We carried out this meta-analysis of the most recent RCTs to analyze the efficacy of IVVC in sepsis or septic shock patients. Eighteen RCTs with 2980 patients were included in the final analysis. The primary outcome analysis revealed IVVC in patients was not associated with improved mortality. There was no statistically difference between IVVC combination therapy and monotherapy. The primary outcome was contrary to the study of Marik et al [8], which suggested that combined treatment could decrease hospital mortality.
Triple therapy of vitamin C, hydrocortisone and thiamine has biological rationality in sepsis, however, our primary outcome was insignificant in contrast to studies which concluded that coadministration offered acceptable outcomes in sepsis or septic shock patients [8, 23]. On the other hand, several large RCTs of vitamin C in sepsis or septic shock did not produce a remarkable reduction in mortality [15, 22, 28, 32]. Despite the large multicenter trial of VITAMINS, there was a concern about the time of combined therapy initiation [28]. Individual trial of combination treatment had shown a decrease in 28-day mortality in the prespecified subgroup of sepsis patients within 48 hours speculate that administering vitamin C as soon as possible may be meaningful [27]. However, this result has not been replicated. Recent meta-analyses [11, 13, 37–39] also failed to find improved mortality among patients with sepsis, no matter combined therapy or monotherapy. Different inclusion criterion, patient populations and timing of administration explain the heterogeneity of different research results to some extent.
We attempted to determine whether there may be some significant differences between high and low dose of vitamin C by performing a subgroup analysis comparing mortality. Interestingly, our results revealed that low dose (< 10g/d) IVVC, compared with high-dose (≥ 10g/d), was associated with improved mortality, which was contrary to previous meta-analysis that high-dose (≥ 10g/d) IVVC was benefit for mortality. CITRIS-ALI trial revealed that compared with placebo, high-dose vitamin C (50mg/kg every 6 hours) had a lower 28-day mortality in exploratory analysis [33], while the larger scale LOVIT trial did not improve 28-day mortality [15]. In addition, a component network meta-analysis demonstrated high-dose (6g/d) and very-high dose vitamin C (> 12g/d) were associated with decreased mortality but with low certainty [40]. The exact mechanism by which low dose vitamin C works has not been clarified. However, these studies populations of previous meta-analysis included critically ill patients. It is unclear whether higher doses confer greater benefits, although low dose was used in VITAMINS trial [28], the vitamin C level reached almost the same concentration at 6 hours [41] as published in CITRIS-ALI trial [33] of a high dosing regimen at 48 hours.
In spite of mortality, this meta-analysis indicated promising results that IVVC could significantly short duration of vasopressor use and improve the change of SOFA score. Our results were contrary to meta-analysis of Cai et al [11] that vitamin C did not produce obvious effect on duration of vasopressor use, however, the meta-analysis comprised 6 cohort studies. Corticosteroids decrease vasopressor requirements in septic shock [42, 43], and the hemodynamic improvement may be due to corticosteroids [32]. Changes in SOFA score would suggest whether organ function has improved. Vitamin C has numerous effects including antioxidant, immune-supporting, anti-inflammatory and improving vascular endothelial cell function [6, 44]. As is well known that vitamin C is a cofactor in the production of biosynthetic enzymes that can increase vasopressin synthesis [6, 45]. Therefore, the vital role of vitamin C may be the reason that vitamin C therapy could shorten duration of vasopressor use and improve the change in SOFA score. Although there were medium degree of heterogeneity in the pool results of duration of vasopressor use and the change of SOFA score, no publication bias were found in the assessment of Egger’s test.
Hypovitaminosis is generally recognized as vitamin C level below 23 µmol/L [25]. A previous study showed that patients with sepsis need 3g/day of vitamin C to reach normal plasma level [34]. Six researches enrolled in our meta-analysis showed the vitamin C level in intervention group at 32-96h were increased significantly, but did not translate into improved clinical mortality. A previous study suggested that some patients may develop vitamin deficiency within 48 hours after discontinuing vitamin C infusion in spite of the dosing regimen [46], which may result in no improvement in mortality.
Adverse effects of vitamin C were rare. Driny et al believed that vitamin C was safe, tolerable and would not cause patients withdraw from the study [35]. On the basis of this study, Fowler et al. reported that different doses of vitamin C would not lead to any serious adverse events [7]. No studies have shown that IVVC increase adverse events, except that Chang’s study was discontinued due to the high incidence of severe hypernatremia after interim analysis, which was attributed to the adverse effects of hydrocortisone [27]. Our meta-analysis also demonstrated that vitamin C in sepsis is considered relatively safe.
We acknowledge some limitations. First of all, 10 RCTs [16, 23, 24, 27, 29–31, 34–36] were all single center studies with small sample size, which may result in selective bias or introduce small sample effect, so as to obtain a large beneficial therapeutic effect conclusion. Second, RCTs in this meta-analysis used different doses of vitamin C. Most RCTs administered 1.5 g vitamin C every 6 hours or 25 mg/kg every 6 hours, whereas other RCTs administered 50 mg/kg every 6 hours or 200 mg/kg daily. We consider that the difference dose of vitamin C was likely to affect the efficacy of the meta-analysis as our subgroup and meta-regressions analysis indicated that dose was associated with mortality. Third, the duration of vitamin C may have obvious impact on the outcome. Many patients in the intervention group did not administrate the entire duration of IVVC [16, 32], which may reduce benefit of the treatment group to some extent. Moreover, some trials [22, 27, 36] early terminated their RCTs for reasons, which may also be one of the reasons for different outcomes.