Liver International ISSN 1478-3223

ORIGINAL ARTICLE

Decompensated cirrhosis may be a risk factor for adverse events in endoscopic retrograde cholangiopancreatography Sumant Inamdar1, Tyler M. Berzin2, Joshua Berkowitz2, Divyesh V. Sejpal2, Mandeep S. Sawhney2, Ram Chutanni2, Douglas K. Pleskow2 and Arvind J. Trindade1 1 Division of Gastroenterology, Hofsta Northwell School of Medicine, Long Island Jewish Medical Center, Northwell Health System, New Hyde Park, NY, USA 2 The Center for Advanced Endoscopy, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA

DOI: 10.1111/liv.13100

Abstract Background and Aims: There are limited data regarding the safety of endoscopic retrograde cholangiopancreatography (ERCP) in cirrhosis. The current literature consists of small series totalling less than 225 patients. Methods: Retrospective matched cohort study of the National Inpatient Sample (NIS) for 2009. We compared adverse events of cirrhotic patients who underwent ERCP (n = 1930) with a matched control group that consisted of randomly selected non-cirrhotic patients who underwent ERCP (n = 5790). An additional control group, to measure cirrhosis-related adverse events, consisted of cirrhotic patients undergoing non-pancreaticobiliary endoscopy. Results: ERCP-associated adverse events of post-ERCP pancreatitis (PEP) (8.3% vs. 5.5%) and bleeding (2.3% vs. 1.0%) were more common in the cirrhosis cohort vs. the non-cirrhosis cohort (all P < 0.05). In subgroup analysis, compensated cirrhotic patients (n = 1308) had a similar adverse event profile to non-cirrhotic controls except for a slightly higher rate of PEP (7.7% vs. 5.5%; P < 0.05). However, decompensated cirrhotic patients (n = 622) had statistically significant higher rates of PEP (9.7% vs. 5.5%) and bleeding (4.3% vs. 1.0%), compared with non-cirrhotic controls respectively (P < 0.05). In regard to cirrhosis-related adverse events, cirrhotic patients undergoing ERCP were more likely to develop bacterial peritonitis vs. cirrhotic patients undergoing non-pancreaticobiliary endoscopy (2.2% vs. 1.1%; P < 0.005). Conclusion: ERCP adverse events were statistically higher among patients with decompensated cirrhosis. This increased risk needs to be confirmed with prospective studies. A thorough risk/benefit assessment should be performed prior to performing ERCP in decompensated cirrhotic patients. Keywords bile duct stones – pancreaticobiliary disease – safety

Endoscopic retrograde cholangiopancreatography (ERCP) has become the mainstay for the management of common bile duct disorders and common bile duct (CBD) obstruction (1). Like other invasive procedures, ERCP has inherent risks that are well documented in the literature. Adverse events associated with endoscopy in general include bleeding, infection, perforation and anaesthesia-related risks (2). Additional adverse events

specific to ERCP include post-ERCP pancreatitis (PEP), biliary infections and air embolism. Given the associated risks of ERCP, appropriate patient selection is of paramount importance. Patients with cirrhosis can be divided into those who have compensated or decompensated disease. Patients with decompensated cirrhosis may have varices, encephalopathy, ascites or jaundice. They usually have

Abbreviations AHRQ, Agency for Healthcare Research and Quality; CBD, common bile duct; ERCP, endoscopic retrograde cholangiopancreatography; HCUP, Healthcare Cost and Utilization Project; ICD9, International Classification of Diseases, Ninth Revision; MELD, Model for End-stage Liver Disease; PEP, post-ERCP pancreatitis. Correspondence Dr Arvind J. Trindade, Long Island Jewish Medical Center, Division of Gastroenterology 270-05 76th Avenue, New Hyde Park, NY 11040, USA Tel: +7184707983; Fax: +7184705509 e-mail: [email protected] Handling Editor: Espen Melum Received 25 August 2015; Accepted 18 February 2016

Liver International (2016) © 2016 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

1

Inamdar et al.

Adverse events of ERCP in cirrhosis

Key Points

• There are limited data on the safety of endoscopic retrograde cholangiopancreatography (ERCP) in patients with cirrhosis. • This is a large retrospective matched cohort study from a National Cohort in the United States of America. • Adverse events from ERCP were more common in the cirrhosis cohort vs. the non-cirrhosis cohort. In subgroup analysis, this was secondary to the decompensated cirrhosis subgroup. • ERCP adverse events were statistically higher among patients with decompensated cirrhosis. This increased risk needs to be confirmed with prospective studies. poor synthetic function of the liver, leading to an elevated coagulation profile or thrombocytopenia; both of which can contribute to bleeding. As a result, a cirrhotic patients undergoing ERCP may be predisposed to adverse events (2, 3). Surprisingly, there are limited data regarding the safety of ERCP among patients with cirrhosis. The current literature consists of less than 225 patients. Only two studies, with a total of 123 patients, examine the safety of ERCP in cirrhotic patients, categorized either by Childs-Pugh class or MELD score (4, 5). Li et al. evaluated 46 patients with cirrhosis and CBD stones who underwent ERCP and measured adverse events in comparison to a non-cirrhotic cohort of 100 patients. They found that patients with Child-Pugh class C cirrhosis had significantly more bleeding than patients with Child-Pugh class A or B (4). This study was limited in that there were only eight patients in the Child C class. Zhang et al. reported on 77 cirrhotic patients who underwent ERCP for CBD stones. Twentyseven percent of patients developed post-procedure adverse events including bleeding (18.2%), PEP (6.1%), infection (1%) and encephalopathy (1%). They found that 47% of patients with a Model for End-stage Liver Disease (MELD) score above 11.5 developed an adverse event (5). Given the lack of data in the literature on the safety of ERCP in patients with cirrhosis, we aimed to analyse a large national cohort of patients from a national database, to evaluate adverse events of ERCP in patients with cirrhosis. Given the available literature suggests increased adverse events in patients with more advanced manifestations of cirrhosis, we compared patients with compensated cirrhosis with those with decompensated disease. Methods Data source

The cohort used for this study was derived from the Nationwide Inpatient Sample (NIS) database for 2009.

2

The NIS is a publicly available nationwide inpatient database sponsored by the Agency for Healthcare Research and Quality (AHRQ) and is part of the Healthcare Cost and Utilization Project (HCUP). Each annual file of the NIS consists of approximately 8 million hospital stays from about 1000 hospitals in 45 states, representing an approximate 20% stratified sample of hospitals in the United States. Using the provided discharge weight files, national estimates may be calculated. The NIS data set includes information for each hospital admission with patient records including clinical information extracted from inpatient and discharge data such as demographic variables, discharge disposition, primary and secondary diagnoses (up to 25), primary and secondary procedures (up to 15), hospital characteristics and information on nearly 100 patient- and hospital-related variables. The NIS database is representative of the national population. It concurs with the National Hospital Discharge Survey and can be reliably used to represent the population of patients undergoing ERCP in the United States (6, 7). Study population

We selected cases and controls based on procedural coding in accordance with the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). These codes have been previously validated and have been shown to have good sensitivity and positive predictive value (8, 9). Our cases consisted of all adult patients (age ≥18 years) with cirrhosis (ICD 9: 571.xx) who were hospitalized and underwent inpatient ERCP (ICD 9 procedure codes: 51.10, 51.11, 52.13, 52.14, 52.21, 52.92, 52.93, 52.94, 52.97, 52.98, 51.14, 51.88, 51.87, 51.86, 51.85, 51.84) in 2009 from the NIS database. We selected two control groups for the study: (A) non-cirrhotic patients who underwent ERCP. This control group was chosen for comparison of adverse events related to ERCP. (B) Cirrhotic patients who underwent an endoscopic procedure (upper endoscopy, ICD 9 code: 45.13, 45.16 and colonoscopy, ICD 9 code: 45.23, 45.24, 45.25) other than ERCP for causes other than gastrointestinal bleeding. This group was chosen to compare adverse events related to cirrhosis after a procedure (e.g. encephalopathy, peritonitis, hepatorenal syndrome and bleeding). Cirrhotic patients who underwent endoscopy for gastrointestinal bleeding were excluded as they are more likely to have oesophageal variceal bleeding, peritonitis and hepatorenal syndrome (10). The controls were selected randomly by SAS 9.2 (SAS Institute Inc., Cary, NC, USA) using proc survey select, in which three controls were randomly selected for each case and matched for age, gender and year

Liver International (2016) © 2016 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

Inamdar et al.

Adverse events of ERCP in cirrhosis

All paents in NIS database who were admied in 2009 and underwent ERCP idenfied (n = 34 976)

All paents in NIS database who were admied in 2009 and had cirrhosis idenfied (n = 154 018)

IDENTIFICATION OF CASES Paents meeng the inclusion and exclusion criteria who had cirrhosis and underwent ERCP in 2009 idenfied (n = 1930)

Excluded paents who underwent gastrointesnal procedure for gastrointesnal bleeding SELECTION OF CONTROLS 3 non cirrhoc paents who underwent ERCP randomly selected as controls

3 cirrhoc paents who underwent other endoscopic procedures randomly selected as controls

Randomly selected non cirrhoc paents meeng the inclusion and exclusion criteria who underwent ERCP in 2009 (n = 5790)

Randomly selected cirrhoc paents meeng the inclusion and exclusion criteria who did not undergo ERCP, but other endoscopic procedures in 2009 (n = 5790)

For each cirrhoc paent who underwent ERCP, 3 non cirrhoc paents who underwent ERCP and 3 cirrhocpaents who did not undergo ERCP but other endoscopic procedures were randomly selected from the source populaon.

Fig. 1. Diagram showing the study design and selection of the study population and controls.

of study. The sampling design used for the study is shown in Figure 1.

Defining ERCP-related adverse events

We examined ERCP-related adverse events such as PEP, ERCP-related haemorrhage, perforation and cholecystitis. Non-procedure-related acute pancreatitis on admission was differentiated from PEP using a methodology used previously to derive PEP from a database (11, 12). Patients with pancreatitis as a part of complication of GI procedure are coded differently from non-procedure-related pancreatitis (ICD9: 997.4). In addition, these patients have a diagnosis of acute pancreatitis that is not a primary (dx1) or secondary diagnosis (dx2). Any patients who met these criteria and underwent an ERCP were classified as patients with post-ERCP pancreatitis (PEP). Patients with a primary diagnosis (dx1) or secondary diagnosis (dx2) as pancreatitis were classified as patients with acute pancreatitis not related to ERCP. ERCP-related haemorrhage was identified by using ICD 9 codes: 998.11, 909.3 and V58.89. We also included patients who did not have a primary diagnosis (dx1) or secondary diagnosis (dx2) of GI bleeding, who underwent ERCP and who were coded to have GI bleed as one of their diagnosis. Perforation after ERCP was defined as ICD 9: 569.83. Cholecystitis (ICD 9: 575.0 and 575.1) after ERCP was identified by using patients who did not have a primary diagnosis (dx1) or secondary diagnosis (dx2) of cholecystitis, underwent ERCP and were coded to have cholecystitis as one of their other diagnosis.

Outcome and predictor variables

Defining cirrhosis-related adverse events

We were interested in examining two separate outcomes: (i) ERCP-related adverse events among cirrhotic and non-cirrhotic patients; and (ii) cirrhosis-related adverse events among patients undergoing ERCP- and non-ERCP-related gastrointestinal procedures. We did not include mortality as an adverse event as patients with cirrhosis have higher inpatient mortality compared with patients without cirrhosis irrespective if a procedure is performed.

We examined cirrhosis-related adverse events such as procedure-associated haemorrhage, bacterial peritonitis, encephalopathy and hepatorenal syndrome. Procedureassociated haemorrhage was defined by using ICD 9 codes: 998.11, 909.3 and V58.89. Patients, who did not have a primary diagnosis (dx1) or secondary diagnosis (dx2) of bacterial peritonitis, underwent a gastrointestinal endoscopic procedure and were coded to have bacterial peritonitis (ICD 9 code: 567.xx) as one of their other diagnosis were identified as bacterial peritonitis as an adverse event. Similarly as described above, we identified patients with encephalopathy (ICD 9 code: 572.2) and hepatorenal syndrome (ICD 9 code: 572.4) as adverse events. We utilized the Elixhauser comorbidity index (13, 14) to adjust for comorbidities in the study. The NIS database contains 29 AHRQ comorbidity measures used by Elixhauser et al. (13). The Comorbidity Software supplied by HCUP was used for the creation and analysis of these comorbidities. This is a well-established comorbidity index used to predict in-patient mortality and has been validated in previous studies (15).

Defining compensated and decompensated cirrhosis

Patients with decompensated cirrhosis have symptomatic complications related to cirrhosis including those related to hepatic insufficiency and those related to portal hypertension. In our study, we considered patients to have decompensated cirrhosis if they had two or more signs of decompensation [varices (ICD9: 456.x), ascites (ICD9: 789.5), hypoalbuminemia (ICD9: 273.8), jaundice (ICD9: 782.4), thrombocytopenia (ICD9: 287.5.8), and hepatic encephalopathy (ICD9: 572.2)], which indicates that they would be equivalent to Child-Pugh class B/C. Liver International (2016) © 2016 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

3

Inamdar et al.

Adverse events of ERCP in cirrhosis

Statistical analyses

The SAS 9.2 statistical software was used to conduct all statistical tests. We used the proc survey commands and incorporated individual discharge-level weights to account for the stratified 2-stage cluster design of the database. In order to calculate nationwide data, we use the appropriate hospital and discharge-level weights published by the HCUP for weighted analysis. Univariate and bivariate analyses were performed to assess the indications and adverse events related to ERCP in cirrhotic patients. Chi-square (v2) test was used to compare categorical data and the Student’s t-test was used to compare continuous data. A P value of less than 0.05 was considered statistically significant. Ethical considerations

Since we used the NIS database for our study which consists of completely de-identified data, there was no risk of loss of confidentiality and the present study was

exempt from institutional review board review. Prior to using the NIS database, we completed a data user agreement with the AHRQ. As per the data user agreement, in order to protect patient confidentiality, we did not present any individual table cell counts of 10 or lower. In such instances, we used the abbreviation IS to indicate that the information was suppressed. Results Patient characteristic and demographics

We identified 1930 cirrhotic patients who underwent ERCP in 2009 in the NIS database, which yields a national estimate of 9603 cirrhotic patients on whom ERCP was performed in the United States in 2009. Table 1 shows the characteristics of cirrhotic patients and the non-cirrhotic controls who received ERCP and also elucidates the characteristics of cirrhotic patients who underwent ERCP (cases) and non-ERCP gastrointestinal procedures (controls). Cirrhotic patients who

Table 1. Patient characteristics among cirrhotic and non-cirrhotic patients who underwent ERCP and among cirrhotic patients who underwent ERCP and cirrhotic controls who did not undergo ERCP ERCP patients Cirrhotic (n = 1930) Demographic characteristics Age, mean (y) (95% CI) 0.05). Decompensated cirrhotic patients (4%) were significantly more likely to develop ERCP-associated haemorrhage (post-sphincterotomy bleeding) compared with compensated cirrhotic (1.3%) and non-cirrhotic patients (1%) (P < 0.001). There was no difference (P = 0.37) in ERCP-associated haemorrhage between compensated cirrhotic and noncirrhotic patients. Post-ERCP pancreatitis (PEP) Liver International (2016) © 2016 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

Adverse events of ERCP in cirrhosis

Table 2. Adverse events among cirrhotic patients undergoing ERCP Cirrhosis-related adverse events

ERCP (n = 1930) %

Non-ERCP (n = 5790) %

Procedureassociated haemorrhage Bacterial peritonitis Encephalopathy Hepatorenal syndrome ERCP-related adverse events

2.3

1.7

0.1201

2.2 2.1 0.8

1.1 2.7 0.8

0.0003 0.1326 0.8266

Cirrhotic (n = 1930) % 8.3

Non-cirrhotic (n = 5790) % 5.5

Decompensated cirrhosis may be a risk factor for adverse events in endoscopic retrograde cholangiopancreatography.

There are limited data regarding the safety of endoscopic retrograde cholangiopancreatography (ERCP) in cirrhosis. The current literature consists of ...
565B Sizes 0 Downloads 14 Views