Surg Endosc DOI 10.1007/s00464-015-4286-7
and Other Interventional Techniques
Intraoperative leak testing has no correlation with leak after laparoscopic sleeve gastrectomy Monica Sethi1 • Jonathan Zagzag1 • Karan Patel1 • Melissa Magrath1 Eduardo Somoza1 • Manish S. Parikh1 • John K. Saunders1 • Aku Ude-Welcome1 • Bradley F. Schwack1 • Marina S. Kurian1 • George A. Fielding1 • Christine J. Ren-Fielding1
•
Received: 15 February 2015 / Accepted: 25 May 2015 Ó Springer Science+Business Media New York 2015
Abstract Background Staple line leak is a serious complication of sleeve gastrectomy. Intraoperative methylene blue and air leak tests are routinely used to evaluate for leak; however, the utility of these tests is controversial. We hypothesize that the practice of routine intraoperative leak testing is unnecessary during sleeve gastrectomy. Methods A retrospective cohort study was designed using a prospectively collected database of seven bariatric surgeons from two institutions. All patients who underwent sleeve gastrectomy from March 2012 to November 2014 were included. The performance of intraoperative leak testing and the type of test (air or methylene blue) were based on surgeon preference. Data obtained included BMI, demographics, comorbidity, presence of intraoperative leak test, result of test, and type of test. The primary outcome was leak rate between the leak test (LT) and no leak test (NLT) groups. SAS version 9.4 was used for univariate and multivariate analyses. Results A total of 1550 sleeve gastrectomies were included; most were laparoscopic (99.8 %), except for one converted and two open cases. Routine intraoperative leak tests were performed in 1329 (85.7 %) cases, while 221 (14.3 %) did not have LTs. Of the 1329 cases with LTs, there were no positive intraoperative results. Fifteen (1 %) patients developed leaks, with no difference in leak rate Presented at the SAGES 2015 Annual Meeting, April 15–18, 2015, Nashville, Tennessee. & Monica Sethi [email protected] 1
Department of Surgery, New York University School of Medicine, 530 First Ave., Suite 10S, New York, NY 10016, USA
between the LT and NLT groups (1 vs. 1 %, p = 0.999). After adjusting for baseline differences between the groups with a propensity analysis, the observed lack of association between leak and intraoperative leak test remained. In this cohort, leaks presented at a mean of 17.3 days postoperatively (range 1–67 days). Two patients with staple line leaks underwent repeat intraoperative leak testing at leak presentation, and the tests remained negative. Conclusion Intraoperative leak testing has no correlation with leak due to laparoscopic sleeve gastrectomy and is not predictive of the later development of staple line leak. Keywords Leak Intraoperative leak test Sleeve gastrectomy Staple line Fistula Methylene blue
Bariatric surgery is widely accepted as the most effective treatment for morbid obesity. In recent years, the landscape of bariatric surgery has changed due to an increased use of the laparoscopic sleeve gastrectomy (LSG) relative to other procedures. In 2012, the LSG represented 36.3 % of all bariatric procedures performed at academic medical centers [1]. The sleeve gastrectomy involves a stapled vertical transection of the stomach and creation of a tubular alimentary channel along the stomach’s lesser curvature. Although primarily restrictive, the LSG may alter gut hormonal milieu through the decreased production of ghrelin and increased production of PYY and GLP-1 [2, 3], typically resulting in weight loss comparable to the gastric bypass and superior to the gastric band [4–6]. When compared to gastric bypass procedures, the LSG has a relatively low risk and complication profile [7, 8]. Nevertheless, a known complication of LSG is staple line leak, which usually occurs at the proximal portion of the staple line and has a reported incidence of 1.1–5.3 % [5].
123
Surg Endosc
Leaks can result in significant morbidity including sepsis, hemodynamic instability, multi-organ failure, and even mortality [5]. Early identification and treatment of leaks is critical, but the majority of leaks present days to weeks after the index operation, making early identification challenging [9]. In an effort to identify staple line leak during the index operation, provocative intraoperative testing, including methylene blue and air leak tests, are often performed. However, there is minimal literature on the utility of these tests, and what is available is highly conflicted. In fact, the 2012 International Sleeve Gastrectomy Expert Panel failed to reach a consensus (48 % consensus) about whether routine intraoperative leak tests should be performed [10]. Additionally, these tests are not benign—they introduce increased instrumentation, with reports in the literature of nasogastric tubes causing esophageal perforation [11], as well as increased costs in the form of resource utilization. The present study seeks to evaluate the ability of intraoperative leak tests to detect staple line leaks due to sleeve gastrectomy. We hypothesize that routine intraoperative leak testing has no correlation with leak due to LSG and is not predictive of the later development of leak. A secondary area of investigation was to determine the operative time and costs associated with intraoperative leak testing.
operation regardless of the setting in which they occurred (i.e., during initial hospitalization, on re-admission, or as an outpatient). All leaks were radiographically confirmed. For each patient who developed enteric leak, we gathered information about the duration from operation to leak presentation, location of the leak, presence of contrast extravasation on either esophagram or CT scan, and the type of treatment. Occasionally, an intraoperative leak test was performed at the time of leak presentation, and the results of these tests were recorded as well. Statistical analysis Univariable comparisons between the LT and NLT group were performed with a two-sample t test, Wilcoxon ranksum test, Chi-square test, or Fisher’s exact test. A propensity score model was used to evaluate the leak rate between the LT and NLT groups, while accounting for other patient characteristics. Diagnostic properties of the intraoperative leak test were also evaluated with a sensitivity and specificity analysis. An adjusted quartile regression analysis was used to evaluate differences in operative time between the LT and NLT groups among surgeons who stopped routinely performing the test. A cost analysis was performed using estimates of operative time and product costs from our institutions. SAS version 9.4 was used for all statistical analyses.
Materials and methods Operative technique and postoperative care We conducted a retrospective cohort study from a prospectively collected database of two high-volume Bariatric Surgery Centers of Excellence. We included all patients who underwent sleeve gastrectomy between March 2012 and November 2014 by one of seven bariatric surgeons. All patients were seen in an office setting within 2 weeks of surgery, and all patients had at least 70-day follow-up. Exclusion criteria consisted of age younger than 18 years at the time of surgery. Data obtained included patient demographics, comorbidities, intraoperative characteristics including performance and results of intraoperative leak testing, as well as postoperative complications. The performance of intraoperative leak testing and the type of test (air or methylene blue) were based on surgeon preference. Data on the performance of a leak test, the type of test, and the amount of methylene blue that was instilled were collected from operative reports. The primary outcome was leak rate between the leak test (LT) and the no leak test (NLT) groups. Secondary outcomes were operative time, hospital length of stay (LOS), and overall 30-day complication rate, which included any postoperative complication, re-admission, and/or re-
123
Entry into the peritoneal cavity was accomplished with the use of a 10-mm optical trocar in the left upper quadrant, a Veress needle, or a supraumbilical cut-down per surgeon preference. Any hiatal hernias were repaired at the time of surgery. A vertical sleeve gastrectomy was performed starting approximately 5 cm proximal to the pylorus, using a linear stapler. Bougie size was not standardized and was based on surgeon preference, ranging from 28 to 40 Fr. One surgeon used an endoscope rather than a bougie for intraoperative sizing of the LSG. In most cases, bioabsorbable staple line reinforcement was used during stapling. Drain placement and suture reinforcement of the staple line were not routinely performed. Some surgeons sutured the omentum to the staple line to prevent rotation of the stomach. Intraoperative leak testing was performed per surgeon preference. In cases with methylene blue testing, at the completion of the sleeve, the bougie was removed and an orogastric or nasogastric tube was placed under direct vision. Between 40 and 150 ml of methylene blue was instilled into the stomach with distal obstruction of the duodenum. Leaks were assessed for by the presence of blue
Surg Endosc
dye in the operative field, and the methylene blue solution was suctioned out of the stomach. One surgeon routinely performed intraoperative leak testing with an air leak test. The test consisted of instilling saline into the central upper abdomen. The stomach was insufflated with air through an endoscope with distal occlusion of the pylorus. Leaks in the staple line were evaluated for by the presence of air bubbles. The irrigation fluid was then suctioned out and the stomach decompressed. Postoperative care followed a standardized guideline. An esophagram was performed on postoperative day (POD) 1. A liquid diet was started if no leak was observed on the esophagram study and if the clinical course was uneventful. Hospital discharge generally occurred between POD 1 and 3 depending on the patient’s ability to maintain hydration, ambulate, and manage pain with oral analgesics. After discharge, patients were followed up at the bariatric surgery outpatient offices by a multidisciplinary team within 2 weeks of discharge, at 30 days, and at 3, 6, and 12 months, and yearly thereafter.
Results During the study period, 1562 patients underwent sleeve gastrectomy. Of these, twelve patients were excluded because they did not have adequate follow-up, leaving 1550 for analysis (315 males, 1235 females). The mean age was 40.2 years (range 18–72), and mean BMI was 44.2 kg/ m2 (range 29.1–96.1). Baseline demographic and clinical data are shown in Table 1. Of the 1550 sleeve gastrectomy cases, the vast majority were laparoscopic (99.81 %), except for one (0.06 %) converted and two (0.13 %) open cases. One hundred and thirty-two (9 %) patients had a history of previous foregut or bariatric surgery, and 898 (58 %) patients had a concurrent procedure performed at the time of the sleeve gastrectomy, which were predominantly hiatal hernia repairs. Additional operative characteristics of the study population, including staple line reinforcement, oversewing, drain placement, and bougie size, are shown in Table 2. The 1550 study patients included 1329 (85.7 %) in the leak test group and 221 (14.3 %) in the no leak test group. Routine intraoperative methylene blue leak testing was performed in 991 (63.9 %) patients, and the air leak test was performed in 338 (21.8 %) patients (Fig. 1). Of the 1329 in the leak test group, there were zero (0 %) positive leak test results. Table 3 shows the outcomes of the study population, as well as a comparison of the LT and NLT groups. The majority of patients (97 %) had an uneventful postoperative course after sleeve gastrectomy. Three percent
(n = 47) had 30-day complications, and there was one mortality, with no significant differences between the LT and NLT groups. There was no difference in the primary outcome of leak rate between the groups (1 vs. 1 %, p = 0.999). There were, however, significant differences on multiple levels between the LT and NLT groups, including several baseline and operative characteristics such as hospital location and bougie size. We controlled for these differences by propensity scoring, and there remained no difference in leak rate between the groups (OR = 0.816, 95 % CI = [0.103, 6.439], p = 0.847). A subanalysis comparing air and methylene blue leak tests also showed no difference in leak rates (1 vs. 1 %, p = 0.535). Of note, in an unadjusted comparison between the groups, the LT group had a longer operative time and longer hospital LOS than the NLT group. After adjusting for other factors in a propensity model, the association between intraoperative leak test and LOS was no longer significant (OR = 1.23, 95 % CI = [0.792, 1.908], p = 0.358). Operative time was controlled for in a separate analysis described later in this section. Additionally, operative volume was not related to leak rate (p = 0.656). Fifteen patients (1 %) developed staple line leaks during the study period. The mean number of days from surgery to leak presentation was 17.3 days (median = 9 days, range 1–67 days) with only two leaks presenting during the index admission. The majority of leaks (n = 13) were located proximally near the gastroesophageal junction; of the remaining two leaks, one was located at the distal staple line and the other was unable to be accurately localized. Treatment modalities ranged from operative drainage and stenting (n = 6), stenting alone (n = 3), nonoperative management with nihil per os and antibiotics (n = 4), and stent followed by revision to Roux-en-Y gastric bypass (n = 2). Thirteen of the patients who developed leak had intraoperative leak testing during the index operation, and as stated, there were no (0 %) positive leak tests. Eleven of these patients had methylene blue tests, and two had air leak tests. Of those who underwent methylene blue testing, three patients had 100 ml instilled, three had 60 ml instilled, and for the remaining five patients, the amount of methylene blue was unknown. Two patients underwent repeat intraoperative leak testing at the time of leak presentation, prior to any definitive treatment. Of those two, one patient with a leak seen at the upper staple line on POD 1 esophagram underwent a repeat methylene blue leak test during diagnostic laparoscopy, which was negative. Despite suture reinforcement, the leak persisted and the patient eventually required conversion to gastric bypass after unsuccessful stent treatment. The other patient presented with symptoms of epigastric pain on POD 46. He was diagnosed with a proximal leak on esophagram
123
Surg Endosc Table 1 Baseline demographic and clinical characteristics of the study population stratified by leak test versus no leak test
Overall N = 1550
Leak test N = 1329
No leak test N = 221
p value
Female, % (n)
80 % (n = 1235)
81 % (n = 1083)
69 % (n = 152)
\0.0001a
Male, % (n)
20 % (n = 315)
19 % (n = 246)
31 % (n = 69)
Gender
Age, years Mean (SD) [Range] 18–29, % (n)
40.2 (11.6) [18, 72] 20 % (n = 310)
39.9 (11.5) [18, 70] 21 % (n = 277)
42.1 (12.3) [18, 72] 15 % (n = 33)
0.010b 0.144a
30–39, % (n)
31 % (n = 480)
31 % (n = 413)
30 % (n = 67)
40–49, % (n)
26 % (n = 398)
26 % (n = 341)
26 % (n = 57)
50–59, % (n)
18 % (n = 280)
17 % (n = 231)
22 % (n = 49)
5 % (n = 82)
5 % (n = 67)
7 % (n = 15)
White, % (n)
77 % (n = 1190)
78 % (n = 1031)
72 % (n = 159)
Black, % (n)
14 % (n = 219)
14 % (n = 190)
13 % (n = 29)
Other, % (n)
1 % (n = 19)
1 % (n = 15)
2 % (n = 4)
Unknown, % (n) Ethnicity
8 % (n = 122)
7 % (n = 93)
13 % (n = 29)
Hispanic, % (n)
57 % (n = 881)
61 % (n = 816)
29 % (n = 65)
Non-Hispanic, % (n)
35 % (n = 542)
32 % (n = 425)
53 % (n = 117)
8 % (n = 127)
7 % (n = 88)
18 % (n = 39)
BVU, % (n)
61 % (n = 946)
68 % (n = 904)
19 % (n = 42)
NYULMC, % (n)
39 % (n = 604)
32 % (n = 425)
81 % (n = 179)
Mean (SD) [Range]
44.2 (7.8) [29.1, 96.1]
44.2 (7.7) [29.1, 88.5]
44.1 (8.4) [29.7, 96.1]
0.814b
BMI \ 50, % (n)
81 % (n = 1263)
82 % (n = 1085)
81 % (n = 178)
0.697a
BMI C 50, % (n)
19 % (n = 287)
18 % (n = 244)
19 % (n = 43)
60?, % (n) Race
Unknown, % (n)
0.013a
\0.0001a
Hospital \0.0001a
BMI, kg/m2
SD standard deviation, BVU Bellevue Hospital Center, NYULMC New York University Langone Medical Center, BMI body mass index a
Chi-square test,
b
two-sample t test
and underwent stent placement by interventional radiology. To confirm the leak’s location during stenting, Gastrografin was injected through an orogastric route into the stomach under fluoroscopy, which was negative despite being able to visualize the leak site on endoscopy. The leak resolved after 5 weeks of stenting. To determine the diagnostic predictive value of intraoperative leak testing, we determined the sensitivity, specificity, as well as positive predictive values (PPV) and negative predictive values (NPV) of the test. The sensitivity and specificity for detecting a leak were 0 % (95 % CI = [0, 24 %]) and 100 %, (95 % CI = [99.7, 100 %]), respectively, resulting in an undefined PPV and an NPV of 99 % (95 % CI = [98.3, 99.5 %]). Based on a preliminary analysis of our data, four surgeons stopped routinely performing intraoperative leak tests during the study period; of note, some would
123
selectively perform intraoperative leak tests during revisional surgeries or technically difficult cases. The operative time for these surgeons was analyzed using a quartile regression model with adjustment for concurrent procedures performed at the index operation, as well as a propensity score to account for the differences between LT and NLT groups. Three patients with open surgery or conversion to open were excluded from analysis. Results showed that operative time reduced by 7.6 min when the leak test was not performed (95 % CI = [-13.2, -2.0], p = 0.0083). At our hospitals, the cost of an additional 7.6 min of operative time and the supplies required for a methylene blue leak test are estimated to total approximately $855.37 per test. Based on these estimates, the total cost for intraoperative methylene blue leak tests during the course of the study was $847,671.67.
Surg Endosc Table 2 Operative characteristics of the study population stratified by leak test versus no leak test Overall N = 1550 Prior foregut/bariatric surgery, % (n)
9 % (n = 132)
Leak test N = 1329
No leak test N = 221
6 % (n = 83)
22 % (n = 49)
[99 % (n = 1327)
100 % (n = 221)
p value \0.0001a
Surgical approach, % (n) [99 % (n = 1453)
Laparoscopic Open Converted to open, % (n)
\1 % (n = 2)
\1 % (n = 2)
0 % (n = 0)
\1 % (n = 1)
0 % (n = 0)
\1 % (n = 1)
0.999b 0.143b
94 % (n = 1453)
93 % (n = 1424)
95 % (n = 211)
0.251a
5 % (n = 73)
5 % (n = 64)
4 % (n = 9)
0.629a
18 % (n = 278)
20 % (n = 262)
7 % (n = 16)
\0.0001a
28–34
19 % (n = 285)
15 % (n = 198)
40 % (n = 87)
\0.0001a
35–39 40?
35 % (n = 540) 46 % (n = 707)
39 % (n = 510) 46 % (n = 605)
14 % (n = 30) 47 % (n = 102)
Staple line reinforcement, % (n) Oversew, % (n) Drain placed, % (n) Bougie size, % (n)
a
Chi-square test,
b
Fisher’s exact test
Fig. 1 Bar plot of the percent of patients by type of intraoperative leak test
Discussion While early recognition and treatment are essential in managing staple line leaks after sleeve gastrectomy, leaks may present weeks or even months after surgery [5, 12]. Patients typically exhibit subtle clinical signs before rapidly deteriorating, rendering timely diagnosis difficult. To detect leaks as early as possible, some surgeons use intraoperative leak testing to assess the integrity of the staple line during LSG. These intraoperative tests, which include methylene blue and air leak tests, are designed to detect leak when tissues are viable and most amenable to repair by re-stapling or suturing [5]. It should be emphasized, however, that a negative intraoperative leak test result does not preclude the possibility of a subsequent staple line leak. Indeed, a negative test result may give the
surgeon a false sense of security and bias the differential diagnosis away from leak. Additionally, these tests are not completely benign and can add risks and unnecessary costs to the surgery. Nevertheless, despite the dearth of scientific studies showing the benefit of intraoperative leak testing, many surgeons still favor these tests and perform them routinely. In this study, we have shown that intraoperative leak tests have a sensitivity of 0 % (95 % CI = [0, 24.7 %]). Although the test specificity was high (100 %, 95 % CI = [99.7, 100 %]), the low sensitivity combined with the occurrence of zero positive results out of 1329 tests suggests that the utility of intraoperative leak testing needs to be questioned. Furthermore, the incidence of detected postoperative or delayed leaks was the same with or without routine intraoperative leak testing. Therefore, the authors believe that routine intraoperative leak testing may be unnecessary during LSG. The most important reason why intraoperative leak testing does not reliably detect leaks after LSG is that most leaks develop after surgery. Intraoperative leak testing can only detect the rare leaks due to technical staple line failure, such as those occurring due to stapler misfire or other intraoperative complications [5, 13]. However, staple line failure tends to be infrequent in the hands of experienced surgeons at high-volume centers. The pathogenesis of postoperative or delayed leaks that occur days to weeks after LSG is not fully understood. Some authors suggest that these postoperative leaks are caused by ischemia, thermal damage, or misplacement of the staple line to involve the distal esophagus in patients with hiatal hernias [14]. These scenarios all lead to intraluminal pressure that exceeds tissue and suture line resistance, thus causing leak. Additionally, narrowing the sleeve
123
Surg Endosc Table 3 Intraoperative outcomes, length of stay, and 30-day outcomes for the study population stratified by leak test versus no leak test (n = 221) Overall N = 1550
Leak test N = 1329
No leak test N = 221
p value
Intraoperative outcomes \1 % (n = 1)
Any occurrence, % (n) OR time, median (IQR)
86 (71, 143)
0 % (n = 0) 89 (72, 107)
\1 % (n = 1) 76 (60, 93)
0.143b \0.0001c
Length of stay \0.0001a
1
8 % (n = 129)
7 % (n = 87)
19 % (n = 42)
2
72 % (n = 1099)
73 % (n = 961)
63 % (n = 138)
3 4?
16 % (n = 243) 4 % (n = 64)
16 % (n = 214) 4 % (n = 53)
13 % (n = 29) 5 % (n = 11)
1 % (n = 15)
1 % (n = 13)
1 % (n = 2)
0.999b
0 % (n = 0)
\1 % (n = 1)
0.143b
Postoperative outcomes Leak, % (n)
\1 % (n = 1)
30-day death Any 30-day occurrence, % (n)
3 % (n = 47)
3 % (n = 36)
5 % (n = 11)
0.069a
Bold represents primary outcome OR operating room, IQR interquartile range a
Chi-square test,
b
Fisher’s exact test,
c
Wilcoxon rank-sum test
at the incisura angularis can lead to a physiologic stricture that may contribute to leaks at the gastroesophageal junction by creating a distal obstruction [15]. Classically, leaks due to ischemia present on POD 5 or 6. This is in line with data from Sakran et al. [5], which cites a median of 7 days to leak presentation. In our study, however, the number of days to leak presentation was longer than previously reported, with a mean of 17.3 days. Additionally, with leaks in our study presenting up to 67 days postoperatively, the pathophysiology of staple line leaks after LSG deserves further study. Another potential reason for the poor sensitivity of this test may be due to testing misuse. Leak testing with methylene blue was originally designed to diagnose both intraoperative leak and leak during the following 7 days by observing methylene blue effluence through abdominal drains [16, 17]. However, drain placement has fallen out of favor, and surgeons who do place drains rarely leave them in place long enough to detect postoperative leak. In 2013, only 39 % of surgeons left a drain in the abdominal cavity after LSG, and the number continues to decline [18]. If a drain is left in place, however, postoperative leak testing with methylene blue may effectively diagnose leak. Some surgeons have also used this method during follow-up to monitor the progress of the leak [6, 19]. Still, this method’s usefulness is limited to leaks that communicate with the surgically placed drains [20]. Ours is not the first study to note the lack of association between intraoperative leak testing and postoperative leak due to LSG. De Aretxabala et al. [12] had negative leak test results in all of the nine leaks included in their study. Casella et al. [21] reported two positive intraoperative
123
methylene blue tests (1 %), which were repaired intraoperatively without further postoperative sequelae; however, the six (3 %) postoperative leaks, even those detected on the first postoperative day, all had negative intraoperative methylene blue tests. In a larger study by Sakran et al., 2834 patients underwent LSG, resulting in 38 leaks (1.5 %). An intraoperative leak test was performed during the original surgery for 33 (75 %) of the 38 leaks, using methylene blue in 25 patients and air testing in eight patients. The intraoperative leak test results were negative in all but one case, in which there was a stapler misfire. The leak persisted despite intraoperative recognition, immediate repair, and a subsequent confirmatory intraoperative methylene blue test [5]. In a recent meta-analysis by Parikh et al., 67 % of 8922 patients underwent intraoperative leak testing. The leak rate was compared between methylene blue leak test, air leak test, ‘‘other’’ leak test, and no leak test, and in an unadjusted GEE regression model, the performance of leak test did not impact the leak rate (p = 0.454) [22]. The meta-analysis did not comment on the number of positive intraoperative leak tests. Some studies have noted that an intraoperative leak test result can be negative even when there is a known leak. Mizrahi et al. were unable to locate three out of four CTproven leaks on diagnostic laparoscopy, despite intraoperative air and methylene blue leak tests. They speculate that this may be due to symptomatic micro-leaks or tissue edema at the leak site, causing the tests to have a negative result [7]. In our study, two of our 15 leaks experienced similar scenarios—negative intraoperative leak test results despite the presence of a known leak.
Surg Endosc
It is also possible that leak tests do not supply enough pressure to cause extravasation of air or methylene blue, whereas active swallowing during an esophagram may provide enough pressure to promote extravasation. Additionally, in our study, we show that the amount of dilute methylene blue solution instilled into the stomach during leak testing varies from 40 to 150 ml; thus, the pressure created during intraoperative leak testing is likely quite variable. Given that there were only eleven patients who developed leaks and had prior intraoperative methylene blue tests (five of whom had an unknown amount of methylene blue), the small sample size limits our ability to assess an association between low volume methylene blue instillation and potential missed leaks, but this may be an area for future study. In a meta-analysis of 4000 patients, only 15 of 29 studies documented their use of intraoperative leak tests. Of those, 62 % used methylene blue [9]. Although most surgeons perform methylene blue over air leak tests, studies from the gastric bypass literature note that air insufflation, especially when delivered through an endoscope, may be more sensitive than methylene blue tests in detecting anastomotic leak [23, 24]. Additionally, endoscopy offers the ability to assess hemostasis intraluminally. Methylene blue can also stain the field, making it difficult to repeat the test if a leak is detected and repaired [25]. On the other hand, the equipment and personnel required for endoscopy lead to additional costs that are not encountered with methylene blue leak testing. In our study, we showed that there is no difference in LSG leak rate between the two types of leak tests, although for air leak testing, we can only comment on the type performed with an endoscope rather than an orogastric tube. Regardless of the type of test, intraoperative leak testing prolongs operative time and is associated with increased costs and resource utilization. The cost of routine intraoperative leak tests for sleeve gastrectomy has not, to our knowledge, been addressed in previous studies. When considering the prevalence of bariatric surgery, and LSG in particular, it becomes apparent that routinely using resources that are not necessary or tests that are not evidence based can be costly and wasteful. In our study, we showed that 991 intraoperative methylene blue tests cost $847,671.67 during our 2.5-year study period. Some have argued that the cost of these studies is justified by the potential cost of preventing a leak [26]. However, in this study, intraoperative leak testing did not detect a single leak, and therefore, there was no cost benefit. It is important to note that in addition to a reduction in operative time, unadjusted analyses demonstrated a reduction in hospital LOS among the NLT group. Unlike operative time, however, the difference in LOS was not significant after adjusting for other factors, such as hospital
location, in the propensity model. Therefore, the difference in LOS is likely attributed to other differences between the groups (i.e., hospital location) and not related to the performance of a leak test. Leak testing is not without its risks. Nasogastric tube has been associated with lung perforation, esophageal perforation, massive bleeding, and other mediastinal complications [11, 27, 28]; intraoperative orogastric tube placement could lead to a similar sequela. It has also been theorized that the shear forces on staple lines engendered by the leak testing process could result in perioperative staple line weakness, although scientific data seem to disprove this theory [29]. During intraoperative leak testing, methylene blue can be systemically absorbed and cause blue urine and prolonged staining of the skin [30]. As per our experience, methylene blue can be spilled on patients’ hair and faces during surgery, causing them distress and requiring costly hair treatments to remove the blue color. Additionally, there are multiple reports in the oncologic and breast surgery literature of methylene blue causing anaphylaxis and skin necrosis [31]. At our hospitals, as our experience has grown and our leak rate decreased, the utility of intraoperative leak tests has been re-evaluated and some surgeons have stopped the practice altogether. Similarly, Bellanger et al. [15], in their review of 529 LSG without a leak, performed 350 cases with no intraoperative detection of leak and subsequently stopped the practice. However, leak testing may be justified in cases of revisional surgery, intraoperative complications, or in the case of a surgeon who is early in the learning curve. Thus, we suggest that surgeons determine whether or not to abandon routine intraoperative leak testing only after they monitor their leak rates and the sensitivity of their own intraoperative leak tests. Our study has several limitations. First, the study design was retrospective; thus, it relies on the accuracy and thoroughness of the database and limits our ability to generalize the results. Second, considering the low occurrence of leaks after LSG, large numbers are required to detect a significant difference in leak rate. In the current literature, most singlecenter studies do not include data on the specific results of intraoperative leak tests, as is the case with large multicenter databases. Therefore, a meta-analysis involving the sensitivity and specificity of intraoperative leak testing would be challenging. Lastly, it is possible that other factors contributed to the decreased operative time among surgeons who stopped routinely performing leak tests. Although we attempted to account for these factors with our adjusted analysis, a randomized study would be the most effective at determining the true effect size. In conclusion, intraoperative leak testing has no correlation with leak during laparoscopic sleeve gastrectomy
123
Surg Endosc
and is not associated with the later development of staple line leak. It also prolongs operative time by 7.6 min, which translates into significantly increased operative costs. Therefore, the better practice is to selectively perform intraoperative methylene blue or air leak tests in cases with intraoperative complications such as staple line misfire, revisional surgery, or in practices with high leak rates. Acknowledgments The authors wish to acknowledge Heekoung Youn, MA for technical assistance and G. Craig Wood, MS for leading the statistical analyses sited in this publication. Disclosures Dr. Ren-Fielding received research Grants (S#06-851, S#10686) from Allergan Medical, and received an honorarium as a speaker and as a member of their Advisory Board. Dr. Fielding received an honorarium as part of the Speaker’s Bureau/Faculty, research Grants (S#06-851, S#10686), and an educational grant from Allergan Medical. Dr. Kurian received a research Grant (S#06-851) and an honorarium as a speaker for Allergan Medical. Dr. Schwack received an honorarium as a speaker for Allergan Medical. Dr. Sethi, Dr. Zagzag, Dr. Somoza, Dr. Parikh, Dr. Saunders, Dr. Ude-Welcome, Mr. Patel and Ms. Magrath have no conflicts of interest or financial ties to disclose.
References 1. Nguyen NT, Nguyen B, Gebhart A, Hohmann S (2013) Changes in the makeup of bariatric surgery: a national increase in use of laparoscopic sleeve gastrectomy. J Am Coll Surg 216:252–257 2. Gagner M, Buchwald JN (2014) Comparison of laparoscopic sleeve gastrectomy leak rates in four staple-line reinforcement options: a systematic review. Surg Obes Relat Dis 10:713–723 3. Nannipieri M, Baldi S, Mari A, Colligiani D, Guarino D, Camastra S, Barsotti E, Berta R, Moriconi D, Bellini R, Anselmino M, Ferrannini E (2013) Roux-en-Y gastric bypass and sleeve gastrectomy: mechanisms of diabetes remission and role of gut hormones. J Clin Endocrinol Metab 98:4391–4399 4. Fridman A, Moon R, Cozacov Y, Ampudia C, Lo Menzo E, Szomstein S, Rosenthal RJ (2013) Procedure-related morbidity in bariatric surgery: a retrospective short- and mid-term follow-up of a single institution of the American College of Surgeons Bariatric Surgery Centers of Excellence. J Am Coll Surg 217:614–620 5. Sakran N, Goitein D, Raziel A, Keidar A, Beglaibter N, Grinbaum R, Matter I, Alfici R, Mahajna A, Waksman I, Shimonov M, Assalia A (2013) Gastric leaks after sleeve gastrectomy: a multicenter experience with 2,834 patients. Surg Endosc 27:240–245 6. Tan JT, Kariyawasam S, Wijeratne T, Chandraratna HS (2010) Diagnosis and management of gastric leaks after laparoscopic sleeve gastrectomy for morbid obesity. Obes Surg 20:403–409 7. Mizrahi I, Tabak A, Grinbaum R, Beglaibter N, Eid A, Simanovsky N, Hiller N (2014) The utility of routine postoperative upper gastrointestinal swallow studies following laparoscopic sleeve gastrectomy. Obes Surg 24:1415–1419 8. Aggarwal S, Kini SU, Herron DM (2007) Laparoscopic sleeve gastrectomy for morbid obesity: a review. Surg Obes Relat Dis 3:189–194 9. Aurora AR, Khaitan L, Saber AA (2012) Sleeve gastrectomy and the risk of leak: a systematic analysis of 4,888 patients. Surg Endosc 26:1509–1515
123
10. Rosenthal RJ, International Sleeve Gastrectomy Expert P, Diaz AA, Arvidsson D, Baker RS, Basso N, Bellanger D, Boza C, El Mourad H, France M, Gagner M, Galvao-Neto M, Higa KD, Himpens J, Hutchinson CM, Jacobs M, Jorgensen JO, Jossart G, Lakdawala M, Nguyen NT, Nocca D, Prager G, Pomp A, Ramos AC, Rosenthal RJ, Shah S, Vix M, Wittgrove A, Zundel N (2012) International Sleeve Gastrectomy Expert Panel Consensus Statement: best practice guidelines based on experience of [12,000 cases. Surg Obes Relat Dis 8:8–19 11. Wu PY, Kang TJ, Hui CK, Hung MH, Sun WZ, Chan WH (2006) Fatal massive hemorrhage caused by nasogastric tube misplacement in a patient with mediastinitis. J Formos Med Assoc 105:80–85 12. de Aretxabala X, Leon J, Wiedmaier G, Turu I, Ovalle C, Maluenda F, Gonzalez C, Humphrey J, Hurtado M, Benavides C (2011) Gastric leak after sleeve gastrectomy: analysis of its management. Obes Surg 21:1232–1237 13. Aggarwal S, Bhattacharjee H, Chander Misra M (2011) Practice of routine intraoperative leak test during laparoscopic sleeve gastrectomy should not be discarded. Surg Obes Relat Dis 7:e24– e25 14. Mittermair R, Sucher R, Perathoner A, Wykypiel H (2014) Routine upper gastrointestinal swallow studies after laparoscopic sleeve gastrectomy are unnecessary. Am J Surg 207:897–901 15. Bellanger DE, Greenway FL (2011) Laparoscopic sleeve gastrectomy, 529 cases without a leak: short-term results and technical considerations. Obes Surg 21:146–150 16. Christian D, Barba C, Burke-Martindale C (2004) A simple bedside evaluation to detect gastroesophageal leaks after gastric bypass surgery. Surg Endosc 18:S193 (abst) 17. Ovnat A, Peiser J, Solomon H, Charuzi I (1986) Early detection and treatment of a leaking gastrojejunostomy following gastric bypass. Isr J Med Sci 22:556–558 18. Gagner M, Deitel M, Erickson AL, Crosby RD (2013) Survey on laparoscopic sleeve gastrectomy (LSG) at the Fourth International Consensus Summit on Sleeve Gastrectomy. Obes Surg 23:2013–2017 19. Jen S, Simillis C, Efthimiou E (2013) A very challenging leak from a sleeve gastrectomy. Surg Obes Relat Dis 9:e56–e59 20. Gonzalez R, Nelson LG, Gallagher SF, Murr MM (2004) Anastomotic leaks after laparoscopic gastric bypass. Obes Surg 14:1299–1307 21. Casella G, Soricelli E, Rizzello M, Trentino P, Fiocca F, Fantini A, Salvatori FM, Basso N (2009) Nonsurgical treatment of staple line leaks after laparoscopic sleeve gastrectomy. Obes Surg 19:821–826 22. Parikh M, Issa R, McCrillis A, Saunders JK, Ude-Welcome A, Gagner M (2013) Surgical strategies that may decrease leak after laparoscopic sleeve gastrectomy: a systematic review and metaanalysis of 9991 cases. Ann Surg 257:231–237 23. Gagner M, Deitel M, Kalberer TL, Erickson AL, Crosby RD (2009) The Second International Consensus Summit for Sleeve Gastrectomy, March 19–21, 2009. Surg Obes Relat Dis 5:476–485 24. Alaedeen D, Madan AK, Ro CY, Khan KA, Martinez JM, Tichansky DS (2009) Intraoperative endoscopy and leaks after laparoscopic Roux-en-Y gastric bypass. Am Surg 75:485–488 (discussion 488) 25. Alasfar F, Chand B (2010) Intraoperative endoscopy for laparoscopic Roux-en-Y gastric bypass: leak test and beyond. Surg Laparosc Endosc Percutan Tech 20:424–427 26. Madan AK, Stoecklein HH, Ternovits CA, Tichansky DS, Phillips JC (2007) Predictive value of upper gastrointestinal studies versus clinical signs for gastrointestinal leaks after laparoscopic gastric bypass. Surg Endosc 21:194–196
Surg Endosc 27. Torrington KG, Bowman MA (1981) Fatal hydrothorax and empyema complicating a malpositioned nasogastric tube. Chest 79:240–242 28. Winterholler M, Erbguth FJ (2002) Accidental pneumothorax from a nasogastric tube in a patient with severe hemineglect: a case report. Arch Phys Med Rehabil 83:1173–1174 29. Causey MW, Fitzpatrick E, Carter P (2013) Pressure tolerance of newly constructed staple lines in sleeve gastrectomy and duodenal switch. Am J Surg 205:571–574 (discussion 574–575)
30. Sucandy I, Szomstein S, Rosenthal RJ (2010) Palmar staining following methylene blue leak test during laparoscopic Roux-enY gastric bypass. J Laparoendosc Adv Surg Tech A 20:563–564 31. Lee JH, Chang CH, Park CH, Kim JK (2014) Methylene blue dye-induced skin necrosis in immediate breast reconstruction: evaluation and management. Arch Plast Surg 41:258–263
123
and Other Interventional Techniques
Intraoperative leak testing has no correlation with leak after laparoscopic sleeve gastrectomy Monica Sethi1 • Jonathan Zagzag1 • Karan Patel1 • Melissa Magrath1 Eduardo Somoza1 • Manish S. Parikh1 • John K. Saunders1 • Aku Ude-Welcome1 • Bradley F. Schwack1 • Marina S. Kurian1 • George A. Fielding1 • Christine J. Ren-Fielding1
•
Received: 15 February 2015 / Accepted: 25 May 2015 Ó Springer Science+Business Media New York 2015
Abstract Background Staple line leak is a serious complication of sleeve gastrectomy. Intraoperative methylene blue and air leak tests are routinely used to evaluate for leak; however, the utility of these tests is controversial. We hypothesize that the practice of routine intraoperative leak testing is unnecessary during sleeve gastrectomy. Methods A retrospective cohort study was designed using a prospectively collected database of seven bariatric surgeons from two institutions. All patients who underwent sleeve gastrectomy from March 2012 to November 2014 were included. The performance of intraoperative leak testing and the type of test (air or methylene blue) were based on surgeon preference. Data obtained included BMI, demographics, comorbidity, presence of intraoperative leak test, result of test, and type of test. The primary outcome was leak rate between the leak test (LT) and no leak test (NLT) groups. SAS version 9.4 was used for univariate and multivariate analyses. Results A total of 1550 sleeve gastrectomies were included; most were laparoscopic (99.8 %), except for one converted and two open cases. Routine intraoperative leak tests were performed in 1329 (85.7 %) cases, while 221 (14.3 %) did not have LTs. Of the 1329 cases with LTs, there were no positive intraoperative results. Fifteen (1 %) patients developed leaks, with no difference in leak rate Presented at the SAGES 2015 Annual Meeting, April 15–18, 2015, Nashville, Tennessee. & Monica Sethi [email protected] 1
Department of Surgery, New York University School of Medicine, 530 First Ave., Suite 10S, New York, NY 10016, USA
between the LT and NLT groups (1 vs. 1 %, p = 0.999). After adjusting for baseline differences between the groups with a propensity analysis, the observed lack of association between leak and intraoperative leak test remained. In this cohort, leaks presented at a mean of 17.3 days postoperatively (range 1–67 days). Two patients with staple line leaks underwent repeat intraoperative leak testing at leak presentation, and the tests remained negative. Conclusion Intraoperative leak testing has no correlation with leak due to laparoscopic sleeve gastrectomy and is not predictive of the later development of staple line leak. Keywords Leak Intraoperative leak test Sleeve gastrectomy Staple line Fistula Methylene blue
Bariatric surgery is widely accepted as the most effective treatment for morbid obesity. In recent years, the landscape of bariatric surgery has changed due to an increased use of the laparoscopic sleeve gastrectomy (LSG) relative to other procedures. In 2012, the LSG represented 36.3 % of all bariatric procedures performed at academic medical centers [1]. The sleeve gastrectomy involves a stapled vertical transection of the stomach and creation of a tubular alimentary channel along the stomach’s lesser curvature. Although primarily restrictive, the LSG may alter gut hormonal milieu through the decreased production of ghrelin and increased production of PYY and GLP-1 [2, 3], typically resulting in weight loss comparable to the gastric bypass and superior to the gastric band [4–6]. When compared to gastric bypass procedures, the LSG has a relatively low risk and complication profile [7, 8]. Nevertheless, a known complication of LSG is staple line leak, which usually occurs at the proximal portion of the staple line and has a reported incidence of 1.1–5.3 % [5].
123
Surg Endosc
Leaks can result in significant morbidity including sepsis, hemodynamic instability, multi-organ failure, and even mortality [5]. Early identification and treatment of leaks is critical, but the majority of leaks present days to weeks after the index operation, making early identification challenging [9]. In an effort to identify staple line leak during the index operation, provocative intraoperative testing, including methylene blue and air leak tests, are often performed. However, there is minimal literature on the utility of these tests, and what is available is highly conflicted. In fact, the 2012 International Sleeve Gastrectomy Expert Panel failed to reach a consensus (48 % consensus) about whether routine intraoperative leak tests should be performed [10]. Additionally, these tests are not benign—they introduce increased instrumentation, with reports in the literature of nasogastric tubes causing esophageal perforation [11], as well as increased costs in the form of resource utilization. The present study seeks to evaluate the ability of intraoperative leak tests to detect staple line leaks due to sleeve gastrectomy. We hypothesize that routine intraoperative leak testing has no correlation with leak due to LSG and is not predictive of the later development of leak. A secondary area of investigation was to determine the operative time and costs associated with intraoperative leak testing.
operation regardless of the setting in which they occurred (i.e., during initial hospitalization, on re-admission, or as an outpatient). All leaks were radiographically confirmed. For each patient who developed enteric leak, we gathered information about the duration from operation to leak presentation, location of the leak, presence of contrast extravasation on either esophagram or CT scan, and the type of treatment. Occasionally, an intraoperative leak test was performed at the time of leak presentation, and the results of these tests were recorded as well. Statistical analysis Univariable comparisons between the LT and NLT group were performed with a two-sample t test, Wilcoxon ranksum test, Chi-square test, or Fisher’s exact test. A propensity score model was used to evaluate the leak rate between the LT and NLT groups, while accounting for other patient characteristics. Diagnostic properties of the intraoperative leak test were also evaluated with a sensitivity and specificity analysis. An adjusted quartile regression analysis was used to evaluate differences in operative time between the LT and NLT groups among surgeons who stopped routinely performing the test. A cost analysis was performed using estimates of operative time and product costs from our institutions. SAS version 9.4 was used for all statistical analyses.
Materials and methods Operative technique and postoperative care We conducted a retrospective cohort study from a prospectively collected database of two high-volume Bariatric Surgery Centers of Excellence. We included all patients who underwent sleeve gastrectomy between March 2012 and November 2014 by one of seven bariatric surgeons. All patients were seen in an office setting within 2 weeks of surgery, and all patients had at least 70-day follow-up. Exclusion criteria consisted of age younger than 18 years at the time of surgery. Data obtained included patient demographics, comorbidities, intraoperative characteristics including performance and results of intraoperative leak testing, as well as postoperative complications. The performance of intraoperative leak testing and the type of test (air or methylene blue) were based on surgeon preference. Data on the performance of a leak test, the type of test, and the amount of methylene blue that was instilled were collected from operative reports. The primary outcome was leak rate between the leak test (LT) and the no leak test (NLT) groups. Secondary outcomes were operative time, hospital length of stay (LOS), and overall 30-day complication rate, which included any postoperative complication, re-admission, and/or re-
123
Entry into the peritoneal cavity was accomplished with the use of a 10-mm optical trocar in the left upper quadrant, a Veress needle, or a supraumbilical cut-down per surgeon preference. Any hiatal hernias were repaired at the time of surgery. A vertical sleeve gastrectomy was performed starting approximately 5 cm proximal to the pylorus, using a linear stapler. Bougie size was not standardized and was based on surgeon preference, ranging from 28 to 40 Fr. One surgeon used an endoscope rather than a bougie for intraoperative sizing of the LSG. In most cases, bioabsorbable staple line reinforcement was used during stapling. Drain placement and suture reinforcement of the staple line were not routinely performed. Some surgeons sutured the omentum to the staple line to prevent rotation of the stomach. Intraoperative leak testing was performed per surgeon preference. In cases with methylene blue testing, at the completion of the sleeve, the bougie was removed and an orogastric or nasogastric tube was placed under direct vision. Between 40 and 150 ml of methylene blue was instilled into the stomach with distal obstruction of the duodenum. Leaks were assessed for by the presence of blue
Surg Endosc
dye in the operative field, and the methylene blue solution was suctioned out of the stomach. One surgeon routinely performed intraoperative leak testing with an air leak test. The test consisted of instilling saline into the central upper abdomen. The stomach was insufflated with air through an endoscope with distal occlusion of the pylorus. Leaks in the staple line were evaluated for by the presence of air bubbles. The irrigation fluid was then suctioned out and the stomach decompressed. Postoperative care followed a standardized guideline. An esophagram was performed on postoperative day (POD) 1. A liquid diet was started if no leak was observed on the esophagram study and if the clinical course was uneventful. Hospital discharge generally occurred between POD 1 and 3 depending on the patient’s ability to maintain hydration, ambulate, and manage pain with oral analgesics. After discharge, patients were followed up at the bariatric surgery outpatient offices by a multidisciplinary team within 2 weeks of discharge, at 30 days, and at 3, 6, and 12 months, and yearly thereafter.
Results During the study period, 1562 patients underwent sleeve gastrectomy. Of these, twelve patients were excluded because they did not have adequate follow-up, leaving 1550 for analysis (315 males, 1235 females). The mean age was 40.2 years (range 18–72), and mean BMI was 44.2 kg/ m2 (range 29.1–96.1). Baseline demographic and clinical data are shown in Table 1. Of the 1550 sleeve gastrectomy cases, the vast majority were laparoscopic (99.81 %), except for one (0.06 %) converted and two (0.13 %) open cases. One hundred and thirty-two (9 %) patients had a history of previous foregut or bariatric surgery, and 898 (58 %) patients had a concurrent procedure performed at the time of the sleeve gastrectomy, which were predominantly hiatal hernia repairs. Additional operative characteristics of the study population, including staple line reinforcement, oversewing, drain placement, and bougie size, are shown in Table 2. The 1550 study patients included 1329 (85.7 %) in the leak test group and 221 (14.3 %) in the no leak test group. Routine intraoperative methylene blue leak testing was performed in 991 (63.9 %) patients, and the air leak test was performed in 338 (21.8 %) patients (Fig. 1). Of the 1329 in the leak test group, there were zero (0 %) positive leak test results. Table 3 shows the outcomes of the study population, as well as a comparison of the LT and NLT groups. The majority of patients (97 %) had an uneventful postoperative course after sleeve gastrectomy. Three percent
(n = 47) had 30-day complications, and there was one mortality, with no significant differences between the LT and NLT groups. There was no difference in the primary outcome of leak rate between the groups (1 vs. 1 %, p = 0.999). There were, however, significant differences on multiple levels between the LT and NLT groups, including several baseline and operative characteristics such as hospital location and bougie size. We controlled for these differences by propensity scoring, and there remained no difference in leak rate between the groups (OR = 0.816, 95 % CI = [0.103, 6.439], p = 0.847). A subanalysis comparing air and methylene blue leak tests also showed no difference in leak rates (1 vs. 1 %, p = 0.535). Of note, in an unadjusted comparison between the groups, the LT group had a longer operative time and longer hospital LOS than the NLT group. After adjusting for other factors in a propensity model, the association between intraoperative leak test and LOS was no longer significant (OR = 1.23, 95 % CI = [0.792, 1.908], p = 0.358). Operative time was controlled for in a separate analysis described later in this section. Additionally, operative volume was not related to leak rate (p = 0.656). Fifteen patients (1 %) developed staple line leaks during the study period. The mean number of days from surgery to leak presentation was 17.3 days (median = 9 days, range 1–67 days) with only two leaks presenting during the index admission. The majority of leaks (n = 13) were located proximally near the gastroesophageal junction; of the remaining two leaks, one was located at the distal staple line and the other was unable to be accurately localized. Treatment modalities ranged from operative drainage and stenting (n = 6), stenting alone (n = 3), nonoperative management with nihil per os and antibiotics (n = 4), and stent followed by revision to Roux-en-Y gastric bypass (n = 2). Thirteen of the patients who developed leak had intraoperative leak testing during the index operation, and as stated, there were no (0 %) positive leak tests. Eleven of these patients had methylene blue tests, and two had air leak tests. Of those who underwent methylene blue testing, three patients had 100 ml instilled, three had 60 ml instilled, and for the remaining five patients, the amount of methylene blue was unknown. Two patients underwent repeat intraoperative leak testing at the time of leak presentation, prior to any definitive treatment. Of those two, one patient with a leak seen at the upper staple line on POD 1 esophagram underwent a repeat methylene blue leak test during diagnostic laparoscopy, which was negative. Despite suture reinforcement, the leak persisted and the patient eventually required conversion to gastric bypass after unsuccessful stent treatment. The other patient presented with symptoms of epigastric pain on POD 46. He was diagnosed with a proximal leak on esophagram
123
Surg Endosc Table 1 Baseline demographic and clinical characteristics of the study population stratified by leak test versus no leak test
Overall N = 1550
Leak test N = 1329
No leak test N = 221
p value
Female, % (n)
80 % (n = 1235)
81 % (n = 1083)
69 % (n = 152)
\0.0001a
Male, % (n)
20 % (n = 315)
19 % (n = 246)
31 % (n = 69)
Gender
Age, years Mean (SD) [Range] 18–29, % (n)
40.2 (11.6) [18, 72] 20 % (n = 310)
39.9 (11.5) [18, 70] 21 % (n = 277)
42.1 (12.3) [18, 72] 15 % (n = 33)
0.010b 0.144a
30–39, % (n)
31 % (n = 480)
31 % (n = 413)
30 % (n = 67)
40–49, % (n)
26 % (n = 398)
26 % (n = 341)
26 % (n = 57)
50–59, % (n)
18 % (n = 280)
17 % (n = 231)
22 % (n = 49)
5 % (n = 82)
5 % (n = 67)
7 % (n = 15)
White, % (n)
77 % (n = 1190)
78 % (n = 1031)
72 % (n = 159)
Black, % (n)
14 % (n = 219)
14 % (n = 190)
13 % (n = 29)
Other, % (n)
1 % (n = 19)
1 % (n = 15)
2 % (n = 4)
Unknown, % (n) Ethnicity
8 % (n = 122)
7 % (n = 93)
13 % (n = 29)
Hispanic, % (n)
57 % (n = 881)
61 % (n = 816)
29 % (n = 65)
Non-Hispanic, % (n)
35 % (n = 542)
32 % (n = 425)
53 % (n = 117)
8 % (n = 127)
7 % (n = 88)
18 % (n = 39)
BVU, % (n)
61 % (n = 946)
68 % (n = 904)
19 % (n = 42)
NYULMC, % (n)
39 % (n = 604)
32 % (n = 425)
81 % (n = 179)
Mean (SD) [Range]
44.2 (7.8) [29.1, 96.1]
44.2 (7.7) [29.1, 88.5]
44.1 (8.4) [29.7, 96.1]
0.814b
BMI \ 50, % (n)
81 % (n = 1263)
82 % (n = 1085)
81 % (n = 178)
0.697a
BMI C 50, % (n)
19 % (n = 287)
18 % (n = 244)
19 % (n = 43)
60?, % (n) Race
Unknown, % (n)
0.013a
\0.0001a
Hospital \0.0001a
BMI, kg/m2
SD standard deviation, BVU Bellevue Hospital Center, NYULMC New York University Langone Medical Center, BMI body mass index a
Chi-square test,
b
two-sample t test
and underwent stent placement by interventional radiology. To confirm the leak’s location during stenting, Gastrografin was injected through an orogastric route into the stomach under fluoroscopy, which was negative despite being able to visualize the leak site on endoscopy. The leak resolved after 5 weeks of stenting. To determine the diagnostic predictive value of intraoperative leak testing, we determined the sensitivity, specificity, as well as positive predictive values (PPV) and negative predictive values (NPV) of the test. The sensitivity and specificity for detecting a leak were 0 % (95 % CI = [0, 24 %]) and 100 %, (95 % CI = [99.7, 100 %]), respectively, resulting in an undefined PPV and an NPV of 99 % (95 % CI = [98.3, 99.5 %]). Based on a preliminary analysis of our data, four surgeons stopped routinely performing intraoperative leak tests during the study period; of note, some would
123
selectively perform intraoperative leak tests during revisional surgeries or technically difficult cases. The operative time for these surgeons was analyzed using a quartile regression model with adjustment for concurrent procedures performed at the index operation, as well as a propensity score to account for the differences between LT and NLT groups. Three patients with open surgery or conversion to open were excluded from analysis. Results showed that operative time reduced by 7.6 min when the leak test was not performed (95 % CI = [-13.2, -2.0], p = 0.0083). At our hospitals, the cost of an additional 7.6 min of operative time and the supplies required for a methylene blue leak test are estimated to total approximately $855.37 per test. Based on these estimates, the total cost for intraoperative methylene blue leak tests during the course of the study was $847,671.67.
Surg Endosc Table 2 Operative characteristics of the study population stratified by leak test versus no leak test Overall N = 1550 Prior foregut/bariatric surgery, % (n)
9 % (n = 132)
Leak test N = 1329
No leak test N = 221
6 % (n = 83)
22 % (n = 49)
[99 % (n = 1327)
100 % (n = 221)
p value \0.0001a
Surgical approach, % (n) [99 % (n = 1453)
Laparoscopic Open Converted to open, % (n)
\1 % (n = 2)
\1 % (n = 2)
0 % (n = 0)
\1 % (n = 1)
0 % (n = 0)
\1 % (n = 1)
0.999b 0.143b
94 % (n = 1453)
93 % (n = 1424)
95 % (n = 211)
0.251a
5 % (n = 73)
5 % (n = 64)
4 % (n = 9)
0.629a
18 % (n = 278)
20 % (n = 262)
7 % (n = 16)
\0.0001a
28–34
19 % (n = 285)
15 % (n = 198)
40 % (n = 87)
\0.0001a
35–39 40?
35 % (n = 540) 46 % (n = 707)
39 % (n = 510) 46 % (n = 605)
14 % (n = 30) 47 % (n = 102)
Staple line reinforcement, % (n) Oversew, % (n) Drain placed, % (n) Bougie size, % (n)
a
Chi-square test,
b
Fisher’s exact test
Fig. 1 Bar plot of the percent of patients by type of intraoperative leak test
Discussion While early recognition and treatment are essential in managing staple line leaks after sleeve gastrectomy, leaks may present weeks or even months after surgery [5, 12]. Patients typically exhibit subtle clinical signs before rapidly deteriorating, rendering timely diagnosis difficult. To detect leaks as early as possible, some surgeons use intraoperative leak testing to assess the integrity of the staple line during LSG. These intraoperative tests, which include methylene blue and air leak tests, are designed to detect leak when tissues are viable and most amenable to repair by re-stapling or suturing [5]. It should be emphasized, however, that a negative intraoperative leak test result does not preclude the possibility of a subsequent staple line leak. Indeed, a negative test result may give the
surgeon a false sense of security and bias the differential diagnosis away from leak. Additionally, these tests are not completely benign and can add risks and unnecessary costs to the surgery. Nevertheless, despite the dearth of scientific studies showing the benefit of intraoperative leak testing, many surgeons still favor these tests and perform them routinely. In this study, we have shown that intraoperative leak tests have a sensitivity of 0 % (95 % CI = [0, 24.7 %]). Although the test specificity was high (100 %, 95 % CI = [99.7, 100 %]), the low sensitivity combined with the occurrence of zero positive results out of 1329 tests suggests that the utility of intraoperative leak testing needs to be questioned. Furthermore, the incidence of detected postoperative or delayed leaks was the same with or without routine intraoperative leak testing. Therefore, the authors believe that routine intraoperative leak testing may be unnecessary during LSG. The most important reason why intraoperative leak testing does not reliably detect leaks after LSG is that most leaks develop after surgery. Intraoperative leak testing can only detect the rare leaks due to technical staple line failure, such as those occurring due to stapler misfire or other intraoperative complications [5, 13]. However, staple line failure tends to be infrequent in the hands of experienced surgeons at high-volume centers. The pathogenesis of postoperative or delayed leaks that occur days to weeks after LSG is not fully understood. Some authors suggest that these postoperative leaks are caused by ischemia, thermal damage, or misplacement of the staple line to involve the distal esophagus in patients with hiatal hernias [14]. These scenarios all lead to intraluminal pressure that exceeds tissue and suture line resistance, thus causing leak. Additionally, narrowing the sleeve
123
Surg Endosc Table 3 Intraoperative outcomes, length of stay, and 30-day outcomes for the study population stratified by leak test versus no leak test (n = 221) Overall N = 1550
Leak test N = 1329
No leak test N = 221
p value
Intraoperative outcomes \1 % (n = 1)
Any occurrence, % (n) OR time, median (IQR)
86 (71, 143)
0 % (n = 0) 89 (72, 107)
\1 % (n = 1) 76 (60, 93)
0.143b \0.0001c
Length of stay \0.0001a
1
8 % (n = 129)
7 % (n = 87)
19 % (n = 42)
2
72 % (n = 1099)
73 % (n = 961)
63 % (n = 138)
3 4?
16 % (n = 243) 4 % (n = 64)
16 % (n = 214) 4 % (n = 53)
13 % (n = 29) 5 % (n = 11)
1 % (n = 15)
1 % (n = 13)
1 % (n = 2)
0.999b
0 % (n = 0)
\1 % (n = 1)
0.143b
Postoperative outcomes Leak, % (n)
\1 % (n = 1)
30-day death Any 30-day occurrence, % (n)
3 % (n = 47)
3 % (n = 36)
5 % (n = 11)
0.069a
Bold represents primary outcome OR operating room, IQR interquartile range a
Chi-square test,
b
Fisher’s exact test,
c
Wilcoxon rank-sum test
at the incisura angularis can lead to a physiologic stricture that may contribute to leaks at the gastroesophageal junction by creating a distal obstruction [15]. Classically, leaks due to ischemia present on POD 5 or 6. This is in line with data from Sakran et al. [5], which cites a median of 7 days to leak presentation. In our study, however, the number of days to leak presentation was longer than previously reported, with a mean of 17.3 days. Additionally, with leaks in our study presenting up to 67 days postoperatively, the pathophysiology of staple line leaks after LSG deserves further study. Another potential reason for the poor sensitivity of this test may be due to testing misuse. Leak testing with methylene blue was originally designed to diagnose both intraoperative leak and leak during the following 7 days by observing methylene blue effluence through abdominal drains [16, 17]. However, drain placement has fallen out of favor, and surgeons who do place drains rarely leave them in place long enough to detect postoperative leak. In 2013, only 39 % of surgeons left a drain in the abdominal cavity after LSG, and the number continues to decline [18]. If a drain is left in place, however, postoperative leak testing with methylene blue may effectively diagnose leak. Some surgeons have also used this method during follow-up to monitor the progress of the leak [6, 19]. Still, this method’s usefulness is limited to leaks that communicate with the surgically placed drains [20]. Ours is not the first study to note the lack of association between intraoperative leak testing and postoperative leak due to LSG. De Aretxabala et al. [12] had negative leak test results in all of the nine leaks included in their study. Casella et al. [21] reported two positive intraoperative
123
methylene blue tests (1 %), which were repaired intraoperatively without further postoperative sequelae; however, the six (3 %) postoperative leaks, even those detected on the first postoperative day, all had negative intraoperative methylene blue tests. In a larger study by Sakran et al., 2834 patients underwent LSG, resulting in 38 leaks (1.5 %). An intraoperative leak test was performed during the original surgery for 33 (75 %) of the 38 leaks, using methylene blue in 25 patients and air testing in eight patients. The intraoperative leak test results were negative in all but one case, in which there was a stapler misfire. The leak persisted despite intraoperative recognition, immediate repair, and a subsequent confirmatory intraoperative methylene blue test [5]. In a recent meta-analysis by Parikh et al., 67 % of 8922 patients underwent intraoperative leak testing. The leak rate was compared between methylene blue leak test, air leak test, ‘‘other’’ leak test, and no leak test, and in an unadjusted GEE regression model, the performance of leak test did not impact the leak rate (p = 0.454) [22]. The meta-analysis did not comment on the number of positive intraoperative leak tests. Some studies have noted that an intraoperative leak test result can be negative even when there is a known leak. Mizrahi et al. were unable to locate three out of four CTproven leaks on diagnostic laparoscopy, despite intraoperative air and methylene blue leak tests. They speculate that this may be due to symptomatic micro-leaks or tissue edema at the leak site, causing the tests to have a negative result [7]. In our study, two of our 15 leaks experienced similar scenarios—negative intraoperative leak test results despite the presence of a known leak.
Surg Endosc
It is also possible that leak tests do not supply enough pressure to cause extravasation of air or methylene blue, whereas active swallowing during an esophagram may provide enough pressure to promote extravasation. Additionally, in our study, we show that the amount of dilute methylene blue solution instilled into the stomach during leak testing varies from 40 to 150 ml; thus, the pressure created during intraoperative leak testing is likely quite variable. Given that there were only eleven patients who developed leaks and had prior intraoperative methylene blue tests (five of whom had an unknown amount of methylene blue), the small sample size limits our ability to assess an association between low volume methylene blue instillation and potential missed leaks, but this may be an area for future study. In a meta-analysis of 4000 patients, only 15 of 29 studies documented their use of intraoperative leak tests. Of those, 62 % used methylene blue [9]. Although most surgeons perform methylene blue over air leak tests, studies from the gastric bypass literature note that air insufflation, especially when delivered through an endoscope, may be more sensitive than methylene blue tests in detecting anastomotic leak [23, 24]. Additionally, endoscopy offers the ability to assess hemostasis intraluminally. Methylene blue can also stain the field, making it difficult to repeat the test if a leak is detected and repaired [25]. On the other hand, the equipment and personnel required for endoscopy lead to additional costs that are not encountered with methylene blue leak testing. In our study, we showed that there is no difference in LSG leak rate between the two types of leak tests, although for air leak testing, we can only comment on the type performed with an endoscope rather than an orogastric tube. Regardless of the type of test, intraoperative leak testing prolongs operative time and is associated with increased costs and resource utilization. The cost of routine intraoperative leak tests for sleeve gastrectomy has not, to our knowledge, been addressed in previous studies. When considering the prevalence of bariatric surgery, and LSG in particular, it becomes apparent that routinely using resources that are not necessary or tests that are not evidence based can be costly and wasteful. In our study, we showed that 991 intraoperative methylene blue tests cost $847,671.67 during our 2.5-year study period. Some have argued that the cost of these studies is justified by the potential cost of preventing a leak [26]. However, in this study, intraoperative leak testing did not detect a single leak, and therefore, there was no cost benefit. It is important to note that in addition to a reduction in operative time, unadjusted analyses demonstrated a reduction in hospital LOS among the NLT group. Unlike operative time, however, the difference in LOS was not significant after adjusting for other factors, such as hospital
location, in the propensity model. Therefore, the difference in LOS is likely attributed to other differences between the groups (i.e., hospital location) and not related to the performance of a leak test. Leak testing is not without its risks. Nasogastric tube has been associated with lung perforation, esophageal perforation, massive bleeding, and other mediastinal complications [11, 27, 28]; intraoperative orogastric tube placement could lead to a similar sequela. It has also been theorized that the shear forces on staple lines engendered by the leak testing process could result in perioperative staple line weakness, although scientific data seem to disprove this theory [29]. During intraoperative leak testing, methylene blue can be systemically absorbed and cause blue urine and prolonged staining of the skin [30]. As per our experience, methylene blue can be spilled on patients’ hair and faces during surgery, causing them distress and requiring costly hair treatments to remove the blue color. Additionally, there are multiple reports in the oncologic and breast surgery literature of methylene blue causing anaphylaxis and skin necrosis [31]. At our hospitals, as our experience has grown and our leak rate decreased, the utility of intraoperative leak tests has been re-evaluated and some surgeons have stopped the practice altogether. Similarly, Bellanger et al. [15], in their review of 529 LSG without a leak, performed 350 cases with no intraoperative detection of leak and subsequently stopped the practice. However, leak testing may be justified in cases of revisional surgery, intraoperative complications, or in the case of a surgeon who is early in the learning curve. Thus, we suggest that surgeons determine whether or not to abandon routine intraoperative leak testing only after they monitor their leak rates and the sensitivity of their own intraoperative leak tests. Our study has several limitations. First, the study design was retrospective; thus, it relies on the accuracy and thoroughness of the database and limits our ability to generalize the results. Second, considering the low occurrence of leaks after LSG, large numbers are required to detect a significant difference in leak rate. In the current literature, most singlecenter studies do not include data on the specific results of intraoperative leak tests, as is the case with large multicenter databases. Therefore, a meta-analysis involving the sensitivity and specificity of intraoperative leak testing would be challenging. Lastly, it is possible that other factors contributed to the decreased operative time among surgeons who stopped routinely performing leak tests. Although we attempted to account for these factors with our adjusted analysis, a randomized study would be the most effective at determining the true effect size. In conclusion, intraoperative leak testing has no correlation with leak during laparoscopic sleeve gastrectomy
123
Surg Endosc
and is not associated with the later development of staple line leak. It also prolongs operative time by 7.6 min, which translates into significantly increased operative costs. Therefore, the better practice is to selectively perform intraoperative methylene blue or air leak tests in cases with intraoperative complications such as staple line misfire, revisional surgery, or in practices with high leak rates. Acknowledgments The authors wish to acknowledge Heekoung Youn, MA for technical assistance and G. Craig Wood, MS for leading the statistical analyses sited in this publication. Disclosures Dr. Ren-Fielding received research Grants (S#06-851, S#10686) from Allergan Medical, and received an honorarium as a speaker and as a member of their Advisory Board. Dr. Fielding received an honorarium as part of the Speaker’s Bureau/Faculty, research Grants (S#06-851, S#10686), and an educational grant from Allergan Medical. Dr. Kurian received a research Grant (S#06-851) and an honorarium as a speaker for Allergan Medical. Dr. Schwack received an honorarium as a speaker for Allergan Medical. Dr. Sethi, Dr. Zagzag, Dr. Somoza, Dr. Parikh, Dr. Saunders, Dr. Ude-Welcome, Mr. Patel and Ms. Magrath have no conflicts of interest or financial ties to disclose.
References 1. Nguyen NT, Nguyen B, Gebhart A, Hohmann S (2013) Changes in the makeup of bariatric surgery: a national increase in use of laparoscopic sleeve gastrectomy. J Am Coll Surg 216:252–257 2. Gagner M, Buchwald JN (2014) Comparison of laparoscopic sleeve gastrectomy leak rates in four staple-line reinforcement options: a systematic review. Surg Obes Relat Dis 10:713–723 3. Nannipieri M, Baldi S, Mari A, Colligiani D, Guarino D, Camastra S, Barsotti E, Berta R, Moriconi D, Bellini R, Anselmino M, Ferrannini E (2013) Roux-en-Y gastric bypass and sleeve gastrectomy: mechanisms of diabetes remission and role of gut hormones. J Clin Endocrinol Metab 98:4391–4399 4. Fridman A, Moon R, Cozacov Y, Ampudia C, Lo Menzo E, Szomstein S, Rosenthal RJ (2013) Procedure-related morbidity in bariatric surgery: a retrospective short- and mid-term follow-up of a single institution of the American College of Surgeons Bariatric Surgery Centers of Excellence. J Am Coll Surg 217:614–620 5. Sakran N, Goitein D, Raziel A, Keidar A, Beglaibter N, Grinbaum R, Matter I, Alfici R, Mahajna A, Waksman I, Shimonov M, Assalia A (2013) Gastric leaks after sleeve gastrectomy: a multicenter experience with 2,834 patients. Surg Endosc 27:240–245 6. Tan JT, Kariyawasam S, Wijeratne T, Chandraratna HS (2010) Diagnosis and management of gastric leaks after laparoscopic sleeve gastrectomy for morbid obesity. Obes Surg 20:403–409 7. Mizrahi I, Tabak A, Grinbaum R, Beglaibter N, Eid A, Simanovsky N, Hiller N (2014) The utility of routine postoperative upper gastrointestinal swallow studies following laparoscopic sleeve gastrectomy. Obes Surg 24:1415–1419 8. Aggarwal S, Kini SU, Herron DM (2007) Laparoscopic sleeve gastrectomy for morbid obesity: a review. Surg Obes Relat Dis 3:189–194 9. Aurora AR, Khaitan L, Saber AA (2012) Sleeve gastrectomy and the risk of leak: a systematic analysis of 4,888 patients. Surg Endosc 26:1509–1515
123
10. Rosenthal RJ, International Sleeve Gastrectomy Expert P, Diaz AA, Arvidsson D, Baker RS, Basso N, Bellanger D, Boza C, El Mourad H, France M, Gagner M, Galvao-Neto M, Higa KD, Himpens J, Hutchinson CM, Jacobs M, Jorgensen JO, Jossart G, Lakdawala M, Nguyen NT, Nocca D, Prager G, Pomp A, Ramos AC, Rosenthal RJ, Shah S, Vix M, Wittgrove A, Zundel N (2012) International Sleeve Gastrectomy Expert Panel Consensus Statement: best practice guidelines based on experience of [12,000 cases. Surg Obes Relat Dis 8:8–19 11. Wu PY, Kang TJ, Hui CK, Hung MH, Sun WZ, Chan WH (2006) Fatal massive hemorrhage caused by nasogastric tube misplacement in a patient with mediastinitis. J Formos Med Assoc 105:80–85 12. de Aretxabala X, Leon J, Wiedmaier G, Turu I, Ovalle C, Maluenda F, Gonzalez C, Humphrey J, Hurtado M, Benavides C (2011) Gastric leak after sleeve gastrectomy: analysis of its management. Obes Surg 21:1232–1237 13. Aggarwal S, Bhattacharjee H, Chander Misra M (2011) Practice of routine intraoperative leak test during laparoscopic sleeve gastrectomy should not be discarded. Surg Obes Relat Dis 7:e24– e25 14. Mittermair R, Sucher R, Perathoner A, Wykypiel H (2014) Routine upper gastrointestinal swallow studies after laparoscopic sleeve gastrectomy are unnecessary. Am J Surg 207:897–901 15. Bellanger DE, Greenway FL (2011) Laparoscopic sleeve gastrectomy, 529 cases without a leak: short-term results and technical considerations. Obes Surg 21:146–150 16. Christian D, Barba C, Burke-Martindale C (2004) A simple bedside evaluation to detect gastroesophageal leaks after gastric bypass surgery. Surg Endosc 18:S193 (abst) 17. Ovnat A, Peiser J, Solomon H, Charuzi I (1986) Early detection and treatment of a leaking gastrojejunostomy following gastric bypass. Isr J Med Sci 22:556–558 18. Gagner M, Deitel M, Erickson AL, Crosby RD (2013) Survey on laparoscopic sleeve gastrectomy (LSG) at the Fourth International Consensus Summit on Sleeve Gastrectomy. Obes Surg 23:2013–2017 19. Jen S, Simillis C, Efthimiou E (2013) A very challenging leak from a sleeve gastrectomy. Surg Obes Relat Dis 9:e56–e59 20. Gonzalez R, Nelson LG, Gallagher SF, Murr MM (2004) Anastomotic leaks after laparoscopic gastric bypass. Obes Surg 14:1299–1307 21. Casella G, Soricelli E, Rizzello M, Trentino P, Fiocca F, Fantini A, Salvatori FM, Basso N (2009) Nonsurgical treatment of staple line leaks after laparoscopic sleeve gastrectomy. Obes Surg 19:821–826 22. Parikh M, Issa R, McCrillis A, Saunders JK, Ude-Welcome A, Gagner M (2013) Surgical strategies that may decrease leak after laparoscopic sleeve gastrectomy: a systematic review and metaanalysis of 9991 cases. Ann Surg 257:231–237 23. Gagner M, Deitel M, Kalberer TL, Erickson AL, Crosby RD (2009) The Second International Consensus Summit for Sleeve Gastrectomy, March 19–21, 2009. Surg Obes Relat Dis 5:476–485 24. Alaedeen D, Madan AK, Ro CY, Khan KA, Martinez JM, Tichansky DS (2009) Intraoperative endoscopy and leaks after laparoscopic Roux-en-Y gastric bypass. Am Surg 75:485–488 (discussion 488) 25. Alasfar F, Chand B (2010) Intraoperative endoscopy for laparoscopic Roux-en-Y gastric bypass: leak test and beyond. Surg Laparosc Endosc Percutan Tech 20:424–427 26. Madan AK, Stoecklein HH, Ternovits CA, Tichansky DS, Phillips JC (2007) Predictive value of upper gastrointestinal studies versus clinical signs for gastrointestinal leaks after laparoscopic gastric bypass. Surg Endosc 21:194–196
Surg Endosc 27. Torrington KG, Bowman MA (1981) Fatal hydrothorax and empyema complicating a malpositioned nasogastric tube. Chest 79:240–242 28. Winterholler M, Erbguth FJ (2002) Accidental pneumothorax from a nasogastric tube in a patient with severe hemineglect: a case report. Arch Phys Med Rehabil 83:1173–1174 29. Causey MW, Fitzpatrick E, Carter P (2013) Pressure tolerance of newly constructed staple lines in sleeve gastrectomy and duodenal switch. Am J Surg 205:571–574 (discussion 574–575)
30. Sucandy I, Szomstein S, Rosenthal RJ (2010) Palmar staining following methylene blue leak test during laparoscopic Roux-enY gastric bypass. J Laparoendosc Adv Surg Tech A 20:563–564 31. Lee JH, Chang CH, Park CH, Kim JK (2014) Methylene blue dye-induced skin necrosis in immediate breast reconstruction: evaluation and management. Arch Plast Surg 41:258–263
123
