IV vitamin C for sepsis: why the dose decides whether it protects mitochondria — or stresses them

Vitamin C is one of the most contested substances in intensive care. Sepsis patients are frequently and severely deficient in it — nearly 90 percent have low levels — and as an antioxidant it should, in theory, protect the mitochondria whose failure drives organ shutdown in sepsis. A new systematic review and meta-analysis in Frontiers in Nutrition, pooling 14 randomized controlled trials and 1,958 patients, lands on a nuanced verdict: IV vitamin C may reduce 28-day mortality, but only in a narrow window — and outside that window, the same molecule may do the opposite of what's intended.

What did the 14-trial analysis find?

The pooled result: IV vitamin C reduced 28-day mortality versus control (relative risk 0.70), shortened the duration of vasopressor drugs, and lowered organ-failure scores — with no significant safety signal. But the headline comes with a large asterisk: the benefit was driven by small single-center studies and was not confirmed in the large, low-bias multicenter trials, which reported neutral or potentially harmful effects. Overall certainty of evidence was rated low.

The subgroup patterns are where it gets interesting. Benefit showed up in the short-duration subgroup (three days or less: relative risk 0.67) and, by the authors' reading, at moderate rather than extreme doses. The sickest patients — those with the deepest antioxidant depletion — appeared to benefit most, consistent with replacement therapy for deficiency rather than a drug effect in everyone.

The dose problem: antioxidant one day, pro-oxidant the next

Here is the mechanism the paper spells out, and it runs straight through mitochondria. At supra-physiological concentrations, ascorbate can reduce metal ions such as iron and catalyze the Fenton reaction, generating hydroxyl radicals — flipping from antioxidant to pro-oxidant. "This pro-oxidant switch could theoretically offset the antioxidant benefits of vitamin C at very high doses," the authors write, "potentially explaining the lack of efficacy observed in the high-dose and very high-dose subgroups."

And the mitochondrial connection is explicit: mitochondrial dysfunction is a key driver of organ failure in sepsis, and excessive oxidative stress can worsen mitochondrial damage — so high-dose vitamin C, through its pro-oxidant effects, "could theoretically worsen, rather than improve, mitochondrial function." The authors describe a U-shaped or threshold dose-response: moderate doses may help, excessive doses may harm. That framing also fits the LOVIT trial, in which high-dose vitamin C was associated with persistent organ dysfunction, and the 2026 Surviving Sepsis Campaign's weak recommendation against routine IV vitamin C in sepsis.

What this means for the bigger picture

This is a watchdog story with a constructive edge. The analysis doesn't kill the idea of IV vitamin C — it refines it: measure baseline levels, replace deficiency, keep doses moderate and courses short, and test the hypothesis in large, rigorous trials. What it does push back on is the "more is better" instinct that still surrounds high-dose vitamin C infusions. In sepsis, where mitochondria are already under siege, the difference between an antioxidant and a pro-oxidant may be a number on the dosing pump.

Our grade: contested and preliminary. The mortality signal is real but fragile — built on small studies, contradicted by the biggest trials — and the dose-response story, while mechanistically elegant, is still hypothesis-generating.

Where the evidence stands

• Established: Sepsis patients are frequently vitamin-C deficient; mitochondrial dysfunction is a key driver of organ failure in sepsis.
• Contested: A mortality benefit (RR 0.70) driven by small single-center trials, not confirmed in large multicenter RCTs; low certainty of evidence.
• Preliminary: The U-shaped dose-response — benefit at moderate doses, possible pro-oxidant mitochondrial harm at high doses — is mechanistically plausible but still theoretical.