Twenty-seven RCTs were included (the total number of participants was not specified).
The mean quality score was 8.15 (range: 6 to 11).
Early EN versus PN (7 studies, including 5 studies in which patients were randomised within 48 hours of admission).
The meta-analysis showed that EN was associated with a significant reduction in infectious morbidity (RR 0.46, 95% CI: 0.29, 0.74, p=0.001; 7 studies; no significant heterogeneity) and hospital LOS (WMD -3.94 days, 95% CI: -5.86, -2.02, p<0.0001; 4 studies; heterogeneity was significant, p=0.03) compared with PN. There was no significant difference between treatments in mortality (based on 7 studies) or organ failure (based on 5 studies).
EN or PN versus STD.
EN versus STD: there was insufficient information from one small study (n=27) that lasted 4 days.
PN versus STD: PN given within 24 hours of admission significantly increased LOS (16 versus 10 days, p<0.04) and non significantly increased the time to a clear liquid diet (10 versus 6 days, p=0.8) compared with STD (based on 1study in 55 patients with generally mild pancreatitis). PN given later (24 to 48 hours after 'liquid resuscitation') significantly reduced overall complications (52.4% versus 91.3%, p<0.01), hospital LOS (28.6 versus 39.1 days, p<0.05) and mortality (14.3% versus 43.5%, p<0.05) compared with STD (based on 1 study with 44 patients with severe acute pancreatitis).
Post-operative EN versus PN or STD. EN versus PN: there was insufficient information from one small study (n=22).
EN versus STD: the meta-analysis showed that EN reduced mortality compared with STD in patients who had required surgery for complications of acute pancreatitis, but the reduction was not statistically significant (RR 0.26, 95% CI: 0.06, 1.09, p=0.06; based on 2 studies with 71 patients).
Addition of supplements.
EN plus supplements versus EN alone: individual studies reported benefits (including reductions in mortality, LOS and complications) from adding arginine, glutamine, omega-3 polyunsaturated fatty acids and probiotics to EN compared with EN alone.
PN plus supplements versus PN alone: supplementing PN with parenteral glutamine did not significantly reduce overall complications (RR 0.68, 95% CI: 0.42, 1.09, p=0.11; based on 3 studies). Two studies reported reduced hospital LOS with glutamine supplementation, but in only one was the reduction statistically significant.
Tolerance to EN.
One small study (n=28) reported a non statistically significant reduction in the proportion of patients with no pain relapse (0% versus 27%) and LOS (12 versus 21 days) for nasojejunal feeding compared with oral feeding. One study (n not stated) reported no significant differences between nasogastric versus nasojejunal feeding. One small study (14 patients following surgery for the complications of acute pancreatitis) reported that bolus EN significantly increased the volume, bicarbonate content and enzyme output of the pancreas in comparison with continuous EN infusion (p<0.05).