Current issues of ACP Journal Club are published in Annals of Internal Medicine


Pain, CT scores, and high serum lipase predicted pain relapse in refeeding patients with acute pancreatitis

ACP J Club. 1997 Sep-Oct;127:51. doi:10.7326/ACPJC-1997-127-2-051

Source Citation

Lévy P, Heresbach D, Pariente EA, et al. Frequency and risk factors of recurrent pain during refeeding in patients with acute pancreatitis: a multivariate multicentre prospective study of 116 patients. Gut. 1997 Feb;40:262-6.



To determine the risk factors associated with pain relapse during refeeding in patients with acute pancreatitis (AP).


A multicenter, multidimensional, prospective study.


11 medical or surgical centers in France.


116 patients (mean age 51 y, 64% men) who had AP defined as acute epigastric pain requiring emergency hospitalization with a serum amylase or lipase level > 5 times the normal upper limit; or evidence of AP on ultrasonography or computed tomography (CT). Patients were included if oral feeding had to be stopped for > 48 hours.

Assessment of risk factors

Clinical (sex, age, cause of AP, delay between onset of symptoms or hospitalization and discontinuation of feeding, period of pain and hospitalization, and vomiting), biochemical (maximum levels of serum amylase, lipase, and C reactive protein; Ranson score; and levels of serum amylase and lipase before refeeding), radiologic (pancreatic calcifications, Balthazar CT score, enlarged pancreatic duct, and abdominal fluid collection), and therapeutic data (nasogastric suction; radiologic or endoscopic procedures; parenteral nutrition; use of antibiotics, gastric antisecretory drugs, and somatostatin analogues; and the delay between onset of symptoms or the end of the period of pain and the time of refeeding) were recorded for each patient.

Main outcome measure

Pain relapse defined as acute pain requiring oral feeding to be discontinued again.

Main results

24 patients (21%) had pain relapse. Pain relapse occurred on days 1 and 2 after refeeding in 12 patients (50%). Unidimensional analysis showed that a longer duration of initial pain (P < 0.002), serum lipase levels that were > 3 times normal 1 day before refeeding (P < 0.03), and higher Balthazar CT scores (P < 0.002) occurred more often in patients who had relapse. Therapeutic procedures and delay between onset of symptoms or the end of the period of pain and refeeding were not associated with pain relapse. Multidimensional analysis found that Balthazar CT score, period of pain, and serum lipase concentration 1 day before refeeding independently predicted pain relapse. At a threshold logistic score of 0.5, sensitivity, specificity, and accuracy to predict pain relapse were 37%, 95%, and 83%, respectively.


Long periods of pain, high Balthazar CT scores, and serum lipase concentrations > 3 times normal 1 day before refeeding were independently associated with an increased risk for pain relapse during refeeding in patients with acute pancreatitis.

Sources of funding: Laboratoires Solvay-Pharma.

For article reprint: Dr. P. Lévy, Department Medico-Chirurgical de Pathologie Digestive, 6 Place Port-au-Prince, 75013 Paris, France. FAX 33-1-47-37-0533.


A recently published practice guideline states that refeeding after an episode of AP is generally appropriate when abdominal pain and tenderness have resolved, bowel sounds have returned, and the patient is hungry (1). Elevations of serum pancreatic enzymes may also be considered in the decision to refeed, but isolated elevation of these enzymes in an otherwise-recovered patient has not been considered to be an absolute contraindication to refeeding. The potential danger of premature refeeding is pain or AP relapse, and subsequent prolongation of hospital stay. The study by Lévy and colleagues attempts to quantify clinical features that might allow more accurate decisions on the timing of refeeding. The logistic score generated includes 3 features that clinicians might use in making qualitative decisions about refeeding: elevated levels of serum lipase, prolonged pain, and more severe changes on CT. Although this logistic score reached acceptable specificity, the sensitivity was only 37% (almost two thirds of patients who developed recurrent pain after refeeding did not reach the cutoff). A second problem with the logistic score is that it included CT, which may not be done on every patient and may not be necessary in patients with clinically mild pancreatitis (1). What may be more useful to clinicians is that a lipase level > 3 times normal 1 day before refeeding was nearly as sensitive as the overall score.

This study found that premature refeeding is associated with a markedly prolonged hospital stay, although it cannot be determined whether the period of hospitalization would have been shortened by delaying refeeding. Although Lévy and coworkers begin the process of quantifying a currently qualitative decision, the model will need to be validated in other settings to determine its overall utility and ability to shorten hospital stay.

For now, we should continue to rely on a more qualitative judgment about refeeding using the criteria reported by Banks (1), although the factors identified in this study are reasonable additional considerations.

Chris E. Forsmark, MD
University of FloridaGainesville, Florida, USA


1. Banks PA. Practice guidelines in acute pancreatitis. Am J Gastroenterol. 1997; 92:377-86.