IV Vitamin C in Sepsis: A Latest Systematic Review and Meta-Analysis

Sepsis is a high-incidence disease and demands intensive care. Finding effective treatment is the key to cure sepsis. Studies have shown a lower level of vitamin C in patients with sepsis. Therefore, vitamin C supplementation has become one of the measures to treat sepsis. However, the clinical studies of vitamin C in the treatment of sepsis have been controversial. We performed a meta-analysis to evaluate vitamin C's efficacy and safety in the treatment of sepsis. We searched four electronic databases: PubMed, Embase, Web of Science, and the Cochrane Library, and two researchers independently screened 24 eligible RCTs published in English. Our review demonstrates that intravenous (IV) vitamin C might improve short-term mortality (RR, 0.82; 95% CI, 0.65–1.02; P=0.07; and I2 = 45%) and overall mortality (RR, 0.86; 95% CI, 0.74–1.01; P=0.06; and I2 = 51%) of patients with sepsis. Moreover, the SOFA score of patients with sepsis improved significantly after treatment with vitamin C for over 72 hours (RR, 0.26; 95% CI, 0.09–0.42; P=0.002; and I2 = 0%). The main results of our study were moderate-quality evidence. More high-quality, multicenter RCTs are needed to provide more substantial evidence on the efficacy and safety of IV vitamin C for sepsis.


Introduction
Sepsis is a life-threatening condition caused by a dysregulated host response to infection and could lead to organ dysfunction [1]. Sepsis is a common disease in the intensive care unit, which is the leading cause of death from infection and major healthcare problems. Te occurrence of sepsis was close to 50 million cases, while about 20% of all-cause deaths worldwide were sepsis-related [2]. A meta-analysis estimated the hospital mortality for sepsis to be close to 27% [3]. Te mortality for sepsis can be as high as 46.7%, which is much higher than that in other critically ill patients [2]. Despite diferences in medical expenditure between countries and regions [4], sepsis results in high healthcare costs due to its high incidence and the need for high-level medical care for critically ill individuals. Te total medical expenditures related to sepsis are the highest in the United States (€51,671 × 10 6 ) [5]. Early diagnosis and efective treatment of sepsis can reduce patients with severe sepsis, thereby reducing medical costs [6].
Vitamin C, also called ascorbic acid, can impact the expression of coagulation and proinfammatory genes. It can also regulate the immune system, maintain circulating cytokine homeostasis, and exhibit anti-infammatory and antioxidant properties [7]. Te level of vitamin C is often lower in patients with sepsis [8,9]; therefore, vitamin C supplementation has become one of the measures to treat sepsis. Clinical studies of vitamin C in the treatment of sepsis have shown mixed results on the mortality of patients. Some studies suggest that vitamin C could signifcantly reduce mortality in patients with sepsis [10][11][12], while others did not show any efect [13,14]. Te use of vitamin C in the treatment of sepsis is controversial. Tere have been many high-quality randomized controlled studies on the treatment of sepsis with vitamin C in the last two years [13,[15][16][17]. We attempt to evaluate the efcacy and safety of vitamin C supplementation in patients with sepsis through a systematic review and meta-analysis of randomized controlled trials (RCTs).

Methods
Tis systematic review and meta-analysis were conducted based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), 2020 [18].
Te research protocol has been registered on the international prospective register of systematic reviews (PROSPERO), with CRD42022350810 as the registration number.

Data Sources and Search Strategy.
Four electronic databases, including PubMed, Embase, Web of Science, and the Cochrane Library, were searched independently by two researchers (Chengli Wen and Qinxue Hu) to explore eligible literature published in English from the establishment of the databases until August 2022. Te databases were searched by medical subject headings and keywords, which included "vitamin C," "ascorbic acid," "sepsis," and "septic shock."

Eligibility Criteria.
Inclusion criteria for the study were as follows: (1) study design: RCT; (2) participants: adult patients with sepsis or septic shock; (3) intervention: vitamin C administered intravenously (IV), given alone or in combination with thiamine and/or corticosteroids; and (4) result: mortality or length of hospital/ICU stay. Exclusion criteria were as follows: (1) study designs other than RCTs; (2) studies involving animals or minors; (3) non-IV mode of vitamin C administration; and (4) outcomes without mortality or length of hospital/ICU stay.

Data Extraction.
Te literature was screened and extracted based on the eligibility criteria independently by two researchers (Chengli Wen and Qinxue Hu). In case of disagreements, the data were discussed with a third researcher (Hui Liu). Te information obtained from the eligible studies included the frst author of the studies, the publication year, the country, study design, sample size, intervention details, mortality (28-day, 30-day, 90-day, 180day, ICU, hospital, and all-cause mortalities), duration of vasopressors, ∆Sequential Organ Failure Assessment (SOFA) (72 h), new onset of acute kidney injury (AKI), duration of mechanical ventilation, adverse events, length of ICU, and hospital stay.

Study Quality Assessment.
Te quality of the studies was independently assessed by two researchers (Chengli Wen and Qinxue Hu) based on the Cochrane Collaboration's tool and modifed Jadad scale. Te disagreements between the two authors were resolved in consultation with the third researcher (Hui Liu). Cochrane is a tool for measuring risk assessment that assigns a high-risk, low-risk, or unclear-risk assessment on key points, including the generation of random sequence, the allocation concealment method, participants' blinding, data collectors, healthcare professionals, outcome assessors, and data analysts, and incomplete outcome data [19]. Te modifed Jadad scale score was determined from four aspects: (a) randomization (0: not randomized or the randomization method was inappropriate, 1: the study was described as randomized, and 2: the study used a suitable randomization method), (b) allocation concealment (0: the study did not explain the method of concealed allocation, 1: the study illustrated the allocation concealment method, and 2: the technique of concealed allocation was explained adequately), (c) double blinding (0: no blinding or improper blinding technique, 1: the study employed double-blind technique, and 2: the appropriate double blinding technique was described), and (d) withdrawals and dropouts (0: the study did not describe the follow-ups and 1: the study described the withdrawals and dropouts). Te modifed Jadad scale scores of 1-3 and 4-7 were considered low-quality and high-quality studies, respectively [20].

Outcome Measures.
Te primary outcome was the overall mortality, defned as mortality at the end of the follow-up for each study. Diferent studies followed up for diferent periods; therefore, the overall mortality could be one of the 28-day, 30-day, 90-day, 180-day, hospital, ICU, and all-cause mortalities in literature. Te secondary outcomes included 28-day, 30-day, 90-day, 180-day, ICU, and hospital mortalities, as well as new onset of AKI, adverse events, duration of vasopressors, ∆SOFA (72 h), and duration of mechanical ventilation.

Efect
Measures. Dichotomous results were reported using relative hazard ratios (RR), while continuous results were reported using standardized uniform diferences (SMD), as the units of the results were diferent.

Synthesis Methods for Data.
For data processing, the method reported by Wan et al. [21]and Luo et al. [22] was used; the median was converted into the mean depending on the study's sample size. When the sample size was greater than 50, the method provided by Luo et al. was chosen; otherwise, the method by Wan et al. was employed. Similar interventions were compared using the random efect metaanalysis. RevMan 5.3 (Cochrane IMS, Oxford, United Kingdom) was used for all analyses with a randomefect model. Stata 14.0 MP-Parallel Edition (College Station, TX, USA) was used to do the funnel plot, Egger's test, and meta-regression. Heterogeneity was measured using the chisquared test and I 2 statistics. Te value I 2 greater than 50% and P less than 0.05 indicated statistical heterogeneity. When heterogeneity was evident and the number of literature was more than 20, meta-regression, subgroup, and sensitivity analyses were used to identify the sources of the heterogeneity. Furthermore, Egger's test and the funnel plot were used to detect publication bias when the number of literature was greater than or equal to 10. All estimates were reported with a 95% confdence interval (CI).

Subgroup Analyses.
Subgroup analyses were performed to fnd the sources of heterogeneity in the clinical outcomes. Te subgroups included (1) protocol: vitamin C monotherapy compared to the vitamin C combination treatment group (vitamin C combined with thiamine and/or hydrocortisone); (2) the quality of the studies: low-quality studies compared with high-quality studies; (3) blind method: according to the blind method, the studies were divided into four groups: double-blind, single-blind, open-label, and unclear; and (4) publication year: according to the year of publication, the studies were divided into four groups: publication in and before 2019, in 2020, in 2021, and in 2022.

Study Identifcation and Selection.
Initially, 969 pieces of literature were obtained from the aforementioned electronic databases, of which 211 were from PubMed, 263 from Embase, 329 from Web of Science, and 166 from Cochrane Central. In order to collect all relevant literature, one study was manually added after reading the relevant systematic review and meta-analysis. Eventually, a total of 970 studies were included in the screening. Of them, 489 duplicate records were removed, and 425 studies were omitted after reading the title and abstract, as they did not follow the inclusion criteria. Subsequently, we removed other 32 studies, including two with no full-text or abstract, fve with no outcome measures, two systematic reviews, six reviews, six study protocols, and eleven retrospective studies. Finally, 24 eligible pieces of literature were included in the metaanalysis, with a total of 3,759 patients enrolled in the study [8,[10][11][12][13][14][15][16][17][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37]. Te selection fow diagram is presented in Figure 1.

Results of Study Quality Assessment.
Assessment of the quality of the study based on the Cochrane Collaboration's tool is shown in Figure 2, and the scores of the modifed Jada scale are presented in Table 1. According to the Jada scores, most studies were of high quality [8, 10, 11, 13-17, 24, 26, 28-30, 32, 33, 35] and only eight were of low quality [12,23,25,27,31,34,36,37].

30-Day
Mortality. An analysis of fve studies reporting 30-day mortality [14,15,27,28,33] showed that vitamin C did not improve the 30-day mortality of the patients (RR, 1.02; 95% CI, 0.85-1.24; P � 0.80; I 2 � 0%, Figure 6). Furthermore, there was no heterogeneity among the included studies, and the result was credible.  Figure 4: Te funnel plot for publication bias, comparing the overall mortality in the vitamin C and control groups. A blue dot represents a single study, and the funnel dotted line represents 95% confdence intervals.  Figure 7). Tis result was credible, as there was no heterogeneity among the included studies.

Adverse Events.
As we should also pay attention to the safety and efectiveness of the treatment, we searched the literature for evidence of adverse events. Te adverse events included hyperglycemia, hypernatremia, hospital-acquired infection, fuid over-load and hyperglycemia, hemorrhagic shock, and worsening kidney function. Only 3 literature were included in the study [        adverse events in the vitamin C treatment group was higher than that in the control group (RR, 1.42; 95% CI, 0.94-2.15; P � 0.09; I 2 � 0%, Figure 12); however, the diference was not statistically signifcant.

Duration of Vasopressors.
Twelve studies reported the outcome of the duration of vasopressors, and a total of 1198 patients were included [11, 12, 15-17, 23, 26, 27, 29, 31, 36, 37]. Te analysis of the included studies showed that vitamin C could signifcantly reduce the duration of vasopressors, but the two groups demonstrated signifcant heterogeneity (RR, −0.72; 95% CI, −1.00-−0.44; P < 0.00001; I 2 � 81%, Figure 13). Tere was no signifcant publication bias in the included studies (Supplementary Figure 6), and the Egger test (P � 0.892) also showed the same result. Sensitivity analysis showed that the heterogeneity of the included literature was not signifcantly changed when any one study was removed, and the result remained changed (Supplementary Figure 7).

Duration of Mechanical Ventilation.
A total of seven studies reported this outcome [11,16,17,23,25,30,31]. Te result showed that vitamin C could reduce the duration of mechanical ventilation (RR, −0.28; 95% CI, −0.75-0.19; P � 0.24; I 2 � 88%, Figure 14). Signifcant heterogeneity existed in the included studies, and due to the small number of included studies, we could not detect publication bias. After sensitivity analysis, we found that only when the report from Mahmoodpoor et al. [17] was removed, the heterogeneity of the included studies signifcantly decreased, but the result also altered (RR, 0.03; 95% CI, −0.14-0.20; P�0.74 ; I 2 � 11, Supplementary Figure 8). Terefore, we considered that the result was unstable.

Length of Hospital
Stay. Ten pieces of literature reported the outcome of the length of hospital stay [15,16,23,24,27,29,30,32,33,36]. Our analysis showed that vitamin C did not signifcantly reduce the length of the hospital stay (RR, 0.11; 95% CI, −0.04-0.26; P � 0.15; I 2 � 53%, Figure 17). Moreover, the included studies exhibited signifcant heterogeneity. Te funnel plot (Supplementary Figure 10) and Egger test (P � 0.630) showed no signifcant publication bias in the included studies. Moreover, sensitivity analysis showed that when only the study by Raghu et al. was removed, the heterogeneity changed signifcantly. However, the result remained unchanged (Supplementary Figure 11), indicating the stability of the results.

Discussion
We performed a systematic review and meta-analysis to evaluate the efcacy and safety of IV vitamin C in the treatment of sepsis. Tis meta-analysis has the largest number of included literature and cases to date. Te primary outcome analysis showed that IV vitamin C tended to reduce overall mortality in patients with sepsis compared with the control group. However, the diference between the two groups was not statistically signifcant (Figure 3), which is consistent with the result of Patel et al. [38]. Te heterogeneity among the included studies was signifcant. Results of statistically signifcant studies are more likely to be reported and published than nonsignifcant and invalid results, which may lead to publication bias. Te included studies had signifcant publication bias, as shown by the funnel plot ( Figure 4) and the Egger tests (P � 0.003).  In the sensitivity analysis, the heterogeneity was signifcantly decreased after excluding the report by Raghu and Ramalingam [36] with a signifcant publication bias; the I 2 decreased from 51% to 26%, but the outcome remained unchanged. After removing the three reports [11,36,37] with possible publication bias, the result remained       International Journal of Clinical Practice unchanged, with I 2 decreasing from 51% to 0% (Supplementary Figure 12). Tese results indicate that vitamin C could improve overall mortality in patients with sepsis. We also performed the meta-regression based on the quality of studies, study protocol, blinding method, and publication year, and the P values were 0.013, 0.027, 0.746, and 0.053, respectively. Te meta-regression and subgroup analysis showed that the quality of studies (P � 0.04; I 2 � 77.3%) could be the primary source of the heterogeneity, while the protocol (P � 0.13; I 2 � 56.1%) and the publication year (P � 0.03; I 2 � 66.7%) could also contribute to the heterogeneity. Consistent with Patel et al. [38], our subgroup analysis suggested that compared with the combination treatment group, vitamin C monotherapy showed a decreasing trend in overall mortality. However, it difered from the result of a high-quality randomized controlled study published in 2022 by Lamontagne et al. [13]. Teir results indicated that vitamin C monotherapy did not reduce overall mortality. Terefore, the efcacy of vitamin C monotherapy in reducing overall mortality is uncertain and needs to include a large number of high-quality studies. Furthermore, low-quality studies indicated that vitamin C could improve the overall mortality of patients with sepsis, while high-quality studies did not show the same trend. Low-quality studies were mainly published in 2021. In the subgroup of years of publication, only the studies published in 2021 showed that vitamin C reduced overall mortality in patients with sepsis; the literature published in other time groups did not show the same trend. In the blind method subgroups, the unclear group showed an exciting result, but the scores of the included studies were less than or equal to 3. In other words, all the included studies were of low quality, indicating that low-quality literature can signifcantly afect the bias in outcomes. Te fnding that vitamin C could improve overall mortality in patients with sepsis was stable, but the results of recently published high-quality studies did not show the same trend [13,33]. According to the GRADE system [39], the evidence that vitamin C could improve overall mortality in patients with sepsis was of moderate quality.
Te 28-day and 30-day mortalities were defned as shortterm mortality, while 90-day and 180-day mortalities were defned as long-term mortality. IV vitamin C tended to reduce 28-day mortality in patients with sepsis, while the 30day mortality remained unafected. Vitamin C also showed no efcacy in decreasing 90-day and 180-day mortalities. Terefore, our result suggested that vitamin C did not improve the long-term mortality of sepsis. In addition, we also found that vitamin C improves ICU and hospital mortalities, but the diference was not statistically signifcant    compared with the control group. As sepsis is an acute disease, most patients' ICU and hospital stays are relatively short (generally less than 30 days), consistent with our included studies. Terefore, we can conclude that vitamin C might play a role in improving short-term mortality. Tis result is consistent with the meta-analysis of Li et al. [40].
Te SOFA score is used to evaluate organ function damage in patients with sepsis [41]. It is the sum of respiratory status, liver function, renal function, coagulation function, circulatory status, and nervous system score. Te SOFA score decreases with the improvement in organ functions; therefore, the △SOFA score would have a positive value. We demonstrated that vitamin C treatment for over 72 hours in patients with sepsis could improve their organ damage. Tis is consistent with the results of the metaanalysis published in 2021 and 2022 [40,42]. However, vitamin C did not show a positive trend for reducing the occurrence of AKI. Vitamin C might reduce SOFA score by improving the coagulation function and circulatory status and reducing infammatory response [7].
Our results showed that vitamin C could reduce the length of ICU stay, while the hospital stay did not indicate the same trend. All these diferences, however, were not statistically signifcant. Te included literature was highly heterogeneous regarding the duration of vasopressors and mechanical ventilation; therefore, the results were not credible.
Te incidence of adverse reactions in the vitamin C treatment group was comparable to the control group. Major adverse efects included hyperglycemia, hypernatremia, fuid overload, and hospital-acquired infection. Te reports concerning these adverse events used a combination of vitamin C with thiamine and hydrocortisone [28,32,33]. Te largest number of adverse events was reported by Moskowitz et al. [28], with vitamin C combined with thiamine and hydrocortisone in the intervention group and a placebo in similar volumes (0.9% sodium chloride) in the control group. Hyperglycemia, hypernatremia, and fuid overload are adverse reactions of hydrocortisone [43,44], while no adverse efects have been reported with vitamin C monotherapy. Moreover, the occurrence of hospitalacquired infection in the treatment and control groups was comparable [28]. Tere is no substantial evidence that IV vitamin C increases the incidence of adverse events in patients with sepsis.

Te Strength and Limitations.
Tere are many strengths of this research. Our study was conducted according to our previously developed protocol published in PROSPERO. We conducted a comprehensive literature search to avoid bias as much as possible. We also examined the sources of heterogeneity in the main study results. Moreover, the design types of the included literature stipulated that only RCTs were included, and further inclusion and exclusion criteria were formulated. However, there are still many limitations to our study. First, the inclusion of low-quality literature might have afected the results; the heterogeneity was evident, which may have afected the credibility of the results.
Second, the included studies varied in intervention and the dosages of vitamin C and the regional distribution and ethnic group of the enrolled populations varied, which may increase sources of heterogeneity. Tird, due to diferent follow-up time, the outcome measures are not exactly the same, which may also be the source of heterogeneity. Fourth, due to diferent follow-up time, we could not analyze the efect of IV vitamin C on the time of free vasopressors, mechanical ventilation, and ICU stay. Finally, although some of our outcomes, such as overall mortality, 28-day mortality, and length of ICU stay, showed positive efects, 95% confdence intervals intersected the null line, which may impact interpretation of the results.

Conclusion
Our analysis showed that intravenous vitamin C might improve short-term mortality and overall mortality of sepsis. Moreover, the SOFA score of patients with sepsis improved signifcantly after treatment with vitamin C over 72 hours. Te main results of our study were moderate-quality evidence. More high-quality, multicenter RCTs are needed to provide stronger evidence on the efcacy and safety of IV vitamin C for patients with sepsis.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.
ventilation. After removal of the study of Mahmoodpoor et al. [17], the heterogeneity of the included studies signifcantly decreased, and the result was also changed (E). Supplementary Figure 9: Funnel plot for publication bias, which compared the length of ICU stay in the vitamin C group and control group. Supplementary Figure 10: Funnel plot for publication bias which compared the hospital stay in the vitamin C group and control group. Supplementary