Skip to main content

Foundational Trials

VICTAS

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of Vitamin C, Thiamine, and Hydrocortisone on Ventilator- and Vasopressor-Free Days in Patients With Sepsis: The VICTAS Randomized Clinical Trial
  • Acronym: VICTAS
  • Year: 2021
  • Journal published in: JAMA
  • Citation: Sevransky JE, Rothman RE, Hager DN, Bernard GR, Brown SM, Buchman TG, et al. Effect of Vitamin C, Thiamine, and Hydrocortisone on Ventilator- and Vasopressor-Free Days in Patients With Sepsis: The VICTAS Randomized Clinical Trial. JAMA. 2021;325(8):742-750.

Context & Rationale

  • Background
    • High-dose intravenous vitamin C, thiamine, and hydrocortisone (“metabolic resuscitation” / HAT therapy) attracted attention after an influential retrospective before–after cohort reported large mortality reductions in septic shock treated with this combination.1
    • Vitamin C monotherapy trials in severe infection/ARDS suggested possible biologic effects but yielded uncertain clinical benefit signals (and highlighted multiplicity/endpoint interpretation issues).2
    • Subsequent randomised trials of combination regimens (with or without hydrocortisone) generally showed neutral effects on patient-centred outcomes, leaving equipoise regarding efficacy, timing, and whether benefit (if any) is mediated by specific components or phenotypes.3456
  • Research Question/Hypothesis
    • In adults with suspected infection and sepsis-induced respiratory and/or cardiovascular dysfunction requiring ICU admission, does early intravenous vitamin C + thiamine + hydrocortisone improve ventilator- and vasopressor-free days (VVFDs) through day 30 compared with placebo?
  • Why This Matters
    • If effective, the regimen is inexpensive and widely available, with potential to reduce organ-support exposure and downstream morbidity.
    • If ineffective (or harmful), de-implementation is important given opportunity costs, workflow burden (q6-hourly dosing), and safety considerations (e.g., glucose measurement interference with high-dose vitamin C).9

Design & Methods

  • Research Question: Whether intravenous vitamin C + thiamine + hydrocortisone, started early in sepsis-induced acute respiratory and/or cardiovascular dysfunction, increases VVFDs through day 30 versus placebo.
  • Study Type: Multicentre, randomised, double-blind, placebo-controlled trial with an adaptive sample-size design; conducted in ICU-requiring adults across 43 hospitals in the United States; patients enrolled from emergency department or ICU locations.78
  • Population:
    • Key inclusion (summary): age ≥18 years; suspected infection (blood culture ordered AND antibiotics initiated); ICU admission planned; acute respiratory dysfunction and/or cardiovascular dysfunction persisting at randomisation; randomised within 24 hours of meeting organ dysfunction criteria (full eligibility in protocol/supplement).7
    • Respiratory dysfunction (operational definition): PaO2/FiO2 ≤300 OR SpO2/FiO2 ≤315, AND requiring one of: (i) invasive mechanical ventilation, (ii) non-invasive positive-pressure ventilation, or (iii) high-flow nasal cannula ≥40 L/min AND FiO2 ≥0.40 (with persistence at randomisation).
    • Cardiovascular dysfunction (operational definition): any vasopressor for >1 hour to maintain MAP ≥65 mmHg despite ≥1 L crystalloid (with persistence at randomisation).
    • Key exclusions: Not fully enumerated in the main report (refer to published protocol for detailed exclusions).7
  • Intervention:
    • Vitamin C 1.5 g IV + thiamine hydrochloride 100 mg IV + hydrocortisone sodium succinate 50 mg IV, administered every 6 hours, initiated within 4 hours of randomisation and continued for up to 96 hours (or until ICU discharge or death).
    • Study hydrocortisone (or matching placebo) was withheld when open-label corticosteroids totalled ≥200 mg hydrocortisone-equivalent daily dose.
  • Comparison:
    • Matching placebo (IV) administered every 6 hours on the same schedule and duration, with identical rules regarding withholding study drug when open-label corticosteroids reached ≥200 mg hydrocortisone-equivalent daily dose.
  • Blinding: Double-blind (participants, clinicians, investigators, and outcome assessors); investigational pharmacies prepared masked study drug/placebo.
  • Statistics: Adaptive sample-size design (maximum N=2000) targeting detection of a 1.5-day improvement in VVFDs (design assumptions included ~25% 30-day mortality and survivor-time effects); interim analyses at N=200/300/400 (early stopping for high predictive probability of a 30-day mortality benefit only), and at N=500/1000/1500 (stopping for VVFD efficacy or futility). One-sided type I error was controlled at 2.5%; the primary VVFD hypothesis used a one-sided P threshold of 0.022 after interim-adjustment; gatekeeping required primary VVFD significance before formal testing of 30-day mortality (one-sided α=0.024). Primary analysis was intention-to-treat; prespecified sensitivity and per-protocol analyses were performed.8
  • Follow-Up Period: Primary outcome through day 30; survival follow-up to day 180 (telephone), with follow-up completion reported by January 2020.

Key Results

This trial was stopped early. Enrolment was administratively terminated at 501 patients after the funder declined a request for additional funding (no prespecified interim stopping boundary was met); a post hoc conditional power analysis estimated a 30.7% predictive probability of ultimately demonstrating VVFD efficacy had enrolment continued to N=2000.

Outcome Vitamin C + Thiamine + Hydrocortisone (n=252) Placebo (n=249) Effect p value / 95% CI Notes
Primary: Ventilator- and vasopressor-free days (VVFDs) through day 30, median (IQR) 25 (0 to 29) 26 (0 to 28) Median difference −1 day P=0.85; 95% CI −4 to 2 Death assigned 0 VVFDs; last known status carried forward after discharge; one-sided primary hypothesis with interim-adjusted α.
30-day all-cause mortality, No. (%) 55 (22%) 60 (24%) Not reported Not reported Gatekeeping: formal mortality testing contingent on primary VVFD significance (not met).
Mortality before ICU discharge, No. (%) 52 (20.6%) 49 (19.7%) Difference 0.9% P=0.79; 95% CI −8.0 to 6.1 Exploratory outcome (main analysis).
Mortality at 180 days (among those with observed status), No. (%) 94/232 (40.5%) 85/225 (37.8%) Difference 2.7% P=0.53; 95% CI −11.3 to 5.8 Observed-status analysis; follow-up incomplete for a subset.
Time from randomisation to ICU discharge, days, median (IQR) 4 (2 to 8) 4 (2 to 8) Difference 0 days P=0.82; 95% CI −2 to 1 Exploratory outcome (main analysis).
Time from randomisation to hospital discharge, days, median (IQR) 10 (6 to 17) 9 (5 to 17) Difference 1 day P=0.66; 95% CI −3 to 2 Exploratory outcome (main analysis).
Delirium- and coma-free days through day 30, median (IQR) 4 (2 to 5) 4 (2 to 5) Difference 0 days P=0.45; 95% CI 0 to 1 Exploratory outcome; low medians reflect substantial delirium/coma burden among survivors.
Kidney replacement therapy–free days through day 30, median (IQR) 30 (0 to 30) 30 (0 to 30) Difference 0 days P=0.58; 95% CI 0 to 0 Exploratory outcome (main analysis).
SOFA score change from baseline to day 4, median (IQR) 5 (3 to 7) 5 (2 to 7) Difference 0 P=0.10; 95% CI −1 to 0 Exploratory outcome; interpretation depends on directionality as defined in the trial report.
Harms: adverse events possibly related to study drug 2 events 0 events Not reported Not reported Two events were judged possibly related (haemorrhagic shock; worsening kidney function); no serious adverse events were judged related.
  • There was no evidence of benefit on the primary endpoint (VVFDs 25 (0–29) vs 26 (0–28); median difference −1 day; 95% CI −4 to 2; P=0.85).
  • The trial’s early administrative termination (N=501 vs planned maximum N=2000) increased imprecision; however, post hoc conditional power suggested a low probability (30.7%) that continuation to N=2000 would have yielded VVFD efficacy.
  • Per-protocol analysis was also neutral (VVFDs 26 (0–29) vs 26 (0–28); median difference 0; 95% CI −3 to 2), with no supportive signal across multiple exploratory outcomes.

Internal Validity

  • Randomisation and allocation: Central randomisation (1:1), stratified by site; allocation concealed via numbered study kits prepared by investigational pharmacies.
  • Dropout or exclusions: 501 randomised (252 intervention; 249 placebo) and included in the main intention-to-treat analyses; 180-day mortality status was observed in 457 participants (232 intervention; 225 placebo) for the reported observed-status analysis.
  • Performance/detection bias: Double-blind design; primary outcome and most secondary outcomes were objective (death; ventilator/vasopressor use; discharge), reducing detection bias.
  • Protocol adherence: Per-protocol populations were 212/252 (84.1%) in intervention vs 205/249 (82.3%) in placebo (major protocol deviations excluded), indicating moderate deviation burden with similar direction across arms.
  • Baseline characteristics: Groups were similar in severity and support needs (e.g., baseline SOFA median 9 (IQR 6–12) in both arms; 44% in each arm had both respiratory and cardiovascular dysfunction).
  • Heterogeneity: Broad sepsis population across 43 hospitals with mixed infection sources and organ dysfunction patterns; this improves generalisability but may dilute any phenotype-specific effect.
  • Timing: Randomised within 24 hours of organ dysfunction onset; median time from meeting criteria to randomisation was 15 hours (IQR 8–22) in intervention vs 14 hours (IQR 8–20) in placebo.
  • Dose: Fixed “high-dose” vitamin C (6 g/day) plus thiamine (400 mg/day) plus hydrocortisone (200 mg/day), q6-hourly for up to 96 hours or ICU discharge.
  • Separation of the variable of interest: Active regimen vs placebo was clear for vitamin C and thiamine; hydrocortisone separation was partially attenuated because open-label corticosteroids were used in 33.0% (83/252) vs 32.1% (80/249), and study hydrocortisone/placebo was withheld if open-label dose reached ≥200 mg/day.
  • Adjunctive therapy use: Not reported in sufficient detail to quantify differential co-interventions beyond corticosteroid exposure.
  • Outcome assessment: VVFDs combined survival and freedom from organ support; death was scored as 0; for hospital discharge before day 30, last known support status was carried forward to day 30.
  • Statistical rigor: Prespecified adaptive interim monitoring with adjusted one-sided α for the primary endpoint; gatekeeping for mortality reduced multiplicity; nevertheless, early administrative termination curtailed the adaptive design’s intended operating characteristics.
  • Safety measurement nuance: High-dose vitamin C can interfere with some point-of-care glucose meters; VICTAS used an approved device not affected by high vitamin C, which is important for internal validity and patient safety.9

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong given robust randomisation/blinding, objective endpoints, and reasonable protocol delivery; the principal threat is early administrative termination (reduced precision and altered operating characteristics) alongside partial corticosteroid “contamination” from open-label use.

External Validity

  • Population representativeness: ICU-level adults with suspected infection and significant organ dysfunction (vasopressors and/or advanced respiratory support), broadly reflective of moderate–severe sepsis populations in high-income settings.
  • Applicability: Regimen uses globally available drugs; however, implementation requires q6-hourly dosing and awareness/mitigation of vitamin C–glucometry interference, which may be challenging in resource-limited settings or where validated glucose devices are unavailable.
  • Exclusions and system factors: Eligibility required planned ICU admission and specific organ dysfunction criteria; thus, findings may not extrapolate to ward-managed sepsis or very early ED-only sepsis without organ support.

Conclusion on External Validity: Generalisability is moderate for ICU sepsis populations requiring organ support; translation to non-ICU settings or systems without robust medication-delivery and glucose-monitoring infrastructure is more limited.

Strengths & Limitations

  • Strengths:
    • Double-blind, placebo-controlled, multicentre design across 43 hospitals, enhancing internal validity and clinical relevance.
    • Patient-centred primary endpoint integrating survival and freedom from key organ supports (ventilation/vasopressors).
    • Prespecified adaptive design and multiplicity control (interim-adjusted α; gatekeeping for mortality).
  • Limitations:
    • Stopped early for administrative/funding reasons (N=501 vs planned maximum N=2000), reducing precision and leaving residual uncertainty about smaller effects.
    • Open-label corticosteroid use in ~one-third of patients in both arms, potentially attenuating separation of the hydrocortisone component.
    • Some missing long-term survival status by day 180 (reported using observed-status analyses), which may introduce bias if missingness was non-random.

Interpretation & Why It Matters

  • Clinical practice
    • VICTAS provides no support for routine use of combined vitamin C + thiamine + hydrocortisone to increase organ-support–free days in ICU sepsis (VVFDs median 25 vs 26; P=0.85).
    • Given workflow burden and safety considerations (e.g., glucose measurement interference), off-label use should be avoided outside trials or structured protocols with appropriate monitoring.
  • Trial methodology
    • The adaptive design sought efficiency, but early administrative termination illustrates a practical vulnerability of complex platform/adaptive strategies to funding continuity.
    • VVFDs is a stringent, patient-centred composite; the absence of signal across multiple exploratory outcomes reduces the plausibility of a large unmeasured benefit.

Controversies & Subsequent Evidence

  • Governance, transparency, and early administrative termination: A JAMA editor’s note and related correspondence detailed concerns about funding decisions and DSMB membership disclosures in VICTAS, underscoring the importance of trial governance and transparency in publicly funded critical care RCTs.11
  • Editorial interpretation: The accompanying editorial emphasised that VICTAS (stopped early) and other contemporaneous RCTs did not demonstrate clinically meaningful benefit for combination therapy, and cautioned against routine adoption without clearer evidence of benefit and safety in defined populations.10
  • Safety nuance (glucometry): High-dose vitamin C can cause spuriously elevated point-of-care glucose readings with some meters, risking inappropriate insulin and hypoglycaemia; this safety concern was explicitly raised in the sepsis literature and is a key implementation issue when extrapolating trial protocols to routine care.9
  • Subsequent randomised evidence: Multiple RCTs testing vitamin C-based regimens (combination and partial combinations) have largely shown neutral effects on mortality and organ-support outcomes, aligning with VICTAS’s null primary result.34561314
  • Vitamin C monotherapy (signal of harm/neutrality): The large LOVIT trial of high-dose IV vitamin C in ICU sepsis reported no improvement in outcomes and raised concern for potential harm on a composite endpoint, further weakening the biological/clinical case for routine vitamin C-based strategies.12
  • Meta-analytic synthesis: Component network meta-analysis and subsequent systematic reviews have not shown convincing mortality benefit for vitamin C-based regimens in sepsis/septic shock, and suggest uncertainty (or null) for patient-centred outcomes across heterogeneous trials.151617
  • Guidelines: Major sepsis guidelines recommend against routine IV vitamin C in sepsis/septic shock, and a rapid practice guideline on IV vitamin C in critically ill adults similarly does not support routine use outside research settings; corticosteroid guidance has evolved separately and does not endorse adding vitamin C/thiamine as standard care.181920

Summary

  • VICTAS tested q6-hourly IV vitamin C (6 g/day), thiamine (400 mg/day), and hydrocortisone (200 mg/day) for up to 96 hours versus placebo in ICU sepsis with respiratory and/or cardiovascular dysfunction.
  • The trial was stopped early for administrative/funding reasons at N=501 (planned maximum N=2000).
  • There was no improvement in the primary outcome: VVFDs through day 30 were 25 (0–29) vs 26 (0–28); median difference −1 day; 95% CI −4 to 2; P=0.85.
  • Secondary/exploratory outcomes (ICU/hospital length of stay, delirium/coma-free days, kidney replacement therapy–free days, SOFA change, and longer-term mortality) did not show supportive benefit signals.
  • Findings align with the broader randomised evidence base and inform guidelines recommending against routine IV vitamin C-based regimens in sepsis outside trials.

Overall Takeaway

VICTAS is a landmark test of the “metabolic resuscitation” hypothesis in ICU sepsis, demonstrating no improvement in a stringent, patient-centred organ-support composite (VVFDs) despite biologic plausibility and widespread clinical interest. Although stopped early for administrative reasons, the absence of signal across the primary and multiple exploratory outcomes—together with concordant subsequent RCTs and meta-analyses—supports guideline recommendations against routine vitamin C-based regimens in sepsis outside clinical trials.

Overall Summary

  • In ICU sepsis with acute respiratory and/or cardiovascular dysfunction, vitamin C + thiamine + hydrocortisone did not increase VVFDs at 30 days and has not been supported by the broader post-VICTAS evidence base.

Bibliography