Skip to main content

Foundational Trials

PEPaNIC

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Early versus Late Parenteral Nutrition in Critically Ill Children
  • Acronym: PEPaNIC
  • Year: 2016
  • Journal published in: New England Journal of Medicine
  • Citation: Fivez T, Kerklaan D, Mesotten D, et al. Early versus late parenteral nutrition in critically ill children. N Engl J Med. 2016;374:1111-1122

Context & Rationale

  • Background
    • Paediatric critical care nutrition practice had been driven by the premise that early macronutrient replacement mitigates catabolism, supports immune function, and prevents cumulative energy/protein deficits.
    • However, parenteral nutrition (PN) carries recognised risks (catheter-related complications, hyperglycaemia, hepatic dysfunction, and potentially infection), and the metabolic “stress response” early in critical illness may render aggressive nutrient delivery biologically discordant.
    • Adult evidence (notably EPaNIC) challenged the early-PN paradigm by suggesting improved outcomes when PN was withheld during the first week, but paediatric equipoise persisted because children have different physiology, smaller reserves, and distinct case-mix (e.g., neonates, congenital heart disease).
  • Research Question/Hypothesis
    • In critically ill children at nutritional risk, does withholding supplemental PN during the first 7 days of paediatric ICU admission (while providing early enteral nutrition and micronutrients) reduce new infections and accelerate recovery compared with early initiation of PN to meet calculated caloric targets?
  • Why This Matters
    • PN timing is a high-frequency, high-cost ICU decision with direct implications for infection risk, duration of organ support, length of stay, and downstream neurodevelopmental outcomes in children.
    • Because “standard care” varied widely internationally, a definitive pragmatic randomised trial had potential to reset paediatric nutrition guidance and implementation pathways.

Design & Methods

  • Research Question: Among critically ill children at nutritional risk, does late initiation of PN (withholding PN until day 8) improve clinical outcomes versus early initiation of PN within 24 hours?
  • Study Type: Multicentre, randomised, controlled, parallel-group, open-label, superiority trial in tertiary paediatric ICUs (international).
  • Population:
    • Setting: Tertiary paediatric ICUs in Belgium (Leuven), the Netherlands (Rotterdam), and Canada (Edmonton).
    • Age: 0–17 years (including term neonates).
    • Clinical eligibility: Expected paediatric ICU stay >24 hours; inability to receive oral feeding at admission.
    • Nutritional risk enrichment: STRONGkids score ≥2 at admission (moderate/high nutritional risk).
    • Key exclusions: Not reported in the index manuscript excerpted here; protocol excluded predefined populations (e.g., prematurity, anticipated death very early, and conditions requiring specialised nutrition pathways). 1
  • Intervention:
    • Late PN (withhold PN days 1–7): No parenteral macronutrients during the first 7 days after paediatric ICU admission.
    • Fluids and glucose: Dextrose 5%/saline mixture used to match hydration needs; dextrose 10% permitted to correct hypoglycaemia (≤40 mg/dL).
    • Micronutrients: Intravenous vitamins, minerals, and trace elements from day 2 until ≥80% of caloric targets were achieved enterally.
    • After day 7: Supplemental PN initiated on day 8 if enteral intake remained insufficient to meet protocol targets.
  • Comparison:
    • Early PN: Supplemental PN initiated within 24 hours after admission when enteral nutrition did not achieve caloric targets.
    • Macronutrients: Protocolised PN initiation beginning with glucose/amino acids (day 1) and addition of lipids from day 2; titration to achieve caloric targets (with centre-specific products and practice pathways).
    • Micronutrients: Intravenous vitamins, minerals, and trace elements provided similarly (from day 2 until ≥80% of targets achieved enterally).
  • Blinding: Open-label for clinicians and families; infection diagnosis adjudicated by blinded infectious disease specialists; outcomes included objective endpoints (e.g., mortality, duration of organ support).
  • Statistics: Planned N=1,440 to address co-primary outcomes: (i) detect an absolute reduction in paediatric ICU infection incidence from 20% to 15% with ≥80% power (one-sided α=0.05) and (ii) detect a 1-day reduction in mean ICU length of stay with 90% power (two-sided α=0.05); primary analyses were intention-to-treat, using multivariable logistic regression (new infection) and time-to-event modelling (time to live discharge). 1
  • Follow-Up Period: Outcomes assessed during paediatric ICU and hospital stay; mortality tracked to 90 days after enrolment.

Key Results

This trial was not stopped early. Enrolment and follow-up proceeded to the planned sample size.

Outcome Late PN (withhold PN days 1–7) Early PN Effect p value / 95% CI Notes
New infection (any), during paediatric ICU stay 77/717 (10.7%) 134/723 (18.5%) OR 0.48 95% CI 0.35 to 0.66; P<0.001 Adjusted analysis; OR is late PN vs early PN (OR<1 favours late PN).
Time to live discharge from paediatric ICU Mean 6.5 ± 0.4 days Mean 9.2 ± 0.8 days HR 1.23 95% CI 1.11 to 1.37; P<0.001 Adjusted analysis; HR>1 indicates earlier live ICU discharge with late PN; mean difference 2.7 days shorter (95% CI 1.3 to 4.3) reported.
New airway infection 30/717 (4.2%) 59/723 (8.2%) OR 0.52 95% CI 0.33 to 0.81; P=0.004 Adjusted analysis; late PN vs early PN.
New bloodstream infection 41/717 (5.7%) 65/723 (9.0%) OR 0.57 95% CI 0.38 to 0.87; P=0.01 Adjusted analysis; late PN vs early PN.
Hypoglycaemia (≤40 mg/dL) within first 7 days 65/717 (9.1%) 35/723 (4.8%) Not reported P=0.001 Trade-off signal; protocol allowed dextrose 10% to correct hypoglycaemia.
Duration of mechanical ventilatory support Mean 4.4 ± 0.3 days Mean 6.4 ± 0.7 days HR 1.19 95% CI 1.07 to 1.32; P=0.001 Adjusted analysis; HR>1 indicates earlier weaning with late PN.
Acute kidney injury requiring renal-replacement therapy 18/717 (2.5%) 26/723 (3.6%) OR 0.49 95% CI 0.24 to 0.96; P=0.04 Adjusted analysis; late PN vs early PN.
Death within 90 days after enrolment 38/717 (5.3%) 49/723 (6.8%) OR 0.64 95% CI 0.39 to 1.05; P=0.08 Adjusted analysis; not statistically significant.
Total hospital length of stay (index + transfer hospitals) Mean 18.6 ± 1.0 days Mean 22.6 ± 1.3 days HR 1.21 95% CI 1.08 to 1.34; P<0.001 Adjusted analysis; mean difference 4.1 days shorter (95% CI 1.4 to 6.6) reported.
  • Late PN reduced new infections by an absolute 7.8% (10.7% vs 18.5%) and improved time-to-recovery metrics (ICU discharge and ventilation weaning), without a statistically significant mortality difference.
  • Late PN increased hypoglycaemia (9.1% vs 4.8%), emphasising the need for protocolised glucose monitoring and rapid correction pathways when adopting delayed PN strategies.
  • Benefits persisted in adjusted analyses, supporting a causal interpretation beyond baseline risk-factor imbalance.

Internal Validity

  • Randomisation and Allocation: Central, computer-based allocation with stratification by centre, age group (<1 year vs ≥1 year), and admission subgroup (medical neurologic vs other; surgical cardiac vs other); allocation concealment was operationalised by electronic assignment at enrolment.
  • Drop out or exclusions: 1,440 children were randomised (723 early PN; 717 late PN); intention-to-treat analysis was performed; post-randomisation exclusions were not reported in the index manuscript.
  • Performance/Detection Bias: Open-label design introduces potential for clinician behaviour change (e.g., discharge timing decisions), but major endpoints were objective and infection outcomes were adjudicated by assessors blinded to treatment assignment.
  • Protocol Adherence: The intervention was operationally “hard” (withholding PN macronutrients for 7 days vs initiating within 24 hours), with prespecified rescue for hypoglycaemia (dextrose 10%); detailed quantitative caloric/protein separation was presented graphically rather than as tabulated per-day numeric values.
  • Baseline Characteristics: Groups were clinically similar (e.g., median age 1.4 vs 1.5 years; male sex 57.4% vs 57.9%; emergency admission 53.0% vs 55.8%; baseline PELOD median 21 in both groups).
  • Heterogeneity: Multicentre design increases clinical heterogeneity (case-mix and centre-specific nutrition products); the primary effect estimates remained robust after adjustment, and prespecified subgroup analyses did not demonstrate major qualitative effect reversal.
  • Timing: The timing contrast was clinically meaningful (early PN within 24 hours vs no PN macronutrients during days 1–7, with PN possible from day 8), directly testing a real-world decision point.
  • Dose: Caloric targets were protocolised but based on standard predictive approaches rather than direct measurement of energy expenditure; macronutrient formulations differed by centre according to local practice, potentially affecting the “dose” and biological plausibility of early PN harm/benefit.
  • Separation of the Variable of Interest: Late PN vs early PN produced demonstrable metabolic divergence (hypoglycaemia 9.1% vs 4.8% within 7 days) and downstream clinical separation (ICU stay mean 6.5 ± 0.4 vs 9.2 ± 0.8 days; ventilation mean 4.4 ± 0.3 vs 6.4 ± 0.7 days).
  • Key Delivery Aspects: Both groups received early enteral feeding attempts and micronutrient provision; thus, the contrast was primarily supplemental macronutrients via PN during the first week.
  • Crossover: Not reported.
  • Adjunctive therapy use: Not reported.
  • Outcome Assessment: New infection required antimicrobial therapy and was adjudicated blinded; time-to-discharge outcomes used time-to-event methods and clinically relevant endpoints.
  • Statistical Rigor: Co-primary outcomes were analysed with prespecified multivariable and time-to-event models; effect estimates were presented with confidence intervals; the trial achieved planned sample size and statistical separation on co-primary endpoints.

Conclusion on Internal Validity: Overall, internal validity appears strong given central randomisation, intention-to-treat analysis, large sample size, and blinded adjudication of infections, with the principal limitation being the open-label design (notably for time-to-discharge outcomes).

External Validity

  • Population Representativeness: Broad tertiary PICU case-mix with a high proportion of infants and substantial surgical/cardiac workload (e.g., cardiac surgery accounted for ~40% of admissions in both groups), reflecting many high-income PICU settings.
  • Applicability: Findings are most directly applicable to PICUs where early PN to meet calculated targets is common when enteral nutrition is insufficient; applicability to units with markedly different baseline nutrition practices (or substantially higher prevalence of severe chronic malnutrition) requires contextual judgement.
  • Health-system translation: Implementation requires robust early enteral feeding pathways, micronutrient strategies, and glucose safety protocols to manage the increased hypoglycaemia signal with late PN.

Conclusion on External Validity: Generalisability is moderate-to-high across comparable tertiary PICUs in high-resource systems, with caution for patient populations not well represented (e.g., preterm infants, highly specialised metabolic conditions, and settings with very different baseline PN availability).

Strengths & Limitations

  • Strengths:
    • Large, pragmatic, multicentre international RCT (N=1,440) focused on a highly actionable ICU decision (timing of supplemental PN).
    • Clinically important co-primary outcomes (new infection; time to live ICU discharge) with consistent benefit signals across multiple recovery endpoints.
    • Blinded adjudication of infection outcomes reduced detection bias for a key endpoint.
    • Protocolised micronutrient strategy reduced the risk that findings reflect vitamin/trace element deficiency rather than macronutrient timing.
  • Limitations:
    • Open-label design, which can influence co-interventions and discharge decisions (despite objective components and time-to-event modelling).
    • Macronutrient formulations and glucose management practices varied by centre, introducing clinical heterogeneity in the “early PN dose”.
    • Energy requirements were target-driven rather than measured by indirect calorimetry; this may affect interpretation regarding “underfeeding” versus “overfeeding”.
    • Detailed numeric separation in delivered calories/protein was largely shown graphically rather than in per-day tabulated form in the main manuscript.

Interpretation & Why It Matters

  • Clinical practice
    For most critically ill children at nutritional risk, a default strategy of early enteral nutrition with micronutrients and withholding supplemental PN during the first week improves recovery (infection reduction; shorter ventilation/ICU/hospital stay), provided hypoglycaemia is actively prevented and treated.
  • Trialists & methodologists
    PEPaNIC illustrates how a biologically grounded, system-level exposure (nutrient timing) can yield large effect sizes on infection and length-of-stay outcomes, and underscores the value of blinded endpoint adjudication in otherwise unblinded ICU trials.
  • Risk–benefit framing
    The principal “cost” of late PN was increased hypoglycaemia, shifting implementation from “avoid PN” to “avoid PN early, but execute safe fasting-with-support protocols”.

Controversies & Subsequent Evidence

  • External validity and patient mix: The accompanying editorial highlighted that many children had short paediatric ICU stays, meaning that “withhold PN until day 8” functionally prevents PN exposure in a large subgroup and may be most impactful in settings where early PN is frequently given to short-stay patients. 2
  • Energy prescription and possible overfeeding: Editorial and correspondence emphasised that caloric targets were equation-based (not indirect calorimetry) and could overestimate needs; thus, the early PN arm may have been exposed to early overfeeding rather than “physiological replacement”. 23
  • Hypoglycaemia as an implementation hazard: Correspondence noted increased hypoglycaemia with delayed PN; this is clinically important because paediatric hypoglycaemia is not a “surrogate-only” signal, and protocols must specify monitoring and rescue dextrose pathways when adopting late PN strategies. 3
  • Commentary on practice implications: A focused nutrition commentary (post-publication) framed the PEPaNIC results as supporting “permissive underfeeding” early in paediatric critical illness, while emphasising careful selection and monitoring in vulnerable subgroups. 4
  • Subgroup and extended analyses: A secondary analysis in acutely undernourished children reported that late PN remained associated with fewer infections and faster discharge compared with early PN. 5
  • Term neonates: A prespecified secondary analysis in critically ill term neonates reported outcomes consistent with benefit from withholding PN early, informing neonatal practice debates (distinct from preterm neonatal nutrition). 6
  • Long-term neurodevelopment: Two-year and four-year follow-up studies reported no signal of neurodevelopmental harm and suggested improved long-term developmental outcomes with late PN, shifting interpretation from short-term “length-of-stay only” to potentially meaningful child-centred endpoints. 78
  • Economic and mechanistic extensions: A cost-effectiveness analysis supported late PN as cost-saving with improved outcomes; mechanistic analyses (including fasting-response biology such as ketogenesis) strengthened plausibility that early macronutrients can be maladaptive in early critical illness. 910
  • Guideline incorporation: ESPNIC guidance states that withholding PN for up to one week can be considered in critically ill children, with decisions individualised to nutritional status and illness course. 11
  • North American guidance: SCCM/ASPEN paediatric nutrition guidance acknowledged emerging evidence that a delayed approach to PN appears beneficial, aligning paediatric recommendations more closely with PEPaNIC-derived practice. 12
  • Evidence synthesis: Recent systematic reviews in adjacent populations (including paediatric and neonatal critical illness) continue to evaluate early versus delayed PN strategies and generally reinforce that routine early PN confers limited benefit and requires careful harm surveillance (particularly hypoglycaemia). 131415

Summary

  • In 1,440 critically ill children at nutritional risk, withholding PN during the first week reduced new infections (10.7% vs 18.5%; OR 0.48; 95% CI 0.35 to 0.66; P<0.001).
  • Late PN accelerated recovery: mean ICU stay 6.5 ± 0.4 vs 9.2 ± 0.8 days (HR 1.23; 95% CI 1.11 to 1.37; P<0.001) and shortened mechanical ventilation (mean 4.4 ± 0.3 vs 6.4 ± 0.7 days; HR 1.19; 95% CI 1.07 to 1.32; P=0.001).
  • Mortality to 90 days was not statistically different (5.3% vs 6.8%; OR 0.64; 95% CI 0.39 to 1.05; P=0.08).
  • Late PN increased hypoglycaemia within 7 days (9.1% vs 4.8%; P=0.001), making glucose safety a key implementation requirement.
  • Subsequent analyses, long-term follow-up, and contemporary guidelines broadly support late PN as a default strategy for many PICU patients, while highlighting subgroup-specific judgement and monitoring.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Some Further Reading entries derived from secondary analyses report page ranges as not consistently available from open metadata; journal landing pages contain full citation details.

Overall Takeaway

PEPaNIC reframed early paediatric critical illness nutrition: routine supplemental PN to meet calculated targets in the first week increased infections and prolonged recovery, whereas withholding PN (with micronutrient support and careful glucose management) improved clinically meaningful outcomes. Together with mechanistic and long-term follow-up studies and evolving guidelines, it underpins a modern default of prioritising enteral nutrition and deferring PN until after the first week for many PICU patients, while identifying hypoglycaemia prevention as an implementation-critical safeguard.

Overall Summary

  • Withholding PN for 7 days in at-risk PICU children reduced infections and shortened ICU/hospital stay.
  • No statistically significant mortality difference was demonstrated to 90 days.
  • Late PN increased hypoglycaemia, mandating proactive monitoring and rescue glucose protocols.

Bibliography