Skip to main content

Foundational Trials

DOREMI

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Milrinone as Compared with Dobutamine in the Treatment of Cardiogenic Shock
  • Acronym: DOREMI
  • Year: 2021
  • Journal published in: The New England Journal of Medicine
  • Citation: Mathew R, Di Santo P, Jung RG, Marbach JA, Hutson J, Simard T, et al. Milrinone as compared with dobutamine in the treatment of cardiogenic shock. N Engl J Med. 2021;385(6):516-525.

Context & Rationale

  • Background
    • Cardiogenic shock carries high short-term mortality and frequent multiorgan failure despite contemporary reperfusion, vasopressors, and mechanical circulatory support (MCS).
    • Intravenous inotropes are commonly used to augment cardiac output and end-organ perfusion, yet they are also associated with tachyarrhythmias, myocardial oxygen demand, and hypotension.
    • Dobutamine (predominantly β1/β2 agonist) has rapid titratability and short half-life but may be blunted by chronic β-blockade and can increase heart rate and myocardial oxygen consumption.
    • Milrinone (phosphodiesterase-3 inhibitor; “inodilator”) may improve contractility independent of β-receptors and reduce pulmonary vascular resistance, but has a longer half-life, vasodilatory hypotension risk, and renal clearance considerations.
    • Before DOREMI, evidence comparing milrinone and dobutamine in cardiogenic shock was limited to small studies and extrapolation from non-shock low-output states, leaving substantial practice variation and equipoise.
  • Research Question/Hypothesis
    • In adults with cardiogenic shock requiring intravenous inotrope support, does milrinone reduce in-hospital death and major complications compared with dobutamine?
    • Hypothesis (implicit): milrinone’s pharmacology (β-independent inotropy and vasodilation) could translate into improved patient-centred outcomes versus dobutamine.
  • Why This Matters
    • Inotrope selection is a frequent, high-consequence bedside decision in cardiogenic shock, often made with limited comparative trial evidence.
    • A blinded head-to-head RCT could clarify whether mechanistic differences translate into clinically meaningful benefits or harms, and inform subsequent trial design, practice standardisation, and guidance documents.

Design & Methods

  • Research Question: Among adults with cardiogenic shock requiring intravenous inotrope therapy, is milrinone superior to dobutamine in reducing a composite of in-hospital death and major complications?
  • Study Type: Randomised, double-blind, single-centre, investigator-initiated trial in a cardiac intensive care unit (CICU) (Ottawa, Canada).
  • Population:
    • Setting: cardiac intensive care unit (CICU).
    • Inclusion: adults (≥18 years) with cardiogenic shock requiring intravenous inotropic support, spanning Society for Cardiovascular Angiography and Interventions (SCAI) stages B–E.
    • Operational inclusion triggers (examples): systolic BP <90 mmHg with end-organ dysfunction; systemic/pulmonary congestion despite vasodilators/diuretics; acute coronary syndrome with shock and haemodynamic low output (cardiac index <1.8 L/min/m2 with LVEDP >18 mmHg); need to augment cardiac output despite vasopressor therapy; developing shock where the treating team judged inotropic therapy necessary.
    • Exclusions: out-of-hospital cardiac arrest; pregnancy; milrinone or dobutamine initiated prior to randomisation; concurrent participation in another interventional trial; inability to obtain written informed consent; treating clinician determined patient was not eligible.
  • Intervention:
    • Milrinone infusion, blinded, titrated using a five-stage dosing scale (stage 1: 0.125 μg/kg/min; stage 2: 0.250; stage 3: 0.375; stage 4: 0.500; stage 5: >0.500).
    • No bolus dosing was described in the trial report; titration decisions were clinician-driven within the protocolised staging framework.
    • Other cardiogenic shock therapies (vasopressors, ventilation, diuretics/vasodilators, MCS) were permitted as clinically indicated.
  • Comparison:
    • Dobutamine infusion, blinded, titrated using a matching five-stage dosing scale (stage 1: 2.5 μg/kg/min; stage 2: 5; stage 3: 7.5; stage 4: 10; stage 5: >10).
    • Otherwise usual CICU care for cardiogenic shock, with discretionary co-interventions and rescue therapy when clinically required.
  • Blinding: Double-blind for participants, treating clinicians, investigators, and research staff; the bedside nurse preparing the study infusion was necessarily unblinded (infusion/pump concealment strategies used); unblinding permitted if required for safety.
  • Statistics: A total sample size of 192 patients (96 per group) was planned to detect a 20% absolute reduction in the primary composite outcome (assumed 55% to 35%) with 80% power at a two-sided 5% significance level; primary analysis was intention-to-treat.
  • Follow-Up Period: In-hospital (index admission) outcomes; selected physiologic/organ-function measures collected up to 120 hours after randomisation.

Key Results

This trial was not stopped early. Enrolment reached the planned sample size (192 participants; 96 per group).

Outcome Milrinone (n=96) Dobutamine (n=96) Effect p value / 95% CI Notes
Primary composite outcome (in-hospital) 47 (49%) 52 (54%) RR 0.90 95% CI 0.69 to 1.19; P=0.47 Composite: death, resuscitated cardiac arrest, MCS, nonfatal MI, TIA/stroke, or RRT initiation.
Primary composite outcome (time-to-event analysis) 47 (49%) 52 (54%) HR 0.91 95% CI 0.61 to 1.34; P=Not reported Time-to-event from randomisation; model details per trial report.
In-hospital death (any cause) 35 (37%) 41 (43%) RR 0.85 95% CI 0.60 to 1.21; P=Not reported Direction favoured milrinone but confidence interval included clinically meaningful benefit and harm.
Resuscitated cardiac arrest 7 (7%) 9 (9%) HR 0.78 95% CI 0.29 to 2.07; P=Not reported Fine–Gray competing-risk approach reported in trial methods.
Cardiac transplant or MCS 11 (11%) 14 (15%) HR 0.78 95% CI 0.36 to 1.71; P=Not reported No participant underwent cardiac transplantation (per trial report).
Nonfatal myocardial infarction 1 (1%) 0 Not estimable Not reported Single event occurred in the milrinone group.
Transient ischaemic attack or stroke 1 (1%) 2 (2%) HR 0.50 95% CI 0.05 to 5.50; P=Not reported Neurologist-diagnosed events (per trial report).
Initiation of renal replacement therapy 21 (22%) 16 (17%) HR 1.39 95% CI 0.73 to 2.67; P=Not reported Excluding participants receiving RRT at baseline (per trial footnote).
CICU length of stay (median, IQR), days 4.5 (2.0–7.0) 5.5 (3.0–10.0) Not reported Not reported Skewed time-to-discharge distribution typical in shock cohorts.
CICU length of stay ≥7 days 31 (32%) 42 (44%) RR 0.74 95% CI 0.51 to 1.07; P=Not reported Secondary endpoint; not adjusted for multiple comparisons.
Total time receiving inotropes (median, IQR), hours 36 (18–79) 39 (19–64) Not reported Not reported Suggests similar inotrope exposure duration (separation by duration limited).
Acute kidney injury 86 (92%) 85 (90%) RR 1.02 95% CI 0.94 to 1.12; P=Not reported High incidence reflects severity and definitions used in shock.
Normalisation of lactate level 33 (46%) 36 (56%) RR 0.80 95% CI 0.56 to 1.15; P=Not reported Excluded those with normal lactate at baseline; assessed up to 120 hours.
Arrhythmia leading to medical team intervention 48 (50%) 44 (46%) RR 1.19 95% CI 0.85 to 1.57; P=Not reported Supplementary safety table reports a similar event rate with a RR 1.09 (0.81–1.47).
  • Milrinone did not significantly reduce the primary in-hospital composite compared with dobutamine: 47/96 (49%) vs 52/96 (54%); RR 0.90; 95% CI 0.69 to 1.19; P=0.47.
  • Mortality was numerically lower with milrinone but imprecise: 35/96 (37%) vs 41/96 (43%); RR 0.85; 95% CI 0.60 to 1.21.
  • Prespecified subgroup analyses showed no evidence of heterogeneity; examples: severe LV dysfunction RR 1.07; 95% CI 0.74 to 1.55 vs mild/moderate LV dysfunction RR 0.70; 95% CI 0.46 to 1.08; concomitant vasopressor use at initiation RR 0.98; 95% CI 0.74 to 1.30 vs no vasopressor RR 1.06; 95% CI 0.61 to 1.83 (interaction P values not reported).

Internal Validity

  • Randomisation and Allocation:
    • Randomised 1:1 (96 per group) with stratification by affected ventricle (right vs left/both) and use of a staged dosing scheme for both drugs.
    • Allocation concealment was operationalised through blinded study infusion preparation; the preparing nurse was unblinded, while clinicians/investigators remained blinded.
  • Dropout or Exclusions (post-randomisation):
    • Not reported as a major issue for in-hospital endpoints; outcomes were measured during the index admission.
    • Screening-to-enrolment attrition was substantial: 319 screened; 192 randomised.
    • Key pre-randomisation exclusions: transfer to ICU after inotrope already started (47); clinician deemed ineligible (40); inability to obtain consent (23); out-of-hospital cardiac arrest (13); other interventional trial (4).
  • Performance/Detection Bias:
    • Blinding was feasible and likely effective for most bedside decisions, reducing risk of differential co-interventions.
    • Several primary-outcome components are clinician-triggered (MCS initiation, RRT initiation), introducing potential practice-pattern influence; single-centre setting may mitigate inter-site variability but can embed local thresholds.
  • Protocol Adherence:
    • Exposure duration was similar: median total time receiving inotropes 36 (IQR 18–79) hours with milrinone vs 39 (IQR 19–64) hours with dobutamine.
    • Physiologic trajectories showed broadly similar profiles (example measures at 24 hours from supplementary data): vasoactive-inotropic score 12.5 ± 14.6 vs 12.2 ± 11.4; mean arterial pressure 75.1 ± 10.6 vs 76.0 ± 12.4 mmHg; lactate 2.0 ± 1.7 vs 2.4 ± 2.4 mmol/L.
  • Baseline Characteristics:
    • Groups were broadly comparable in age (68.9 ± 13.8 vs 72.0 ± 11.3 years), sex (women 38% vs 35%), and ischaemic aetiology (69% vs 65%).
    • Notable imbalances (directionally): previous myocardial infarction 41% vs 30%; previous PCI 31% vs 20%; vasopressor use at inotrope initiation 39% vs 50%.
    • Severity was clinically meaningful: SCAI stage C/D predominated; baseline lactate median 2.9 mmol/L in both groups; in-hospital death occurred in 37% vs 43%.
  • Heterogeneity:
    • Prespecified subgroup analyses did not detect effect modification across major clinical strata; however, many subgroups were small (eg right-ventricular shock subgroup n=16), limiting power to detect interaction.
  • Timing:
    • Randomisation timing was variable: mean time from CICU admission to randomisation 23.4 ± 92.6 hours (milrinone) vs 17.9 ± 50.6 hours (dobutamine), suggesting some participants were enrolled late relative to shock onset.
  • Dose:
    • Both drugs were delivered via staged infusion scales (milrinone 0.125–>0.5 μg/kg/min; dobutamine 2.5–>10 μg/kg/min), reflecting a pragmatic, clinician-titrated approach.
    • Bolus dosing was not described; whether this optimally reflects typical milrinone initiation strategies in shock is context-dependent.
  • Separation of the Variable of Interest:
    • Inotrope exposure duration: 36 (18–79) vs 39 (19–64) hours.
    • CICU stay: 4.5 (2.0–7.0) vs 5.5 (3.0–10.0) days.
    • Early haemodynamics (examples from supplementary data): mean arterial pressure at 0 hours 78.1 ± 12.5 vs 74.7 ± 14.7 mmHg; at 24 hours 75.1 ± 10.6 vs 76.0 ± 12.4 mmHg.
  • Outcome Assessment:
    • Most endpoints were objective (death, RRT, arrhythmia requiring intervention), but the initiation thresholds for MCS and RRT can be clinician-mediated and may dilute detectability of modest pharmacologic effects.
  • Statistical Rigor:
    • Intention-to-treat analysis was performed; time-to-event analyses were used for selected components with competing-risk methodology reported.
    • Power was based on a large hypothesised absolute benefit (20%), leaving the trial underpowered for smaller but clinically relevant differences; confidence intervals remained wide.

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong, supported by randomisation and blinding, complete in-hospital outcome ascertainment, and consistent ITT analyses; the main threats are imprecision from sample size and reliance on clinician-triggered components within the composite endpoint.

External Validity

  • Population Representativeness:
    • Participants reflected a tertiary CICU cardiogenic shock cohort with predominately reduced LVEF (median 25% in both groups) and a substantial ischaemic burden (65–69%).
    • SCAI stages B–E were included, but stage C/D predominated; stage E was uncommon.
  • Applicability:
    • Findings are most directly applicable to well-resourced cardiac ICUs using structured shock pathways, where both inotropes are readily available and titratable.
    • Exclusion of out-of-hospital cardiac arrest and those already receiving inotropes before randomisation limits applicability to post-arrest shock and transferred/refractory populations.
    • Generalisation to mixed medical ICUs, resource-limited settings, or systems with different MCS/RRT thresholds may be constrained.

Conclusion on External Validity: Generalisability is moderate: the cohort is clinically recognisable for tertiary CICU cardiogenic shock, but the single-centre design and key exclusions limit transferability to broader shock systems and the most refractory/post-arrest populations.

Strengths & Limitations

  • Strengths:
    • Blinded, randomised head-to-head comparison of the two most commonly used intravenous inotropes in cardiogenic shock.
    • Clinically meaningful, patient-centred primary composite outcome focused on in-hospital events.
    • Pragmatic titration framework likely mirrors real-world clinician behaviour more closely than rigid haemodynamic targets.
    • High event rate, enhancing the efficiency of event capture for the sample size achieved.
  • Limitations:
    • Single-centre design with modest sample size; confidence intervals remain compatible with both clinically relevant benefit and harm.
    • Composite endpoint includes clinician-triggered interventions (MCS, RRT) that may be influenced by local practice thresholds even under blinding.
    • Limited invasive haemodynamic monitoring (pulmonary artery catheter present in 23 participants at randomisation), constraining mechanistic inference about haemodynamic superiority.
    • Some baseline imbalances (eg previous MI and PCI) could contribute to residual confounding despite randomisation, particularly in small trials.

Interpretation & Why It Matters

  • Clinical implication
    • In a tertiary CICU cardiogenic shock population, milrinone was not superior to dobutamine for a composite of death and major in-hospital complications, and safety signals were broadly similar.
    • In practice, inotrope choice remains individualised (eg pulmonary hypertension/right ventricular dysfunction physiology, tachyarrhythmia propensity, chronic β-blocker exposure, renal function, and vasodilatory tolerance) rather than evidence-driven superiority of one agent.
  • Methodological meaning
    • The point estimate for the primary outcome (RR 0.90) suggests any true effect, if present, is likely modest; the trial primarily informs equipoise and bounds plausible benefit/harm.
    • Decision-mediated endpoints (MCS, RRT) are common in shock trials; blinding reduces bias but does not eliminate “practice-threshold” dilution.
  • Research implication
    • DOREMI justifies larger multicentre RCTs powered for smaller effect sizes, and supports focusing on prespecified, mechanism-informed subgroups (eg right ventricular shock, renal dysfunction, β-blocker exposure) while acknowledging interaction tests require far larger samples.

Controversies & Subsequent Evidence

  • Correspondence raised concerns about dosing strategy and generalisability (including pharmacokinetic considerations for milrinone in renal dysfunction and the heterogeneous shock population), and highlighted the trial’s limited precision for subgroup inference.1
  • An accompanying editorial emphasised the wide confidence intervals, the challenges of detecting modest drug-class differences against dominant determinants of shock outcome (aetiology, reperfusion, vasopressors, MCS), and the need for larger pragmatic trials.2
  • Updated systematic reviews incorporating DOREMI reported no clear mortality advantage of milrinone over dobutamine, with low/very low certainty largely driven by the small number of randomised comparisons and imprecision.34
  • Post-hoc analyses from the DOREMI programme explored effect modification by concomitant β-blocker therapy and did not demonstrate a consistent differential benefit between agents within that subgroup (hypothesis-generating).5
  • A secondary analysis focusing on lactate clearance trajectories reported that more rapid lactate normalisation was prognostically important, without establishing superiority of one inotrope strategy for achieving lactate normalisation in the parent trial cohort.6
  • Contemporary heart failure guidance continues to recommend cautious, short-duration inotrope use as bridge/rescue therapy in advanced HF/shock contexts, without endorsing a universal preference for milrinone or dobutamine in cardiogenic shock populations.7

Summary

  • DOREMI was a blinded, randomised CICU trial comparing milrinone versus dobutamine in cardiogenic shock (n=192).
  • No significant difference was observed in the primary in-hospital composite outcome: 49% vs 54%; RR 0.90; 95% CI 0.69 to 1.19; P=0.47.
  • Mortality was numerically lower with milrinone (37% vs 43%) but imprecise: RR 0.85; 95% CI 0.60 to 1.21.
  • Key organ support endpoints (MCS, RRT) and arrhythmia events were similar between groups; confidence intervals were wide for most secondary endpoints.
  • The trial supports ongoing equipoise and patient-specific selection rather than a default preference for either agent.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Several component outcomes were analysed using time-to-event methods with competing-risk considerations; interpret HRs accordingly.
  • Secondary endpoints were not adjusted for multiplicity; emphasis should remain on the primary endpoint estimate and its confidence interval.

Overall Takeaway

DOREMI is a landmark blinded, head-to-head RCT in cardiogenic shock showing that milrinone did not improve a clinically important in-hospital composite outcome compared with dobutamine. Its principal contribution is to define equipoise with patient-centred outcomes and to bound the plausible magnitude of benefit or harm, reinforcing patient-specific inotrope selection while motivating larger multicentre trials powered for modest effects.

Overall Summary

  • Milrinone vs dobutamine: no significant difference in the primary in-hospital composite (49% vs 54%).
  • Mortality numerically favoured milrinone (37% vs 43%) but was imprecise.
  • Safety and organ support outcomes were broadly similar; the trial supports ongoing equipoise and individualised agent selection.

Bibliography