In this trial of IV vitamin C therapy for septic shock, the observed reduction in absolute risk of 28-day mortality in the experimental arm of 13.9% (p = 0.10) did not meet the primary endpoint of reducing absolute 28-day mortality by 20% as hypothesized. Analysis of our secondary outcomes showed a modest increase in fluid administered within 6 hours of study drug initiation in the vitamin C arm that reached statistical significance but is of unclear clinical significance and statistically higher incidence of RRT use in the vitamin C arm. Post hoc analysis examining the relative times of study drug initiation and RRT initiation indicates that the majority of patients requiring RRT in the vitamin C group had it initiated or planned prior to study drug initiation. This suggests that vitamin C was noncausal; however, we cannot assess the extent to which this difference between the groups may have influenced other outcomes.
We did not note any statistically significant differences in ICU mortality, improvements in SOFA or APACHE II scores between the groups, time to resolution of shock, time to lactate clearance, or duration of ICU or hospital stay; however, attrition bias due to uneven mortality between the groups may have reduced any potential differences in the latter four outcomes. Also, because SOFA and APACHE II scores are routinely only determined once per day, this metric may not fully appreciate precipitous deterioration of status leading to death.
We attempted to determine whether there may be some synergistic effect between vitamin C and steroids by performing a subgroup analysis comparing mortality between arms in subgroups of patients receiving or not receiving steroids. In the subgroup of patients receiving steroids, the observed reductions in absolute 28-day and ICU mortality were greater than those in the full study cohort (14.3% and 11.4%, respectively); however, statistical significance was not reached. Additionally, the use of steroids was not protocolized but rather at the discretion of the treating provider, which may introduce selection bias.
Several trials of vitamin C-based therapies for sepsis and septic shock have recently been completed. Among trials of HAT therapy or vitamin C and thiamine, the maximum reduction in 28- or 30-day absolute mortality observed has been 7.5%, and none have shown a statistically significant reduction in mortality.[6,8–12] Individual trials of HAT therapy have suggested a reduction in 28-day mortality when the therapy is initiated within 48 hours of onset of sepsis, and quicker resolution of shock associated with HAT therapy;[4,7,10] however, these findings have not been uniformly reproduced.
The CITRIS-ALI trial of vitamin C monotherapy for patients with sepsis and acute lung injury demonstrated a reduction in 28-day absolute mortality of 16.6% (p = 0.03); however, this was one of multiple secondary outcomes that obfuscates its statistical significance, and the primary outcomes of reductions in SOFA score and plasma inflammatory markers were not met. Subgroup analysis of patients requiring positive-pressure ventilation in our study similarly suggested a heightened mortality benefit for this group relative to the general study population, but potential conclusions from this analysis are tempered by its post hoc nature. Additional trials of vitamin C for patients with septic shock and sepsis-induced respiratory dysfunction employing short-term mortality as a primary outcome could further inform whether vitamin C carries an enhanced therapeutic effect for this population.
Our trial had several limitations. Although it benefitted from its multicenter design, all included centers were in a single geographical area, and our enrollment demographics reflected this with significant underrepresentation of non-White subjects. This limits the generalizability of our findings. Second, the size of our study population was chosen based on a large survival benefit observed in a retrospective study that has not been reproduced in prospective trials. Our study is, therefore, underpowered to detect potentially smaller differences between the groups. Third, we did not measure serum vitamin C levels, so we are unable to assess the degree of preexisting hypovitaminosis in study subjects nor the degree to which vitamin C levels were increased by supplementation. However, prior studies have suggested that even patients with normal vitamin C levels at admission become deficient with onset of septic shock. Fourth, although we enrolled subjects within 24 hours of vasopressor initiation, we were frequently unable to assess the amount of time a subject may have been in untreated shock prior to vasopressor initiation, leading to immeasurable variations in the time between shock onset and study drug initiation. Fifth, there was a statistically significant difference in RRT use between the arms of our study. Although RRT was underway for many of these subjects prior to study drug initiation, it is unclear the extent to which this may have influenced other outcomes.
Crit Care Med. 2022;50(5):e458-e467. © 2022 Lippincott Williams & Wilkins