Risk factors for pancreatitis following endoscopic retrograde cholangiopancreatography

The aim of this study was to assess whether sex, age, ASA grade, previous history of acute pancreatitis, diabetes, hyperlipidaemia, hypercalcaemia, kidney disease and liver cirrhosis influence the risk for developing post‐endoscopic retrograde cholangiopancreatography (ERCP) pancreatitis (PEP). A total of 15 800 ERCP procedures retrieved from the Swedish National Quality Register for Gallstone Surgery and ERCP (GallRiks) for 2006–2014 were identified and cross‐checked with the National Patient Register. Women, patients aged less than 65 years, patients with hyperlipidaemia and those with a previous history of acute pancreatitis had a significantly increased risk of PEP, whereas patients with diabetes had a significantly decreased risk. Risk of pancreatitis following ERCP


Introduction
One of the most feared complications described after endoscopic retrograde cholangiopancreatography (ERCP) is post-ERCP pancreatitis (PEP), which occurs with an incidence of 3⋅5-5 per cent 1,2 . PEP is defined 3 as 'clinical pancreatitis with amylase at least three times the upper limit of normal at more than 24 h after the procedure requiring hospital admission or prolongation of planned admission', whereas its severity has been based mainly on the length of hospital stay.
The risk of developing PEP can be assessed in relation to several variables, including technical factors (manipulation and injection of contrast into the pancreatic duct, cannulation attempts lasting more than 5 min, and biliary balloon sphincter dilatation) and patient-related factors such as female sex, younger age, sphincter of Oddi dysfunction 2 -5 and a previous history of PEP or pancreatitis 6 . The most common causes of acute pancreatitis are biliary stone and alcohol abuse. However, other conditions, including long-term haemodialysis or peritoneal dialysis, are associated with an increased risk 7,8 , and co-morbidities such as peptic ulcer, hepatic disease and diabetes are frequently described 9 .
In particular, patients with type 2 diabetes have a 1⋅91-fold increased risk of developing biliary disease and a 2⋅83-fold increased risk of pancreatitis 10 . An increased risk of pancreatitis has also been shown to be associated with younger age and the presence of hypertriglyceridaemia 11 , and a reduced risk associated with the use of insulin and long-term use of metformin in diabetic patients 12 . Finally, patients with more advanced cirrhosis (Child-Pugh grade B and C) have a higher incidence of ERCP complications than those with Child-Pugh grade A 13 , and an increased risk of postprocedure bleeding, although not of PEP 14 .
The aim of the present study was to investigate the risk of PEP in patients with diabetes, liver cirrhosis, hyperlipidaemia, hypercalcaemia and kidney disease.   GallRiks is regularly externally validated, and the validation process and its national coverage results are published each year 15 -17 . Records include patient-and procedure-related data as well as intraoperative and postoperative complications up to 30 days after ERCP. For the present study, all ERCP procedures registered in GallRiks between 2006 and 2014 for bile duct stones were included. ERCPs conducted for other indications, repeated ERCP (in the same patient) and ERCPs with missing follow-up data were excluded.
PEP was defined as typical abdominal pain, a serum amylase level more than three times the upper limit of normal more than 24 h after ERCP, and the need for hospitalization 3 .
Data on chronic disease (diabetes, liver cirrhosis, hyperlipidaemia, hypercalcaemia and kidney disease) and previous episodes of acute pancreatitis were obtained by cross-checking GallRiks data with that in the National Patient Register using ICD codes ( Table 1).
The Regional Ethics Review Board in Stockholm approved the study (reference number 2015/339-31/1).

Statistical analysis
Univariable and multivariable logistic regression analyses with the endpoint of PEP were performed. In the multivariable analyses, adjustment was made for sex and age (at least 65 years versus less than 65 years). Adjustments in the multivariable analysis were made based on assumptions of cause-effect relationships.
A subgroup analysis was conducted in patients with a previous history of pancreatitis. The mean(s.d.) time between the previous episode of pancreatitis and ERCP was determined and compared in patients who developed PEP following ERCP and those who did not have this complication, using Student's t test. Statistical analysis was performed with SPSS ® version 25 (IBM, Armonk, New York, USA).

Results
Some 15 800 of 57 492 ERCP procedures carried out between 2006 and 2014 that met the study design criteria were analysed (Fig. 1). Patient characteristics and risk factors for PEP are shown in Table 3. Table 2 shows the results of univariable and multivariable analyses with the endpoint of PEP. Univariable analysis found a significantly greater risk of PEP in women (odds ratio (OR) 1⋅33, 95 per cent c.i. 1⋅14 to 1⋅55), patients aged less than 65 years (OR 1⋅68, 1⋅45 to 1⋅94) and those with a previous history of acute pancreatitis (OR 5⋅26, 4⋅53 to 6⋅10). Patients with diabetes had a lower risk of PEP (OR 0⋅55, 0⋅42 to 0⋅72). In multivariable analysis, after adjustment for age and sex, a previous history of acute pancreatitis (OR 5⋅44, 4⋅68 to 6⋅31) and hyperlipidaemia (OR 1⋅32, 1⋅02 to 1⋅70) were found to increase the risk of PEP, whereas diabetes decreased the risk (OR 0⋅64, 0⋅48 to 0⋅85).
In a subgroup analysis of 2567 patients with a previous history of acute pancreatitis, the mean(s.d.) time from the previous episode of pancreatitis to ERCP was 4423(5262) days in patients who developed PEP versus 6990(5071) days in those who did not develop PEP (P = 0⋅037). However, when the previous episode of pancreatitis had occurred more than 30 days before ERCP, this association was no longer significant. In that group, the mean time from pancreatitis to ERCP was 7772(4747) days in patients who did not develop PEP and 7727(4781) days in those who did (P = 0⋅858).

Discussion
This national register-based analysis found that women, patients aged less than 65 years and those with a previous history of acute pancreatitis had a significantly greater risk of PEP, as documented previously by other authors 2 -6 . However, as it is difficult to distinguish a new episode of acute pancreatitis from an exacerbation of an ongoing process, patients with pancreatitis immediately before ERCP were excluded, indicating that an episode of pancreatitis occurring more than 30 days before elective ERCP had no association with the development of PEP.
In accordance with previous studies 11, 18 investigating hypertriglyceridaemia, hyperlipidaemia was also found to increase the risk of PEP. However, other associated co-morbidities such as obesity were not investigated in the present study as data on BMI were not available in the registry. Similarly, other possible conditions influencing the risk of PEP, such as alcohol abuse and medications, are not registered consistently in GallRiks.
Although the literature 7,8,19 documents contrasting results with respect to hypercalcaemia/kidney disease and risk of PEP, it should be noted that only 58 patients in the present cohort had hypercalcaemia and 579 had kidney disease, with no data on the degree of renal failure; thus it would be difficult to draw any firm conclusion regarding the association between hypercalcaemia/kidney disease and PEP.
Similar to previous findings 13,14 , liver cirrhosis was not found to be a risk factor for PEP.
In contrast to previous studies 10,20 , in which diabetes was shown to be associated with acute pancreatitis, a decreased risk of PEP was found in diabetic patients. This was confirmed in the multivariable analysis, after adjustment for age and sex. It has been shown previously 12 that the risk of acute pancreatitis is dependent on the type of diabetes medication received by patients. Although the cohort of diabetic patients consisted of patients on different kinds of diabetic treatment, the registry lacked information on disease severity and treatment; thus these associations were not investigated and need to be validated in future studies.

Disclosure
The authors declare no conflict of interest.