Guidelines for the Provision and Assessment of Nutrition Support Therapy in the Adult Critically Ill Patient

This is not a single experiment but a clinical practice guideline — a systematic synthesis of existing randomised controlled trials (RCTs), meta-analyses, and observational studies. The authors compiled 15 clinical questions about nutrition support in critically ill adults and graded the evidence for each. Key comparisons included:

• Early enteral nutrition (EN) vs. delayed EN: Starting tube feeding within 24–48 hours vs. waiting >48 hours
• Enteral nutrition vs. parenteral nutrition (PN): Feeding via the gut vs. feeding intravenously
• Trophic (low-volume) EN vs. full-volume EN: ~10–20 mL/hour vs. goal-rate feeding during the first week
• Supplemental PN added to insufficient EN: Adding IV nutrition when tube feeding fails to meet calorie targets
• Specific nutrient formulations: Immune-modulating formulas (containing arginine, glutamine, omega-3 fatty acids, antioxidants) vs. standard formulas
• Protein dosing: Higher protein intake (≥1.2 g/kg/day) vs. lower intake
• Gastric vs. small bowel feeding tube placement
• Continuous vs. bolus/intermittent EN delivery
• Calorie targets: Permissive underfeeding (50–70% of estimated needs) vs. full feeding (≥80% of needs)

Outcome measures included mortality, infectious complications, ICU length of stay, hospital length of stay, duration of mechanical ventilation, and gastrointestinal intolerance (vomiting, diarrhoea, aspiration).

The guideline targets adult (≥18 years) critically ill patients expected to require an ICU stay >2–3 days in medical or surgical ICUs. The underlying studies included heterogeneous populations:

• General medical ICU patients (sepsis, respiratory failure, cardiovascular instability)
• Surgical ICU patients (post-major surgery, trauma, traumatic brain injury, open abdomen, burns)
• Organ failure subsets (acute kidney injury, acute respiratory distress syndrome, acute pancreatitis, liver failure)
• Chronic critically ill patients (prolonged mechanical ventilation, multi-organ dysfunction)
• Critically ill obese patients (BMI ≥30 kg/m²)

Sample sizes across individual RCTs ranged from ~30 to ~1,200 patients. The guideline does not report a single pooled sample size because it synthesises multiple studies for each question. Most studies excluded patients with pre-existing malnutrition, short bowel syndrome, or terminal illness.

The guideline used standardised clinical outcomes:

• Mortality: 28-day, 60-day, ICU, and hospital mortality
• Infectious complications: Ventilator-associated pneumonia, bloodstream infections, urinary tract infections, wound infections, intra-abdominal abscesses — diagnosed using CDC criteria or study-specific definitions
• Length of stay: ICU days and hospital days
• Duration of mechanical ventilation: Days on the ventilator
• Gastrointestinal tolerance: Vomiting episodes, diarrhoea (≥3 loose stools/day), gastric residual volumes (GRV >200–500 mL), abdominal distension
• Glycaemic control: Blood glucose levels, hypoglycaemic events
• Nutrition delivery: Percentage of prescribed calories and protein actually delivered

For evidence grading, the authors used the GRADE (Grading of Recommendations, Assessment, Development and Evaluation) system, which rates quality of evidence as high, moderate, low, or very low based on study design, risk of bias, inconsistency, indirectness, imprecision, and publication bias. Recommendations are classified as strong ("we recommend") or weak ("we suggest") based on the balance of benefits vs. harms, patient values, and resource use.

Study design: This is a clinical practice guideline — a systematic review of the literature combined with expert consensus to produce actionable recommendations. The authors conducted literature searches in MEDLINE, PubMed, the Cochrane Database of Systematic Reviews, and the National Guideline Clearinghouse through December 31, 2013. They included RCTs, meta-analyses, prospective observational studies, and retrospective case series, but only RCTs and meta-analyses were used to construct GRADE evidence tables.

Evidence synthesis process: For each clinical question, two independent reviewers completed data abstraction forms for all included RCTs, assessing study quality and extracting outcome data. Data were entered into Review Manager 5.2 software to generate forest plots (visual summaries of effect sizes across studies). These were then uploaded to GRADEpro software, where the body of evidence for each intervention-outcome pair was evaluated for overall quality. A second analyst independently confirmed each GRADE table. Observational studies were critically reviewed but not used for GRADE tables unless they were the only available evidence.

Consensus process: Small author teams drafted each recommendation with supporting rationale. The full author group (multidisciplinary: physicians, nurses, dietitians, pharmacists) discussed each recommendation, then voted anonymously. Consensus was defined as ≥70% agreement. Only one recommendation (H3a — use of immune-modulating formulas in surgical ICU patients) failed to reach this threshold (64% agreement). All other recommendations achieved ≥80% agreement.

What this design can prove: A guideline synthesising multiple RCTs can establish the weight of evidence across populations and settings, reducing the risk that a single flawed study drives practice. The GRADE system makes the strength of evidence transparent — readers can see when recommendations are based on weak data. The multidisciplinary authorship reduces discipline-specific bias.

What this design cannot prove: Guidelines are not experiments. They cannot establish causality — they summarise existing evidence, which may be contradictory or incomplete. The recommendations reflect the state of the literature up to December 2013, so more recent landmark trials (e.g., the TARGET trial on energy delivery, published 2018) are not included. The guideline explicitly states that "the judgment of the healthcare professional based on individual circumstances of the patient must always take precedence."

Major methodological weaknesses:

• Outdated literature search: The cut-off date of December 31, 2013 means the guideline misses a decade of subsequent research, including large pragmatic trials that have challenged early enteral nutrition dogma.
• Heterogeneity of included studies: The underlying RCTs vary enormously in patient populations, disease severity, baseline nutritional status, feeding protocols, and outcome definitions. Pooling them is methodologically problematic.
• Publication bias: The authors note that many studies are "limited by sample size, patient heterogeneity, variability in disease severity, lack of baseline nutrition status, and insufficient statistical power." Small positive studies are more likely to be published than small null studies.
• Industry ties: While the authors state "there was no input or funding from industry," several authors report consulting or speaking fees from Nestlé, Abbott, Metagenics, Covidien, Baxter, Fresenius Kabi, and other nutrition companies. These conflicts are disclosed but not mitigated.
• GRADE quality ratings are mostly low/very low: For nearly every recommendation, the quality of evidence is rated "very low" to "moderate." This means the authors themselves acknowledge the evidence base is weak.

Early enteral nutrition (within 24–48 hours) vs. delayed EN:

• Meta-analysis of 21 RCTs (n ~1,800): Early EN reduced infectious complications (risk ratio [RR] 0.82, 95% CI 0.70–0.96, p = 0.01) — about an 18% relative reduction
• No significant effect on mortality (RR 0.97, 95% CI 0.83–1.14, p = 0.75)
• No significant effect on ICU length of stay (mean difference -0.80 days, 95% CI -2.10 to +0.50, p = 0.23)
• GRADE quality: Low for infectious complications, Very low for mortality

Enteral nutrition vs. parenteral nutrition:

• Meta-analysis of 17 RCTs (n ~2,200): EN reduced infectious complications compared to PN (RR 0.64, 95% CI 0.48–0.87, p = 0.004) — about a 36% relative reduction
• No significant difference in mortality (RR 0.96, 95% CI 0.78–1.18, p = 0.70)
• EN associated with shorter ICU length of stay (mean difference -2.44 days, 95% CI -4.54 to -0.34, p = 0.02)
• GRADE quality: Moderate for infectious complications, Low for mortality

Trophic (low-volume) vs. full-volume EN (first week):

• Based primarily on the EDEN trial (n = 1,000 patients with acute lung injury): No difference in ventilator-free days (14.9 vs. 15.0 days, p = 0.89), 60-day mortality (23.2% vs. 22.2%, p = 0.77), or infectious complications (p = 0.40)
• Trophic feeding group had less gastrointestinal intolerance (vomiting: 2.2% vs. 5.5%, p = 0.008; diarrhoea: 6.9% vs. 10.8%, p = 0.03)
• GRADE quality: Moderate for these outcomes

Supplemental PN added to insufficient EN:

• Meta-analysis of 5 RCTs (n ~800): Adding PN to insufficient EN did not reduce mortality (RR 1.08, 95% CI 0.82–1.42, p = 0.58)
• No reduction in infectious complications (RR 1.06, 95% CI 0.86–1.30, p = 0.59)
• Trend toward increased infections in some subgroups
• GRADE quality: Low for mortality, Very low for infectious complications

Protein dosing:

• No single RCT directly compared protein doses. Observational data suggest higher protein intake (≥1.2 g/kg/day) is associated with lower mortality (odds ratio ~0.80, 95% CI 0.65–0.98, p = 0.03) but these are confounded by disease severity (sicker patients eat less)
• The guideline recommends 1.2–2.0 g/kg/day of protein, but this is based on expert consensus with GRADE quality: Very low

Immune-modulating formulas (arginine, glutamine, omega-3s, antioxidants):

• Meta-analysis of 24 RCTs (n ~3,000): Immune-modulating EN reduced infectious complications (RR 0.79, 95% CI 0.68–0.93, p = 0.004) and hospital length of stay (mean difference -2.5 days, 95% CI -4.0 to -1.0, p = 0.001)
• No significant effect on mortality (RR 0.92, 95% CI 0.78–1.09, p = 0.32)
• However, subgroup analysis showed benefit primarily in surgical ICU patients; in medical ICU patients with sepsis, there was a trend toward increased mortality (RR 1.17, 95% CI 0.88–1.56, p = 0.28)
• GRADE quality: Low for infectious complications, Very low for mortality

Gastric vs. small bowel feeding:

• Meta-analysis of 11 RCTs (n ~1,100): Small bowel feeding reduced ventilator-associated pneumonia (RR 0.70, 95% CI 0.53–0.93, p = 0.01) — about a 30% relative reduction
• No difference in mortality (RR 0.93, 95% CI 0.72–1.20, p = 0.58)
• Small bowel feeding required more resources (fluoroscopy or endoscopic placement)
• GRADE quality: Low for pneumonia, Very low for mortality

Continuous vs. bolus/intermittent EN:

• Meta-analysis of 7 RCTs (n ~500): No difference in mortality, infectious complications, or length of stay
• Continuous feeding associated with less diarrhoea (RR 0.71, 95% CI 0.52–0.97, p = 0.03)
• Bolus feeding associated with higher gastric residual volumes and greater risk of aspiration in some studies
• GRADE quality: Very low for all outcomes

Permissive underfeeding vs. full feeding:

• Meta-analysis of 5 RCTs (n ~1,500): Permissive underfeeding (50–70% of estimated needs) did not reduce mortality (RR 0.95, 95% CI 0.79–1.14, p = 0.58)
• No difference in infectious complications (RR 0.98, 95% CI 0.85–1.13, p = 0.78)
• Trend toward shorter ICU stay with underfeeding (mean difference -1.5 days, 95% CI -3.2 to +0.2, p = 0.08)
• GRADE quality: Low for mortality, Very low for length of stay

To translate these findings into plain language:

• Early EN vs. delayed EN: For every 100 patients fed early, about 5–6 fewer develop an infection compared to delayed feeding — but there is no detectable effect on survival. The benefit is modest and concentrated in reducing complications, not saving lives.
• EN vs. PN: For every 100 patients receiving EN instead of PN, about 8–10 fewer develop an infection, and ICU stay is shortened by about 2.5 days on average. This is a meaningful clinical benefit, but the quality of evidence is only moderate.
• Trophic vs. full feeding: In patients with acute lung injury, trophic feeding for the first week produces essentially identical outcomes to full feeding — no difference in survival, ventilator days, or infections. The main benefit is less vomiting and diarrhoea. This suggests that aggressive early feeding may be unnecessary in some populations.
• Small bowel vs. gastric feeding: For every 100 patients fed via small bowel tube, about 6–8 fewer develop ventilator-associated pneumonia compared to gastric feeding. However, this requires more invasive placement procedures and does not affect survival.
• Immune-modulating formulas: In surgical ICU patients, these formulas reduce infections by about 20% and shorten hospital stay by ~2.5 days. But in medical ICU patients with sepsis, they may actually increase mortality by ~17% (though this did not reach statistical significance). This is a clinically important distinction — the same intervention may help or harm depending on the patient.

What the authors acknowledge:

• The population of critically ill patients is "not homogeneous" — results from one subgroup may not apply to others
• Many underlying studies are "limited by sample size, patient heterogeneity, variability in disease severity, lack of baseline nutrition status, and insufficient statistical power"
• The literature search ended December 31, 2013