PermiT: Permissive Underfeeding or Standard Enteral Feeding in Critically Ill Adults

Arabi. NEJM 2015; 372:2398-2408. doi:10.1056/NEJMoa1502826

Clinical Question

• In critically ill adults, does restriction of non-protein calories (permissive underfeeding) compared to standard feeding reduce mortality at 90 days?

Design

• Randomised, controlled trial
• Computer-generated sealed opaque envelopes
• Block randomisation with stratification according to medical centre
• Non-blinded intervention but with objective end-point
• Clinician-led management of non-nutrition elements of care
• Powered at 80% to detect 8% absolute reduction in mortality from baseline of 25% with standard feeding if 892 patients recruited, allowing for 3% loss

Setting

• 7 hospitals across Saudi Arabia and Canada
• November 2009 to September 2014

Population

• Inclusion: Enteral feeding commenced within 48 hours of ICU admission; expected to stay in ICU ≥ 72 hours; age 18–80 years
• Exclusion: use of total parenteral nutrition; high-dose vasopressors (e.g. noradrenaline > 0.4 µg/kg/min); palliation or condition with expected 6-month mortality ≥ 50%; post cardiac arrest; pregnancy; liver transplant; burns
• 894 patients randomised (14% of screened patients)
  • Demographic and physiological scores were similar between groups (intervention vs control as mean ± standard deviation)
    • Age: 50.2±19.5 vs 50.9±19.4
    • BMI: 29.0±8.2 vs 29.7±8.8
    • Medical: 75% vs 75%
    • Surgical: 4.2% vs 2.7%
    • Trauma: 21% vs 22%
    • APACHE 2: 21.0±7.9 vs 21.0±8.2
    • Mechanical ventilation: 97% vs 97%

Intervention

• Permissive underfeeding
  • Caloric goal was 40–60% of caloric requirement
  • Protein was supplemented with Beneprotein, Nestle Nutrition
  • Supplemented fluid input with enteric 0.9% saline or water
  • Average caloric intake achieved was 835±297 kcal/day (46±14% caloric goal)

Control

• Standard feeding
  • Caloric goal was 70–100% of caloric requirement
  • Average caloric intake achieved was 1299±467 kcal/day (71±22% caloric goal)

 Treatment common to both groups

• Caloric requirements calculated using
  • Penn State equation 2003 if BMI < 30
  • Ireton-Jones equation 1992 if BMI ≥ 30 or spontaneously breathing
• Protein goal was 1.2–1.5g per kg body weight per day
• Continued for 14 days or until ICU discharge, oral feeding, death or palliation
• Glucose goal was 4.4–10 mmol/l (80–180 mg/dl)
• Enteric multivitamins
• Under or over feeding on one day was balanced by targeted over or underfeeding the next day, respectively
• Caloric intake calculations included propofol, intravenous dextrose and parenteral nutrition

Outcome

• Primary outcome: there was no statistically significant difference in 90-day mortality
  • Permissive underfeeding: 27.2%
  • Standard feeding: 28.9%
  • Absolute risk reduction = 1.67% (95% CI -4.27 to 7.59%)
  • Relative risk reduction = 0.94 (95% CI 0.76 to 1.16; P-value = 0.58)
  • Cox proportional-hazard ratio (unadjusted) = 0.91 (95% CI 0.71 to 1.17; P-value = 0.48)
• Secondary outcome: there were no statistically significant differences favouring either group
  • Mortality in ICU
  • 28-day mortality
  • In-hospital mortality
  • 180-day mortality
  • Serial SOFA scores
• Tertiary data:
  • There were no significant differences in electrolytes
  • There was no difference in the calories provided by parental feeding
    • Permissive underfeeding: 3±32 kcal/day
    • Standard feeding: 5±59 kcal/day
    • P-value = 0.38
  • Blood glucose was higher in standard feeding group:
    • Permissive underfeeding: 9.1±5.3 mmol/l
    • Standard feeding: 9.4±5.0
    • P-value = 0.04
  • Insulin requirement was higher in standard feeding group
    • Permissive underfeeding: 15±27 units/day
    • Standard feeding: 22±40 units/day
    • P-value = 0.02
• Post-hoc analysis
  • Incidence of renal replacement therapy
    • Permissive underfeeding: 7.1%
    • Standard feeding: 11.4%
    • Absolute risk reduction: 4.11% (95% CI 0.21 to 8.23%)
    • Number-needed-to-treat with permissive underfeeding to prevent one renal replacement therapy = 24

Authors’ Conclusions

• Moderate caloric feeding with maintenance of full protein requirement, compared to standard full feeding, was not associated with reduced mortality

Strengths

• Pragmatic and well designed trial
• Appropriate statistical plan
• Minimal biases present
• Multi-centre including different ethnic populations
• Highly generalisable given spread of medical, surgical and trauma patients

Weaknesses

• Lack of blinding and clinician-led management of non-nutritional care may have introduced bias but the impact of this is unknown
• Primary outcome of 90-days is a long-duration compared to just 14 days of intervention, such that a direct cause-and-effect link is unlikely to be observed
  • However, no difference was seen in shorter secondary outcomes either
• Only powered to detect 8% absolute risk reduction
  • A smaller effect may still be clinically important
  • This study may have failed to show a true but small treatment effect (false negative)
• 86% of screened patients were not randomised, which may have an impact upon the external validity (generalisability) of the results
• The only notable positive outcome is the reduced incidence of renal replacement therapy
  • As a post-hoc analysis, this should be cautiously interpret and possibly used as theory-generating for future trials