Skip to main content

Foundational Trials

LOVIT

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Intravenous Vitamin C in Adults with Sepsis in the Intensive Care Unit
  • Acronym: LOVIT (Lessening Organ dysfunction with VITamin C)
  • Year: 2022
  • Journal published in: The New England Journal of Medicine
  • Citation: Lamontagne F, Masse MH, Menard J, Sprague S, Pinto R, Heyland DK, et al. Intravenous Vitamin C in Adults with Sepsis in the Intensive Care Unit. N Engl J Med. 2022;386(25):2387-2398.

Context & Rationale

  • Background
    • Critically ill patients with sepsis frequently have low circulating vitamin C concentrations, plausibly reflecting increased consumption, reduced intake/absorption, and altered distribution.
    • Vitamin C has biologically plausible mechanisms in sepsis (antioxidant effects; endothelial stabilisation; cofactor roles in catecholamine synthesis and immune function).
    • Early small trials (vitamin C alone or in combination “metabolic resuscitation” regimens) produced heterogeneous signals for shock resolution, organ dysfunction and mortality, but were often underpowered and/or relied on secondary endpoints.
    • Off-label use became common in some settings despite uncertain efficacy and plausible harms (e.g., oxalate load/renal injury; glucose meter interference; rare haemolysis in G6PD deficiency).
  • Research Question/Hypothesis
    • In ICU adults with sepsis requiring vasopressors, does a 96-hour course of high-dose intravenous vitamin C reduce death or persistent organ dysfunction at day 28 compared with placebo?
  • Why This Matters
    • A large, blinded trial was needed to resolve conflicting early signals in an intervention that is inexpensive, widely available, and potentially rapidly implementable.
    • Demonstrating either benefit or harm would have immediate implications for “adjunctive” sepsis therapy practice and guideline recommendations.
    • The trial design also tests whether biologically plausible antioxidant therapy translates into clinically meaningful, patient-centred outcomes in septic shock.

Design & Methods

  • Research Question: Among ICU adults with sepsis receiving vasopressors, does high-dose intravenous vitamin C (vs placebo) reduce death or persistent organ dysfunction at day 28?
  • Study Type: Randomised, multicentre, investigator-initiated, international, double-blind, placebo-controlled trial in adult ICUs (35 ICUs in Canada, France, and New Zealand); enrolment November 2018 to July 2021.
  • Population:
    • Adults (≥18 years) admitted to ICU for ≤24 hours.
    • Suspected or confirmed infection as the primary reason for ICU admission.
    • Receiving a vasopressor infusion at screening/randomisation.
    • Key exclusions included: known allergy to vitamin C; known G6PD deficiency; pregnancy or breastfeeding; history of kidney stones within the previous year; limitations of life-sustaining treatment or expected death within 48 hours; receipt of intravenous vitamin C during the current hospitalisation (excluding parenteral nutrition); inability to obtain consent within local frameworks.
  • Intervention:
    • Intravenous vitamin C (ascorbic acid) 50 mg/kg every 6 hours for up to 96 hours (maximum 16 doses), initiated as soon as feasible after randomisation.
  • Comparison:
    • Matching placebo intravenous infusion every 6 hours for up to 96 hours, alongside usual sepsis care at clinician discretion.
  • Blinding: Double-blind (participants, clinicians, investigators, and outcome assessors); indistinguishable study preparations dispensed via pharmacy processes.
  • Statistics: A total of 770 patients were required to detect a 10% absolute reduction in the primary outcome (from 50% to 40%) with 80% power at a two-sided 5% significance level; planned enrolment was 800 (with an increase to preserve the planned number of non–COVID-19 participants during the pandemic); primary analysis was intention-to-treat. 12
  • Follow-Up Period: Primary endpoint at day 28; survival and health-related quality-of-life outcomes assessed to 6 months.

Key Results

This trial was not stopped early. Recruitment and follow-up were completed to the prespecified target.

Outcome Vitamin C Placebo Effect p value / 95% CI Notes
Death or persistent organ dysfunction at day 28 (primary) 191/429 (44.5%) 167/434 (38.5%) RR 1.21 95% CI 1.04 to 1.40; P=0.01 Persistent organ dysfunction: vasopressor use, invasive mechanical ventilation, or new renal-replacement therapy at day 28.
Death at day 28 152/429 (35.4%) 137/433 (31.6%) RR 1.17 95% CI 0.98 to 1.40; P=Not reported All-cause mortality.
Persistent organ dysfunction at day 28 39/429 (9.1%) 30/433 (6.9%) RR 1.30 95% CI 0.83 to 2.05; P=Not reported Composite component (reported separately).
ICU death 110/429 (25.6%) 97/433 (22.4%) RR 1.15 95% CI 0.92 to 1.44; P=Not reported All-cause ICU mortality.
Hospital death 132/429 (30.8%) 122/433 (28.2%) RR 1.09 95% CI 0.90 to 1.31; P=Not reported All-cause in-hospital mortality.
Death by 6 months 191/417 (45.8%) 185/426 (43.4%) HR 1.14 95% CI 0.93 to 1.39; P=Not reported Time-to-event analysis.
Days without organ dysfunction in ICU (to day 28) 17.0 (−1 to 25.0) 19.5 (−1 to 25.0) Median diff −2.43 95% CI −7.23 to 2.37 Higher values indicate more days alive and free of organ dysfunction.
Vasopressor-free days (to day 28) 22.0 (15.0 to 25.0) 23.0 (14.0 to 25.0) Median diff −0.55 95% CI −2.44 to 1.35 No clinically meaningful separation.
Ventilator-free days (to day 28) 26.0 (7.0 to 28.0) 28.0 (9.0 to 28.0) Median diff −0.96 95% CI −3.56 to 1.64 No evidence of improved respiratory support liberation.
Stage 3 acute kidney injury 162/429 (37.8%) 164/433 (37.9%) RR 1.00 95% CI 0.85 to 1.19; P=Not reported No signal for differential severe AKI.
Hypoglycaemia 26/429 (6.1%) 22/433 (5.1%) RR 1.25 95% CI 0.73 to 2.14; P=Not reported Includes an index-trial safety signal relating to glucose measurement artefact.
Serious adverse events (adjudicated) 1/429 (0.2%) 0/433 (0%) Not estimated Not reported One life-threatening anaphylaxis event in the vitamin C group.
  • Primary finding: The primary composite outcome was higher with vitamin C (191/429 [44.5%] vs 167/434 [38.5%]; RR 1.21; 95% CI 1.04 to 1.40; P=0.01), with the excess largely attributable to the mortality component (152/429 vs 137/433).
  • Patient-centred secondary outcomes: There was no evidence of improved organ support liberation (e.g., vasopressor-free days 22 [15–25] vs 23 [14–25]; median diff −0.55; 95% CI −2.44 to 1.35) and no improvement in 6-month survival (HR 1.14; 95% CI 0.93 to 1.39).
  • Subgroups and result interpretation:
    • Females: RR 1.39; 95% CI 1.10 to 1.76 (72/151 vs 62/173); males: RR 1.11; 95% CI 0.92 to 1.34 (119/278 vs 104/260).
    • Lower predicted-risk quartiles showed larger harm estimates (quartile 1: RR 2.05; 95% CI 1.08 to 3.90 [22/95 vs 12/98]; quartile 2: RR 1.49; 95% CI 1.09 to 2.03 [55/117 vs 39/118]), but multiplicity and the absence of reported interaction p values limit causal inference.
    • A prespecified adjusted analysis attenuated the primary effect estimate (RR 1.15; 95% CI 0.90 to 1.47; P=Not reported), underscoring sensitivity to modelling choices.

Internal Validity

  • Randomisation and allocation:
    • Central randomisation with allocation concealment; assignment was blinded via identical study preparations and pharmacy-controlled dispensing.
    • Randomisation was stratified by trial site, supporting balance across centres.
  • Dropout or exclusions (post-randomisation):
    • 872 patients underwent randomisation; 8 were deemed ineligible immediately after randomisation and were excluded before receipt of study drug.
    • One participant randomised under deferred consent died before consent could be obtained; allocation and primary outcome status were retained for the primary analysis only (explaining the placebo denominator of 434 for the primary outcome).
    • Primary-outcome status at day 28 was missing for one participant; best-case/worst-case sensitivity analyses were prespecified for such missingness.
  • Performance/detection bias:
    • Double blinding reduces the plausibility of differential co-intervention or outcome ascertainment bias.
    • The primary outcome was largely objective, but the “persistent organ dysfunction” component is contingent on ongoing life-sustaining therapies at day 28 (potentially influenced by care limitation practices), with blinding mitigating this risk.
  • Protocol adherence (delivery as intended):
    • At least one trial infusion: 425/429 (99.1%) in vitamin C vs 427/433 (98.6%) in placebo.
    • Median number of doses received: 16 (IQR 14 to 16) in both groups.
    • Receipt of ≥90% of scheduled doses (among those with ≥1 scheduled dose): 412/423 (97.4%) in vitamin C vs 412/429 (96.0%) in placebo.
    • First dose administered >4 hours after randomisation: 42/429 (9.8%) vs 45/433 (10.4%).
    • Open-label vitamin C exposure: 1/429 (0.2%) vs 1/433 (0.2%).
  • Baseline characteristics and illness severity:
    • Groups were broadly comparable: age 65±14 vs 67±14 years; APACHE II 24±7 vs 24±7; SOFA 12±3 vs 12±3; invasive ventilation at baseline 209/429 (48.7%) vs 213/433 (49.2%).
    • Time from ICU admission to randomisation: 12.0±5.6 vs 12.1±5.6 hours, supporting early treatment in established septic shock physiology.
    • Baseline plasma vitamin C was measured in a subset only (324 vs 322); mean values were 20.6±70.6 vs 19.1±39.7 μmol/L, limiting mechanistic interpretation and effect-modification analyses.
  • Separation of the variable of interest:
    • Study-drug exposure separation was supported by high adherence and minimal open-label contamination (0.2% in each arm).
    • Post-randomisation vitamin C levels were not reported, so pharmacodynamic separation was inferred rather than directly demonstrated.
  • Adjunctive therapy use:
    • Corticosteroids during ICU days 1–4: 281/429 (65.5%) vs 263/433 (60.7%).
    • Thiamine during ICU days 1–4: 112/429 (26.1%) vs 106/433 (24.5%).
    • These balances reduce concern that differential co-interventions explain the primary signal.
  • Statistical rigour:
    • The primary analysis met conventional frequentist significance (P=0.01) and the trial exceeded the minimum planned sample size, supporting nominal power for the prespecified effect size.
    • Sensitivity/adjusted analyses attenuated the effect estimate, underscoring dependence on modelling assumptions and reinforcing cautious interpretation.

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong given double blinding, high protocol adherence, balanced co-interventions, and objective endpoints, but is tempered by post-randomisation exclusions and interpretive complexity of a composite outcome driven predominantly by mortality.

External Validity

  • Population representativeness:
    • Participants were typical of ICU sepsis requiring vasopressors (older adults, high APACHE II/SOFA, ~50% invasively ventilated at baseline).
    • Exclusions (e.g., G6PD deficiency, recent kidney stones, imminent death/limitations of care) are clinically reasonable but restrict applicability to these subgroups.
    • Most enrolment occurred in Canadian ICUs, with smaller contributions from France and New Zealand.
  • Applicability:
    • Findings are most directly applicable to high-income ICU settings delivering contemporary sepsis care, where early initiation of adjuncts is feasible.
    • Generalisation to less severe sepsis (no vasopressor requirement), non-ICU settings, paediatrics, and resource-limited systems is uncertain.
    • The tested regimen was a high-dose, short-course strategy; findings do not directly inform alternative dosing, duration, or tapering approaches.

Conclusion on External Validity: Generalisability is good for adult ICU patients with vasopressor-dependent sepsis in similar health systems, but limited for non-ICU sepsis populations and for different dosing strategies.

Strengths & Limitations

  • Strengths:
    • Large, multicentre, international, double-blind randomised design with high protocol adherence and minimal contamination.
    • Clinically important primary outcome incorporating mortality and sustained organ support dependency.
    • Robust follow-up to 6 months for survival and patient-reported outcomes (where available).
  • Limitations:
    • Post-randomisation exclusions of ineligible patients, and a deferred-consent primary-outcome-only participant, complicate strict intention-to-treat purity.
    • Composite endpoint interpretation is challenging when one component (mortality) drives the effect while the other component shows smaller and directionally discordant differences.
    • Baseline vitamin C levels were measured in a subset only, with no reported post-randomisation pharmacodynamic data.
    • Health-related quality-of-life outcomes at 6 months were available for a subset of survivors, introducing potential non-response bias.

Interpretation & Why It Matters

  • Clinical practice
    • High-dose intravenous vitamin C (50 mg/kg every 6 hours for 96 hours) should not be used routinely as adjunctive therapy for ICU sepsis with vasopressor requirement, given a signal for harm in the primary outcome and no demonstrated improvement in organ support–free days or longer-term survival.
  • Trialists/methodologists
    • LOVIT illustrates how composite outcomes can be driven by a single component (mortality), creating interpretive complexity when other components differ less or in different directions.
    • The trial also demonstrates that “low-risk” nutraceutical adjuncts can generate clinically important safety signals when tested rigorously.
  • Implementation and safety
    • Where vitamin C is used in research contexts, glucose monitoring strategy must account for potential point-of-care assay interference and risk of iatrogenic insulin administration.
    • Future trials (if any) should justify dosing/duration, consider pharmacodynamic monitoring, and predefine discontinuation strategies (including whether tapering is necessary).

Controversies & Subsequent Evidence

  • Why did an intervention with biological plausibility show harm?
    • Published critiques emphasised that high-dose vitamin C is pharmacological rather than nutritional, with potential for unanticipated adverse biology (e.g., pro-oxidant chemistry in relevant microenvironments) and measurement artefacts that can translate into patient harm. 34
  • Discontinuation hypothesis (rebound risk):
    • A secondary analysis proposed that abrupt cessation after a short, high-dose course might contribute to delayed harm, raising the question of whether tapering or longer physiological replacement strategies should be studied instead; this interpretation remains contested and is not supported by direct pharmacokinetic data within LOVIT. 5
  • Frequentist vs Bayesian interpretation:
    • A Bayesian reanalysis explored posterior probabilities for benefit and harm under different priors and emphasised that inference is sensitive to prior choices and to the weight assigned to earlier small trials. 6
    • An accompanying editorial argued that Bayesian approaches can be informative but should not be used to dismiss statistically robust harm signals, particularly in the context of widespread off-label adoption. 7
  • Meta-analytic signal after LOVIT:
    • A large systematic review of parenteral vitamin C in severe infection (including LOVIT) found no consistent mortality benefit and reinforced uncertainty regarding clinically meaningful improvement in organ dysfunction outcomes. 8
    • Subsequent meta-analyses focusing on sepsis/septic shock have generally reported no benefit and have highlighted heterogeneity by regimen (monotherapy vs combination therapy), baseline risk, and outcome timing. 910
  • Guideline responses:
    • A rapid recommendation guideline concluded that intravenous vitamin C should not be used routinely in adult sepsis or septic shock (recommendation against, reflecting very low certainty of benefit and emerging harm concerns). 11
    • Japanese guideline updates similarly moved away from routine vitamin C use in sepsis. 1213
  • Major follow-up trial:
    • The C-EASIE multicentre trial of early vitamin C in sepsis/septic shock did not identify a clinically important benefit, reinforcing lack of efficacy across settings and prompting further caution against empiric use. 14

Summary

  • LOVIT tested high-dose intravenous vitamin C (50 mg/kg every 6 hours for 96 hours) versus placebo in ICU adults with sepsis receiving vasopressors.
  • The trial was not stopped early and achieved high protocol adherence with minimal open-label contamination.
  • The primary outcome (death or persistent organ dysfunction at day 28) was higher with vitamin C (RR 1.21; 95% CI 1.04 to 1.40; P=0.01), driven mainly by mortality.
  • No improvement was seen in vasopressor-free days, ventilator-free days, organ dysfunction–free days, or 6-month survival.
  • Subgroup patterns suggested possible differential harm in some strata (e.g., females; lower predicted-risk quartiles), but these findings remain exploratory.

Overall Takeaway

LOVIT is a landmark ICU sepsis trial because it rigorously tested a widely discussed, inexpensive adjunctive therapy and identified a signal for harm rather than benefit. Its results shifted the evidentiary centre-of-gravity away from empiric high-dose vitamin C use in vasopressor-dependent sepsis, and it catalysed subsequent meta-analyses and guideline updates recommending against routine administration.

Overall Summary

  • High-dose intravenous vitamin C increased the primary composite of death or persistent organ dysfunction at day 28 (RR 1.21; P=0.01).
  • No meaningful improvement in organ support–free days or 6-month survival was demonstrated.
  • Subsequent meta-analyses and guideline responses generally recommend against routine vitamin C in adult sepsis/septic shock.

Bibliography