Skip to main content

Foundational Trials

TARGET Protein

Image: Shutterstock / luchschenF

Publication

  • Title: Augmented Enteral Protein During Critical Illness: The TARGET Protein Randomized Clinical Trial
  • Acronym: TARGET Protein
  • Year: 2025
  • Journal published in: JAMA
  • Citation: Summers MJ, Chapple LS, Karahalios A, Bellomo R, Chapman MJ, Ferrie S, et al; TARGET Protein Investigators; Australian and New Zealand Intensive Care Society Clinical Trials Group. Augmented enteral protein during critical illness: the TARGET Protein randomized clinical trial. JAMA. 2025;334(4):319-328.

Context & Rationale

  • Background
    • Critical illness is characterised by severe catabolism, muscle wasting, and persistent functional disability among survivors.
    • ICU nutrition guidelines recommend relatively high protein delivery targets, yet routine practice often achieves substantially less protein than recommended.
    • Observational studies linking higher protein delivery to improved outcomes are vulnerable to confounding and time-dependent (survivorship) bias.
    • Randomised evidence for “more protein” has historically been limited and heterogeneous in dose, timing, and co-interventions, leaving uncertainty about benefit, harm, and subgroups.
  • Research Question/Hypothesis
    • Does routine substitution to an isocaloric, higher-protein enteral formula (vs a usual-protein formula) improve a patient-centred 90-day outcome in adult ICU patients prescribed enteral nutrition?
    • Hypothesis: augmented enteral protein delivery would increase the number of days free of the index hospital and alive at day 90.
  • Why This Matters
    • High-protein formulas are widely used, relatively low-cost, and easy to implement at unit level, but the clinical value of routine protein augmentation is uncertain.
    • A pragmatic, unit-level strategy that increases protein without materially changing calories is an important test of whether protein itself is a causal lever for recovery.
    • Using “days alive and out of hospital” integrates mortality, length of stay, and readmissions, aligning nutrition trials with outcomes that matter to patients and health systems.

Design & Methods

  • Research Question: In adult ICU patients prescribed enteral nutrition, does an isocaloric, higher-protein enteral formula (vs a usual-protein formula) increase days free of the index hospital and alive at day 90?
  • Study Type: Investigator-initiated, pragmatic, cluster-randomised, cross-sectional, double cross-over, open-label trial in 8 mixed adult ICUs in Australia and New Zealand; staggered commencement in 2 strata; outcomes derived from ICU registry and hospital administrative data; protocol and statistical analysis plan published a priori.12
  • Population:
    • Setting: 8 adult ICUs; each ICU participated for 12 months with 3-month cluster periods; 4 ICUs commenced 23 May 2022 and 4 ICUs commenced 23 Aug 2022 (delayed start stratum).
    • Inclusion criteria: Age ≥16 years; prescribed enteral nutrition during index ICU admission, or prescribed enteral nutrition for the first time during a subsequent ICU admission within the same index hospital admission.
    • Exclusion criteria: Clinician judged trial formula contraindicated; ≥12 hours of non-trial enteral nutrition already delivered in ICU; previously participated in TARGET Protein.
    • Consent model: Waiver of consent / opt-out (Australia) and “consent to continue” (New Zealand), reflecting a unit-level practice intervention.
  • Intervention:
    • Augmented protein enteral formula (Nutrison Protein Intense; 100 g protein/L; 1.0 kcal/mL) as the default formula during intervention periods.
    • Initiated when clinicians commenced enteral nutrition; rate and caloric targets determined by usual local practice (no mandated energy prescription).
    • Continued while clinically indicated up to ICU discharge, death, or day 90; patients who remained in ICU across a crossover date continued their original assigned formula; patients readmitted within 90 days resumed their original formula.
    • Protein supplements and parenteral nutrition were permitted if clinicians considered them necessary (recorded).
  • Comparison:
    • Usual protein enteral formula (Nutrison Protein Plus; 63 g protein/L; 1.0 kcal/mL) as the default formula during control periods.
    • Otherwise identical delivery context: clinician-led enteral feeding rates/targets; rescue feeding strategies and co-interventions permitted and recorded.
  • Blinding: Unblinded at bedside (formula-level intervention); primary outcome and mortality derived from administrative/registry data, limiting detection bias for these endpoints; open-label delivery creates risk of performance bias for nutrition delivery and discharge-related outcomes.
  • Statistics: Sample size for an 8-cluster, 4-period cluster crossover design estimated that cluster-period size ≥60 provides 80% power and ≥80 provides 90% power to detect a 1-day difference in days free of index hospital and alive at day 90 (assumptions: within-cluster within-period correlation 0.02; cluster autocorrelation 0.8; 2-sided significance threshold P<.05); primary analysis was intention-to-treat using a quantile regression mixed model (median difference) with bootstrap 95% CIs; sensitivity and Bayesian analyses were pre-specified.12
  • Follow-Up Period: 90 days.

Key Results

This trial was not stopped early.

Outcome Augmented protein Usual protein Effect p value / 95% CI Notes
Days free of index hospital and alive at day 90 (primary outcome) 62 (0 to 77) 64 (0 to 77) Median difference −1.97 days 95% CI −7.24 to 3.30; P=0.46 Quantile mixed model; death assigned 0; survivors analysed separately below.
Days free of index hospital at day 90 (survivors only) 74 (61 to 81) 74 (61 to 81) Median difference −0.01 days 95% CI −1.25 to 1.22; P=0.74 Separates length-of-stay signal from mortality.
Alive at day 90 1221/1681 (72.6%) 1269/1716 (74.0%) RR 0.99 95% CI 0.95 to 1.03; P=0.47 No mortality signal at 90 days.
Length of ICU stay (time to live ICU discharge) 6.6 (3.1 to 18.0) days 6.2 (3.0 to 15.0) days Cause-specific HR 0.93 95% CI 0.88 to 1.00; P=0.04 Competing risk analysis; HR <1 indicates slower discharge.
Length of hospital stay (time to live hospital discharge) 21.4 (10.2 to 80.0) days 21.1 (10.1 to 68.9) days Cause-specific HR 0.96 95% CI 0.90 to 1.02; P=0.15 Competing risk analysis; no clear between-group difference.
Duration of invasive ventilation (among ventilated patients) 84.0 (35.0 to 178.9) h 78.0 (33.2 to 161.0) h Mean difference 6.8 h 95% CI −3.0 to 16.5; P=0.17 No clear effect on ventilator duration.
Tracheostomy 134/1681 (8.0%) 121/1716 (7.1%) RR 1.15 95% CI 0.66 to 2.01; P=0.57 Wide CI; event rate low.
New kidney replacement therapy during index ICU admission 122/1681 (7.3%) 127/1716 (7.4%) RR 0.97 95% CI 0.81 to 1.16; P=0.69 No overall difference; renal subgroup findings described below.
  • Protein separation was achieved without meaningful calorie separation: All feeding days protein 1.12 (0.71–1.53) vs 0.69 (0.40–0.95) g/kg IBW/day; calories 14 (9–19) vs 14 (8–19) kcal/kg IBW/day; excluding enrolment day protein 1.33 (0.88–1.78) vs 0.84 (0.49–1.11) g/kg IBW/day.
  • The primary outcome was neutral: median hospital-free alive days at 90 days were 62 vs 64 (median difference −1.97 days; 95% CI −7.24 to 3.30; P=0.46), with similar 90-day survival (72.6% vs 74.0%).
  • Pre-specified subgroup signals: new kidney replacement therapy at formula commencement (median difference −13.19 days; 95% CI −50.00 to 23.62; interaction P<.001) and invasive ventilation at formula commencement (median difference −3.44 vs +2.70 days; interaction P=0.02); interpret as hypothesis-generating in the context of multiplicity.

Internal Validity

  • Randomisation and allocation:
    • Cluster-level randomisation of ICUs to intervention sequences reduced contamination risk that commonly undermines individual randomisation in nutrition trials.
    • Allocation concealment at patient level is not applicable; allocation was revealed to sites in advance for operational reasons, creating potential for anticipatory behaviour at cluster transition points.
    • Staggered commencement (delayed start stratum) and period effects were explicitly modelled in the primary analysis.
  • Dropout / exclusions:
    • Enrolled: 3411; analysed for the primary outcome: 3397 (14 withdrew consent for retention: 12 augmented, 2 usual).
    • Primary outcome availability was high given registry/administrative data capture; missingness primarily affected weight-based delivery metrics (IBW/ABW not available for all participants).
  • Performance / detection bias:
    • Open-label delivery could influence feeding behaviours (rates, interruptions) and co-interventions.
    • Primary outcome and mortality ascertainment were largely objective and system-derived, reducing detection bias for those endpoints.
    • Discharge-related outcomes (ICU/hospital discharge) remain susceptible to system pressures and clinician behaviour, independent of physiological recovery.
  • Protocol adherence:
    • Protocol deviations occurred in 151/1681 (9.4%) augmented vs 95/1716 (5.6%) usual; “non-trial formula used based on clinician choice” occurred in 93/1681 (5.5%) vs 64/1716 (3.7%).
    • Audit episodes (n=568): correct trial formula used 376/568 (66%); alternate trial formula 4/568 (0.7%); non-trial formula 21/568 (3.7%); not receiving enteral nutrition at the audit time point 166/568 (29%).
    • Patients remaining in ICU across crossover dates continued their original assigned trial formula; this design feature reduces within-patient contamination at period boundaries.
  • Baseline characteristics:
    • Groups were well balanced: age 60.8 (48.2–70.7) vs 60.5 (48.2–70.3) years; APACHE II score 19 (14–25) vs 19 (14–25); BMI 27.7 (24.1–32.5) vs 27.4 (23.8–32.2).
    • High acuity cohort: invasive ventilation at formula commencement 1385/1681 (82.4%) vs 1355/1716 (79.0%); vasopressor infusion 930/1681 (55.3%) vs 944/1716 (55.0%).
  • Timing:
    • Early delivery: time from ICU admission to trial nutrition commencement 19.0 (9.2–37.7) h vs 19.3 (9.5–39.8) h.
  • Dose and separation of the variable of interest:
    • Trial formula volume: 696.0 (408.0–950.8) vs 676.2 (405.0–956.7) mL/day; trial formula duration 87.0 (36.4–187.0) vs 84.0 (34.0–182.4) h.
    • All feeding days (median): protein 1.12 (0.71–1.53) vs 0.69 (0.40–0.95) g/kg IBW/day; calories 14 (9–19) vs 14 (8–19) kcal/kg IBW/day.
    • Excluding enrolment day (median): protein 1.33 (0.88–1.78) vs 0.84 (0.49–1.11) g/kg IBW/day; calories 17 (11–22) vs 17 (10–22) kcal/kg IBW/day.
    • Protein supplement and parenteral nutrition use were uncommon and similar: protein supplements 12/1681 (0.7%) vs 11/1716 (0.6%); parenteral nutrition 114/1681 (6.8%) vs 120/1716 (7.0%).
  • Heterogeneity:
    • Intracluster correlation coefficient for the primary outcome was small (0.004), suggesting limited residual clustering after adjustment.
    • Pre-specified subgroup interactions were present for new kidney replacement therapy and invasive ventilation at formula commencement; credibility was assessed (ICEMAN), supporting cautious interpretation rather than definitive subgroup claims.
  • Outcome assessment and definitions:
    • Primary outcome definition is explicit and robust to ascertainment: death or ongoing hospitalisation at day 90 yields 0 hospital-free alive days.
    • Hospital readmissions were captured within the index hospital only; transfers or readmissions to other hospitals could not be counted, potentially affecting precision of “hospital-free days”.
  • Statistical rigour:
    • Skewed outcome addressed using a quantile mixed model; uncertainty estimated with bootstrap CIs; secondary and Bayesian analyses were pre-specified.2
    • Multiple secondary outcomes and subgroup tests increase the risk of chance findings; subgroup credibility assessment partially mitigates (but does not eliminate) this risk.

Conclusion on Internal Validity: Overall, internal validity appears moderate to strong given pragmatic cluster randomisation, objective ascertainment of the primary outcome, minimal attrition, and clear protein separation; key threats include open-label delivery (performance bias), discharge-related outcome susceptibility to system factors, and uncertainty about whether achieved doses represent “high protein” under all weight-scaling conventions.

External Validity

  • Population representativeness:
    • Broad adult ICU inclusion with few exclusions beyond formula contraindication and prior prolonged non-trial enteral feeding; high rates of mechanical ventilation and vasopressor use align with common ICU enteral nutrition recipients.
    • Excludes patients with clinician-perceived contraindication to trial formulas and those already established on alternative enteral formulas for >12 hours, which may under-represent complex GI/surgical feeding phenotypes.
  • Applicability:
    • Intervention is a pragmatic substitution to a specific isocaloric formula pair; likely generalisable where 1.0 kcal/mL polymeric formulas are standard and baseline protein delivery is similar.
    • Less directly generalisable to settings where routine practice already achieves higher protein doses, where protein modules/parenteral amino acids are used systematically, or where calorie targets and delivery differ materially.
    • Does not test high-protein strategies combined with structured early mobilisation, anabolic pharmacology, or tailored approaches based on metabolic/renal phenotype.

Conclusion on External Validity: Generalisability is good to adult ICUs in similar high-resource systems using comparable enteral feeding practices; extrapolation is more limited for markedly different nutrition delivery paradigms (higher baseline protein, early parenteral amino acids) or for phenotypes requiring prolonged/high-intensity protein exposure.

Strengths & Limitations

  • Strengths:
    • Large, pragmatic, multicentre evaluation (n=3397) with unit-level implementation and minimal disruption to routine care.
    • Cluster crossover design minimised between-unit confounding and contamination typical of individual randomisation for formula-level interventions.
    • Isocaloric formulas enabled a focused test of protein augmentation rather than conflating protein with energy delivery.
    • Patient-centred primary outcome with objective ascertainment (administrative/registry linkage) and 90-day follow-up.
    • A priori protocol and SAP, with explicit modelling of clustering/period effects and pre-specified subgroup credibility assessment.12
  • Limitations:
    • Open-label delivery; clinician-driven feeding rates, interruptions, and co-interventions could dilute or distort treatment effects.
    • Achieved protein dose depends on weight scalar: by ABW, median protein delivery remained below many guideline targets (1.04 vs 0.64 g/kg ABW/day excluding enrolment day).
    • Median exposure to the trial formula was relatively short (87 vs 84 hours), potentially limiting capacity to influence longer-term recovery outcomes.
    • Primary outcome is constrained to the index hospital: readmissions/transfers outside the index hospital were not captured, potentially affecting precision of “hospital-free days”.
    • Discharge-related outcomes are sensitive to health-system capacity and practice, which may not reflect patient-level physiological recovery.

Interpretation & Why It Matters

  • Clinical practice
    • Routine substitution to an isocaloric higher-protein enteral formula increased delivered protein substantially but did not improve 90-day hospital-free survival.
    • These data do not support a “one-size-fits-all” default to high-protein formulas for unselected adult ICU patients solely to improve downstream hospital-free survival.
  • Mechanism and timing
    • Null clinical effects despite improved protein delivery are compatible with an impaired anabolic response early in critical illness, or a need for phenotype- and timing-specific dosing rather than universal early escalation.
    • Observed increases in urea without outcome benefit highlight that “more nitrogen delivered” is not a surrogate for “more functional recovery”.
  • Trial design impact
    • Demonstrates that registry-embedded, cluster crossover trials can efficiently evaluate unit-level nutrition strategies with robust follow-up and pragmatic applicability.

Controversies & Subsequent Evidence

  • Interpretation of delivered “dose” depends on weight scalar and feeding adequacy:
    • Correspondence argued that delivered protein (median 1.04 vs 0.64 g/kg ABW/day excluding enrolment day) and delivered energy were below many guideline recommendations, potentially testing “moderate vs low” protein rather than “high vs standard”.3
    • The trialists responded that the comparison reflected real-world practice in participating ICUs; using IBW, protein delivery aligned more closely with guideline targets (1.33 vs 0.84 g/kg IBW/day excluding enrolment day), and energy delivery was similar between groups, making calorie confounding less plausible.4
  • Renal phenotype and nitrogen handling remain central uncertainties:
    • Augmented protein increased serum urea by day 10 (13.0 [8.3–21.3] vs 10.6 [7.2–17.6] mmol/L), consistent with increased nitrogen load; clinical significance remains uncertain.
    • Pre-specified subgroup interaction suggested fewer hospital-free alive days in patients with new kidney replacement therapy at formula commencement (median difference −13.19 days; 95% CI −50.00 to 23.62; interaction P<.001), with moderate credibility using ICEMAN criteria.
    • Post hoc work from EFFORT Protein reports potential harm from higher protein dosing in acute kidney injury and explores urea trajectories as prognostic/biological signals, supporting ongoing equipoise and renal-phenotype targeting.567
  • Consistency with subsequent and contemporary randomised evidence:
    • EFFORT Protein (higher vs usual protein dosing) and PRECISe (high protein strategy focused on functional recovery) both reported no improvement in key clinical endpoints with higher protein strategies, and PRECISe reported lower health-related quality of life in the high-protein arm.89
    • An updated systematic review and meta-analysis with trial sequential analysis reports ongoing uncertainty about benefit or harm of higher versus lower protein delivery, reinforcing the need for trials that target timing, dose, and patient phenotype rather than uniform escalation.10
  • Guideline implications:
    • Contemporary ICU nutrition guidelines continue to recommend protein targets but acknowledge low certainty and heterogeneity; TARGET Protein adds high-quality evidence against universal early protein augmentation by formula substitution alone in an unselected ICU population.1112

Summary

  • Pragmatic cluster crossover trial (n=3397) compared an isocaloric high-protein enteral formula (100 g/L) with a usual-protein formula (63 g/L) across 8 adult ICUs.
  • Protein delivery increased materially (excluding enrolment day: 1.33 vs 0.84 g/kg IBW/day), while energy delivery remained similar between groups.
  • No improvement in the primary endpoint (days free of the index hospital and alive at day 90) and no difference in 90-day survival.
  • ICU discharge was modestly slower in the augmented protein group (cause-specific HR 0.93; 95% CI 0.88 to 1.00), without a clear hospital discharge or mortality signal.
  • Renal subgroup interaction and higher urea highlight uncertainty about protein dosing in renal dysfunction and support phenotype- and timing-specific research rather than universal escalation.

Overall Takeaway

In a large, pragmatic, multicentre cluster crossover trial, increasing enteral protein delivery through substitution to an isocaloric higher-protein formula improved protein intake but did not improve 90-day hospital-free survival or mortality. The results argue against routine, universal early protein augmentation by formula substitution alone, and they sharpen the case for phenotype- and timing-specific strategies—particularly in patients with renal dysfunction—rather than uniform escalation.

Overall Summary

  • Augmenting protein via an isocaloric high-protein enteral formula increased delivered protein but did not improve hospital-free alive days or 90-day survival in unselected adult ICU patients.

Bibliography