mean PE diagnosis at postoperative day 10 (range 4–30) [25]. The delayed presentation of PE and the decrease in the VTE rate with prolonged anticoagulation have led to the practice of extended prophylaxis, commonly for 10 days [26,27]. In a prospective study comparing 132 and 176 bariatric patients who underwent in-hospital-only or 10-day prophylaxis, respectively, the VTE rate was significantly higher in the short prophylaxis group (4.5% versus 0; P = .006) [27]. Practice patterns in terms of prolonged anticoagulation are unknown. A survey of bariatric surgeons suggested that 48.3% continue anticoagulation after discharge, but the response rate to the survey was not provided [28]. Furthermore, 2 recent prospective randomized trials of anticoagulation modalities for bariatric patients limited the duration of
prophylaxis to in-hospital or o10 days, respectively [29,30]. This is in contrast to a systematic review that recommended 1-month prophylaxis in patients undergoing major open or laparoscopic abdominal or pelvic surgery [15]. These studies collectively have suggested that the landscape of prophylaxis for bariatric surgery patients is unclear and that more information is needed. Our study reports the delayed nature of PE presentation and therefore implies that anticoagulation for 410 days may be beneficial, similar to what is suggested for other major abdominal surgery. Our study is limited by the retrospective design and selection bias. Specifically, the ACS-NSQIP database is not bariatric specific, and therefore, it is possible that postoperative occurrences may be missed or miscoded. Our definition of a GIL was based on the presence of organ space infection. Although an organ space infection may not require the presence of GIL, this association has been previously used to assess for GIL in bariatric and other gastrointestinal surgery studies [14,16,17]. This study is further limited by sample size. Specifically, the difference seen in the timing of PE after RYGB and SG was not statistically significant, but we are unable to draw definitive conclusions on this finding because our study was underpowered to adequately assess this observed difference. Furthermore, although patient follow-up in ACS-NSQIP is consistent, patients are only followed up for 30 days, and thus we were unable to assess for complications after this time point. There were minimal differences in baseline characteristics between patients who developed any of the complications examined (PE and GIL) before or after hospital discharge. This study suggests that older and chronically ill patients are more likely to develop GIL before hospital discharge, but our ability to predict which patients will develop delayed complications was limited on the basis of the data provided by this study. Conclusions Our study reveals that GIL and PE occur mostly after hospital discharge following bariatric surgery, with no
Timing of Major Complications after Bariatric Surgery / Surgery for Obesity and Related Diseases ] (2015) 00–00
significant differences between laparoscopic RYGB and SG in terms of timing. These findings do not support the routine use of contrast studies to evaluate for GIL after bariatric surgery. Furthermore, prolonged anticoagulation beyond 10 days may be beneficial in providing prophylaxis against PE for this patient population. Future directions would need to involve the cost-effectiveness of such changes in postoperative care after bariatric surgery. Disclosures The authors have no commercial associations that might be a conflict of interest in relation to this article.
[14]
[15]
[16] [17]
[18]
References [19] [1] Courcoulas AP, Yanovski SZ, Bonds D, et al. Long-term outcomes of bariatric surgery: a National Institutes of Health symposium. JAMA Surg 2014;149(12):1323–9. [2] Birkmeyer NJ, Dimick JB, Share D, et al. Michigan Bariatric Surgery Collaborative. Hospital complication rates with bariatric surgery in Michigan. JAMA 2010;304(4):435–42. [3] DeMaria EJ, Pate V, Warthen M, Winegar DA. Baseline data from American Society for Metabolic and Bariatric Surgery-designated Bariatric Surgery Centers of Excellence using the Bariatric Outcomes Longitudinal Database. Surg Obes Relat Dis 2010;6(4):347–55. [4] Flum DR, Belle SH, King WC, et al. Longitudinal Assessment of Bariatric Surgery (LABS) Consortium. Perioperative safety in the longitudinal assessment of bariatric surgery. N Engl J Med 2009;361 (5):445–54. [5] Jamal MH, Corcelles R, Shimizu H, et al. Thromboembolic events in bariatric surgery: a large multi-institutional referral center experience. Surg Endosc 2015;29(2):376–80. [6] Finks JF, English WJ, Carlin AM, et al. Michigan Bariatric Surgery Collaborative; Center for Healthcare Outcomes and Policy. Predicting risk for venous thromboembolism with bariatric surgery: results from the Michigan Bariatric Surgery Collaborative. Ann Surg 2012;255 (6):1100–4. [7] Smith MD, Patterson E, Wahed AS, et al. Thirty-day mortality after bariatric surgery: independently adjudicated causes of death in the longitudinal assessment of bariatric surgery. Obes Surg 2011;21 (11):1687–92. [8] Omalu BI, Ives DG, Buhari AM, et al. Death rates and causes of death after bariatric surgery for Pennsylvania residents, 1995 to 2004. Arch Surg 2007;142(10):923–8. [9] Hamilton EC, Sims TL, Hamilton TT, Mullican MA, Jones DB, Provost DA. Clinical predictors of leak after laparoscopic Roux-en-Y gastric bypass for morbid obesity. Surg Endosc 2003;17(5):679–84. [10] Carter JT, Tafreshian S, Campos GM, et al. Routine upper GI series after gastric bypass does not reliably identify anastomotic leaks or predict stricture formation. Surg Endosc 2007;21(12):2172–7. [11] Madan AK, Stoecklein HH, Ternovits CA, Tichansky DS, Phillips JC. Predictive value of upper gastrointestinal studies versus clinical signs for gastrointestinal leaks after laparoscopic gastric bypass. Surg Endosc 2007;21(2):194–6. [12] ASMBS Clinical Issues Committee. ASMBS guideline on the prevention and detection of gastrointestinal leak after gastric bypass including the role of imaging and surgical exploration. Surg Obes Relat Dis 2009;5(3):293–6. [13] Carucci LR, Turner MA, Conklin RC, DeMaria EJ, Kellum JM, Sugerman HJ. Roux-en-Y gastric bypass surgery for morbid obesity:
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
5
evaluation of postoperative extraluminal leaks with upper gastrointestinal series. Radiology 2006;238(1):119–27. Jacobsen HJ, Nergard BJ, Leifsson BG, et al. Management of suspected anastomotic leak after bariatric laparoscopic Roux-en-y gastric bypass. Br J Surg 2014;101(4):417–23. Rasmussen MS, Jørgensen LN, Wille-Jørgensen P. Prolonged thromboprophylaxis with low molecular weight heparin for abdominal or pelvic surgery. Cochrane Database Syst Rev 2009;(1):CD004318. Weller WE, Rosati C. Comparing outcomes of laparoscopic versus open bariatric surgery. Ann Surg 2008;248(1):10–5. Kang CY, Halabi WJ, Chaudhry OO, et al. Risk factors for anastomotic leakage after anterior resection for rectal cancer. JAMA Surg 2013;148(1):65–71. Leslie DB, Dorman RB, Anderson J, et al. Routine upper gastrointestinal imaging is superior to clinical signs for detecting gastrojejunal leak after laparoscopic Roux-en-Y gastric bypass. J Am Coll Surg 2012;214(2):208–13. Lee S, Carmody B, Wolfe L, et al. Effect of location and speed of diagnosis on anastomotic leak outcomes in 3828 gastric bypass cases. J Gastrointest Surg 2007;11(6):708–13. Moon RC, Shah N, Teixeira AF, Jawad MA. Management of staple line leaks following sleeve gastrectomy. Surg Obes Relat Dis 2015;11 (1):54–9. Juza RM, Haluck RS, Pauli EM, Rogers AM, Won EJ, LynSue JR. Gastric sleeve leak: a single institution’s experience with early combined laparoendoscopic management. Surg Obes Relat Dis 2015;11(1):60–4. Zellmer JD, Mathiason MA, Kallies KJ, Kothari SN. Is laparoscopic sleeve gastrectomy a lower risk bariatric procedure compared with laparoscopic Roux-en-Y gastric bypass? A meta-analysis. Am J Surg 2014;208(6):903–10. Doraiswamy A, Rasmussen JJ, Pierce J, Fuller W, Ali MR. The utility of routine postoperative upper GI series following laparoscopic gastric bypass. Surg Endosc 2007;21(12):2159–62. Hamad GG, Choban PS. Enoxaparin for thromboprophylaxis in morbidly obese patients undergoing bariatric surgery: findings of the prophylaxis against VTE outcomes in bariatric surgery patients receiving enoxaparin (PROBE) study. Obes Surg 2005;15 (10):1368–74. Abou-Nukta F, Alkhoury F, Arroyo K, et al. Clinical pulmonary embolus after gastric bypass surgery. Surg Obes Relat Dis 2006;2 (1):24–8. Borkgren-Okonek MJ, Hart RW, Pantano JE, et al. Enoxaparin thromboprophylaxis in gastric bypass patients: extended duration, dose stratification, and antifactor Xa activity. Surg Obes Relat Dis 2008;4(5):625–31. Raftopoulos I, Martindale C, Cronin A, Steinberg J. The effect of extended post-discharge chemical thromboprophylaxis on venous thromboembolism rates after bariatric surgery: a prospective comparison trial. Surg Endosc 2008;22(11):2384–91. Pryor HI 2nd, Singleton A, Lin E, Lin P, Vaziri K. Practice patterns in high-risk bariatric venous thromboembolism prophylaxis. Surg Endosc 2013;27(3):843–8. Imberti D, Baldini E, Pierfranceschi MG, et al. Prophylaxis of venous thromboembolism with low molecular weight heparin in bariatric surgery: a prospective, randomized pilot study evaluating two doses of parnaparin (BAFLUX Study). Obes Surg 2014;24(2):284–91. Steele KE, Canner J, Prokopowicz G, et al. The EFFORT trial: preoperative enoxaparin versus postoperative fondaparinux for thromboprophylaxis in bariatric surgical patients: a randomized doubleblind pilot trial. Surg Obes Relat Dis 2014;pii:S1550-S7289(14) 00391–8.
