Abstract
Background/Aims
Recently, the European Society of Gastrointestinal Endoscopy (ESGE) proposed criteria for “difficult biliary cannulation” during endoscopic retrograde cholangiopancreatography (ERCP). This study aimed to investigate the clinical relevance of the ESGE criteria from the perspective of post-ERCP pancreatitis (PEP).
Methods
An ERCP database was prospectively maintained between November 2014 and December 2015 across six teaching hospitals in South Korea. The ESGE criteria (biliary cannulation time, the number of cannulation attempts, and inadvertent pancreatic duct [PD] manipulation) were recorded in this database as well as other technical factors. Logistic regression analysis was used to identify risk factors for PEP. Then, the PEP prediction model was investigated using decision tree analysis.
Results
We analyzed 1,067 consecutive patients with naïve papilla. The overall rate of PEP was 6.6%. Multivariate analysis revealed that female sex (odds ratio [OR], 1.860; 95% confidence interval [CI], 1.124 to 3.078), a selective biliary cannulation duration >5 minutes (OR, 3.282; 95% CI, 1.641 to 6.566), and inadvertent PD manipulation (OR, 2.614; 95% CI, 1.480 to 4.617) were significant factors affecting PEP. Decision tree analysis revealed that biliary cannulation time (χ2=49.857, p<0.001) and inadvertent PD manipulation (χ2=8.556, p=0.010) were decisive factors. PEP occurred in 3.9%, 11.8%, and 16.2% of patients with biliary cannulation duration lasting 3 to 5 minutes, >5 minutes, and >5 minutes with inadvertent PD manipulation, respectively.
Conclusions
Biliary cannulation time and inadvertent PD manipulation could be relevant indicators of PEP, and 5 minutes might be used as a cutoff value for the implementation of the rescue cannulation technique. (Gut Liver 2021;15-465)
Keywords: Pancreatitis; Cholangiopancreatography, endoscopic retrograde; Risk factors; Decision trees; Practice guideline
INTRODUCTION
Endoscopic retrograde cholangiopancreatography (ERCP) is a technically challenging procedure for several pancreatobiliary disorders. Reportedly, selective biliary cannulation, which is a prerequisite for successful ERCP, was difficult in 5% of cases even when performed by experienced endoscopists, and this rate rose to approximately 18% of cases when performed across low-volume centers.1
Difficult biliary cannulation associated with repeated and prolonged attempts increases the risks of ERCP-related adverse events, particularly the risk of post-ERCP pancreatitis (PEP).2,3 Recently, the European Society of Gastrointestinal Endoscopy (ESGE) defined “difficult biliary cannulation” as any of the following factors: (1) more than five contacts with the papilla prior to successful cannulation; (2) more than 5 minutes of cannulation time following visualization of the papilla; and (3) more than one inadvertent pancreatic duct (PD) cannulation or opacification.4,5 However, the definition of “difficult biliary cannulation” varies across studies and thus interstudy comparisons are impractical. Therefore, the aim of this study was to investigate the clinical relevance of the ESGE’s criteria for difficult cannulation from the perspective of post-ERCP pancreatitis.
MATERIALS AND METHODS
ERCP database from six endoscopy centers in Daegu, South Korea (Kyungpook National University Chilgok Hospital, Kyungpook National University Hospital, Keimyung University Dongsan Medical Center, Yeungnam University Hospital, Daegu Catholic University Hospital, and Daegu Fatima Hospital) was prospectively maintained between November 2014 and December 2015. This database has been registered on the Clinical Research Information Service as clinical research (cris.nih.go.kr number, KCT0003606). All ERCP procedures performed with the intention of accessing the common bile duct to cannulate the naïve papilla were included in this study. However, patients with any of the following criteria were excluded from the study: pregnant women, age <18 years, underlying chronic pancreatitis, main PD dilatation measuring >5 mm, uncontrolled coagulopathy (international normalized ratio >1.5 or platelet count <50,000/mm3), or medically unstable cardiopulmonary disability for conscious sedation.
The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the Medicity Daegu Joint Institutional Review Board (IRB No. 2018-12-008). All authors had access to the study data and reviewed and approved the final manuscript.
1. Endoscopic procedure
ERCPs were performed at each participating hospital by experienced endoscopists who had performed >200 therapeutic ERCPs per year. Cannulation techniques (conventional catheter-first with subsequent sphincterotome or sphincterotome-first with/without a preloaded guidewire) and the types of devices used for therapeutic procedures (balloon or basket for stone removal, plastic or self-expandable metal stent [covered or uncovered] for biliary drainage) were left to the endoscopist’s discretion. Moreover, peri-procedural management such as the amount of hydration or type of fluid and prophylactic drug was also left to the strategy of participant hospitals. Endoscopist-directed-nurse administered sedation with midazolam, fentanyl, and/or propofol was used at all participating hospitals.
2. Procedural outcomes and definitions
The ERCP database was maintained using a standardized reporting form at all six participating hospitals. Indications, the cannulation technique, therapeutic interventions, results, and adverse events related to the procedure were recorded in all cases. The following data were recorded upon completion of each ERCP: total procedure time, biliary cannulation time, cannulation attempts, contact with papilla, inadvertent PD catheter insertion, PD contrast injection, and PD guidewire passage. Procedure-related adverse events such as PEP, perforation, bleeding, or others were also recorded based on the definitions proposed by Cotton et al.,6,7 if these occurred within the index hospitalization period. PEP was defined as the occurrence of new-onset abdominal pain that was associated with an increase in serum amylase at least three times the upper limit of normal at the next day after the procedure.6 ERCP-related bleeding was categorized into three subtypes: intraprocedural, immediate, and delayed bleeding. Intraprocedural bleeding was defined when it occurred during the procedure and additional manipulation was required for hemostasis. Immediate bleeding was defined when a hemoglobin drop of at least 2 g/dL was observed within 24 hours after the procedure and delayed bleeding was defined when it occurred after 24 hours later. Biliary cannulation time was defined as the duration between the first visualization of the papilla and deep cannulation. Cannulation time was recorded using a built-in stop-watch in the endoscopy system. Cannulation contact was defined as each continuous contact with the papilla with the intention of deep biliary cannulation. Repositioning following a loss of contact with the papilla was defined as a fresh attempt, and the number of cannulation attempts was recorded. In case of inadvertent PD catheter and contrast agent or guidewire passage into the PD occurred, the number of these events was also recorded. Furthermore, inadvertent PD catheter insertion, PD contrast injection, and PD guidewire passage were categorized as inadvertent PD manipulation because 2016 ESGE guidelines do not distinguish between these criteria in detail.
3. Statistical analysis
The independent t-test or the Mann-Whitney U-test was used for continuous variables, and the chi-square or the Fisher exact test was used for categorical variables. Univariate and multivariate analyses with logistic regression were performed to evaluate patient- and procedure-related variables to determine the independent risk factors associated with PEP. All variables showing a p-value <0.05 in the univariate analysis were subjected to multivariate analysis. The results were expressed as odds ratios with 95% confidence intervals. All p-values were two-sided, and a p-value <0.05 was considered statistically significant. All statistical analyses were performed using the SPSS statistics software, version 20.0 (IBM Corp., Armonk, NY, USA).
Furthermore, decision tree analysis with the Chi-squared Automatic Interaction Detector algorithm was conducted for the hierarchical evaluation of the risk factors of PEP including the criteria proposed by ESGE. In addition to previously known risk factors for PEP including female sex, a history of precut sphincterotomy, and PD contrast injection, other factors that were observed to be significant on univariate analysis were included in the decision tree analysis.
RESULTS
1. Baseline characteristics and technical outcomes
Table 1 summarizes the baseline, clinical, and procedural characteristics (clinical and technical aspects) of the 1,067 consecutive patients with naïve papilla who were included in this study. Common bile duct stones (73.0%) were the most common indication for ERCP in our study. The second was malignant biliary stricture (238 patients, 22.1%), including cholangiocarcinoma (157 patients), pancreatic cancer (38 patients), gallbladder cancer (25 patients), ampulla of Vater cancer (14 patients), and bile duct metastasis (four patients). Selective bile duct cannulation at the first attempt succeeded in 761 patients (70.4%). The overall success rate for biliary cannulation was 97.4% (1,040 patients; median time, 5 minutes; range, 1 to 65 minutes). A periampullary diverticulum was identified in 29.5% of the patients. Precut sphincterotomy (5.6%) or infundibulotomy (7.6%) was performed as a rescue technique for selective bile duct cannulation. Procedure-related adverse events including PEP, post-ERCP bleeding, perforation, and cholangitis occurred in 70 (6.6%), 120 (11.2%), seven (0.6%), and 14 (1.3%) patients, respectively. The grade of PEP was mild in 61 (87.1%), moderate in seven (10.0%), and severe in two (2.9%). Three subtypes of post-ERCP bleeding was as follows: intraprocedural bleeding was occurred in 114 (95%), immediate in three (2.5%), and delayed in three (2.5%).
Table 1.
Baseline Characteristics of All Enrolled Patients
| Characteristic | Value (n=1,067) |
|---|---|
| Age, yr | 70 (19–101) |
| Female sex | 451 (42.3) |
| Indication for ERCP | |
| Choledocholithiasis | 790 (73.0) |
| Malignant biliary stricture | 238 (22.1) |
| Benign biliary stricture | 20 (1.9) |
| Post-cholecystectomy bile leakage | 11 (1.0) |
| Others | 10 (0.9) |
| Presence of periampullary diverticulum | 315 (29.5) |
| Type 1 | 46 (14.6) |
| Type 2 | 142 (45.1) |
| Type 3 | 127 (40.3) |
| Procedure time, min | |
| Overall | 16 (3–85) |
| Biliary cannulation | 5 (1–65) |
| Bile duct access | |
| Successful first attempt | 761 (70.4) |
| Failed biliary cannulation | 24 (2.2) |
| Contacts with papilla | |
| None | 597 (56.0) |
| <5 times | 357 (33.5) |
| 5–10 times | 78 (7.3) |
| ≥10 times | 35 (3.3) |
| Inadvertent PD catheter insertion | |
| None | 860 (80.6) |
| <5 times | 169 (15.8) |
| 5–10 times | 24 (2.2) |
| ≥10 times | 14 (1.3) |
| Inadvertent PD contrast injection | |
| None | 909 (85.2) |
| <5 times | 149 (14.0) |
| 5–10 times | 9 (0.8) |
| ≥10 times | 0 |
| Inadvertent PD guidewire passage | |
| None | 767 (71.9) |
| <5 times | 258 (24.2) |
| 5–10 times | 30 (2.8) |
| ≥10 times | 12 (1.1) |
| Precut sphincterotomy | 60 (5.6) |
| Infundibulotomy | 81 (7.6) |
| Endoscopic papillary balloon dilation | 131 (12.3) |
| Normal sphincterotomy | 846 (79.3) |
| Prophylactic drugs | 784 (73.5) |
| Nafamostat mesylate | 362 (33.9) |
| Ulinastin | 228 (21.4) |
| Gabexate mesylate | 194 (18.2) |
| Post-ERCP adverse events | 190 (17.8) |
| Pancreatitis | 70 (6.6) |
| Mild | 61 (87.1) |
| Moderate | 7 (10.0) |
| Severe | 2 (2.9) |
| Bleeding | 120 (11.2) |
| Intraprocedural | 114 (95.0) |
| Immediate | 3 (2.5) |
| Delayed | 3 (2.5) |
| Perforation | 7 (0.6) |
| Cholangitis | 14 (1.3) |
Data are presented as median (range) or number (%).
ERCP, endoscopic retrograde cholangiopancreatography; PD, pancreatic duct.
2. Risk factors for post-ERCP pancreatitis
According to PEP development, the enrolled patients were categorized into two groups (PEP and non-PEP group). Female sex, overall procedure time (>10 minutes), biliary cannulation time (>5 minutes), the number of contacts with the papilla (>5 times), inadvertent PD manipulation (PD contrast injection, guidewire passage, and/or cannulation), and PD stent placement were statistically significant on univariate analysis. However, overall procedure time (>10 minutes), the number of contacts with the papilla (>5 times) and PD stent placement turned out to be insignificant on multivariate analysis (Table 2).
Table 2.
Univariate and Multivariate Analyses of Risk Factors for Post-ERCP Pancreatitis (n=1,067)
| Variable | Univariate analysis | Multivariate analysis | |||||
|---|---|---|---|---|---|---|---|
| With PEP (n=70) |
Without PEP (n=997) | p-value | Odds ratio | 95% CI | p-value | ||
| Age (≤70 yr) | 39 (55.7) | 499 (50.1) | 0.364 | ||||
| Female sex | 39 (55.7) | 412 (41.3) | 0.018 | 1.860 | 1.124–3.078 | 0.016 | |
| Periampullary diverticulum | 18 (25.7) | 297 (29.8) | 0.470 | ||||
| Failed BD access | 3 (4.3) | 21 (2.1) | 0.220 | ||||
| Overall procedure time (>10 min) | 66 (94.3) | 755 (75.7) | <0.001 | 1.916 | 0.615–5.973 | 0.262 | |
| BD cannulation (>5 min) | 56 (80.0) | 418 (41.9) | <0.001 | 3.282 | 1.641–6.566 | 0.001 | |
| Contact with papilla (>5 times) | 14 (20.0) | 99 (9.9) | 0.008 | 0.862 | 0.441–1.684 | 0.664 | |
| Inadvertent PD manipulation | 46 (65.7) | 322 (32.3) | <0.001 | 2.614 | 1.480–4.617 | 0.001 | |
| PD catheter insertion (≥1 attempt) | 26 (37.1) | 181 (18.2) | <0.001 | ||||
| PD contrast injection (≥1 event) | 19 (27.1) | 139 (13.9) | 0.003 | ||||
| PD guidewire passage (≥1 event) | 40 (57.1) | 260 (26.1) | <0.001 | ||||
| Precut sphincterotomy | 5 (7.1) | 55 (5.5) | 0.568 | ||||
| EPBD | 4 (5.7) | 127 (12.7) | 0.083 | ||||
| Mechanical lithotripsy | 4 (5.7) | 63 (6.3) | 0.840 | ||||
| ERPD | 9 (13.0) | 54 (5.4) | 0.010 | 0.963 | 0.430–2.153 | 0.926 | |
| Prophylactic drug | 59 (84.3) | 725 (72.7) | 0.034 | 1.463 | 0.735–2.912 | 0.279 | |
Data are presented as number (%).
ERCP, endoscopic retrograde cholangiopancreatography; PEP, post-ERCP pancreatitis; CI, confidence interval; BD, bile duct; PD, pancreatic duct; EPBD, endoscopic papillary balloon dilation; ERPD, endoscopic retrograde pancreatic drainage.
3. Decision tree analysis for PEP prediction model
Decision tree analysis showed that the biliary cannulation time (χ2=49.857, p<0.001) and inadvertent PD manipulation (χ2=8.556, p=0.010) were significantly associated with PEP. However, the number of contacts with the papilla was not significant for discrimination. Moreover, the final decision tree model showed that the biliary cannulation time was the most significant factor (Fig. 1).
Fig. 1.
A decision tree model for the prediction of post-ERCP pancreatitis generated by classification and regression tree analysis using our ERCP database.
ERCP, endoscopic retrograde cholangiopancreatography; PEP, post-ERCP pancreatitis.
We identified that 3 and 5 minutes of biliary cannulation time were the cutoff values for the development of PEP (3.9% vs 11.8%, respectively). Furthermore, if inadvertent PD manipulation occurred in a patient with biliary cannulation time of >5 minutes, the risk of PEP increased to 16.2% (Fig. 1). However, for the subgroup of moderate-to-severe grade PEP, inadvertent PD manipulation was the only significant factor with a 1.9% risk of moderate-to-severe PEP in the decision tree model (Fig. 2).
Fig. 2.
A decision tree model for the prediction of post-ERCP pancreatitis in the moderate-to-severe grade subgroup, generated by classification and regression tree analysis using our ERCP database.
ERCP, endoscopic retrograde cholangiopancreatography; PEP, post-ERCP pancreatitis.
DISCUSSION
This study revealed that the biliary cannulation time and inadvertent PD manipulation were significant factors for PEP development among the 3 ESGE’s criteria for difficult biliary cannulation (more than 5 minutes of cannulation time, more than five contacts with the papilla, and/or more than one inadvertent PD cannulation/opacification).4,5 Furthermore, decision tree analysis showed that biliary cannulation time is the most relevant parameter of cannulation difficulty from the perspective of PEP development. PEP occurred in 11.8% of all cases when the biliary cannulation time was >5 minutes. Moreover, the risk of PEP increased to 16.3% in patients in whom inadvertent PD manipulation occurred when the biliary cannulation time was >5 minutes. However, the number of contacts with the papilla was not a significant factor for PEP in multivariate analysis and decision tree analysis. This result may be explained by the fact that the number of contacts with the papilla plays the role of a covariate associated with the biliary cannulation time. This finding is also supported by the study of Tian et al.8 which demonstrated that there was a positive correlation between cannulation time and the number of attempts, and cannulation time was a more objective and accurate assessment tool for cannulation difficulty. Furthermore, the low number of contacts with papilla does not necessarily mean easy cannulation. Therefore, among the 3 ESGE’s criteria for difficult cannulation proposed in 2016, the biliary cannulation time and inadvertent PD manipulation would be indicators of PEP development; and then 5 minutes could be used as a cutoff value for the implementation of rescue cannulation technique.
Recently, Mariani et al.9 reported that early precut sphincterotomy (in which rescue needle-knife sphincterotomy was performed whenever bile duct cannulation failed after 5 minutes or after three passes of inadvertent PD cannulation), reduced the incidence of PEP. This finding suggests that precut sphincterotomy by itself is not a risk factor, but prolonged cannulation might be a more significant risk factor for PEP. These findings were in agreement with our study outcomes for reducing the cannulation time to prevent PEP. When decision tree analysis was performed only for moderate-to-severe PEP, inadvertent PD manipulation was the only factor that showed statistical significance, which might indicate that instrumentation and manipulation-induced mechanical injury to the PD is the most important pathophysiological contributor to moderate-to-severe PEP rather than electrosurgery-induced thermal injury associated with sphincterotomy. Furthermore, two recent meta-analyses of randomized controlled trials have also demonstrated the benefit of early precut sphincterotomy to prevent PEP.10,11 Therefore, endoscopists should try to reduce the time of selective biliary cannulation as short as possible and, if not, rescue techniques such as precut sphincterotomy or double guidewire technique should be considered in an early phase of the procedure in the case of difficult biliary cannulation.
Considering the association between inadvertent PD manipulation and PEP, which was also demonstrated in this study, the insertion of a prophylactic indwelling PD stent is a reasonable recommendation. However, pancreatic stent and drugs for PEP prophylaxis were ineffective in this study. This finding might be explained by the observational study design in that certain patients with a seemingly high probability for PEP might be more frequently treated with prophylactic medications or PD stent based on personal criteria.
The limitations of this study are as follows: (1) this study was designed as an observational study although the data were collected prospectively. However, we used a relatively large multicenter ERCP database containing detailed information and technical data regarding procedures performed by highly qualified and experienced endoscopists. (2) The method of and devices used for biliary cannulation as well as the strategy of peri-procedural management such as the amount of hydration or type of fluid and prophylactic drug were not standardized across all participating hospitals and operators. However, all endoscopists who performed the procedures included in this study were highly experienced, which was reflected by the ERCP-related technical outcomes of this study (the success rate of biliary cannulation was 97.5%, and the median biliary cannulation time was 5 minutes). Therefore, the variability in cannulation techniques and devices might not have significantly affected our study results.
Nonetheless, the strength of our study is that multicenter ERCP registry from high-volume endoscopy centers was used for the evaluation of difficult cannulation criteria. Thus, the results accurately reflect in detail the technical aspects of the procedure performed by experienced endoscopists. Furthermore, the use of decision tree analysis to identify the risk factors for PEP provided a clear understanding of the hierarchy of the related variables.12-14 Decision tree analysis has been frequently used in many medical types of research because it provides an optimal prediction model with identifying relevant factors for primary outcomes. Furthermore, the first discriminating variable of final decision tree model is considered to have the strongest statistical association with the primary outcome variable.15,16
In conclusion, among the 3 ESGE’s criteria for difficult cannulation, biliary cannulation time and inadvertent PD manipulation would be indicators of PEP development. Furthermore, the biliary cannulation time was the most significant factor for PEP. Therefore, the cannulation time should be as short as possible for the prevention of PEP development; and then 5 minutes might be used as a cutoff value for the implementation of rescue cannulation technique.
ACKNOWLEDGEMENTS
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP; Ministry of Science, ICT & Future Planning) (number: 2017R1C1B5018180).
Footnotes
CONFLICTS OF INTEREST
C.M.C. is an editorial board member of the journal but did not involve in the peer reviewer selection, evaluation, or decision process of this article. No other potential conflicts of interest relevant to this article were reported.
AUTHOR CONTRIBUTIONS
Study conception and design: Y.S.L., C.M.C. Data collection: Y.S.L., C.M.C., K.B.C., J.H., M.K.J., S.B.K., K.H.K., D.W.L., J.H., D.K., H.K. Analysis and interpretation of the data: Y.S.L., C.M.C. Drafting of the article: Y.S.L., C.M.C. Critical revision of the article for important intellectual content: C.M.C., K.B.C., D.K., H.K., T.N.K., H.G.K. Final approval of the article: all authors.
REFERENCES
- 1.Williams EJ, Taylor S, Fairclough P, et al. Are we meeting the standards set for endoscopy? Results of a large-scale prospective survey of endoscopic retrograde cholangio-pancreatograph practice. Gut. 2007;56:821–829. doi: 10.1136/gut.2006.097543corr1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Halttunen J, Meisner S, Aabakken L, et al. Difficult cannulation as defined by a prospective study of the Scandinavian Association for Digestive Endoscopy (SADE) in 907 ERCPs. Scand J Gastroenterol. 2014;49:752–758. doi: 10.3109/00365521.2014.894120. [DOI] [PubMed] [Google Scholar]
- 3.Freeman ML, DiSario JA, Nelson DB, et al. Risk factors for post-ERCP pancreatitis: a prospective, multicenter study. Gastrointest Endosc. 2001;54:425–434. doi: 10.1067/mge.2001.117550. [DOI] [PubMed] [Google Scholar]
- 4.Dumonceau JM, Andriulli A, Elmunzer BJ, et al. Prophylaxis of post-ERCP pancreatitis: European Society of Gastrointestinal Endoscopy (ESGE) guideline: updated June 2014. Endoscopy. 2014;46:799–815. doi: 10.1055/s-0034-1377875. [DOI] [PubMed] [Google Scholar]
- 5.Testoni PA, Mariani A, Aabakken L, et al. Papillary cannulation and sphincterotomy techniques at ERCP: European Society of Gastrointestinal Endoscopy (ESGE) clinical guideline. Endoscopy. 2016;48:657–683. doi: 10.1055/s-0042-108641. [DOI] [PubMed] [Google Scholar]
- 6.Cotton PB, Lehman G, Vennes J, et al. Endoscopic sphincterotomy complications and their management: an attempt at consensus. Gastrointest Endosc. 1991;37:383–393. doi: 10.1016/s0016-5107(91)70740-2. [DOI] [PubMed] [Google Scholar]
- 7.Cotton PB, Garrow DA, Gallagher J, Romagnuolo J. Risk factors for complications after ERCP: a multivariate analysis of 11,497 procedures over 12 years. Gastrointest Endosc. 2009;70:80–88. doi: 10.1016/j.gie.2008.10.039. [DOI] [PubMed] [Google Scholar]
- 8.Tian C, Gamboa A, Chaudhury B, Willingham FF, Keilin S, Cai Q. Cannulation time is a more accurate measure of cannulation difficulty in endoscopic retrograde cholangiopancreatography than the number of attempts. Gastroenterol Rep (Oxf) 2013;1:193–197. doi: 10.1093/gastro/got024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Mariani A, Di Leo M, Giardullo N, et al. Early precut sphincterotomy for difficult biliary access to reduce post-ERCP pancreatitis: a randomized trial. Endoscopy. 2016;48:530–535. doi: 10.1055/s-0042-102250. [DOI] [PubMed] [Google Scholar]
- 10.Sundaralingam P, Masson P, Bourke MJ. Early precut sphincterotomy does not increase risk during endoscopic retrograde cholangiopancreatography in patients with difficult biliary access: a meta-analysis of randomized controlled trials. Clin Gastroenterol Hepatol. 2015;13:1722–1729. doi: 10.1016/j.cgh.2015.06.035. [DOI] [PubMed] [Google Scholar]
- 11.Tang Z, Yang Y, Yang Z, Meng W, Li X. Early precut sphincterotomy does not increase the risk of adverse events for patients with difficult biliary access: a systematic review of randomized clinical trials with meta-analysis and trial sequential analysis. Medicine (Baltimore) 2018;97:e12213. doi: 10.1097/MD.0000000000012213. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hess KR, Abbruzzese MC, Lenzi R, Raber MN, Abbruzzese JL. Classification and regression tree analysis of 1000 consecutive patients with unknown primary carcinoma. Clin Cancer Res. 1999;5:3403–3410. [PubMed] [Google Scholar]
- 13.Garzotto M, Beer TM, Hudson RG, et al. Improved detection of prostate cancer using classification and regression tree analysis. J Clin Oncol. 2005;23:4322–4329. doi: 10.1200/JCO.2005.11.136. [DOI] [PubMed] [Google Scholar]
- 14.Austin PC. A comparison of regression trees, logistic regression, generalized additive models, and multivariate adaptive regression splines for predicting AMI mortality. Stat Med. 2007;26:2937–2957. doi: 10.1002/sim.2770. [DOI] [PubMed] [Google Scholar]
- 15.Rokach L, Maimon O. Data mining with decision trees: theory and applications. World Scientific Publishing; Hackensack: 2014. [DOI] [Google Scholar]
- 16.De Melo NB, Bernardino ÍM, de Melo DP, Gomes DQC, Bento PM. Head and neck cancer, quality of life, and determinant factors: a novel approach using decision tree analysis. Oral Surg Oral Med Oral Pathol Oral Radiol. 2018;126:486–493. doi: 10.1016/j.oooo.2018.07.055. [DOI] [PubMed] [Google Scholar]