Original article
Pancreas, biliary tract, and liver
A Model to Predict the Severity of Acute Pancreatitis Based on Serum Level of Amylase and Body Mass Index

https://doi.org/10.1016/j.cgh.2015.03.018Get rights and content

Background & Aims

Most patients with acute pancreatitis (AP) develop mild disease, but up to 20% develop severe disease. Many clinicians monitor serum levels of amylase and lipase in an attempt to predict the disease course, but this strategy has not been recommended by practice guidelines. We performed a retrospective analysis to determine whether the percentage changes in amylase and lipase were associated with the severity of disease that developed in patients with AP.

Methods

We analyzed data collected from 182 consecutive patients with AP (21 with severe AP) admitted to the Cleveland Clinic from January 2008 through May 2010 (discover cohort). The association between 11 different factors and the severity of AP were assessed by univariable analysis; multivariable models were explored through stepwise selection regression. The percentage change in the serum level of amylase was calculated as follows: ([amylase day 1 – amylase day 2]/amylase day 1) × 100. The percentage change in amylase and body mass index (BMI) were combined to generate a z-score (z = -5.9 + [0.14 × BMI] + [0.01 × percentage change in amylase]), which was converted into a probability distribution called the change in amylase and BMI (CAB) score. The CAB score was validated using the AP database at the University of Pittsburgh Medical Center (140 patients, 35 with severe AP); we calculated p-scores for each patient and estimated the area under the receiver operating characteristics curve values.

Results

Univariable analysis identified the percentage change in the serum level of amylase and other factors to be associated significantly with the severity of AP (P = .017). The CAB score was best at identifying patients who developed severe AP, with an area under the receiver operating characteristics curve value of 0.79 in the discovery cohort (95% confidence interval, 0.71–0.87) and 0.731 in the validation cohort (95% confidence interval, 0.61–0.84).

Conclusions

We developed a model to identify patients most likely to develop severe AP based on the percentage changes in serum level of amylase during the first 2 days after admission to the hospital and BMI.

Section snippets

Methods

The diagnosis of AP was based on American College of Gastroenterology criteria and required the presence of at least 2 of the 3 following factors: (1) abdominal pain characteristic of AP, (2) serum amylase and/or lipase levels 3 or more times the upper limit of normal, and (3) CT findings characteristic of AP.1 Only patients with at least 2 serum amylase and lipase levels measured within the first 48 hours after admission were included in the study. Patients were classified as having SAP if

Results

Baseline characteristics of the discovery and the validation cohorts are presented in Table 1. The rule-out selection bias, the included and excluded patients from the discovery cohort, were compared and had similar age, etiology, Charlson comorbidity index, and BISAP score on day 1 of admission.

Discussion

This was a dual-center study that aimed to design a multivariable logistic regression model that accurately predicts SAP. We designed a novel multivariable model based on percentage changes in serum amylase levels during the first 2 days of admission and BMI, which accurately predicted the severity of AP. The initial findings using the discovery cohort subsequently were validated by an independent cohort of AP patients.

Serum amylase levels have long been used for the diagnosis of AP. In early

References (33)

  • B. Spanier et al.

    Incidence and mortality of acute and chronic pancreatitis in the Netherlands: a nationwide record-linked cohort study for the years 1995-2005

    World J Gastroenterol

    (2013)
  • V.S. Swaroop et al.

    Severe acute pancreatitis

    JAMA

    (2004)
  • D. Yadav et al.

    Trends in the epidemiology of the first attack of acute pancreatitis: a systematic review

    Pancreas

    (2006)
  • D.C. Whitcomb

    Clinical practice. Acute pancreatitis

    N Engl J Med

    (2006)
  • J.H. Ranson et al.

    Prognostic signs and the role of operative management in acute pancreatitis

    Surg Gynecol Obstet

    (1974)
  • B.U. Wu et al.

    The early prediction of mortality in acute pancreatitis: a large population-based study

    Gut

    (2008)
  • Cited by (0)

    Conflicts of interest The authors disclose no conflicts.

    View full text