PYTHON: Early versus On-Demand Nasoenteric Tube Feeding in Acute Pancreatitis

Bekker. N Engl J Med 2014; 371:1983-93. doi:10.1056/NEJMoa1404393

Clinical Question

  • In patients with severe pancreatitis, does early enteral feeding compared with on-demand feeding reduce death or major infection?


  • Randomised-controlled superiority trial
  • Two groups of patients in a 1:1 ratio
  • Randomisation performed centrally and stratified for severity and treatment centre
  • Treating clinicians non-blinded
  • Outcome adjudicators blinded to patient group
  • Sample size of 208 patients calculated to give 80% power to detect a reduction in primary outcome from 40% (on demand feeding) to 22% (early enteric feeding) based on previous data


  • Six university medical centres and thirteen large teaching hospitals in the Netherlands
  • August 2008 – June 2012


  • Inclusion: Adults who were predicted to have severe pancreatitis
    • Pancreatitis was diagnosed based on two or more of:
      • Typical abdominal pain
      • Characteristic findings on cross-sectional abdominal imaging
      • Serum lipase or amylase more than three times upper limit of normal
    • Severity
      • APACHE II ≥ 8
      • Imrie or modified Glasgow score ≥ 3
      • CRP over 150mg/l within 24 hours of admission
  • Exclusion: Recurrent acute or chronic pancreatitis; Pancreatitis due to ERCP or malignancy; Home enteral or parenteral nutrition; Pregnancy; Assessed more than 24 hours after presentation; Presented more than 96 hours after symptom onset
  • 867 patients screened, 208 enrolled and randomised, 205 analysed (3 incorrectly diagnosed)


  • “Early” group:
    • Nasojejunal tube placed within 24 hours of randomisation and feeding commenced
    • Tubes placed endoscopically or radiologically
    • Feeding was started at 20ml/hr for the first 24 hours, then 45ml/hr from 24-48 hours, then 65ml/hr from 48-72 hours, and full feeding based on actual body weight after this
    • If symptoms of delayed passage developed, the feeding rate was decreased by 50%, and gradually increased again the following day, but stopped if symptoms worsened
    • When patients were able to tolerate an oral diet, tube feeding was gradually stopped and replaced


  • “On-demand” group:
    • Oral diet was provided at 72 hours
    • Supplementation by tube feeding after further 24 hours if unable to tolerate oral intake
    • Prior to this received only IV fluids unless they requested oral food


  • Primary outcome:
    • Composite outcome of death or major infection within 6 months showed no statistically significant difference
      • Early group: 30%
      • On-demand group: 27%
      • Absolute risk increase: 3% (95% CI -9 to 15)
      • NNH: 36
      • p-value: 0.76
    • Death within six months was not significant statistically difference (11% early vs 7% on-demand, p=0.33)
    • Major infection within six months (pneumonia, bacteraemia, infected pancreatic necrosis) was not significant statistically difference (25% early vs 26% on-demand, p=0.87)
  • Secondary outcome:
    • Days to full oral diet was significantly reduced in on-demand group (9 days early vs 6 days on-demand, p=0.001)
    • Rate of NJ tube placement was significantly higher in early group (100% early vs 31% on-demand)
    • No significant difference in:
      • Necrotising pancreatitis (62%)
      • Rate of ITU admission
      • Frequency of gastrointestinal events
      • Length of SIRS
  • Post hoc analyses showed no significant primary outcome difference for:
    • The most unwell patients (APACHE ≥ 13)
    • Those who already had SIRS at randomisation
    • BMI under 25 or ≥ 35
    • Health care utilisation (except with regards tube placement)

Authors’ Conclusions

  • Early enteral feeding in these patients did not improve outcomes when compared to delayed on-demand feeding at 72 hours
  • This result disagrees with current guidelines which advise feeding to provide a protective effect on intestinal integrity and systemic nutritional status
  • It was unusual for tube feeding to be required


  • Succeeded in randomising the treatment of a large number of patients with severe illness.
  • Chosen outcomes are clinically important
  • The statistical methods are valid and well-illustrated
  • Long follow-up
  • Low drop-out rate
  • Interesting result with potential to change practice


  • May have been too small and underpowered
  • Feeding may also been initiated too late or too gently in disease process
  • Scoring systems poor at predicting eventual severity so cases of non-severe pancreatitis may have been included
  • Unblinded treating clinicians
  • NJ tubes used, when nasogastric tubes are easier to place and may be as effective
  • Not able to inform management of other forms of pancreatitis

The Bottom Line

  • Allowing patients with severe pancreatitis three to four days to initiate oral intake appears to be safe and effective
  • This can reduce the discomfort, cost and complications associated with tube feeding in these patients
  • Although well-conducted this was a relatively small trial, and future data may change this picture

External Links


Summary author: John Kiely
Summary date: 12 June 2015
Peer-review editor: @DuncanChambler

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.