Quick Links

Recommendations Summary

CI: Enteral versus Parenteral Nutrition and Critical Illness 2006

Click here to see the explanation of recommendation ratings (Strong, Fair, Weak, Consensus, Insufficient Evidence) and labels (Imperative or Conditional). To see more detail on the evidence from which the following recommendations were drawn, use the hyperlinks in the Supporting Evidence Section below.


  • Recommendation(s)

    CI: Enteral vs Parenteral Nutrition

    If the critically ill ICU patient is hemodynamically stable with a functional GI tract, then EN is recommended over PN. Patients who received EN experienced less septic morbidity and fewer infectious complications than patients who received PN. In the critically ill patient, EN is associated with significant cost savings when compared to PN. There is insufficient evidence to draw conclusions about the impact of EN or PN on LOS and mortality.

    Rating: Strong
    Conditional

    • Risks/Harms of Implementing This Recommendation

      • Inadvertent misplacement of the enteral feeding tube
      • Feeding less than 25% of goal calories or persistence of a negative energy deficit for more than one week may increase the risk for nosocomial infections and complications in critical illness (Rubinson L et al, 2004; Villet S et al, 2005)
      • Overfeeding calories (compared with goal) may increase the risk for hyperglycemia and contribute to a greater risk of associated metabolic complications over time.

    • Conditions of Application

      • Studies that compared EN against PN interventions under conditions of tight glucose control are not available
      • Studies that examined combined EN and PN interventions were not included in this phase of the evidence analysis.

    • Potential Costs Associated with Application

      Enteral nutrition is significantly less expensive than PN.

    • Recommendation Narrative

      Mortality

      • Five studies, two of positive quality (Young et al, 1987; Abou-Assi et al, 2002) and three of neutral quality (Cerra et al, 1988; Hadfield et al, 1995; Woodcock et al, 2001) found no mortality difference between patient who received EN vs. PN
        • These studies may not have been adequately powered to decisively determine this effect.
      • Other studies (Adams et al, 1986; Hadley et al, 1986; Kudsk et al, 1992; Borzotta et al, 1994; Kalfarentzos et al, 1997) reported very few deaths, which may or may not have been related to the EN or the PN
      • Heyland et al, 2003, in a positive quality meta-analysis, reviewed data from 14 RCTs and reported no difference in mortality between patients who received EN vs. PN
      • Another positive quality meta-analysis (Braunschweig et al, 2001) studied 27 PRCTs and also reported no difference in mortality among patients who received EN or PN.

      Infectious Complications

      • One positive quality PRCT (Kudsk et al, 1992) and one neutral quality RCT (Moore et al, 1989) examined the effect of early EN vs. early PN on infection rates and septic morbidity in patients with trauma and major injury
        • These studies demonstrated an overall decrease in septic morbidity in patients who received EN, compared to those who received PN.
      • Borzotta et al, 1994, in another positive quality PRCT, reported no difference in infection rate in a head injury patient who received EN vs. PN
      • Two positive quality RCTs (Kalfarentzos et al, 1997; Abou-Assi et al, 2002) showed that patients with acute pancreatitis who received EN had fewer infections and a lower rate of septic morbidity than did patients with the same diagnosis who received PN
      • One positive quality meta-analysis (Braunschweig et al, 2001) of 20 studies (total 508 EN and 525 PN patients) showed that EN was associated with significantly lower risk of infection (RR: 0.66; 95% CI: 0.56, 0.79) than was PN
      • Another positive quality meta-analysis demonstrated that patients who received EN were less likely to experience infectious complications than patients who received PN (Heyland et al, 2003)
      • Two neutral quality meta-analyses (Moore et al, 1992; Trice et al, 1997) suggested that EN was associated with fewer complications than was PN
      • Several older, likely underpowered neutral studies (Adams et al, 1986; Hadley et al, 1986) and a more recent one (Woodcock et al, 2001) failed to demonstrate a significant difference in septic morbidity among patients receiving EN vs. PN.

       Length of Hospital Stay

      • Three positive quality RCTs (Abou-Assi et al, 2002; Borzotta et al, 1994; Kudsk et al, 1992) and one neutral quality study (Adams et al,  1986) showed no difference in LOS in patients who received EN vs. PN
        • These underpowered studies were limited in methodology and do not provide sufficient evidence to support the assumption that EN vs. PN reduces LOS.
      • In a neutral quality meta-analysis (Moore et al, 1992), a small subset analysis demonstrated a significant difference in LOS in the penetrating trauma group
        • However, overall there was no difference in LOS among patients who received EN vs. PN.

        Cost of Medical Care

      • All studies supported the conclusion that EN therapy costs less than PN therapy
      • One positive quality RCT (Abou-Assi et al, 2002) reported a significant decrease in hospital cost in acute pancreatitis patients who received EN vs. PN; 1.8% vs. 8.4%, respectively (P<0.0001)
      • Based on the two positive quality RCTs (Borzotta et al, 1994; Kalfarentzos et al, 1997) and two neutral quality RCTs (Adams et al, 1986; Cerra et al, 1988), there is a significant daily per patient cost savings when EN vs. PN is used
      • A neutral quality meta-analysis (Trice et al, 1997) suggested that PN was four to 12.5 times more expensive than EN.

    • Recommendation Strength Rationale

      • Two positive quality meta-analyses (Heyland et al, 2003; Braunschweig et al, 2001) strongly recommend EN over PN in critically ill patients          
      • Studies represented a variety of critically ill and injured patients 
      • Conclusion statements are Grade I and II.