Skip to main content

Foundational Trials

VIOLET

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Early High-Dose Vitamin D3 for Critically Ill, Vitamin D-Deficient Patients
  • Acronym: VIOLET (Vitamin D to Improve Outcomes by Leveraging Early Treatment)
  • Year: 2019
  • Journal published in: New England Journal of Medicine
  • Citation: National Heart, Lung, and Blood Institute PETAL Clinical Trials Network; Ginde AA, Brower RG, Caterino JM, Finck L, Banner-Goodspeed VM, Grissom CK, et al. Early high-dose vitamin D3 for critically ill, vitamin D-deficient patients. N Engl J Med. 2019 Dec 26;381(26):2529-2540.

Context & Rationale

  • Background
    • Vitamin D deficiency is common among critically ill adults and observational studies have linked low 25-hydroxyvitamin D concentrations with infection, lung injury, organ dysfunction, and mortality (causality uncertain).
    • Prior randomised evidence was limited; the VITdAL-ICU trial did not shorten hospital length of stay overall but raised the possibility of benefit in patients with severe deficiency in subgroup analyses.1
    • High-dose (bolus) cholecalciferol is a pragmatic intervention, but has uncertain biological relevance in acute critical illness given inflammation-related changes in vitamin D binding and metabolism.
  • Research Question/Hypothesis
    • Among critically ill adults with vitamin D deficiency identified early in the course of illness, does a single early high-dose enteral cholecalciferol bolus reduce 90-day all-cause mortality versus placebo?
  • Why This Matters
    • Vitamin D is inexpensive, widely available, and routinely measured; a true mortality benefit would have immediate system-wide implications.
    • Conversely, a well-conducted negative trial would discourage low-value testing and empiric high-dose supplementation in ICU populations.
    • The trial’s “rapid screening + early treatment” paradigm directly tested a clinically implementable pathway rather than late supplementation.

Design & Methods

  • Research Question: In critically ill adults with vitamin D deficiency identified early after an ICU admission decision, does a single high-dose enteral vitamin D3 bolus reduce 90-day all-cause mortality compared with placebo?
  • Study Type: Randomised, multicentre, double-blind, placebo-controlled trial within the NHLBI PETAL Clinical Trials Network (United States); screening and enrolment from emergency department, hospital ward, operating theatre, or transfer hospitals; randomisation targeted within 12 hours of the ICU admission decision.
  • Population:
    • Key inclusion: adults with an ICU admission decision plus ≥1 qualifying acute condition (e.g., pneumonia, sepsis, shock, mechanical ventilation for acute respiratory failure, aspiration, smoke inhalation, pancreatitis, lung contusion).
    • Vitamin D criterion at enrolment: point-of-care 25-hydroxyvitamin D concentration <20 ng/mL.
    • Key exclusions (examples): inability to take enteral medication; inability to obtain timely consent/surrogate; hypercalcaemia; history of kidney stones; withholding/withdrawal of life-sustaining treatment; expected survival <48 hours; >72 hours since hospital presentation; other protocol-defined exclusions.
  • Intervention:
    • Single enteral dose of cholecalciferol 540,000 IU (liquid), administered orally or via nasogastric/orogastric tube within 2 hours after randomisation.
    • No mandated co-interventions; clinicians were asked to avoid additional vitamin D testing or supplementation for 1 month to minimise contamination.
  • Comparison:
    • Matched placebo (single enteral dose, identical administration window and route).
  • Blinding: Double-blind (participants, clinicians, investigators, and outcome assessors); allocation delivered via a central electronic system with concealment by design.
  • Statistics: Planned enrolment of 3000 patients (assuming 90-day mortality 20% with placebo and 15% with vitamin D3) to provide 87% power at a two-sided 5% significance level; design included 3 interim analyses and assumed ~80% of point-of-care–positive patients would have deficiency confirmed by reference testing; primary analysis was prespecified in the “primary analysis population” with confirmed baseline deficiency, with analyses conducted according to randomised assignment (intention-to-treat within the prespecified analysis population).
  • Follow-Up Period: 90 days after randomisation (primary endpoint assessed to day 90; additional outcomes at day 28 and day 7 for selected physiological endpoints).

Key Results

This trial was stopped early. After the first interim analysis, the data and safety monitoring board recommended stopping for futility; the predictive probability of showing vitamin D3 superiority with full enrolment was <2% (1360 patients randomised vs 3000 planned).

Outcome Vitamin D3 Placebo Effect p value / 95% CI Notes
Death from any cause to day 90 (primary analysis population; confirmed deficiency) 125/531 (23.5%) 109/528 (20.6%) Not reported P=0.26 Primary endpoint; denominators exclude those lost to follow-up for day-90 status.
Death from any cause to day 28 (confirmed deficiency) 92/531 (17.3%) 69/528 (13.1%) Risk difference 4.3; 95% CI -0.1 to 8.6 Absolute difference (percentage points) reported in trial table.
Death in hospital to day 90 (confirmed deficiency) 92/538 (17.1%) 72/539 (13.4%) Risk difference 3.7; 95% CI -0.5 to 8.0 Hospital mortality component reported separately from day-90 all-cause death.
Alive and at home under previous level of care at day 90 (confirmed deficiency) 348/528 (65.9%) 345/526 (65.6%) Risk difference 0.3; 95% CI -5.4 to 6.0 Composite “patient-centred” recovery endpoint.
Ventilator-free days to day 28 (confirmed deficiency) 21.3 ± 11.3 (n=523) 22.1 ± 10.5 (n=534) Mean difference -0.8; 95% CI -2.1 to 0.5 Higher is better; no signal of benefit.
25-hydroxyvitamin D at day 3 (ng/mL; confirmed deficiency subset with data) 46.9 ± 23.2 (n=145) 11.4 ± 5.6 (n=133) Mean difference 35.5; 95% CI 31.5 to 39.6 Strong biochemical separation demonstrating delivery/absorption in measured subset.
Hypercalcaemia to day 14 (confirmed deficiency) 14/513 (2.7%) 11/523 (2.1%) Not reported P=0.51 Safety endpoint; event definition per trial.
Kidney stones to day 90 (confirmed deficiency) 0/507 (0.0%) 3/507 (0.6%) Not reported P=0.25 Rare events; limited precision.
Death from any cause to day 90 (screened-deficient population; point-of-care positive) 159/682 (23.3%) 137/656 (20.9%) Not reported P=0.28 Prespecified broader analysis; consistent with primary analysis population.
  • Despite rapid administration (within 2 hours after randomisation) and clear biochemical separation at day 3 (46.9 ± 23.2 vs 11.4 ± 5.6 ng/mL), early bolus vitamin D3 did not reduce 90-day mortality (23.5% vs 20.6%; P=0.26).
  • Key clinical secondary outcomes showed no improvement, with point estimates often numerically favouring placebo (e.g., 28-day mortality 17.3% vs 13.1%; risk difference 4.3; 95% CI -0.1 to 8.6).
  • Clinically important harms were not apparent; hypercalcaemia was uncommon (2.7% vs 2.1%; P=0.51) and kidney stones were rare (0.0% vs 0.6%; P=0.25).

Internal Validity

    • Randomisation and allocation concealment: central electronic randomisation with permuted blocks stratified by site; double-blind trial design supports robust concealment.
    • Dropout/exclusions: 2 randomised participants were excluded (inadequate consent and withdrawal of consent); day-90 vital status was unavailable for 7 vs 12 participants in the confirmed-deficiency cohort (primary endpoint denominators 531 vs 528), which is low and balanced.
    • Performance/detection bias: blinding reduces co-intervention and ascertainment bias; primary endpoint (all-cause mortality) is objective.
    • Protocol adherence: among confirmed-deficiency participants, study drug was administered to 532/538 (98.9%) vs 532/540 (98.5%); wrong-drug administration was rare (1 placebo-in-error in vitamin D3 arm; 3 vitamin D3-in-error in placebo arm).
    • Baseline characteristics: well balanced (e.g., age 55.5 ± 15.9 vs 54.6 ± 16.7 years; total SOFA 5.6 ± 3.6 vs 5.4 ± 3.7; baseline 25-hydroxyvitamin D 11.2 ± 4.8 vs 11.0 ± 4.7 ng/mL).
    • Heterogeneity: broad inclusion across multiple acute critical illness phenotypes (pneumonia/sepsis/shock/acute respiratory failure) increases generalisability but could dilute benefit if effects are phenotype- or severity-specific.
    • Timing: screening and randomisation targeted within 12 hours of ICU admission decision; study drug administered within 2 hours after randomisation, consistent with an “early treatment” hypothesis.
    • Dose: large single bolus (540,000 IU) ensures rapid rise in 25-hydroxyvitamin D but does not test maintenance supplementation; biological relevance of bolus dosing during acute inflammation remains uncertain.
    • Separation of the variable of interest: day-3 25-hydroxyvitamin D was 46.9 ± 23.2 ng/mL vs 11.4 ± 5.6 ng/mL (mean difference 35.5; 95% CI 31.5 to 39.6), demonstrating strong biochemical separation.
    • Outcome assessment: mortality to day 90 across locations; follow-up mechanisms included record review and telephone, with low loss to follow-up.
    • Statistical rigor: prespecified interim monitoring using Bayesian futility thresholds; early stopping reduces precision and the ability to exclude small but clinically relevant effects.

Conclusion on Internal Validity: Moderate. Randomisation, blinding, objective primary outcome, and strong biochemical separation support causal inference, but early stopping for futility reduced power/precision and the prespecified confirmed-deficiency primary analysis population necessarily departs from a pure “all randomised” ITT framework.

External Validity

    • Population representativeness: enrolment reflected common ICU entry syndromes (pneumonia, sepsis, shock, acute respiratory failure) and targeted early critical illness, but required rapid consent/surrogate processes and enteral access.
    • Important exclusions: patients unable to take enteral medication, those with hypercalcaemia or kidney stone history, and those with imminent death or limitations of life-sustaining treatment were excluded; these groups may have different risk–benefit profiles.
    • Applicability across settings: results are most applicable to high-resource systems able to implement rapid screening and early enteral bolus dosing; generalisability to settings with profound malnutrition, different baseline vitamin D status distributions, or delayed ICU access is uncertain.

Conclusion on External Validity: Generalisability is reasonable for early ICU admissions in similar high-resource hospitals, but extrapolation to patients without enteral access, with contraindications to high-dose vitamin D, or with extreme deficiency/longer-duration critical illness is limited.

Strengths & Limitations

  • Strengths:
    • Pragmatic “screen early, treat early” design aligned with real-world implementation.
    • Double-blind randomised methodology with objective primary outcome.
    • Strong physiological separation in measured 25-hydroxyvitamin D concentrations by day 3.
    • Low loss to follow-up for vital status to day 90.
  • Limitations:
    • Stopped early for futility, reducing power and narrowing inference to larger effect sizes.
    • Primary analysis population required confirmation of deficiency by reference assay after randomisation (prespecified but can be conceptually challenging for ITT purists).
    • Single bolus strategy does not test sustained supplementation during recovery.
    • Outcome table reporting emphasised P value for the primary endpoint without an accompanying effect estimate/CI in the main table.

Interpretation & Why It Matters

  • Clinical implications
    • Routine early high-dose (540,000 IU) bolus vitamin D3 for critically ill patients selected by point-of-care low 25-hydroxyvitamin D is not supported.
    • The trial supports de-implementation of indiscriminate vitamin D testing or empiric high-dose bolus supplementation during acute critical illness outside nutritional support practices and standard deficiency management.
    • The results shift attention towards refined targeting (e.g., severe deficiency and/or particular phenotypes) and towards regimens that test sustained supplementation strategies rather than single bolus dosing.

Controversies & Subsequent Evidence

    • Illness severity and target phenotype: correspondence highlighted concerns that enrolled patients may have been too heterogeneous and, in aggregate, not severe enough (or not “vitamin D–responsive” enough) to show benefit, and questioned whether benefit—if present—would be restricted to severe deficiency or specific inflammatory phenotypes.3
    • Biomarker validity in critical illness: correspondence also emphasised that total 25-hydroxyvitamin D behaves as a negative acute-phase reactant and may not reflect functional vitamin D biology during critical illness, complicating both eligibility targeting and mechanistic interpretation.3
    • Bolus dosing strategy: the single bolus corrected measured 25-hydroxyvitamin D rapidly but may not replicate potential benefits of sustained repletion; this remains an open design question rather than a demonstrated failure of all supplementation approaches.
    • Relationship to prior RCTs: VITdAL-ICU did not demonstrate overall benefit but suggested possible signal in severe deficiency, motivating larger definitive testing.1
    • Subsequent syntheses: post-VIOLET meta-analyses pooling randomised trials have generally not shown a consistent mortality benefit from vitamin D supplementation in critically ill adults; reported effects on other outcomes (e.g., ICU stay, ventilation) are variable and sensitive to trial selection and heterogeneity.45
    • Ongoing definitive testing: the VITDALIZE programme is designed to test high-dose vitamin D3 in critically ill adults with severe deficiency using a multicentre double-blind phase III approach, addressing targeting and regimen questions at scale.2
    • Guideline positioning: paediatric Surviving Sepsis Campaign guidelines suggest against acute repletion of vitamin D deficiency in children with sepsis/septic shock (low certainty evidence), reflecting the broader uncertainty and limited RCT signal in critical illness.6
    • Nutrition/micronutrient guidance: ESPEN’s practical micronutrient guideline notes that the limited number of interventional trials constrains certainty for most recommendations and emphasises structured monitoring and targeted management strategies for vitamins (including vitamin D) during nutritional support rather than indiscriminate high-dose use.7

Summary

  • VIOLET tested an implementable “rapid screening + early bolus treatment” strategy: a single 540,000 IU enteral vitamin D3 dose versus placebo in critically ill adults with point-of-care vitamin D deficiency.
  • The trial was stopped early for futility after the first interim analysis (1360 randomised vs 3000 planned), limiting precision for modest effects.
  • In the prespecified confirmed-deficiency cohort, vitamin D3 did not reduce 90-day mortality (23.5% vs 20.6%; P=0.26) or improve major secondary clinical outcomes.
  • Vitamin D3 produced strong biochemical separation by day 3 (46.9 ± 23.2 vs 11.4 ± 5.6 ng/mL), indicating effective delivery/absorption in measured participants.
  • Clinically important harms were not evident; hypercalcaemia was uncommon and kidney stones were rare, though safety event rates were low and CIs were not always provided.

Overall Takeaway

VIOLET is a landmark “de-implementation” trial: it rigorously tested an attractive, low-cost, early supplementation strategy at scale and found no signal of meaningful benefit on survival or key patient-centred outcomes. Its early stopping for futility and strong biochemical separation together suggest that, at least for a single early bolus regimen in this broadly defined early-critical-illness population, correcting low 25-hydroxyvitamin D does not translate into improved clinical outcomes.

Overall Summary

  • Early high-dose vitamin D3 bolus rapidly raised 25-hydroxyvitamin D concentrations but did not improve 90-day survival or key recovery outcomes in vitamin D–deficient critically ill adults.

Bibliography