Preoperative Placement of Inferior Vena Cava Filters and Outcomes After Gastric Bypass Surgery

Transcription

1 ORIGINAL ARTICLES Preoperative Placement of Inferior Vena Cava Filters and Outcomes After Gastric Bypass Surgery Nancy J. O. Birkmeyer, PhD,* David Share, MD, MPH, Onur Baser, PhD,* Arthur M. Carlin, MD, Jonathan F. Finks, MD,* Carl M. Pesta, DO, Jeffrey A. Genaw, MD, and John D. Birkmeyer, MD*; for the Michigan Bariatric Surgery Collaborative Objective: To assess relationships between inferior vena cava (IVC) filter placement and complications within 30 days of gastric bypass surgery. Summary of Background Data: IVC filters are increasingly being used as prophylaxis against postoperative pulmonary embolism in patients undergoing bariatric surgery, despite a lack of evidence of effectiveness. Methods: On the basis of data from a prospective clinical registry involving 20 Michigan hospitals, we identified 6376 patients undergoing gastric bypass surgery between 2006 and We then assessed relationships between IVC filter placement and complications within 30 days of surgery. We used propensity scores and fixed effects logistic regression to control for potential selection bias. Results: A total of 542 gastric bypass patients (8.5%) underwent preoperative IVC filter placement, most of whom (65%) had no history of venous thromboembolism. The use of IVC filters for gastric bypass patients varied widely across hospitals (range, 0% 34%). IVC filter patients did not have reduced rates of postoperative venous thromboembolism (adjusted odds ratio OR, 1.28; 95% confidence interval CI, ), serious complications (adjusted OR, 1.40; 95% CI, ), or death/permanent disability (adjusted OR, 2.49; 95% CI, ). More than half (57%) of the IVC filter patients in the latter group had a fatal pulmonary embolism or complications directly related to the IVC filter itself, including filter migration or thrombosis of the vena cava. In subgroup analyses, we were unable to identify any patient group for whom IVC filters were associated with improved outcomes. Conclusions: Prophylactic IVC filters for gastric bypass surgery do not reduce the risk of pulmonary embolism and may lead to additional complications. (Ann Surg 2010;252: ) Approximately 200,000 patients now undergo bariatric (weight loss) surgery in the United States each year. Given their underlying morbid obesity and high prevalence of mobility limitations and other comorbidities, bariatric surgery patients are at relatively high risk for postoperative venous thromboembolism (VTE). 1 3 Pulmonary embolism remains the second leading cause (after anastomotic leak) of perioperative death in bariatric surgery, 4 accounting for From the *Departments of Surgery, Michigan Surgical Collaboration for Outcomes Research and Evaluation, University of Michigan, Ann Arbor, MI; Department of Family Medicine, University of Michigan, Ann Arbor, MI; Department of Surgery, Henry Ford Hospital, Detroit, MI; and Department of Surgery, Henry Ford Macomb Hospital, Warren, MI. Supported by a longitudinal research contract from Blue Cross and Blue Shield of Michigan. Reprints: Nancy J. O. Birkmeyer, PhD, Department of Surgery, Michigan Surgical Collaborative for Outcomes Research and Evaluation (M-SCORE), University of Michigan, 211 North 4th Av, Suites 2 A& B, Ann Arbor, MI nbirkmey@med.umich.edu. Copyright 2010 by Lippincott Williams & Wilkins ISSN: /10/ DOI: /SLA.0b013e3181e61e4f almost 40% of perioperative deaths occurring in bariatric surgery patients. For this reason, professional societies in bariatric surgery recommend that some form of VTE prophylaxis be used in all patients undergoing bariatric surgery. 5,6 Whether VTE prophylaxis in bariatric surgery should include preoperative inferior vena cava (IVC) filters is a matter of debate. IVC filters are medical devices that mechanically trap thromboemboli originating in the lower extremities or pelvis to prevent pulmonary embolism. Prior studies of the safety and efficacy of IVC filters for VTE prophylaxis in bariatric surgery are mainly limited to small, single-center case series Furthermore, IVC filters are costly 15 and associated with their own risks including filter migration, IVC thrombosis, and IVC injury. 16 Guidelines from the professional organizations in bariatric surgery do not provide specific guidance about the use of IVC filters for bariatric surgery. 5,6 However, many believe that their use has been increasing with the development of retrievable IVC filters. 17 Because IVC filters are associated with substantial risks and costs, evidence of their effectiveness is important. In this study, we used a large, prospective, externally audited clinical registry to evaluate variability in the use and outcomes of IVC filters in gastric bypass surgery patients. METHODS Study Population This study is based on analysis of data from the Michigan Bariatric Surgery Collaborative (MBSC). As described in greater detail elsewhere, the MBSC is a regional consortium of hospitals and surgeons performing bariatric surgery in Michigan. 18,19 The MBSC now enrolls more than 95% of the patients undergoing bariatric surgery in the state of Michigan (approximately 5000 patients per year), in its clinical registry. Participating hospitals submit data for all of their bariatric surgery patients including those undergoing gastric bypass, laparoscopic gastric banding, biliopancreatic diversion with or without duodenal switch, and sleeve gastrectomy procedures. In the MBSC, medical records are abstracted for each patient at the end of the perioperative period (30 days after surgery). Information collected includes demographic variables, preoperative clinical characteristics, and conditions as well as perioperative process of care and outcomes. The medical record reviews are performed by centrally trained, nurse data abstractors using a standardized and validated instrument. Each participating hospital is site visited annually to verify the accuracy and completeness of their MBSC clinical registry data. For this study, we identified all patients undergoing gastric bypass surgery between June 2006 and October 2008 which includes 6376 patients from 20 hospitals. We included patients undergoing either open or laparoscopic gastric bypass procedures and excluded patients undergoing revisional surgery. We excluded patients undergoing laparoscopic gastric banding (37% of patients in the registry) because of low rates of perioperative complications among patients Annals of Surgery Volume 252, Number 2, August

2 Birkmeyer et al Annals of Surgery Volume 252, Number 2, August 2010 FIGURE 1. Propensity-adjusted rates of complications in gastric bypass patients with preoperative IVC filter placement compared with those without IVC filters. undergoing these procedures. We excluded patients undergoing biliopancreatic diversion with or without duodenal switch and sleeve gastrectomy procedures because of the relative infrequency of use of these procedures (less than 2% and 3%, respectively). Data Collected Patient characteristics collected include patient demographics, weight and medical history, weight-related, and other comorbidities (25 variables). Data are also collected on perioperative process of care including VTE prophylaxis, surgical approach (laparoscopic/open/conversion from laparoscopic to open), other surgical procedures performed at the time of the bariatric procedure, placement of intraoperative drains, and time in operating room (incision to closure). Data are collected on 12 different types of bariatric surgery-related complications. Perioperative complications are grouped according to severity as grade I (nonlife threatening), grade II (potentially life-threatening), or grade III (life-threatening complications associated with residual and permanent disability or death). Grade II complications include abdominal abscess (requiring percutaneous drainage or reoperation), bowel obstruction (requiring reoperation), leak (requiring percutaneous drainage or reoperation), bleeding (requiring transfusion 4 units, endoscopy, reoperation, or splenectomy), respiratory failure (requiring 2 7 days intubation), renal failure (requiring dialysis while patient is hospitalized during the perioperative period), wound infection/dehiscence (requiring reoperation), and VTE (deep vein thrombosis or pulmonary embolism). Grade III complications include myocardial infarction or cardiac arrest, renal failure requiring long-term dialysis, respiratory failure requiring 7 days intubation or tracheostomy, and death. The MBSC Endpoints Committee graded the severity of any perioperative complications that are not prespecified in these categories. Statistical Analysis For this study we focused on 3 primary outcomes: perioperative VTE (a physician diagnosis of deep vein thrombosis or pulmonary embolism), serious complications (combined grade II and III complications), and death/permanent disability (grade III complications). Our primary analysis was conducted at the patient level with preoperative IVC filter placement as the exposure variable. We used propensity scores to control for selection bias resulting from IVC filters being placed in higher-risk patients. This method is recommended over conventional methods of risk-adjustment when event rates are low, and there are multiple potential confounding variables. 20 To estimate propensity scores, 21,22 the log odds of the probability that a patient received an IVC filter was modeled as a function of potential confounding variables included in our dataset using multivariate logistic regression. First, factors associated with the placement of an IVC filter in univariate analysis (using a P 0.05 level of significance) were entered into a stepwise logistic regression model. The log odds of the probability that a patient received an IVC filter were modeled as a function of the factors that were retained in the model which included: history of VTE, body mass index (BMI) 50, age 50; cholelithiasis, and mobility problems. We then calculated each patient s probability, or propensity score, for the use of an IVC filter. Next we stratified patients into groups such that within each group treated and control patients have on an average the same propensity score and a similar distribution of confounders that make up its estimation by testing the balancing property. 23,24 This process resulted in 13 propensitystratified groups of patients (see Supplemental Digital Content 1, Appendix Figure 1, available at: Finally, adjusted odds ratios (OR) were calculated using fixed effects logistic regression to estimate average treatment effects across the 13 propensity-stratified blocks of patients. In addition to our primary analyses, we performed subgroup analyses to identify specific groups of patients that might benefit from IVC filter placement. We selected subgroups based on a survey asking MBSC surgeons to identify patients for whom they often consider for placement of a preoperative IVC filter. Risk factors identified by the surgeons included prior history of VTE, BMI 50, immobility, age 50 years, and open procedure. In sensitivity analysis, we also assessed outcomes with regard to hospital practices. For this analysis we compared outcomes at hospitals with high use of IVC filters ( 10% of gastric bypass patients) with hospitals with low use of IVC filters ( 10% of bariatric surgery patients). As with our patient level analysis, we used fixed-effects logistic regression to calculate propensity-adjusted OR comparing high IVC filter use hospitals to low IVC filter use hospitals with regard to our primary outcomes. RESULTS Overall, 542 patients (8.5%) had prophylactic IVC filters placed before gastric bypass surgery. Patients with and without preoperative IVC filter placement differed with regard to numerous clinical characteristics (Table 1). Patients in the IVC filter group tended to be older, male, heavier, to have problems with mobility (able to walk only with a cane, walker, or requiring a wheelchair or Lippincott Williams & Wilkins

3 Annals of Surgery Volume 252, Number 2, August 2010 IVC Filter Placement and Gastric Bypass Outcomes TABLE 1. Patient and Case Characteristics and Perioperative Outcomes Among Gastric Bypass Patients With and Without Preoperative IVC Filter Placement and in Hospitals Placing IVC Filters in 10% Compared With Hospitals That Place IVC Filters in 10% of Their Gastric Bypass Patients Characteristic No IVC Filter N 5834 Gastric Bypass Patients IVC Filter N 542 P <10% N 4043 Proportion of Hospital s Gastric Bypass Patients With an IVC Filter >10% N 2333 Demographics/medical history Age 50 yr; % Male gender; % Private insurance; % Current smoker; % BMI 50; % Mobility problems; % Prior VTE; % Comorbidity Lung disease; % Cardiovascular disease; % Hyperlipidemia; % GERD; % Peptic ulcer disease; % Cholelithiasis; % Urinary incontinence; % Renal failure; % Diabetes; % Liver disorder; % Sleep apnea; % Musculoskeletal disorder; % History of hernia repair; % Psychological; % Case characteristics Laparoscopic; % OR time ( 3 h); % Outcomes Postoperative VTE; % Serious complications; % Death/permanent disability; % IVC indicates inferior vena cava; BMI, body mass index; VTE, venous thromboembolism; OR, odds ratio; GERD, gastroesophageal reflux disease. P scooter), and to have a history of VTE. They were less likely to have private health insurance. IVC filter patients had a greater likelihood of obesity related comorbidity including significantly higher rates of lung and cardiovascular diseases, renal failure, diabetes, liver disorders, and sleep apnea. With regard to case characteristics, they were less likely to have their procedures performed laparoscopically and their cases were more likely to have duration of 3 hours. Before risk-adjustment, IVC filter patients had significantly higher rates of perioperative complications including postoperative VTE (2.03% for the IVC filter patients vs. 0.53% for the patients without IVC filters, P ), combined serious complications (7.56% vs. 3.62%, P ), and death/permanent disability (1.85% vs. 0.51%, P ). Unadjusted outcomes were worse for IVC filter patients within nearly all important subgroups (Table 2). Patients with a prior history of VTE were the only subgroup with similar rates of complications among patients with and without preoperative IVC filter placement. Complication rates were significantly higher among IVC filter patients for the remaining subgroups including patients with mobility problems, BMI 50, age 50 years, and patients undergoing open procedures. Following propensity adjustment (Fig. 1), patients receiving IVC filters remained at higher risk of postoperative VTE (OR 1.28; 95% confidence interval CI, ) and serious complications (OR 1.40; 95% CI, ) although these differences were no longer statistically significant. IVC filter patients risks of death/permanent disability remained higher (OR 2.49; 95% CI, , P 0.053) following propensity adjustment. Of the 10 IVC filter patients suffering death/permanent disability, 3 experienced pulmonary embolism and 2 had complications directly related to the filter itself including fatal IVC thrombosis and IVC filter migration to the heart. IVC filter use varied from 0% to 35% across the 20 hospitals in the MBSC that perform gastric bypass procedures (Fig. 2). Four hospitals did not use IVC filters in any of their gastric bypass patients while 6 hospitals used IVC filters in 10%. Case mix differences were relatively small comparing patients at the 6 hospi Lippincott Williams & Wilkins 315

4 Birkmeyer et al Annals of Surgery Volume 252, Number 2, August 2010 TABLE 2. Subgroup Analysis of Outcomes for Gastric Bypass Patients With and Without Preoperative IVC Filter Placement Patient Subgroup Outcome Event Rate (%) No IVC IVC OR (95% CI) Prior history of VTE (n 304) Postoperative VTE ( ) Serious complication ( ) Death/permanent disability ( ) Mobility problems (n 334) Postoperative VTE Not estimable Serious complication ( ) Death/permanent disability ( ) BMI 50 (n 2352) Postoperative VTE ( ) Serious complication ( ) Death/permanent disability ( ) Age 50 (n 2336) Postoperative VTE ( ) Serious complication ( ) Death/permanent disability ( ) Open procedure (n 663) Postoperative VTE ( ) Serious complication ( ) Death/permanent disability ( ) IVC indicates inferior vena cava; VTE, venous thromboembolism; OR, odds ratio; CI, confidence interval. tals with the greatest use of IVC filters ( 10% of gastric bypass patients) to the other 14 hospitals (Table 1). The high IVC use hospitals had more patients 50 years of age and with prior history of VTE and greater rates of some obesity-related comorbidities but lower rates of others. Rates of laparoscopic gastric bypass were lower and cases with 3-hour duration were higher at the high IVC filter use hospitals. While no longer statistically significant following propensity adjustment, patients at the high IVC filter use hospitals (Fig. 3) remained at greater risk of serious complications (OR 1.23; 95% CI, ) and death/permanent disability (OR 1.52; 95% CI, ). Patients at the high IVC filter use hospitals had significantly higher risks of postoperative VTE in the propensity-adjusted analysis (OR 1.56; 95% CI, ). DISCUSSION Prophylactic IVC filter use is common among patients undergoing bariatric surgery. If trends in the use of IVC filters in Michigan can be extrapolated, nearly 20,000 gastric bypass patients had preoperative IVC filter placement for VTE prophylaxis in the United States last year. In this large, prospective cohort study we found no clinical benefit associated with this practice. Although adjusting for differences in patient case mix accounted for much of the increased risk of major complications among IVC filter patients overall, we were unable to identify any subgroup of patients that benefited from their use. Furthermore, half of the complications resulting in death or permanent disability among IVC filter patients were directly related to the device itself. Despite the increasing use of IVC filters in the United States, the scientific literature demonstrating their effectiveness is lacking. Although bariatric surgery is one of the main factors involved in the recent increase in the use of IVC filters, studies of the safety and efficacy of IVC filters for VTE prophylaxis in bariatric surgery are few and limited to small, retrospective, single center, case series Only 3 of these studies included a comparison group. 7,8,11 The largest of these studies included 2085 gastric bypass patients, 234 (11%) of whom had a preoperative IVC filter placed. 11 Rates of pulmonary embolism were 0.81% in the IVC filter patients compared with 0.59% in the non-ivc filter placement patients, P None of the prior studies have shown a significant clinical benefit of prophylactic IVC filter placement for bariatric surgery patients. In interpreting our findings, it is important to consider the strengths and limitations of this study. Although ours is the largest study of this topic to date, major complications including postoperative VTE and death are relatively rare after bariatric surgery. Thus, our study had limited statistical power to demonstrate significant harms associated with the use of IVC filters while controlling for important confounding variables. Although a limitation of our study, this also underscores how safe IVC filters or any other intervention designed to prevent these complications would need to be to favorably affect the risk-benefit ratio in bariatric surgery. A second limitation of our study is its lack of detail with regard to type of IVC filter used. In our study, 32% of the IVC filters placed were known to be retrievable, 46% were permanent, and the type of device was unknown in 22%. Our clinical registry lacks data on the specific device used, and there may be IVC filter designs, within the broad classes of permanent and retrievable, that are FIGURE 2. Variability in the use of preoperative IVC filters in gastric bypass patients in Michigan hospitals Lippincott Williams & Wilkins

5 Annals of Surgery Volume 252, Number 2, August 2010 IVC Filter Placement and Gastric Bypass Outcomes FIGURE 3. Propensity-adjusted rates of complications in hospitals that place IVC filters in 10% of their gastric bypass patients compared with hospitals that place IVC filters in 10% of their gastric bypass patients. associated with greater or lesser rates of filter related complications such as filter migration or embolism. Our data also lack long-term follow-up data and to the extent that filter-related complications happen after the perioperative period, our study will have underestimated the potential harms of IVC filters. The third and most important limitation of our study is potential selection bias. To minimize the risk of confounding, we used propensity score methods to adjust for measurable demographic and clinical characteristics that are associated with the use of IVC filters. This method accounts only for known confounders, however, our clinical registry includes data regarding most known risk factors for postoperative VTE and other outcomes. One notable exception to this being data regarding hypercoaguable states, which our clinical registry does not include. While most patients with this risk-factor would have had a history of VTE, this is not a certainty. However, our hospital level analysis demonstrating worse outcomes among hospitals with higher rates of preoperative IVC filter placement, despite fairly similar patient populations, also supports the contention that selection bias is not a likely explanation for our findings. Finally, selection bias is an unlikely explanation for our results given that many of the most serious complications in IVC filter patients were either the result of pulmonary embolism or were directly related to the IVC filter itself. Others have reported complications among patients receiving IVC filters. For example, in the last decade more than 2500 adverse events involving IVC filters have been reported to the Food and Drug Administration s Manufacturer and User Facility Device Experience Database including 838 injuries (31% of reported events) and 93 fatalities (3.5% of reported events). 25 There have also been more than 30 published case reports of migration of IVC filters into the heart or pulmonary arteries. More common long-term complications of IVC filters include deep vein thrombosis, insertion site thrombosis, and IVC thrombosis Aside from their potential risks, IVC filters also have financial downsides. Based on Blue Cross and Blue Shield of Michigan claims data for patients undergoing gastric bypass surgery in the state in 2006, payments (facility and professional) are more than $13,500 higher for IVC filter patients. We estimate that excess costs attributable to preoperative IVC filter placement in bariatric surgery patients may total $270 million dollars in the United States annually. Despite the lack of evidence demonstrating their effectiveness, the use of IVC filters in the United States has increased dramatically in the last 2 decades Given the statistical power constraints in this study, we cannot rule out the possibility that IVC filters may benefit a small number of bariatric surgery patients at particularly high risk of pulmonary embolism. However, our findings raise serious questions about their value as prophylaxis for pulmonary embolism in the large majority of patients undergoing gastric bypass surgery. We believe that the use of IVC filters for VTE prophylaxis in bariatric surgery should be limited pending evidence of benefit for high-risk patients in randomized clinical trials. REFERENCES 1. Anderson F, Spencer F. Risk factors for venous thromboembolism. Circulation. 2003;107: Goldhaber S, Grodstein F, Stampfer M. A prospective study of risk factors for pulmonary embolism in women. JAMA. 1997;277: Samama M. An epidemiologic study of risk factors for deep vein thrombosis in medical outpatients: the Sirius study. Arch Intern Med. 2000;160: Omalu B, Ives D, Buhari A, et al. Death rates and causes of death after bariatric surgery for Pennsylvania residents, 1995 to Arch Surg. 2007;142: Clinical issues committee of the American Society for Metabolic and Bariatric Surgery. Prophylactic measures to reduce the risk of venous thromboembolism in bariatric surgery patients position statement. Surg Obes Relat Dis. 2007;3: Society of American Gastrointestinal and Endoscopic Surgeons (SAGES) Guidelines Committee. Guidelines for deep venous thrombosis prophylaxis during laparoscopic surgery. Surg Endosc. 2007;21: Gargiulo N, Veith F, Lipsitz E, et al. Experience with inferior vena cava filter placement in patients undergoing open gastric bypass procedures. J Vasc Surg. 2006;44: Halmi D, Kolesnikov E. Preoperative placement of retrievable inferior vena cava filters in bariatric surgery. Surg Obes Relat Dis. 2007;3: Kardys C, Stoner M, Manwaring M, et al. Safety and efficacy of intravascular ultrasound-guided inferior vena cava filter in super obese bariatric patients. Surg Obes Relat Dis. 2008;4: Keeling W, Haines K, Stone P, et al. Current indications for preoperative inferior vena cava filter insertion in patients undergoing surgery for morbid obesity. Obes Surg. 2005;15: Obeid F, Bowling W, Fike J, et al. Efficacy of prophylactic inferior vena cava filter placement in bariatric surgery. Surg Obes Relat Dis. 2007;3: Piano G, Ketteler E, Prachand V, et al. Safety, feasibility, and outcome of retrievable vena cava filters in high-risk surgical patients. J Vasc Surg. 2007;45: Lippincott Williams & Wilkins 317

6 Birkmeyer et al Annals of Surgery Volume 252, Number 2, August Schuster R, Hagedorn J, Curet M, et al. Retrievable inferior vena cava filters may be safely applied in gastric bypass surgery. Surg Endosc. 2007;21: Trigilio-Black C, Ringley C, McBride C, et al. Inferior vena cava filter placement for pulmonary embolism risk reduction in super morbidly obese undergoing bariatric surgery. Surg Obes Relat Dis. 2007;3: Chau Q, Cantor SB, Caramel E, et al. Cost-effectiveness of the bird s nest filter for preventing pulmonary embolism among patients with malignant brain tumors and deep venous thrombosis of the lower extremities. Support Care Cancer. 2003;11: Linsenmaier U, Rieger J, Schenk F, et al. Indications, management, and complications of temporary inferior vena cava filters. Cardiovasc Intervent Radiol. 1998;21: Brender E. Use of emboli-blocking filters increases, but rigorous data are lacking. JAMA. 2006;295: Birkmeyer N, Birkmeyer J. Strategies for improving surgical quality: should payers reward excellence or effort? N Engl J Med. 2006;354: Birkmeyer N, Share D, Campbell D, et al. Partnering with payers to improve surgical quality: the Michigan Plan. Surgery. 2005;138: Cepeda M, Boston R, Farrar J, et al. Comparison of logistic regression versus propensity score when the number of events is low and there are multiple confounders. Am J Epidemiol. 2003;158: Rosenbaum P, Rubin D. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70: Rubin D. Estimating causal effects of treatments in randomized and nonrandomized studies. J Educ Psychol. 1974;66: Becker S, Ichino A. Estimation of average treatment effects based on propensity scores. Stata J. 2002;2: Rosenbaum P, Rubin D. Reducing bias in observational studies using subclassification on the propensity score. J Am Stat Assoc. 1984;79: US Food and Drug Administration; Center for Devices and Radiological Health; Department of Health and Human Services. Manufacturer and User Facility Device Experience Database (MAUDE). Available at: accessdata.fda.gov/scripts/cdrh/cfdocs/cfmaude/search.cfm. Accessed December 1, Joels C, Sing R, Heniford B. Complications of inferior vena cava filters. Am Surg. 2003;69: Patel S, Patel R. Inferior vena cava filters for recurrent thrombosis: current evidence. Tex Heart Inst J. 2007;34: Streiff M. Vena caval filters: a comprehensive review. Blood. 2000;95: Athanasoulis C, Kaufman J, Halpern E, et al. Inferior vena cava filters: review of a 26-year single-center clinical experience. Radiology. 2000;216: Stein P, Kayali F, Olson R. Twenty-one year trends in the use of inferior vena cava filters. Arch Intern Med. 2008;164: Yunus T, Tariq N, Callahan R, et al. Changes in inferior vena cava filter placement over the past decade at a large community-based academic health center. J Vasc Surg. 2008;47: Lippincott Williams & Wilkins