Safe Cholecystectomy Multi-Society Practice Guideline and State of the Art Consensus Conference on Prevention of Bile Duct Injury During Cholecystectomy

L. Michael Brunt, MD; Daniel J. Deziel, MD; Dana A. Telem, MD, MPH; Steven M. Strasberg, MD; Rajesh Aggarwal, MD; Horacio Asbun, MD; Jaap Bonjer, MD; Marian McDonald, MD; Adnan Alseidi, MD; Mike Ujiki, MD; Taylor S. Riall, MD, PhD; Chet Hammill, MD; Carol-Anne Moulton, MD; Philip H. Pucher, MD; Rowan W. Parks, MD; Mohammed T. Ansari, MD, MMedSc, MPhil; Saxon Connor, MD; Rebecca C. Dirks, MD; Blaire Anderson, MD; Maria S. Altieri, MD; Levan Tsamalaidze, MD; Dimitrios Stefanidis, MD, PhD

Disclosures

Annals of Surgery. 2020;272(1):3-23. 

In This Article

Guideline Recommendations

Question 1: Should the critical view of safety (CVS) versus other techniques (eg, infundibular, top down, or intraoperative cholangiography) be used to mitigate the risk of BDI during LC?

Recommendation: In patients undergoing LC, we suggest that surgeons use the CVS for anatomic identification of the cystic duct (CD) and artery (expert opinion).

Summary of Evidence

No direct comparative evidence was identified to support the CVS over other methods for anatomic identification. The evidence for intraoperative cholangiogram (IOC) versus no IOC is addressed under guideline question 4. Restricting the meta-analysis to single-arm cohort studies of sample size ≥400 cases, the pooled incidence of BDI was 2 in 1 million cases when CVS was used (n = 4 studies with n = 5446 cases)[39–42] versus a pooled BDI incidence of 1.5 in 1000 cases when the infundibular technique was used (n = 3 studies with 10,060 cases).[43–45] This comparison; however, was deemed high risk of bias by the guideline development group (GDG) because of concerns regarding exchangeability of populations between the 2 intervention groups.

Narrative Synthesis

Forty-five full-text articles identified by the search methodology were reviewed that included 3 systematic reviews.

Use of CVS. The use of the CVS is based on 2 lines of indirect evidence. First, several large single institutional studies with a sufficient number of cases to potentially encounter BDI report lower than expected rates of BDI with routine use of the CVS.[39,41,42,46] In addition, as noted above in the summary of evidence, the pooled BDI incidence from studies reporting use of the CVS is lower than that from studies that describe identification primarily by what would be considered an infundibular approach.[39–45,47–49]

A second line of indirect evidence for use of the CVS comes from case series of BDIs that have analyzed the circumstances of the injury which have demonstrated that BDIs usually occurred in cases when the CVS was not attained. An analysis of the operative reports from 528 BDI cases (396 excluding CD leaks, 479 laparoscopic cases) treated at a single institution in Amsterdam found documentation of the CVS in only 33 cases (6.3%).[50] Similarly, an examination of operative reports from 21 patients with BDI identified the infundibular method in all cases.[51] One study that reviewed 65 videos of LC with complications found that videos from 11 BDI cases had not achieved the CVS whereas a control group without complications after LC 72% had achieved the CVS.[52]

The ability to achieve the CVS in 85%–95% of attempted cases indicates that it is implementable.[39–41] Moreover, there is no substantial evidence that reasonable efforts to achieve the CVS have been associated with undesirable effects. The literature search identified only 1 report of BDI occurring during an attempt to achieve the CVS.[53]

A systematic review (R-AMSTAR score[25] 23/44) of methods for intraoperative assessment of biliary anatomy for prevention of BDI included 4 studies on the CVS with 4593 patients.[54] One retrospective study compared patients undergoing LC by the infundibular approach to those using the CVS.[55] Although no BDIs occurred, the study was seriously underpowered to assess BDI as an outcome metric. It was also critically flawed in that it assumed a particular method was used for identification based on the experience of the surgeon without objective documentation of the actual method used.

Use of Fundus-first Dissection. Use of the "fundus-first" or "top-down" method for intraoperative anatomic identification during LC was examined in a review (R-AMSTAR score 18/44) of operative techniques during operations that were considered difficult based on conversion rate or iatrogenic complications (without specification of BDI occurrence).[56] Eleven studies with a total of 1100 patients from 1995 to 2009 were reported. The authors did not detail the studies reviewed, but concluded that there was evidence for the safety and efficacy of this technique. The 11 studies referenced in this review were individually re-examined in full-text format.[46,57–66]

The selection criteria for patients undergoing laparoscopic fundus-first cholecystectomy were heterogeneous. Nine studies with a total of 934 patients reported no BDI.[57–62,64–66] However, in the largest series of 500 patients, the operative technique involved identification and tape ligation of the CD before fundus-first dissection of the gallbladder.[58] When reported, conversion to open cholecystectomy occurred in 16 of 312 (5.1%) cases and complications in 10 of 285 (3.5%) cases. One additional single-center study, not included in the review described, reported no complications among 81 patients undergoing laparoscopic fundus-first dissection performed after the CD was clipped and the cystic artery divided.[67]

A prospective trial evaluated patients with contracted gallbladders as determined by preoperative sonographic criteria.[68] Patients were randomized by alternate numbers to either conventional LC or to laparoscopic fundus-first dissection. Ten patients were excluded and 2 additional patients were added based on intraoperative finding of a contracted gallbladder. Among the 31 patients undergoing standard LC, there were 10 conversions, 7 with complications and 2 with BDI. There were no conversions or BDIs among the 33 patients undergoing fundus-first operations and 1 patient had a complication.

Two case series of BDIs discuss potential implications of a fundus-first gallbladder dissection. A series of 182 BDIs (30 during LC) from 4 university hospitals in China noted that all BDIs occurred when dissection commenced in the region of the hepato-cystic triangle.[69] Thus, dissection from the fundus was considered the optimal choice during cholecystectomy. On the other hand, a series of 8 devastating vasculo-biliary injuries found that all occurred with a fundus first dissection after conversion of laparoscopic operation for severely inflamed gallbladders.[70]

Justification. Given the available evidence, the GDG relied on expert opinion to make this recommendation. Although studies were not designed to directly answer the effectiveness of the CVS, the anticipated effect would have been large. Event analysis of BDIs implicate methods of anatomic identification other than the CVS as potential causes of injury in essentially all cases evaluated. CVS is attainable in a majority of cases when attempted routinely. The safe extent of dissection in any case must always be tempered by surgical judgment, but there is no evidence that reasonable efforts to achieve the CVS have been harmful, and, therefore, use of CVS is logical. When operative conditions are difficult, and the CVS cannot be reasonably achieved, alternative methods for either anatomic definition or conclusion of the operation are critical to prevention of BDI or other injury.

Question 2: Should the fundus-first (top down) technique versus subtotal cholecystectomy (STC) be used to mitigate the risk of BDI when the CVS cannot be achieved during LC?

Recommendation: When the CVS cannot be achieved and the biliary anatomy cannot be clearly defined by other methods (eg, imaging) during LC, we suggest that surgeons consider STC over total cholecystectomy by the fundus-first (top down) approach (expert opinion).

Summary of Evidence

No direct comparative evidence addressed this key question. A number of noncomparative case series report low BDI risk with the application of both these techniques. An indirect comparison of this evidence was not considered appropriate by the GDG because the majority of noncomparative studies were underpowered, and the comparison was flawed due to confounding by surgical expertise, patient population selection, and subjectivity in the judgment of severity of inflammation. One study implicated the top-down approach under difficult conditions as a risk factor for BDI and vascular injury.[70]

Narrative Synthesis

Thirty-five full-text articles identified by the search methodology are summarized below. Data regarding use of laparoscopic fundus-first dissection has been summarized in the section on Question #1.

Laparoscopic STC

Reviews and Observational Cohort Studies. Three reviews provide generalized information about the outcomes of laparoscopic STC. A systematic review and meta-analysis of 30 studies of STC for difficult gallbladders included 898 laparoscopic operations, 99 laparoscopic operations converted to open, and 234 primary open operations.[71] One BDI was reported (0.08%). Subgroup analysis of 471 completed laparoscopic cases found bile leaks in 149, retained stones in 8, morbidity in 9, and 1 mortality. A review of 28 studies of 1280 patients undergoing STC included 154 laparoscopic subtotal cholecystectomies.[72] Conversion occurred in 13 of 95 reported cases; there was no detailed information on BDI or other complications. A review of operative techniques in difficult cholecystectomy included 12 studies with 822 patients having laparoscopic STC.[56] The authors stated that there was evidence for the safety and efficacy of this technique.

Full-text review of articles cited in the above reviews was performed and the following outcomes for laparoscopic STC were tabulated: BDI, 1/657 (0.15%); conversion 22/637 (3.5%); morbidity 164/637 (25.7%); mortality 2/637 (0.3%); bile leak 161/580 (27.8%); retained stones 14/458 (3.1%).[44,45,63,73–80] Six additional cohort studies of laparoscopic STC that were unique to those cited in the reviews were identified and included 796 patients with the following outcomes reported: BDI 1/370 (0.2%); conversion 96/778 (12.3%); morbidity 45/340 (13.2%); mortality 0/575; bile leak 26/301 (8.6%).[46,81–85]

Comparative Studies

No studies directly compared laparoscopic STC and fundus-first dissection as separate and distinct operations. However, laparoscopic STC has been compared to operative alternatives in several formats.

Laparoscopic STC Versus Laparoscopic Total Cholecystectomy (LC)

An analysis from the University Health System Consortium database compared outcomes of 487 patients undergoing laparoscopic STC to LC.[86] A 1:1 propensity score match was used to account for differences in clinical and demographic factors. After matching, there were no differences in hospital length of stay (LOS), readmission rates, or mortality, but subtotal operations had higher total direct costs. BDI was not specified. A prospective, nonrandomized trial compared results between patients with preoperative predictors for difficult cholecystectomy based on an objective scoring system.[87] Sixty-five patients who underwent a version of subtotal LC had no BDI and 1 conversion. Sixty patients with standard total LC incurred 2 BDIs (P = 0.134) and 6 conversions (P = 0.04).

Laparoscopic STC Versus Converted or Primary Open STC

In a meta-analysis of operations for difficult gallbladders, laparoscopic STC compared favorably to open STC with less risk for sub-hepatic collection [odds ratio (OR) 0.4, 95% confidence interval (CI) 0.2–0.9], retained stones (OR 0.5, 95% CI 0.3–0.9), wound infection (OR 0.07, CI 0.04–0.2), reoperation (OR 0.5, 95% CI 0.3–0.9), and mortality (OR 0.2, 95% CI 0.05–0.9), but was associated with more bile leaks (OR 5.3, 95% CI 3.9–7.2).[71]

A retrospective cohort analysis of the Nationwide Inpatient Sample database of 10,872 "damage control" gallbladder operations that included laparoscopic and open STC and "trocar cholecystostomy."[88] There were 360 BDIs reported with a 13.6% morbidity rate and 7.4% in-hospital mortality rate. Laparoscopic STC was attempted in 5164 cases with 2204 cases converted (42.7%). The specific number of BDIs in each cohort was not reported, but no statistical difference between converted and nonconverted cases was observed (P = 0.159). A single health system comparison of laparoscopic STC (n = 24) and primary open STC (n = 41) found similar postoperative complications with no BDIs, but longer postoperative LOS for open procedures.[82]

Laparoscopic and Open STC Versus Total Cholecystectomy

A retrospective multi-center study from The Netherlands of 191 subtotal cholecystectomies reported 175 laparoscopic subtotals with 45 conversions (25.7%).[85] A subgroup of STC patients (laparoscopic and converted) was compared to a contemporaneous group of 152 patients with converted total cholecystectomy. There was 1 BDI in each group. The STC group had significantly higher rates for CD leak (13% vs 2%), biliary event recurrence (15% vs 3%), readmissions (18% vs 8%), and re-interventions (34% vs 18%). The converted total cholecystectomy group had significantly more frequent wound infections (17% vs 6%) and increased LOS (median 5 vs 4 days).

A single health system study reported a 2:1 propensity-matched comparison of 130 patients with standard cholecystectomy to 65 patients with STC (laparoscopic, converted, and primary open).[82] Subtotal operations were associated with significantly more surgical site infections, re-interventions, and longer LOS.

Justification

The GDG reasoned that when inflammation prevents anatomic identification by the CVS or by imaging, avoiding dissection of ductal structures would have an anticipated large effect in reducing BDI. A STC approach that does not enter the hepatocystic triangle was considered preferable to pursuit of total cholecystectomy under these circumstances. The GDG unanimously agreed that the terms "subtotal cholecystectomy" and "top-down" are not mutually exclusive and, in fact, forms of STC can be performed with a dissection that proceeds from the region of the fundus.[82,86] Therefore, under challenging conditions, BDI risk can be minimized via either subtotal or top down cholecystectomy if dissection in the hepatocystic triangle is avoided.

Question 3: Should video documentation of the CVS (alone or in addition to operative notes) versus photo documentation (alone or in addition to operative notes) be used for limiting the risk or severity of BDI during LC?

No recommendation could be provided for this question due to a lack of agreement of the expert panel and concerns regarding feasibility, acceptability, and medico-legal considerations.

Summary of Evidence

No evidence was found that directly compared the impact of the various alternatives for CVS documentation requirements on outcomes of BDI and other complications of surgery. Indirect evidence on the proxy outcome of the quality of documentation of CVS consisted of 5 observational studies with 368 patients.[89–93] Accurate documentation of the CVS by description in operative notes is poor.[90–92] Photographs of the CVS from 2 views have been superior to photographs with 1 view.[93] IOC was judged superior to single view photographs of the CVS for documentation of biliary anatomy.[89] Video documentation has been reported to be superior to operative notes[92,93] and to CVS photographs with 2 views.[90] No correlation has been established with any clinical outcomes.

Justification

Because direct evidence for BDI was not found, the GDG considered indirect linked evidence.

The GDG opined that surgeons should strive to obtain the CVS during LC to minimize BDI risk (see recommendation to Question 1). The GDG felt that encouraging a more accurate method for documentation of the CVS might promote its wider use and enhance patient safety. The GDG agreed that the quality of CVS documentation, as judged by external independent assessors, is a relevant although uncertain proxy for BDI risk.

There was no consensus on this recommendation at the consensus conference. There was considerable discussion about medico-legal implications and push back from surgeons. Alternate wording that was considered included surgeons "be encouraged to document CVS by doublet photography or video." Another suggestion was that surgeons document in written reports how the anatomy was identified (because this is often omitted or unclear). The concept would not be to establish a mandate, but rather to increase over time use of doublet photography (or video documentation) in surgical practice. It was noted that the value of documentation as a teaching or quality assurance tool is evident and has previously been shown to improve the quality of healthcare delivered.[94] Ultimately, due to stakeholder concerns regarding acceptability, a recommendation was not made.

Question 4: Should intraoperative biliary imaging (eg, intraoperative cholangiography, ultrasound) versus no intraoperative biliary imaging be used for mitigating the risk of BDI during LC?

Recommendation: In patients with acute cholecystitis (AC) or a history of AC, we suggest the liberal use of IOC during LC to mitigate the risk of BDI (conditional recommendation, very low certainty of evidence). Surgeons with appropriate experience and training may use laparoscopic ultrasound (LUS) imaging as an alternative to IOC during LC.

In patients with uncertainty of biliary anatomy or suspicion of BDI during LC, we recommend that surgeons use intraoperative biliary imaging (in particular intraoperative cholangiography) to mitigate the risk of BDI. (strong recommendation, very low certainty of evidence).

Given that the evidence for the benefit of IOC in elective non-acute cholecystectomy is inconclusive, no recommendation addressing this scenario could be made.

Summary of Evidence

Evidence comparing IOC versus no intraoperative biliary imaging was considered in panel deliberations. No conclusive evidence was found for other types of biliary imaging. Randomized controlled trial (RCT) evidence (n = 9) for IOC was too underpowered for meaningful synthesis. Pooled evidence from 14 studies[8,9,95–106] that included 2.5 million patients demonstrated findings favoring IOC over no IOC in most of the studies providing adjusted estimates of effect, but the benefit was greatest in the subgroup analysis for patients with AC. For the subgroup of patients with intraoperatively suspected BDI, the use of IOC led to an almost 3-fold increase in the odds of recognition of BDI compared with non-use of IOC.

Narrative Synthesis

A systematic review of randomized trials that compared IOC to no IOC in patients undergoing cholecystectomy identified 8 studies (including one of open cholecystectomy) that met inclusion criteria (R-AMSTAR score 31.5).[107] Only 2 major BDIs were identified in 1715 patients and there were 2 avulsion injuries of the CD after IOC. Overall, the BDI rate was 0.2% and major BDI rate 0.1%. One additional randomized trial of 371 patients had 1 BDI in each group and was similarly inconclusive.[108] Because BDI is an uncommon event, further randomized trials would need an estimated minimum of 4500 patients per arm to show a difference in outcomes regarding use of IOC or other interventions and would not be practical to perform.[109]

The incidence of BDI during cholecystectomy with and without IOC was pooled from 14 large studies of mostly administrative data of 2,540,700 cholecystectomies.[8,9,95–106] The studies were at moderate to high risk of bias due to the nature of administrative coding data. Overall, the analysis showed a reduced incidence of BDI with the use of IOC [OR 0.78 (0.63–0.96), P < 0.0001]. When risk-adjusted, a similar though slightly lesser effect was observed (OR 0.81, 0.62–1.07). A 1987–2001 study from the Swedish inpatient registry reported 613 BDIs in 152,776 cholecystectomies and showed a 34% risk reduction in 94,569 patients in whom IOC was performed (multivariate analysis OR 0.66, 95% CI 0.54–0.79).[105] The GallRiks National Swedish registry for gallstones surgery analyzed outcomes in 51,041 cholecystectomies from 2005 to 2010.[8] A total of 74 BDIs, defined as any damage to the bile duct or biliary tree including bile leaks, were identified (1.5%). The use of IOC or intent to use IOC was associated with a reduced risk of BDI only in patients with AC (OR 0.44, 95% CI 0.30–0.63) or a history of AC (OR 0.59, 95% CI 0.35–1.00). No association between cholangiogram use and rate of BDI was found in patients who did not have AC or a history of AC (OR 1.06, 95% CI 0.75–1.49).

The rate of intraoperative recognition of injury was assessed by meta-analysis of 8 studies that included 1256 BDI's and compared IOC versus no IOC.[13,96,98,100,102,110–112] The use of IOC was associated with increased intraoperative recognition of BDI compared to no IOC (OR 2.92, 95% CI 1.55–5.68, P = 0.014).

Ultrasound has not been evaluated as extensively as IOC. Two systematic reviews found that LUS and IOC have similar success rates for visualization of the biliary anatomy.[54,113] One retrospective cohort study found 11 BDIs in 594 cases without LUS versus 0 in 248 cases with LUS (P = 0.04).[114] A prospective multicenter cohort study reported no BDI and only 3 bile leaks in 1381 patients.[115] These studies do not directly address LUS versus IOC and the incidence of BDI but do suggest that ultrasound in experienced hands has excellent results in delineating biliary anatomy.

Justification

Pooled evidence from 14 studies of >2.5 million patients demonstrated findings favoring IOC over no IOC in most studies reporting adjusted risk estimates, but the benefit was greatest in patients with AC. The GDG also deliberated on the alternative approach of using LUS imaging. No evidence was found to support or refute its use. In trained hands, ultrasound may be an appropriate alternative to IOC.

Multiple studies have shown that IOC is associated with a higher rate of intraoperative recognition of BDI. Although the certainty of evidence is very low, this is a consistent finding across studies. The GDG concluded that the subgroup of patients with uncertain anatomy or strong intraoperative suspicion of BDI will have a significantly higher baseline risk for BDI and at a higher risk for life-threatening complications of undetected/unrepaired BDI. For this subgroup of patients, the GDG invoked 1 of the 5 paradigmatic GRADE scenarios "Life-threatening situation" for strong recommendations based on "low or very low certainty evidence."[34] The potential benefit of IOC is early recognition and avoidance of potentially increasing the severity of BDI. This benefit includes avoidance of excision of a portion of the bile duct and a higher level of injury which is often more difficult to repair and reconstruct.

Question 5A: Should intraoperative near-infrared (NIR) biliary imaging versus IOC be used for limiting the risk or severity of BDI during LC?

No recommendation was made as current evidence comparing near infrared cholangiography for identification of biliary anatomy during cholecystectomy to IOC is insufficient.

Summary of Evidence

There was no direct evidence available for BDI or other outcomes for NIR versus IOC. The outcomes for ductal visualization were considered indirect evidence for BDI. Four observational studies provided data for the surrogate outcomes of visualization of CD, common bile duct (CBD), and hepatic duct.[116–119] Pooling of the data for visualization of each structure yielded inconsistent findings: RR for CD visualization was 1.1 (95% CI 1.00–1.35; 4 studies; total patients = 430). RR for CBD was 1.0 (95% CI 0.97–1.03; 4 studies; total patients = 430). RR for common hepatic duct (CHD) was 0.76 (95% CI 0.58–1.01, 4 studies; total patients = 300), P = NS.[120]

Narrative Synthesis

In the systematic review of mostly prospective cohort studies by Vlek et al[120] (R-AMSTAR-29), 19 studies were reviewed and data pooled from 4 observational studies for surrogate identification of the CD, CBD, and CHD as shown above. Most of the studies reviewed were highly subject to bias and did not compare ICG visualization to white light. There was considerable heterogeneity in the populations studied. For example, both complicated and uncomplicated gallstone disease was pooled and there were different definitions of complicated disease within the studies.

A review by Pesce et al (R-AMSTAR-27)[121] analyzed 16 studies involving ICG in both laparoscopic and robotic cholecystectomies. Detection rates of the biliary anatomy in 590 pooled patients were as follows: CD-96.2% (94.7%–97.7%); CHD-78.1% (74.8%–81.4%); CD-CHD junction-72.0% (69.0%–75.0%); and CBD-86.0% (83.3%–88.8%). No comparison was done with IOC or white light.

Question 5B: Should intraoperative NIR biliary imaging with white light versus white light biliary imaging alone be used for limiting the risk or severity of BDI during LC?

Recommendation: We suggest that the use of NIR imaging may be considered as an adjunct to white light alone for identification of biliary anatomy during cholecystectomy (conditional recommendation, very low certainty of evidence). The GDG noted that relying on NIR imaging must not be a substitute for good dissection and identification technique (expert opinion).

Summary of Evidence

There was no direct evidence for BDI or other decision-making outcomes. One RCT (N = 640) demonstrated findings favoring NIR for the visualization of various ductal structures (CD, CHD and the CBD) pre and post-dissection. These were considered proxy evidence for BDI.

Narrative Synthesis

The results of a single-blind multicenter, RCT of NIR and white light (N = 321) to white light alone (N = 318) were not available before the consensus conference but has recently been published.[122] The trial compared detection rates of 7 biliary structures: CD, right hepatic duct (RHD), CHD, and CBD, cystic-common bile duct junction, cystic-gallbladder junction (CGJ), and accessory ducts. Detection rates were statistically and significantly higher for the NIR group for each structure before dissection: OR ranges from 2.3 (95% CI 1.6–3.2) for the CGJ to 3.6 (95% CI 1.6–9.3) for the RHD. After dissection, similar differences were observed for all structures except the CD and CGJ, which were not significantly different.

Factors that affected detection rates were also assessed and body mass index (BMI), level of inflammation, patient age, and surgeon experience were found to be significant. Overall, BMI reduced visualization by 6% per BMI unit increase; however, visualization remained better with NIR compared to white light. The detection rate was approximately 40%–70% less among patients with moderate-to-severe versus minimal inflammation. Every increased year of patient age reduced detection of the RHD and CHD by 4% and 2% after dissection, and the CGJ by 1% both before and after dissection. Surgeon experience in years was directly proportional to detection rates as well. There were 2 "mild" BDIs out of 639 patients, both in the white light only group and resulted in conversion to open operation (P = 0.25). A limitation of the study is that surgeons were not blinded to the method of imaging and anatomic structure detection. Of note, a second randomized trial comparing NIR to white light is currently underway (FALCON trial, NCT 02558556).

Justification

The 1 RCT (N = 640 patients) demonstrated findings favoring NIR for the visualization of various ductal structures (CD, CHD, and the CBD) pre- and post-dissection. The risk of bias was rated as serious due to outcome ascertainment bias as the surgeons were nonblinded. The indirectness of the findings was rated very serious because of concerns about generalizability of the findings to general surgical practice and concerns about the surrogacy of ductal visualization for BDI. Importantly, the GDG noted that relying on NIR must not be a substitute for good dissection and identification technique.

Recommendations for Future Studies: NIR cholangiography should be assessed in large trials compared to white light alone and to IOC with risk stratification and risk-adjustment. Because BDI is unlikely to be captured in such trials, we additionally suggest these trials focus on proxy outcomes such as visualization of ductal structures, ability to obtain the CVS, and complications including conversion to open cholecystectomy. In particular, this technology should be studied in difficult cholecystectomy patient populations that include those with AC or a history of AC, severe chronic cholecystitis, and obese patients.

Question 6: Should surgical (complexity) risk stratification versus alternative or no risk stratification be used for mitigating the risk of BDI associated with LC?

Recommendations: A1: For patients with AC, we suggest that surgeons may use the Tokyo Guidelines 18 (TG18), American Association of Surgery for Trauma (AAST) classification, or another effective risk stratification model for grading for severity of cholecystitis and for patient management (expert opinion)

A2: During operative planning of LC and intraoperative decision-making, we suggest that surgeons consider factors that potentially increase the difficulty of LC such as male sex, increased age, chronic cholecystitis, obesity, liver cirrhosis, adhesions from previous abdominal surgery, emergency cholecystectomy, CD stones, enlarged liver, cancer of gallbladder, and/or biliary tract, anatomic variation, bilio-digestive fistula, and limited surgical experience. (expert opinion)

Summary of Evidence

No articles directly compared the incidence of BDI when a risk stratification system was used versus when a risk stratification system was not used. There are many prognostic risk association studies, but these do not evaluate the effectiveness of using this prognostic information to alter management and mitigate the risk of BDI. There is; however, evidence to indicate that the presence of AC increases the risk of mortality and BDI. In addition, grading/risk stratification systems for the severity of AC exist, including the TG 2013 and 2018[94,123,124] and the AAST severity grading.[125] Furthermore, there is evidence that the risk of BDI increases with the severity of inflammation as proposed by the TG 13/18.

Narrative Synthesis

The association between TG13 severity of cholecystitis and BDI was best demonstrated in a 2016 a case-control study by Tornqvist et al.[126] In this study, the TG13 risk stratification model was applied retrospectively to 158 patients who suffered a BDI during LC and 623 matched controls who underwent LC without BDI. The odds of BDI doubled among patients with AC (OR 1.97 95% CI 1.05–3.72) and the risk increased as the grade of inflammation increased. When compared to no cholecystitis, the odds of BDI increased from 0.96 (95% CI 0.41–2.25) in Tokyo grade I cholecystitis, to 2.41 (95% CI 1.21–4.80) in grade II, and 8.43 (95% CI 0.97–72.9) in grade III compared to no cholecystitis. The mortality was also increased with grade from 5.4% for grade III, 0.8% for grade II, and 1.2% for grade I cholecystitis (not risk-adjusted).

The TG are currently the only risk stratification model that risk stratifies and guides management of patients with AC based on the severity grade. Controversy does exist regarding the validity of the TG. A study by Hernandez et al compared the AAST versus TG13 severity grading systems for predicting postoperative outcomes in patients with AC and showed that the AAST criteria had better area under the curve (AUC) for predicting mortality (AUC 0.86 vs 0.73), complications (AUC 0.76 vs 0.63), and need for cholecystostomy tubes (AUC 0.80 vs 0.68), all P < 0.05.[125] Another study demonstrated that TG13 missed 35% of patients with gangrenous/AC on final pathology.[127]

A systematic review of factors that make LC difficult[56] included 91 articles with 3 meta-analyses of randomized trials, 5 controlled randomized trials, 8 well-designed controlled studies, 13 well-designed experimental studies, and 63 descriptive retrospective studies. There was a total of 324,553 patients. Factors associated with difficult LC included male sex, increased age, AC, chronic cholecystitis, obesity, liver cirrhosis, adhesions from previous abdominal surgery, emergency cholecystectomy, CD stone, large liver, big gallbladder, cancer of the gallbladder and/or biliary tract, anatomic variation, bilio-digestive fistula, and surgeon experience. The association of these risk factors with BDI was not specifically examined.

Justification

No evidence was found to support evidence-based recommendations. The GDG suggestions were; therefore, based on existing evidence of association. Given that risk stratification models demonstrate an association of more severe AC with higher rates of BDI, and complications and given the minimal risk of using these models, the GDG felt that guiding surgical intervention based on risk assessment could potentially mitigate the risk and severity of BDI.

Question 7: Should risk stratification that accounts for cholecystolithiasis versus no/alternate risk stratification be used for mitigating the risk of BDI associated with LC?

No recommendation was made as no risk prediction models exist that incorporate the presence or absence of gallstones as a factor that increases BDI or difficulty of LC.

No studies were identified that addressed this question specifically as regards to the primary outcome of BDI or secondary outcomes of conversion, complications, or mortality, or proxy outcome of quality of the CVS.

Question 8: Should immediate cholecystectomy defined as performed within 72 hours of symptom onset be used in AC versus delayed cholecystectomy? Delayed cholecystectomy is defined either as: (a) between 72 hours and 10 days after symptom onset; (b) 6–12 weeks after symptom onset; (c) greater than 12 weeks after symptom onset.

Recommendation: In patients presenting with mild AC (according to TG), we suggest surgeons perform LC within 72 hours of symptom onset (conditional recommendation, very low certainty of evidence). For patients with moderate and severe cholecystitis there is insufficient evidence to make a recommendation, particularly as it relates to the outcome of BDI.

Summary of Evidence

Data were analyzed from a systematic review that consisted of 14 randomized trials[128] and 1 large population study.[129] The included randomized studies did not include AC severity grading raising the possibility that the subgroup of patients with moderate and severe cholecystitis were underrepresented. Another large population study showed that grading of severity was critically important in evaluating the rate of BDI in cholecystectomy for AC.[126] In that study, patients with AC had a significantly higher BDI rate than patients without AC primarily due to a higher BDI risk in patients with moderate cholecystitis (TG grade 2) RR 0.53 (95% CI 0.31–0.90). Patients with mild AC (TG grade 1) did not have a higher incidence of BDI. In severe AC (TG grade 3) there was a nonsignificant trend towards an increased rate of injury. In patients who had 1 or more prior attacks there was a significantly higher incidence of BDI. Also, there is concern that the existing evidence may be limited in its generalizability by the fact that there is variability in how the diagnosis of AC is made, and there is reason to be concerned that the administrative diagnostic codes may have included patients with acute biliary pain but no acute inflammation.

Narrative Synthesis

The effect of the timing of cholecystectomy in patients with AC on the occurrence of BDI and other outcomes has been the subject of multiple studies that consist of 18 RCTs, more than 70 observational trials, 8 meta-analyses of randomized controlled trials, and a systematic review of the meta-analyses. Of the 8 meta-analyses reviewed, the one by Cao et al[128] (R-AMSTAR score 33/44) was selected for inclusion on the basis that it was recent, included almost all pertinent RCTs (14 in all), and the authors provided unpublished explanatory information. Of the multiple nonrandomized studies evaluated, only the population-based propensity score-matched study by de Mestral et al[129] was included. Several other timing studies not included in the GRADE analysis are discussed in Appendix 7, http://links.lww.com/SLA/B992.[130–134]

Serious methodological problems in regard to standardization and other issues resulted in a low level of certainty in regard to conclusions. Awareness of these methodologic pitfalls is essential to interpretation of results and conclusions and to guide future studies. These include:

Underpowering Due to Very Low Incidence Variables. A study adequately powered to detect a 3-fold drop in BDI from 0.3% to 0.1% at a 95% confidence limit would require 9000 patients, which is about 5 times more patients than are present in the largest available meta-analysis of RCTs. Mortality is a similarly affected variable. Only population-based studies containing many thousands of patients are likely to have the number of events required to study a problem with such a low event rate.

Failure to Demonstrate Parity in Pre-existing Health Status Between Early and Delayed Groups. Age, weight, and comorbidities can influence outcomes in patients undergoing cholecystectomy. These and other potential confounders have a potential biasing effect and are in need of parity in comparative studies. For instance, in a recent large population-based study, patients undergoing early cholecystectomy were younger, more frequently female, less likely to have biliary tract obstruction, and had a lower co-morbidity burden.[129] These biases may be correctable by propensity score matching as was done in this study.

Questionable Diagnosis of AC by Diagnostic Coding. The TG established the first standard diagnostic criteria for AC consisting of 1 symptom, 1 sign, and a confirming radiologic finding – the TG Triad. The 2013 version (TG13) of these diagnostic criteria were validated by comparison to pathologic findings and found to have sensitivity and specificity of 91.2% and 96.9%, respectively.[135] Many studies, especially observational studies use hospital discharge diagnosis of AC, which is based on admission diagnosis, operative report diagnosis, and pathologic report diagnosis. Coding may not be accurate when the admission and operative diagnoses are AC and the pathologic diagnosis is not. The frequency of use of the standard diagnostic triad in studies of AC by our analysis was adequate in RCTs (18 studies, of which 13 were evaluable and 10 used the triad),[128] but inadequate in many observational studies[133] (of 45 evaluable studies, 22 used the triad).

Inappropriate Inclusion of Cases Done During the Period of Maximal Inflammation in the "Delayed" Group. The debate over early versus delayed cholecystectomy is a question of whether it is better to operate in an "early" acute inflammatory phase after onset of symptoms, or to operate in a "delayed" manner after the acute inflammation has lessened. Although there is not universal agreement on the time intervals covered by the 2 terms, there is agreement that there is an unfavorable intermediate period during which cholecystectomy is more difficult due to the predictable severity of inflammation. The span of this unfavorable intermediate period is at a minimum between 10 days and 4 weeks. Studies that refer to surgeries done during that unfavorable time as "delayed" are not waiting for acute inflammation to subside, that is, they are not adequately "delayed" which biases the results in favor of early cholecystectomy.

Failure to Consider the Relationship Between Severity of AC and BDI. Until recently, severity grade has not been considered in any timing studies.[134] Tornqvist et al in 2016[126] studied the relation between severity of AC and BDI at cholecystectomy using the 2013 TG13 for diagnosis and severity grade of AC. The adjusted risk of BDI was doubled among patients with AC (OR 1.97, 95% CI 1.05–3.72) compared to controls without AC. Patients with mild AC (TG13 grade 1) had no increased risk of BDI (OR 0.96, 95% CI 0.41–2.25), but patients with moderate AC (TG grade 2) had more than twice the risk (OR 2.41, 95% CI 1.21–4.80). Patients with severe AC (TG13 grade 3) had a trend toward an even higher risk of AC (OR 8.43, 95% CI 0.97–72.9), but the number of patients was small. Patients with a prior history of attack(s) of AC had a significantly higher incidence of BDI also. This finding is consistent with multiple reports that prior attacks of AC are a risk factor for difficult operative conditions.[136,137]

Failure to consider severity grade in timing studies may result in confounding by inclusion of a higher proportion of patients with moderate cholecystitis in the delayed group because of a tendency to avoid early cholecystectomy when it is perceived that there is marked local inflammation. Also, a tendency to select patients with mild cholecystitis may result in studies which do not reflect accurately the influence of moderate inflammation on BDI.

Comparison of Outcomes Between Early and Delayed Cholecystectomy

Seven outcomes were selected for analysis with BDI as the major outcome of interest.

BDI. All meta-analyses of RCTs were underpowered. Therefore, the large, propensity score-matched population-based study of de Mestral et al[129] from the Province of Ontario, Canada was selected for analysis. It reported on 7110 propensity matched pairs of early and delayed cholecystectomy patients who were matched by patient and surgeon characteristics. Early was defined as within 7 days of hospital admission and delayed was defined as delayed beyond the early time frame (interquartile range 4–12 weeks). After matching, BDI was significantly less common in the early group [OR 0.53 (95% CI 0.31–0.90)]. The certainty of the evidence is very low as the diagnosis was made by diagnostic code and severity of inflammation was not considered. The lowest quartile of the delayed group was operated before 4 weeks in the intermediate period of least desirable inflammatory conditions.

Mortality. Mortality was evaluated in 8 RCTs in the meta-analysis of Cao et al (total of 1293 patients).[128] There were only 2 deaths, 1 in the early group and 1 in the late group. The relative risk was 1.03 (95% CI 0.05–20.50, P = NS). The certainty of the evidence was very low. Severity was not graded, there was a low number of events, and there was a moderate risk of bias because most of the larger studies included in the meta-analysis were given a low 3/5 score on the Jadad scale.[128] In 1 RCT by Gutt et al[138] comprising 618 of 1608 patients in the meta-analysis, the delayed group was operated between 1 and 7 weeks after presentation. Consequently, there is the possibility that many patients in the delayed group of this large RCT were operated before 4 weeks, that is, in the intermediate period of unfavorable inflammatory conditions. In the population-based study by de Mestral et al, the mortality rate in the early group was 0.46% and in the late group 0.64%.[129] The relative risk was 0.73 (95% CI 0.47–1.15, P = 0.21 NS).

Conversion to Open Cholecystectomy. This variable was evaluated in 12 RCTs in the meta-analysis of Cao et al[128] (total of 1452 patients). The conversion rate was 13.4% in the early group and 15.4% in the delayed group. The relative risk was 0.86 (95% CI 0.65–1.13), P = NS. The certainty of the evidence was also very low for this variable for the same reasons as above for mortality.

Patients With Complications. Nine RCTs in the meta-analysis of Cao with 1276 patients were available for study.[128] Complications occurred in 15% of patients in the early group (92/625) and 30% in the delayed group (192/643). The relative risk for the number of patients with complications in the early compared with the delayed group was 0.66 (95% CI 0.42–1.03, P = 0.07). In the 10 RCTs that reported total complications, there were 158 events in 630 patients in the early group and 273 events in 646 patients in the delayed group (P = 0.03). The certainty of the evidence was very low for this variable for the same reasons as above for mortality and conversion to open. There was also wide CI and suboptimal information size and results were inconsistent across trials included in the meta-analysis (I-squared = 50.82, P = 0.04).

Duration of Surgery. Duration of surgery was reported for 1276 patients in 10 RCTs in the Cao meta-analysis.[128] The mean duration of surgery was 99 minutes in the early group which was slightly but not significantly longer than in the delayed group (86 minutes, P = 0.23). The certainty of the evidence was very low.

Length of Total Hospitalization. In the meta-analysis, there were 11 RCTs with 1383 patients related to this variable.[128] Mean total length of hospital stay was 4.1 days in the early group versus 7.3 days in the delayed group (P < 0.001). The certainty of the evidence was low as severity was not graded and there was moderate risk of bias. The large treatment effect raised the certainty of evidence one level.

Wound Infection. In the meta-analysis of Cao et al[128] wound infections were reported in 8 RCTs containing 1145 patients. In the early group, the incidence of wound infection was 4.2%, and in the delayed group, it was 6.2%. The relative risk in early and delayed groups was 0.57 (95% CI 0.35–0.93, P = 0.02).

Re-presentation Before Scheduled Delayed Cholecystectomy. In the meta-analysis, 4 RCTs reported on re-presentation in the delayed group after initial successful nonoperative treatment.[128] Of patients treated nonoperatively initially, 18.5% re-presented before having delayed cholecystectomy. Eight studies reported the need for urgent surgery before the scheduled date in the delayed group. Emergency LC was needed earlier than the planned date in 9.7% of patients initially treated nonoperatively.

Justification

The evidence indicates that patients with mild AC do not have an increased rate of BDI whereas patients with moderate AC have double the incidence of BDI. However, because studies have not matched patient groups appropriately for the presence of moderate cholecystitis, it cannot be determined whether early or delayed cholecystectomy is superior in terms of the risk of BDI in that group of patients. Because BDI is the main outcome measure examined in the consensus conference and is a highly morbid, costly complication which necessitates invasive procedures and has a long-term effect on the quality of life of patients, it supersedes all other outcome measures in importance. Therefore, despite the superiority of early cholecystectomy in terms of total patient hospitalization, wound infection, and the need for unexpected care whereas waiting for cholecystectomy, the recommendation for early cholecystectomy can only be made in patients who have mild cholecystitis. The recommendation does not imply that delayed cholecystectomy is superior in patients with moderate severity AC. Further studies that incorporate the severity of the disease are needed to answer this question.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....