CALORIES (2014): Early Enteral vs. Parenteral Nutrition in ICU

Clinical Question

In critically ill patients, is early parenteral nutrition administration superior to early enteral feeding with respect to all-cause mortality?

Bottom Line

No significant differences in 30-day all-cause mortality were found between the early enteral and early parenteral arms.

Major Points

Enteral nutritional support has been the standard of care for critically ill patients who can tolerate it, due in large part to the widespread belief that it is physiologically superior and spares the patient the risks that are commonly associated with parenteral nutrition. However, enteral feeding is associated with its own unique profile of adverse effects and risks, most notably mechanical complications and associated underfeeding. In 2011, the Early Parenteral Nutrition Completing Enteral Nutrition in Adult Critically Ill Patients (EPaNIC) trial demonstrated a decrease in ICU mortality and number of ICU days for patients for whom parenteral support was delayed until hospital day 8. However, other randomized controls trials have demonstrated equivocal or even improved outcomes with early parenteral support with respect to nosocomial infection rate and mortality. This study provides a necessary direct comparison between these routes of delivery in critically ill patients who are candidates for being fed via either route.

The enteral and parenteral groups had near-identical mortality rates at 30 days. Most secondary outcomes were similar between the groups, with the exceptions of incidences of vomiting and hypoglycemia which both occurred more frequently in patients fed enterally. Notably, parenteral support was not associated with an increased risk of infectious complications. It is important to note that while more calories were delivered to the parenteral arm of the trial, neither group met the study's pre-specified caloric goal of 25 kcal/kg/day. The results of this study suggest that early parenteral nutrition administration may not be harmful when compared to standard enteral feeding.

Guidelines

American Society for Parenteral and Enteral Nutrition/Society of Critical Care Medicine Guidelines (2016) :

• Recommend initiation of enteral support in critically ill patient within 24-48 hours.
• Recommend use of enteral over parenteral nutrition in critically ill patients who can be fed with the former.
• For critically ill patients deemed low nutrition risk who cannot be fed enterally, recommend against use of parenteral support in first week.
• For critically ill patients deemed high nutrition risk who cannot be fed enterally, recommend initiation of parenteral support as soon as feasible.

European Society for Clinical Nutrition and Metabolism Guidelines (2018):

• Recommend initiation of enteral support in all eligible critically ill patients within 48 hours.
• In all critically ill patients with contraindications to enteral feeding, recommend initiation of parenteral support within three to seven days.

Design

• Multicenter, randomized, pragmatic, controlled trial
• N=2388
  • Parenteral (1191)
  • Enteral (1197)
• Setting: 33 intensive care units in England
• Enrollment: June 2011 to March 2014
• Analysis: Intention-to-treat
• Primary outcome: 30-day all-cause mortality
• Secondary outcomes:
  • Days free of organ-specific support
  • Infectious complications
  • Non-infectious complications
  • Duration of stay in ICU
  • Duration of stay in hospital
  • ICU mortality
  • In-hospital mortality
  • 90-day mortality

Population

Inclusion Criteria

• Unplanned ICU admission
• Planned ICU stay of at least 3 days; nutritional support duration of at least 2 days
• Age at least 18 years

Exclusion Criteria

• Nutritional support in preceding 7 days
• Indwelling gastrostomy or jejunostomy
• Pregnancy

Baseline Characteristics

(Presented as parenteral vs. enteral, means unless otherwise specified)

• Age (yrs): 63.3 vs. 62.9
• Male sex (n(%)): 689(57.9) vs. 725(60.6)
• Severe coexisting illness (n/total(%))
  • Liver: 29/1181 (2.5) vs. 34/1193 (2.8)
  • Renal: 20/1181 (1.7) vs. 15/1193 (1.3)
  • Respiratory: 34/1181 (2.9) vs. 23/1193 (1.9)
  • Cardiovascular 11/1181 (0.9) vs. 14/1193 (1.2)
  • Immunodeficiency 78/1181 (6.6) vs. 95/1193 (8.0)
• Surgery within 24 hrs of admission (n(%)): 162(13.6) vs. 167(14.0)
• APACHE II scores:
  • Acute physiology score: 15.1 vs. 15.2
  • Total score: 19.6 vs. 19.6
  • Predicted risk of death (median): 0.34 vs. 0.34
• ICNARC scores:
  • Physiology score: 25.6 vs. 25.8
  • Predicted risk of death (median): 0.42 vs. 0.43
• Mechanical ventilation (n/total(%)): 979/1178 (83.1) vs. 993/1185 (83.8)
• SOFA score: 9.5 vs. 9.6
• Subjective assessment of severe malnutrition (n(%)): 151(12.7) vs. 152(12.7)
• Actual or estimated BMI: 27.7 vs. 28.2
• Degree of malnutrition (n/total(%))
  • High: 74/1152 (6.4) vs. 81/1161 (7.0)
  • Moderate: 8/1152 (0.7) vs. 10/1161 (0.9)
  • None: 1070/1152 (92.9) vs. 1070/1161 (92.2)

Interventions

• Within 36 hours of ICU admission, subjects were randomized to one of the following feeding strategies:
  • Parenteral nutrition via central venous catheter with dedicated lumen
  • Enteral feeding via nasogastric vs. nasojejunal tube
• Energy target was 25 kcal/kg for each arm, with aim to reach goal feeding within 48-72 hours of feeding initiation
• Patients were fed exclusively via pre-specified route for 5 days, or until these pre-determined end-points:
  • Discharge from ICU
  • Death
  • Transition to exclusive oral feeding
• Glycemic control goal for subjects was target blood glucose < 180 mg/dL
• Note pragmatic design to study: choice of feeding formula/composition left to discretion of local practitioners

Outcomes

Comparisons are intensive therapy vs. standard therapy.

Primary Outcomes

(Parenteral vs. enteral unless otherwise specified:)

Death within 30 days (n/total (%))

393/1188 (33.1) vs. 409/1195 (34.2) (RR 0.97; 95% CI 0.86-1.08; P=0.57)

Secondary Outcomes

(Note: events counted up to 30 days unless otherwise specified)

Days free from advanced respiratory support (means)

14.3 vs. 14.3 (P=0.94)

Days free from advanced cardiovascular support (means)

18.9 vs. 18.5 (P=0.44)

Days free from renal support (means)

19.1 vs. 18.8 (P=0.66)

Days free from neurologic support (means)

19.2 vs. 18.9 (P=0.57)

Days free from gastrointestinal support (means)

13.0 vs. 13.2 (P=0.81)

Number of treated infectious complications per patient (means)

0.22 vs. 0.21 (P=0.72)

Hypoglycemic episodes (n/total (%))

44/1191 (3.7) vs. 74/1197 (6.2) (P=0.006)

Elevated liver enzymes (n/total (%))

212/1191 (17.8) vs. 179/1197 (15.0) (P=0.07)

Nausea requiring treatment (n/total (%))

44/1191 (3.7) vs. 53/1197 (4.4) (P=0.41)

Abdominal distention (n/total (%))

78/1191 (6.5) vs. 99/1197 (8.3) (P=0.12)

Vomiting (n/total (%))

100/1191 (8.4) vs. 194/1197 (16.2) (P<0.001)

New or worsening pressure ulcers (n/total (%))

181/1190 (15.2) vs. 179/1195 (15.0) (P=0.91)

Median number of ICU days

8.1 vs. 7.3 (P=0.15)

Median number of hospital days

17 vs. 16 (P=0.32)

Death in ICU (n/total (%))

317/1190 (26.6) vs. 352/1197 (29.4) (RR 0.91, 95% CI 0.80-1.03, P=0.13)

Death in hospital (n/total (%))

431/1185 (36.4) vs. 450/1186 (37.9) (RR 0.96, 95% CI 0.86-1.06, P=0.44)

Death at 90 days (n/total (%))

442/1184 (37.3) vs. 464/1188 (39.1) (RR 0.96, 95% CI 0.86-1.06, P=0.40)

Adverse Events

(Parenteral vs. enteral)

Serious adverse events (n (%))

58 (4.9) vs. 58 (4.8) (P=1.00)

Severe hypoglycemic episode (n (%))

5 (0.4) vs. 3 (0.3) (P=1.00)

Insulin requirement during intervention period (n/total (%))

694/1184 (58.6) vs. 668/1191 (56.1)

Regurgitation or aspiration event (n (%))

2 (0.2) vs. 4 (0.3) (P=1.00)

Criticisms

• Neither arm was fed to the pre-specified energy threshold of 25 kcal/day, calling into question whether the study was adequately powered from a treatment dose perspective. It is possible that one approach is superior in this patient population, but that these differences were not observed due to relative underfeeding.

• The pragmatic design of this trial allowed for excellent subject accrual and powering, but also raises the concern for generalizability given that nutritional support dosages and compositions were not thoroughly standardized.

• This study was non-blinded. Many of the secondary outcomes are susceptible to observer bias, though notably the primary outcome of interest was not.

Funding

This study was funded by the United Kingdom National Institute for Health Research.