Peptide-based versus polymeric enteral formulas and gastrointestinal tolerance in hospitalised adults: a systematic review and meta-analysis of randomised trials.

BACKGROUND & AIMS: Guidelines suggest switching from a polymeric to a more expensive peptide-based enteral formula when diarrhoea occurs. Previous systematic reviews were restricted to acute gastrointestinal injury or acute pancreatitis. We assessed whether peptide-based formulas reduce diarrhoea and other manifestations of gastrointestinal intolerance in hospitalised adults receiving tube feeding. METHODS: We searched MEDLINE, Europe PMC, CENTRAL, LILACS, trial registers and citation networks to 17 September 2026 (Embase was not accessible), without language restrictions, for randomised trials comparing peptide-based (hydrolysed-protein) with intact-protein polymeric formulas. Screening, full-text eligibility, data extraction and risk-of-bias assessment (RoB 2) were performed by a large language model; every excluded record was re-screened in a second, independent model pass, every extracted item was tied to a verbatim quotation, and the authors verified against the source documents every item entering the analyses. Screening was not duplicated by human reviewers, and this is reported as the review's principal limitation. Risk ratios (RR) for diarrhoea were pooled with random-effects models, with Hartung-Knapp and sparse-data sensitivity analyses; certainty was rated with GRADE. Subgroup analyses by formula composition were post hoc. RESULTS: Of 3,196 records, 27 studies were included and 12 (914 participants) reported diarrhoea. There was no evidence that peptide-based formulas reduce diarrhoea (RR 0.89, 95% CI 0.53-1.52; I2 64%; very low certainty). A lower risk was observed in two open-label trials, at high risk of bias, of pectin-containing formulas (RR 0.34, 95% CI 0.14-0.78) but not in the other 9 trials (RR 1.25, 95% CI 0.84-1.85; subgroup difference p = 0.006); in some groupings of these trials the estimate suggested more diarrhoea with peptide-based formulas. Excluding trials at high risk of bias gave RR 1.12, 95% CI 0.67-1.88. Mortality showed no clear difference (RR 0.82, 95% CI 0.53-1.24; 11 trials; very low certainty). Other gastrointestinal outcomes were sparsely reported. CONCLUSIONS: Randomised evidence does not show that peptide-based formulas reduce diarrhoea in hospitalised adults, and its certainty is very low. The lower risk in three small trials of pectin-containing formulas is exploratory, imprecise and cannot be attributed to the peptide, and a possible increase in diarrhoea with some peptide-based formulas cannot be excluded.