A total of 271 patients met inclusion criteria. Of these, 104 were excluded leaving 167 who were approached for consent. One hundred twenty-six subjects agreed to participate and were enrolled. One subject withdrew from the study following enrollment but prior to randomization, and one subject withdrew following randomization but prior to initiation of study drug. This resulted in 124 subjects receiving study drug: 60 in the vitamin C cohort and 64 in the placebo cohort. Of these subjects, 112 completed the protocol and 12 were withdrawn prematurely from the drug portion of the study (11 upon transition to comfort measures only and one on transfer to a nonparticipating hospital). All of these 12 subjects received at least one dose of study drug and were included in the final analysis (Figure 1).
Baseline characteristics between the two groups including demographics, baseline laboratory values, initial disease acuity, time to antibiotics, preenrollment fluid administration, ventilatory needs, and past medical history were similar (Table 1; and Table S2, https://links.lww.com/CCM/G946). Pneumonias and infections of the gastrointestinal/biliary system and urinary system were the predominant sources of sepsis in both groups with similar distribution of infectious sources between the groups. Although the source of septic shock was not able to be identified in all cases, the rates at which this occurred (13.3% of vitamin C group subjects and 10.9% of placebo group) were similar to those observed in prior sepsis epidemiological studies.[18,19] The time from vasopressor administration to study drug initiation was similar in the vitamin C and placebo groups with median times of 11.4 hours (interquartile range [IQR], 5.4–17.3 hr) versus 8.8 hours (IQR, 5.4–17.4 hr), respectively (p = 0.21). Thirty patients (50%) received steroids in the vitamin C group, whereas 42 (65.6%) received steroids in the placebo group (p = 0.08).
Primary and Secondary Outcomes
Sixteen of 60 subjects (26.7%) in the vitamin C arm had died at 28 days versus 26 of 64 subjects (40.6%) in the placebo arm (p = 0.10) (Table 2). Kaplan-Meier survival analysis showed a similar result with a log-rank score of 0.10 (Figure 2). There was also no statistically significant difference between the two arms in ICU mortality (23.3% vs 31.1%; p = 0.32).
Comparison of survival in the 28-d study follow-up period: Kaplan-Meier survival analysis comparing survival over 28 d following study drug initiation for patients in the vitamin C and placebo groups. Log-rank testing p value is 0.10.
Improvement in Sequential Organ Failure Assessment (SOFA) and Acute Physiology and Chronic Health Evaluation (APACHE) II scores over the 96-hour study period was similar between arms, as were time to lactate clearance, vasopressor duration, and mechanical ventilation duration (Table 2). Subjects in the vitamin C arm received more IV fluids within the first 6 hours after study drug initiation (1.07 L [IQR, 0.72–1.64 L] vs 0.76 L [IQR, 0.36–1.26 L]; p = 0.03); however, the clinical significance of this 310-mL difference is unclear, and by 24 hours, there were no statistically significant differences in total volume administered (3.59 L [IQR, 2.52–5.02 L] vs 3.37 L [IQR, 1.98–4.70 L]; p = 0.47) nor net total fluid balance (2.12 L [IQR, 1.10–3.59 L] vs 2.05 L [IQR, 1.24–3.11 L]; p = 0.93). ICU length of stay (2.9 d [IQR, 1.8–7.5 d] vs 2.6 d [IQR, 1.5–5.3 d]; p = 0.47) and hospital length of stay (8.9 d [IQR, 4.0–20.0 d] vs 6.3 d [IQR, 3.8–12.5 d]; p = 0.15) following study drug initiation were similar between the two arms. Although there was no statistically significant difference between the arms in serum creatinine changes over the study drug period, there was an increased need for RRT in the vitamin C arm compared with the placebo arm (16.7% vs 3.3%; p = 0.02).
To further characterize the increased use of RRT in the vitamin C arm, we performed a post hoc analysis comparing the time of study drug initiation with the time that RRT was planned or initiated (Table S3, https://links.lww.com/CCM/G946). In the vitamin C group, six of the 10 patients receiving RRT already had it established or immediately planned prior to initiation of study drug. In the placebo group, one of the two patients did. Only four patients required unplanned initiation of RRT post study drug administration in the vitamin C group compared with one patient in the placebo group (p = 0.17). Additionally, although there were no immediate plans for RRT, all five of these patients had oliguric (less than 500-mL urine output over 24 hr) acute kidney injury prior to study initiation.
Subgroup Analysis by Corticosteroid Administration
Corticosteroids were given to 72 subjects as part of usual clinical care. Overall, there were no statistically significant differences in 28-day or ICU mortality between the vitamin C and placebo arms in either the steroid or nonsteroid subgroups (Tables S4 and S5, https://links.lww.com/CCM/G946). There was an increased incidence in requirement of RRT in the vitamin C arm relative to that of the placebo arm in both the steroid (23.3% vs 4.8%; p = 0.02) and nonsteroid (10.0% vs 0%; p = 0.27) subgroups. These findings reflect the increased incidence of RRT requirement in the vitamin C arm overall. Otherwise, there were no significant differences between the vitamin C and placebo arms in acuity score improvement, duration of pressors, duration of mechanical ventilation, or lactate clearance within each subgroup (Tables S4 and S5, https://links.lww.com/CCM/G946). In the steroid-receiving subgroup, there was a trend toward increased hospital length of stay in the vitamin C arm compared with the placebo arm (10.8 d [IQR, 4.2–20.5 d] vs 6.3 d [IQR, 2.6–13.0 d]; p = 0.06), which was not seen in the nonsteroid subgroup (7.9 d [IQR, 3.8–19.7] vs 6.7 d [IQR, 4.7–12.1]; p = 0.58). ICU length of stay did not significantly differ between the arms in either subgroup.
Subgroup Analyses by Respiratory Status and Severity Index Score at Enrollment
In light of recent findings of the Vitamin C Infusion for Treatment in Sepsis Induced Acute Lung Injury (CITRIS-ALI) trial, we wished to further investigate the effects of vitamin C on subjects with respiratory failure. Although we did not collect data for all criteria necessary to determine an ARDS diagnosis, we were able to stratify our study population by the presence of positive-pressure ventilation at enrollment (either mechanical ventilation or non-invasive support) and by the presence of hypoxemic respiratory failure at enrollment, defined as positive-pressure ventilation and a PaO 2/FIO 2 ratio of 300 or less. A post hoc subgroup analysis of 28-day and ICU mortality showed a statistically significant reduction in 28-day mortality in patients with positive-pressure ventilation at baseline (36.3% vs 60.0%; p = 0.05); however, the reduction observed in patients with hypoxemic respiratory failure did not achieve statistical significance (33.3% vs 56.0%; p = 0.11), nor did reductions in ICU mortality (Table 3). Given the post hoc nature of this analysis, these results should be interpreted as hypothesis generating only.
To investigate whether respiratory status at enrollment is simply a surrogate for disease severity in the above analysis, we dichotomized the total population by enrollment SOFA and APACHE II scores and compared these populations with those obtained by respiratory status stratification using Spearman rank correlation testing. We found low-to-moderate correlation in all comparisons (Table S6, https://links.lww.com/CCM/G946). To further explore the degree to which severity of illness at enrollment could influence the efficacy of vitamin C therapy, we performed post hoc subgroup analyses of 28-day and ICU mortalities using the stratifications by enrollment severity index scores described above (Tables S7 and S8, https://links.lww.com/CCM/G946). Although greater reductions in absolute mortality were seen in the groups with higher acuity, the differences were not statistically significant.
Overall, 27 adverse events were reported, 15 in the vitamin C arm and 12 in the placebo arm. Distribution of these events by system and event type shows no significant differences between the vitamin C and placebo arms (Table S9, https://links.lww.com/CCM/G946). Most events were attributed to sequalae of septic shock. Only three events were reported as being possibly related to study drug: one report of nausea (mild severity, vitamin C arm), one report of bradycardia (mild severity, placebo arm), and one report of loose stools (moderate severity, placebo arm).
Crit Care Med. 2022;50(5):e458-e467. © 2022 Lippincott Williams & Wilkins