Skip to main content

Foundational Trials

Hayes

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Elevation of Systemic Oxygen Delivery in the Treatment of Critically Ill Patients
  • Acronym: Not applicable
  • Year: 1994
  • Journal published in: The New England Journal of Medicine
  • Citation: Hayes MA, Timmins AC, Yau EHS, Palazzo M, Hinds CJ, Watson D. Elevation of systemic oxygen delivery in the treatment of critically ill patients. N Engl J Med. 1994;330(24):1717-1722.

Context & Rationale

  • Background
    The “supranormal oxygen delivery” concept arose from observational and interventional perioperative work suggesting that survivors of major stress states achieve higher cardiac index (CI) and oxygen delivery (DO2) than non-survivors, and that protocolised targeting of these values (often using pulmonary artery catheters and inotropes) could reduce complications and mortality in selected high‑risk surgical cohorts.12
  • Research Question/Hypothesis
    In a heterogeneous, high‑risk ICU population (including sepsis syndrome/shock, ARDS, and high‑risk surgical/medical admissions) who fail to reach predefined supranormal targets with volume expansion alone, does dobutamine‑driven augmentation of CI/DO2/oxygen consumption (VO2) improve clinical outcomes versus standard haemodynamic support?
  • Why This Matters
    The trial tested a widely discussed physiological strategy at a time when pulmonary artery catheter‑guided resuscitation and catecholamine escalation were common; it directly examined whether raising systemic oxygen delivery in established critical illness translates into improved VO2, less organ failure, and lower mortality, versus exposing patients to potential harms of high-dose inotropes/vasopressors.

Design & Methods

  • Research Question: Does dobutamine‑driven elevation of systemic oxygen delivery/consumption to predefined “supranormal” targets improve outcomes in high‑risk critically ill patients compared with standard care?
  • Study Type: Prospective, randomised, controlled, two-centre ICU trial (open-label); consecutive screening over two years; treatment initiated in ICU after enrolment.
  • Population:
    • Setting: Intensive care units at two hospitals; consecutive high-risk admissions screened over a two-year period.
    • Key inclusion features: High-risk surgical patients (criteria derived from Shoemaker et al.); and non-surgical patients with life-threatening cardiorespiratory illness (acute respiratory failure or septic shock).
    • Randomisation trigger: After initial fluid resuscitation, failure to reach all three targets (CI >4.5 L/min/m2, DO2 >600 mL/min/m2, VO2 >170 mL/min/m2).
    • Exclusions: Age <16 years; pregnancy; neurosurgery; pre-existing cardiac disease; haematologic cancer.
    • Non-randomised cohort: Patients achieving all three targets with volume expansion alone were not randomised (n=9) but were followed.
  • Intervention:
    • Haemodynamic platform (both groups): Urinary, peripheral arterial, central venous, and pulmonary arterial catheterisation; dopamine 2 μg/kg/min to all; arterial oxygen saturation maintained >90%; haemoglobin maintained >10 g/dL; blood/albumin/synthetic colloids as needed.
    • Volume strategy: IV fluids titrated to an “optimal” pulmonary artery occlusion pressure (PAOP) defined by the plateau of left ventricular stroke-work index versus PAOP.
    • Dobutamine target strategy: Dobutamine 5–200 μg/kg/min to increase CI and DO2 until all three targets were achieved simultaneously; dose reduced/stopped for sinus tachycardia >130 bpm, tachyarrhythmia, or ECG evidence of myocardial ischaemia; then re-titrated to the highest achievable CI/DO2/VO2.
    • Vasopressor strategy (both groups): Norepinephrine 0.05–20 μg/kg/min if required to maintain mean arterial pressure (MAP) 80 mmHg while avoiding excessive peripheral vasoconstriction (systemic vascular resistance index >1500 dyn·s·cm−5/m2).
  • Comparison:
    • Standard support threshold for dobutamine: Dobutamine administered only if CI <2.8 L/min/m2.
    • Otherwise identical platform: Same catheterisation, dopamine infusion, oxygenation/haemoglobin targets, fluid optimisation approach, and norepinephrine strategy.
  • Blinding: Unblinded (treatment allocation and titration strategy visible to clinicians).
  • Statistics: Power calculation: 130 patients per group required to detect a 15-percentage-point absolute mortality reduction (from 33% to 18%) with 80% power at 5% significance; interim analysis after each block of 50 randomised patients; non-parametric (Mann–Whitney) and chi-square tests; 48-hour incremental area-under-curve summaries for serial physiological variables; analysis set not explicitly labelled but outcomes reported for all randomised patients.
  • Follow-Up Period: Physiological measurements at admission and after volume optimisation (1, 2, 4, 8, 12, 16, 20, 24 hours; then 6-hourly for 24 hours; then at least every 8 hours); clinical outcomes to ICU and hospital discharge (or death).

Key Results

This trial was stopped early. Interim analyses were scheduled after each block of 50 randomised patients; at the second interim analysis, both ICU and in-hospital mortality were higher in the treatment group and the investigators judged the probability of demonstrating benefit to be low; reporting is for the 100 randomised patients enrolled before discontinuation.

Outcome Supranormal-target dobutamine strategy (n=50) Control (dobutamine only if CI <2.8) (n=50) Effect p value / 95% CI Notes
ICU mortality 25/50 (50%) 15/50 (30%) Risk difference +20 percentage points 95% CI 1.2 to 38.8 percentage points; P=0.04 Excess deaths largely attributed to multiple organ failure (see below)
In-hospital mortality 27/50 (54%) 17/50 (34%) Risk difference +20 percentage points 95% CI 0.9 to 39.1 percentage points; P=0.04 Median predicted risk of death (APACHE II-based): 34% in both randomised groups
Days in ICU 10 (1–48) 10 (1–64) Not reported Not reported Median (range)
Mechanical ventilation 8 days (0–41); 46 ventilated 8 days (0–54); 44 ventilated Not reported Not reported Median (range) days; number ventilated
Days in hospital 19 (1–187) 23.5 (1–244) Not reported Not reported Median (range)
Cause of death: multiple organ failure 17 patients 9 patients Not reported Not reported Other causes: intractable hypotension 4 vs 4; cardiac event 4 vs 2 (treatment vs control)
Norepinephrine use; maximal dose 31/50; 1.2 μg/kg/min (0.02–16.6) 34/50; 0.23 μg/kg/min (0.05–10) Higher maximal dose in treatment P=0.029 Median (range)
Dobutamine use; maximal dose 50/50; 25 μg/kg/min (2.5–200) 21/50; 10 μg/kg/min (2.5–200) Higher exposure in treatment P<0.001 17/50 treatment patients received ≥50 μg/kg/min at some point
Lactate (mmol/L), median (25th–75th) Baseline 2.2 (1.8–3.5); 48h 1.7 (1.23–2.5) Baseline 2.1 (1.5–3.3); 48h 1.5 (1.1–2.1) No between-group difference at baseline or 48h Baseline P=0.69; 48h P=0.2 Within-group lactate at 48h lower than baseline in both groups (P<0.05)
Physiological separation over first 48h (incremental AUC) CI and DO2 higher; VO2 similar CI and DO2 lower; VO2 similar Improved DO2 without VO2 increase CI P<0.001; DO2 P=0.0012; VO2 P=0.12 Oxygen-extraction ratio AUC lower in treatment (P=0.045); MAP and lactate AUC not different (P=0.42 and P=0.34)
Dose-limiting complications in treatment group 24/50 affected Not reported Not reported Not reported Tachycardia >130 bpm (12); ECG ischaemia (8); hypertension (5); tachyarrhythmias (2)
  • Dobutamine-driven targeting increased CI and DO2 (P<0.001 and P=0.0012) but did not increase VO2 (P=0.12) and did not lower lactate versus control (baseline P=0.69; 48h P=0.2).
  • Mortality was higher with the supranormal-target strategy: ICU mortality 50% vs 30% (P=0.04) and in-hospital mortality 54% vs 34% (P=0.04), despite identical median predicted risk of death (34%) in both randomised groups.
  • Nine patients achieved all three targets with volume expansion alone (not randomised); all nine survived to leave hospital.

Internal Validity

  • Randomisation and Allocation: Randomisation used a table of random numbers after fluid resuscitation among those failing all three targets; allocation concealment was not described.
  • Attrition and exclusions: Outcomes are reported for all 100 randomised patients; no post-randomisation exclusions were reported; a separate non-randomised cohort (n=9) achieved targets with fluids alone and all survived.
  • Performance/detection bias: Open-label titration of inotropes/vasopressors makes performance bias plausible; primary outcomes (mortality) are objective, while attribution of cause of death (e.g., multiple organ failure) is more interpretive.
  • Protocol adherence/feasibility: In the treatment group, 35/50 patients did not achieve all three targets simultaneously despite dobutamine; dose increments were limited by complications in 24/50 (tachycardia, ischaemia, hypertension, tachyarrhythmia).
  • Baseline comparability: Randomised groups were similar at enrolment for age (median 63.5 vs 62 years), APACHE II (median 18 vs 18), APACHE III (median 58 vs 63), and organ-failure score (median 1 vs 1).
  • Separation of the variable of interest: Dobutamine exposure differed (50/50 with median maximal 25 μg/kg/min vs 21/50 with median maximal 10 μg/kg/min; P<0.001); norepinephrine maximal dose was higher in the treatment group (median 1.2 vs 0.23 μg/kg/min; P=0.029); CI and DO2 AUC were higher in treatment, but VO2 AUC was not.
  • Crossover/contamination: Control patients could receive dobutamine if CI <2.8 L/min/m2 (21/50 did, with maximal dose up to 200 μg/kg/min), potentially diluting between-group differences in catecholamine exposure.
  • Statistical considerations: Planned sample size (130 per group) was not reached; the trial was stopped early after interim analyses, which can magnify apparent treatment effects and limits precision for subgroup inference.

Conclusion on Internal Validity: Moderate: randomisation and complete follow-up for hard endpoints support internal validity, but open-label delivery, feasibility limits (failure to reach targets in 35/50), and early stopping reduce confidence in the stability of the effect estimate.

External Validity

  • Population representativeness: High-risk, heterogeneous ICU cohort (sepsis syndrome/shock, ARDS, major surgery/trauma); exclusion of patients with pre-existing cardiac disease and haematologic cancer may limit applicability to contemporary ICUs with higher baseline cardiac comorbidity burden.
  • Intervention context: Management relied on pulmonary artery catheterisation, a fixed dopamine infusion, haemoglobin target >10 g/dL, and explicit CI/DO2/VO2 targets—elements that differ from many modern protocols and monitoring practices.
  • Applicability: Findings are most directly applicable to established critical illness in settings capable of invasive haemodynamic monitoring and high-dose inotrope titration; they are less directly generalisable to perioperative “pre-emptive” optimisation strategies and to resource-limited environments.

Conclusion on External Validity: Generalisability is context-limited (two-centre, PA catheter–based era), but the central message—systemic DO2 augmentation may not translate into improved tissue utilisation and may cause harm in established critical illness—remains conceptually relevant.

Strengths & Limitations

  • Strengths:
    • Prospective randomised controlled design in a heterogeneous ICU population directly testing a prevalent physiological strategy.
    • Explicit haemodynamic targets and dosing ranges (dobutamine 5–200 μg/kg/min; norepinephrine 0.05–20 μg/kg/min) with prespecified safety limits.
    • Objective primary outcomes (ICU and hospital mortality) with complete reporting for all randomised patients.
    • Dense physiological sampling and prespecified 48-hour incremental AUC summaries to quantify exposure and separation.
  • Limitations:
    • Stopped early with only 50 patients per arm (vs planned 130 per arm), limiting precision and increasing susceptibility to random high-magnitude effects.
    • Unblinded intervention delivery; co-interventions and bedside decision-making could be influenced by allocation.
    • Feasibility constraints: 35/50 treatment patients could not achieve all three targets simultaneously; dose-limiting tachycardia/ischaemia common.
    • Era- and technology-specific care (routine pulmonary artery catheterisation, fixed dopamine infusion, haemoglobin target >10 g/dL) may not map cleanly to contemporary practice.

Interpretation & Why It Matters

  • Physiology–outcome disconnect
    Raising systemic DO2 (and CI) did not raise VO2, and did not improve lactate trajectories versus control, suggesting that in established critical illness “oxygen delivery” may be a poor surrogate for tissue oxygen utilisation and cellular recovery.
  • Clinical signal
    Despite identical median predicted mortality risk (34%), the supranormal-target strategy had higher ICU and hospital mortality (50% and 54%) compared with control (30% and 34%), with excess deaths largely attributed to multiple organ failure.
  • Practice implications today
    This trial helped de-emphasise routine pursuit of supraphysiologic haemodynamic targets in general ICU populations and strengthened the rationale for using inotropes selectively (e.g., cardiac dysfunction with persistent hypoperfusion) rather than as a default “oxygen delivery” escalation.

Controversies & Subsequent Evidence

  • Discordant results between single-centre perioperative goal-directed haemodynamic trials (often showing reduced complications/mortality) and multicentre/heterogeneous ICU cohorts prompted debate about timing (pre-emptive perioperative optimisation vs established organ failure), protocol fidelity, and whether systemic DO2 targets meaningfully reflect microcirculatory/mitochondrial dysfunction.34
  • A subsequent NEJM randomised trial of goal-oriented haemodynamic therapy in critically ill patients similarly failed to show outcome benefit from protocolised haemodynamic targeting, reinforcing the concern that late systemic DO2 augmentation may not translate into improved clinical outcomes in ICU case-mix populations.5
  • Methodologically focused syntheses highlighted heterogeneity across trials (population, timing, targets, monitoring, co-interventions) and questioned the generalisability of supranormal targets outside carefully selected cohorts.6
  • A broad meta-analysis of haemodynamic optimisation trials found an overall mortality reduction when combining studies, driven primarily by perioperative/trauma contexts; effects were not uniform across sepsis/organ-failure populations, supporting the notion that “who” and “when” matter as much as “what target”.7
  • Modern guideline statements for sepsis/shock emphasise individualised haemodynamic targets, dynamic assessment of fluid responsiveness, serial evaluation of tissue perfusion, and selective use of inotropes—rather than fixed supranormal systemic oxygen transport goals.8910
  • Contemporary perioperative consensus has become more cautious about routine goal-directed haemodynamic therapy and fixed inotrope infusions, narrowing indications to selected settings and underscoring uncertainty about broad protocol generalisability.11

Summary

  • Two-centre ICU RCT testing supranormal CI/DO2/VO2 targets using dobutamine (5–200 μg/kg/min) versus standard care with dobutamine only for CI <2.8 L/min/m2.
  • Stopped early after interim analyses; among 100 randomised patients, ICU mortality was 50% vs 30% and hospital mortality 54% vs 34% (treatment vs control; both P=0.04).
  • Physiological separation achieved (higher CI/DO2) but VO2 and lactate trajectories did not improve versus control.
  • High catecholamine exposure and frequent dose-limiting complications (tachycardia/ischaemia) constrained target attainment; 35/50 treatment patients failed to achieve all three targets simultaneously.
  • The trial challenged routine pursuit of supranormal systemic oxygen transport goals in established critical illness and influenced subsequent research and guideline caution.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Key methodological nuance: randomisation occurred only after “optimal” volume expansion and failure to reach all three targets, enriching the cohort for patients with limited physiological reserve and/or potential supply-independent oxygen consumption.
  • Interpretation should separate “raising DO2” as a systemic physiological achievement from “improving tissue oxygen utilisation” as a clinical mechanism; Hayes is a prototypical example of divergence between the two.

Overall Takeaway

Hayes demonstrated that protocolised escalation of dobutamine to achieve supranormal systemic oxygen transport targets in established critical illness can increase cardiac output and oxygen delivery without improving oxygen consumption or lactate, and was associated with higher ICU and hospital mortality. As a landmark counterpoint to earlier perioperative “supranormal” trials, it helped shift critical care away from fixed systemic DO2 targets towards individualised resuscitation guided by tissue perfusion and clinical context.

Overall Summary

  • Systemic DO2 augmentation ≠ improved VO2 or lactate in established ICU illness.
  • Higher catecholamine exposure and frequent dose-limiting adverse effects constrained target attainment.
  • Mortality signal favoured standard care, contributing to modern scepticism about routine supranormal haemodynamic goals in ICU populations.

Bibliography

  • 1.Shoemaker WC, Appel PL, Kram HB, Waxman K, Lee TS. Prospective trial of supranormal values of survivors as therapeutic goals in high-risk surgical patients. Chest. 1988;94(6):1176-1186. Link
  • 2.Boyd O, Grounds RM, Bennett ED. A randomized clinical trial of the effect of deliberate perioperative increase of oxygen delivery on mortality in high-risk surgical patients. JAMA. 1993;270(22):2699-2707. Link
  • 3.Takala J. Highs and lows in high-risk surgery: the controversy of goal-directed haemodynamic management. Crit Care. 2005;9(6):642-644. Link
  • 4.De Backer D. Optimal management of the high risk surgical patient: beta stimulation or beta blockade? Crit Care. 2005;9(6):645-646. Link
  • 5.Gattinoni L, Brazzi L, Pelosi P, et al. A trial of goal-oriented hemodynamic therapy in critically ill patients. N Engl J Med. 1995;333(16):1025-1032. Link
  • 6.Heyland DK, Cook DJ, King D, Kernerman P, Brun-Buisson C. Maximizing oxygen delivery in critically ill patients: a methodologic appraisal of the evidence. Crit Care Med. 1996;24(3):517-524. Link
  • 7.Poeze M, Greve JWM, Ramsay G. Meta-analysis of hemodynamic optimization: relationship to methodological quality. Crit Care. 2005;9(6):R771-R779. Link
  • 8.Evans L, Rhodes A, Alhazzani W, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock 2021. Intensive Care Med. 2021;47(11):1181-1247. Link
  • 9.Cecconi M, De Backer D, Antonelli M, et al. Consensus on circulatory shock and hemodynamic monitoring. Task force of the European Society of Intensive Care Medicine. Intensive Care Med. 2014;40(12):1795-1815. Link
  • 10.Monnet X, Messina A, Greco M, et al. ESICM guidelines on circulatory shock and hemodynamic monitoring 2025. Intensive Care Med. 2025;51(11):1971-2012. Link
  • 11.Edwards M, Kunst G, Forni LG, et al. Perioperative Quality Initiative consensus statement on goal-directed haemodynamic therapy. Br J Anaesth. 2025;135:547-560. Link