Skip to main content

Foundational Trials

Heidegger

Image: Shutterstock / luchschenF

Publication

  • Title: Optimisation of energy provision with supplemental parenteral nutrition in critically ill patients: a randomised controlled clinical trial
  • Acronym: Not applicable
  • Year: 2013
  • Journal published in: The Lancet
  • Citation: Heidegger CP, Berger MM, Graf S, et al. Optimisation of energy provision with supplemental parenteral nutrition in critically ill patients: a randomised controlled clinical trial. Lancet. 2013;381(9864):385-393.

Context & Rationale

  • Background
    • Enteral nutrition (EN) is first-line in critical illness, but delivery commonly falls short of intended energy and protein targets.
    • Cumulative energy/protein deficits were associated (observationally) with worse outcomes, including infectious complications.
    • Early “full” parenteral nutrition (PN) remained contentious because of overfeeding risk, hyperglycaemia, and potential infectious harm.
    • A key unresolved question was whether a delayed “step-up” strategy—adding supplemental PN only when EN remains insufficient—could improve outcomes without the harms attributed to early PN.
  • Research Question/Hypothesis
    • In ICU patients receiving <60% of a prespecified energy target by day 3 despite EN, does adding supplemental PN from day 4–8 to reach 100% of the target reduce subsequent nosocomial infections compared with EN alone?
  • Why This Matters
    • Nosocomial infection is a major determinant of antibiotic exposure, ventilator dependence, length of stay, and cost in the ICU.
    • Nutrition strategy (route and timing) is a ubiquitous ICU decision with substantial practice variation; rigorous trial data were needed to support (or refute) a “supplemental PN” approach.

Design & Methods

  • Research Question: In adult ICU patients underfed by EN at day 3 (<60% of energy target), does adding supplemental PN for days 4–8 to meet 100% of the energy target reduce nosocomial infection compared with EN alone?
  • Study Type: Randomised, controlled, investigator-initiated, two-centre (Swiss university hospital ICUs), open-label trial; stratified randomisation; infection ascertainment and statisticians masked.
  • Population:
    • Setting: mixed medical-surgical ICUs in two Swiss university hospitals (Geneva and Lausanne).
    • Eligibility at randomisation (day 3): receiving EN but failing to reach 60% of calculated/measured energy target.
    • Expected ICU stay >5 days and expected survival >7 days.
    • Energy target: indirect calorimetry on day 3 when feasible; if not feasible, Harris–Benedict equation or a simplified 25 kcal/kg estimate.
    • Nosocomial infections were counted from day 9 to day 28 and defined using Centres for Disease Control and Prevention criteria across five prespecified categories (respiratory, bloodstream, urogenital, abdominal, other).
  • Intervention:
    • EN continued plus supplemental PN (SPN) from day 4 to day 8.
    • SPN titrated daily to achieve 100% of the prespecified energy target (effectively “filling the gap” left by EN), using an all-in-one PN admixture.
    • PN delivered via central venous catheter if present (otherwise peripheral venous access).
    • Co-interventions relevant to nutrition delivery: daily accounting of delivered calories/protein; glycaemic management via insulin with an algorithmised approach.
  • Comparison:
    • EN alone during the intervention window (days 4–8), with ongoing attempts to optimise EN as per routine care.
    • No supplemental PN during the intervention window (per protocol).
  • Blinding: Open-label to clinicians/patients; infection data were collected by an investigator from the alternate centre who was unaware of group assignment and diagnoses were confirmed by an independent infectious diseases specialist; statisticians were masked.
  • Statistics: Power calculation: 296 patients required to detect a 33% relative reduction in nosocomial infection rate (assumed 50% baseline), with 80% power (alpha not reported). Primary analysis: intention-to-treat; time-to-event methods (Cox proportional hazards) for first nosocomial infection (day 9–28); secondary outcomes via regression models (with multiplicity control reported using Benjamini–Hochberg).
  • Follow-Up Period: 28 days (primary infection outcome assessed days 9–28).

Key Results

This trial was not stopped early. Recruitment completed as planned (n=305).

Outcome Supplemental PN + EN (n=153) EN alone (n=152) Effect p value / 95% CI Notes
Nosocomial infection (day 9–28) — primary 41/153 (27%) 58/152 (38%) HR 0.65 95% CI 0.43 to 0.97; P=0.0338 Time-to-first infection; infections defined by CDC criteria; diagnoses confirmed by infectious diseases specialist.
Energy delivery (days 4–8) 28 ± 6 kcal/kg/day 20 ± 5 kcal/kg/day Not reported P<0.0001 Process separation variable: near-target delivery achieved via SPN.
Protein delivery (days 4–8) 1.2 ± 0.3 g/kg/day 0.8 ± 0.2 g/kg/day Not reported P<0.0001 Protein delivery increased alongside energy delivery (cannot disentangle energy vs protein effects).
Days with antibiotic therapy (day 9–28 follow-up) 6 (7) 8 (8) β −2.3 95% CI −4.1 to −0.5; P=0.0010 Regression coefficient as reported; adjusted model.
Mechanical ventilation hours in patients without nosocomial infection (day 9–28 follow-up) 15 (59) 29 (61) β −1.3 95% CI −2.1 to −0.4; P=0.0028 Post-hoc restriction to those without infection (as reported); interpret cautiously.
28-day all-cause mortality 20/153 (13%) 28/152 (18%) HR 0.6 95% CI 0.3 to 1.2; P=0.1193 Not statistically significant.
ICU mortality 8/153 (5%) 12/152 (8%) HR 0.6 95% CI 0.2 to 1.6; P=0.2118 Not statistically significant.
  • Supplemental PN produced clear biological separation: mean energy delivery 28 ± 6 vs 20 ± 5 kcal/kg/day and protein delivery 1.2 ± 0.3 vs 0.8 ± 0.2 g/kg/day during days 4–8.
  • The primary endpoint was reduced: 27% vs 38% experienced a nosocomial infection between days 9 and 28 (HR 0.65; 95% CI 0.43 to 0.97; P=0.0338).
  • Mortality differences were not statistically significant at 28 days; antibiotic exposure during follow-up was lower in the SPN group (6 (7) vs 8 (8) days; β −2.3; 95% CI −4.1 to −0.5; P=0.0010).

Internal Validity

    • Randomisation and allocation concealment: stratified allocation with block size of four; sequentially numbered, sealed, opaque envelopes used, supporting concealment.
    • Dropouts/exclusions (post-randomisation): intervention discontinued in 20/153 in the SPN arm (protocol violations 16; GI complications 2; withdrawal 2) and 10/152 in EN (protocol violations 7; GI complication 1; withdrawal 2); per-protocol population 133 vs 142, but the primary analysis was intention-to-treat.
    • Performance/detection bias: open-label delivery could influence co-interventions (e.g., antibiotic thresholds, line care, feeding advancement); mitigation included blinded cross-centre infection data collection and independent infectious diseases confirmation, plus masked statisticians.
    • Protocol adherence and separation of exposure: substantial separation in nutrition delivery during the intervention: 28 ± 6 vs 20 ± 5 kcal/kg/day and 1.2 ± 0.3 vs 0.8 ± 0.2 g/kg/day.
    • Baseline comparability: groups broadly similar in severity (SAPS II 49 vs 47; APACHE II 22 vs 23), age (61 vs 60 years), and surgical admissions (46% vs 45%); some baseline diagnostic imbalances existed but randomisation and adjusted models reduce (not eliminate) concern.
    • Timing: delayed enrolment/randomisation (day 3) enriched for patients with persistent feeding deficit; infections were counted from day 9–28, reducing contamination by early infections likely unrelated to the intervention window.
    • Dose: targeting 100% of measured/prespecified energy expenditure (indirect calorimetry preferred) reduces overfeeding risk relative to formula-only approaches, but energy and protein were both increased, limiting mechanistic attribution.
    • Crossover/contamination: supplemental PN in the control group during days 4–8 was not part of the protocol; PN use after day 8 was not reported, which could affect longer-term separation.
    • Outcome assessment: the primary outcome (nosocomial infection) requires clinical adjudication; prespecified CDC definitions and independent confirmation strengthen credibility, but complete objectivity is not achievable in an open-label design.
    • Statistical rigour: prespecified sample size was met (305 vs 296 planned); intention-to-treat analysis used for primary outcome; time-to-event modelling appropriate for infection timing; multiplicity control applied for secondary outcomes.

Conclusion on Internal Validity: Moderate-to-strong: allocation concealment and meaningful exposure separation support causal inference, but open-label delivery and a partly adjudicated primary outcome leave residual risk of performance and ascertainment bias.

External Validity

    • Population representativeness: the cohort represents a selected ICU subgroup—patients still receiving <60% of energy targets by EN at day 3 and expected to require prolonged ICU care—rather than the general ICU population.
    • Setting and feasibility: two well-resourced Swiss ICUs with access to indirect calorimetry and structured nutrition delivery; infection reduction may be harder to reproduce where PN delivery/line care and calorimetry are less standardised.
    • Applicability: findings most applicable to high-risk, persistently underfed patients in whom EN optimisation alone is unlikely to meet targets; results should not be extrapolated to routine early PN in unselected ICU patients.

Conclusion on External Validity: Generalisability is focused rather than broad: the results support a “step-up” SPN strategy for persistently underfed ICU patients, particularly in centres capable of safe PN delivery and avoidance of overfeeding.

Strengths & Limitations

  • Strengths: concealed allocation; clinically meaningful primary endpoint with prespecified CDC definitions; deliberate separation window (days 4–8 intervention; infections counted days 9–28); indirect calorimetry preferred for targeting; meaningful separation in delivered energy/protein; masked statisticians and partially blinded infection ascertainment/confirmation.
  • Limitations: open-label care with potential co-intervention bias; two-centre design may limit external validity; protocol violations and discontinuations were more frequent in SPN; primary endpoint depends on clinical adjudication; intervention increased both energy and protein, limiting mechanistic attribution; no statistically significant mortality benefit demonstrated.

Interpretation & Why It Matters

  • Clinical significance
    • Among ICU patients still substantially underfed by EN on day 3, a time-limited SPN strategy to reach energy targets reduced nosocomial infection and antibiotic exposure without a clear signal of mortality benefit.
    • The trial operationalises a pragmatic “step-up” paradigm: tolerate early EN deficits, but correct persistent deficits after several days, ideally guided by measured energy expenditure.
    • Because protein delivery increased in parallel with energy, the benefit cannot be attributed uniquely to calories or to PN as a route.

Controversies & Subsequent Evidence

  • Editorial interpretation framed the trial as supporting delayed, targeted SPN (rather than routine early PN), with indirect calorimetry as a safeguard against overfeeding. 1
  • Correspondence raised concerns that the control group remained markedly underfed and that the observed effect might reflect correction of energy/protein deficit rather than a route-specific benefit; an isocaloric/isoprotein comparator was not tested. 236
  • Open-label delivery plus an adjudicated infection endpoint was identified as a potential bias vector (co-interventions and ascertainment), albeit mitigated by blinded cross-centre data collection and independent confirmation. 5
  • Authors emphasised that enrolment targeted patients with persistent EN failure at day 3 and infections were counted from day 9 to reduce misclassification of early infections unrelated to the intervention window. 7
  • Subsequent RCTs suggest timing and patient selection are critical when reconciling PN trials: early full PN in broadly eligible ICU patients (EPaNIC) differed materially from delayed SPN in underfed patients; pragmatic route comparison (CALORIES) found no mortality difference between early EN and early PN. 89
  • In patients with short-term barriers to EN, early PN has been tested without an apparent excess of infectious harm in a randomised trial, supporting context-dependent use when EN is not feasible. 10
  • A contemporary multicentre shock population trial comparing early enteral vs parenteral nutrition (NUTRIREA-3) reinforces that “route” questions remain nuanced and highly population-dependent. 11
  • Guidelines generally endorse early EN when feasible and a “step-up” approach to supplemental PN when EN remains insufficient after several days, with emphasis on patient nutritional risk and safe delivery processes. 1213
  • Meta-analytic syntheses of supplemental PN trials report heterogeneous effects and reinforce that trial context (timing, calorimetry, baseline deficits, protein co-delivery, and infection definitions) materially shapes estimated benefit. 14

Summary

  • In 305 ICU patients underfed by EN at day 3, adding supplemental PN for days 4–8 reduced nosocomial infections during days 9–28: 27% vs 38% (HR 0.65; 95% CI 0.43 to 0.97; P=0.0338).
  • The intervention achieved substantial nutrition separation: 28 ± 6 vs 20 ± 5 kcal/kg/day and 1.2 ± 0.3 vs 0.8 ± 0.2 g/kg/day during the intervention window.
  • No statistically significant difference in 28-day mortality was shown (13% vs 18%; HR 0.6; 95% CI 0.3 to 1.2).
  • Antibiotic exposure during follow-up was lower with SPN (6 (7) vs 8 (8) days; β −2.3; 95% CI −4.1 to −0.5; P=0.0010).
  • Findings primarily inform a targeted “step-up” SPN strategy in persistently underfed patients, not routine early PN in unselected ICU populations.

Overall Takeaway

This trial is a landmark demonstration that, in a carefully selected ICU subgroup with persistent EN under-delivery, a delayed “step-up” supplemental PN strategy aimed at measured energy targets can reduce nosocomial infections. Its clinical message is not “early PN for all”, but rather that persistent underfeeding after several days may be modifiable—and that timing, targeting (ideally by indirect calorimetry), and safe PN delivery processes are central to any benefit.

Overall Summary

  • Delayed supplemental PN (days 4–8) in persistently underfed ICU patients reduced subsequent nosocomial infections (day 9–28) without a clear mortality benefit.

Bibliography