Effects of different ascorbic acid doses on the mortality of critically ill patients: a meta-analysis

Background Low levels of ascorbic acid (AA) have been detected in critically ill patients in which AA supplementation leads to promising outcomes. However, the ability of AA to reduce mortality in critically ill patients remains controversial. In this study, we have performed a meta-analysis to evaluate the effects of AA dose on the mortality of critically ill adults. Methods Electronic databases were searched for trials in which AA had been intravenously administered to critically ill patients regardless of the dose or the co-administration of antioxidant agents. The predefined primary outcome included all-cause mortality at final follow-up. Results The included trials enrolled a total of 1210 patients. Intravenous (IV) AA doses of 3–10 g/d reduced the mortality of critically ill patients (OR 0.25; 95% CI (0.14–0.46); p < 0.001; I2 = 0.0%), while low (< 3 g/d) and high AA doses (≥ 10 g/d) had no effect (OR 1.44; 95% CI (0.79–2.61); p = 0.234; I2 = 0.0% versus OR 1.12; 95% CI (0.62–2.03); p = 0.700; I2 = 0.0%). AA was associated with a decreased duration of vasopressor support and mechanical ventilation, but did not influence fluid requirement or urine output during the first 24 h of admission. The number of patients suffering from acute kidney injury and the length of intensive care unit or hospital stays were also unaffected by the AA. Conclusion Intravenous AA reduces the duration of vasopressor support and mechanical ventilation; 3–10 g AA results in lower overall mortality rates. Given the limitations of the primary literature, further studies are required to fully clarify the effectiveness of AA during the management of critically ill patients. Electronic supplementary material The online version of this article (10.1186/s13613-019-0532-9) contains supplementary material, which is available to authorized users.

Despite promising preliminary results, the benefits of AA remain controversial. Marik and coworkers [20] stated that AA as part of a "cocktail" therapy can reduce mortality in critically ill patients (8.5% vs. 40.4%, p < 0.01), while Lin et al. [19] reported no significant effects of AA infusion (26% vs. 23%, p = 0.8). Notably, the dose of AA varied between these studies, which may account for the discrepancies. Furthermore, the small sample size and single-center nature of the studies questions their reproducibility.
In this review, we provide a comprehensive meta-analysis (MA) of all studies in which AA has been intravenously administered to critically ill patients. We aimed to identify whether the dose of AA impacts mortality and other clinical parameters in this setting, including resuscitation fluid requirement, urine output, acute kidney injury (AKI), vasopressor requirement, the duration of mechanical ventilation, and the length of ICU and/or hospital stay.

Materials and methods
This study was performed and prepared according to the guidelines proposed by Cochrane Collaboration in the Cochrane Handbook for Systematic Reviews of Interventions (http://www.cochr ane handbook.org) and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PROSMA) statement [25,26].

Search strategy
We searched articles of all languages published from inception to November 2018 in PubMed, Embase, Ovid and the Cochrane Central Register of Controlled Trials using the following keywords along with MeSH terms: "ascorbic acid" and "sepsis" or "critical illness" or "Intensive Care Unit" or "burn". We collected all studies in which AA was intravenously administered to adult patients with critical illness.

Study selection criteria
The following trails were included 1. Performed on adults with critical illness. 2. Intravenous AA regardless of the dose vs. placebo or no-intervention. 3. Primary outcome was mortality at the final followup.
The following trails were excluded 1. Performed on children.

Data extraction
Data were independently extracted by the first and third authors. Extracted data consisted of the first author name, year of publication, type of the study, study population, number of patients, AA dose, antioxidant agent, treatment initiation, treatment duration, mortality at follow-up, and other clinical parameters. We resolved disagreements through discussions until a consensus was reached.

Outcome measurements and definitions
The primary outcome was all-cause mortality at final follow-up. Secondary outcomes included resuscitation fluid requirement, urine output, patients suffering from AKI, vasopressor requirement, duration of mechanical ventilation, and length of ICU and/or hospital stay.

Assessment of risk of bias
We used the Cochrane Collaboration tool to assess the risk of bias in randomized controlled trials (RCTs) [26,27]. Domains containing random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of the outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other bias were assessed. The remaining observational trials were assessed using the ROBINS-I tool [28]. Domains include bias due to confounding, bias in selection of participants into the study, bias in classification of interventions, bias due to deviation from intend intervention, bias due to missing data, bias in measurement of outcomes, and bias in selection of the reported results. We rated each domain of the trials as low risk, unclear, or high risk. Trials were considered low risk when each independent domain was rated as low risk. Any domain rated as unclear or high risk increased the overall risk score.

Statistical analysis
Data were analyzed using Statistics/Data Analysis 15.1.
The results of dichotomous data were presented as forest plots through the odds ratios (ORs) with 95% confidence intervals (CIs). Forest plots using SMD with 95% CI were performed for the assessment of continuous data. We quantified heterogeneity via the I 2 statistic. Data were pooled through random (M-H heterogeneity) models if the value of I 2 was greater than 50%, regarded as heterogeneity [29]. A p value ≤ 0.05 was considered statistically significant, except when otherwise specified.

Study characteristics
Of the 12 included trials, eight were RCTs and four were retrospective studies. All studies were published from 1997 to 2018. The characteristics of each trial are shown in Table 1. The studies recruited patients with severe sepsis or septic shock [15,16,20,24], burn shock [17][18][19], critical injury [21,22], post-operation [14,16,36], trauma [36] and those in need of contrast-enhanced CT in the ICU [23]. The sample sizes ranged from 20 to 595. The included studies enrolled a total of 1210 patients of which 624 were administrated IV AA, and 586 were control subjects. The dose of AA ranged from 450 mg/d to 66 mg/kg/h.

Risk of bias and quality of evidence
The risk of bias is summarized in Fig. 2. No trials were considered low risk, and no specific details were used for assessment of the blinding outcomes. The trial performed by Tanaka and colleagues [18] despite being classed as a randomized study was deemed a high risk of bias as participants were allocated to groups according to months. Studies by Ramzkon et al. [21] and Galley et al. [24] lacked specific allocations and were classed as an unclear risk. In total, 11 trials were defined as an unclear risk and a single trial was deemed high risk.  Fig. S1). As the dose of AA varied between the trials, subgroup analysis was performed. Doses < 3 g/d were defined as low, ≥ 10 g/d as high, and 3-10 g/d as medium. The medium AA dose was found to reduce the mortality of critically ill patients (OR 0.25; 95% CI (0.14-0.46); p < 0.001; I 2 = 0.0%, Fig. 3 Fig. 4). Sensitivity analysis was performed through the removal of each single trial and the reanalysis of the remaining trials according to sepsis subgroups. Upon excluding the Galley et al. study [24], the analysis was homogeneous and AA decreased the mortality of patients with sepsis (OR 0.16; 95% CI (0.07-0.37); p < 0.001; I 2 = 0.0%, Additional file 2: Fig. S2).

Length of ICU and hospital stay
Three trials [16,22,23] compared the length of ICU stays between AA and control groups. AA was found to have no influence on the length of ICU residence (SMD = 0. Two trials [18,22] were included to assess the impact of the length of hospital stay on which AA had no influence (SMD = − 0.35; 95% CI (− 0.73-0.04); p = 0.080; I 2 = 0.0%, Fig. 6). As the data from these trials were homogeneous, a fixed Mantel-Haenszel model was employed.

Duration of mechanical ventilation
From the analysis among trials [14,16,18], a lower duration of mechanical ventilation (SMD = − 0.59; 95% CI   Fig. 3 Forest plot of the effect of IV AA on mortality at the final follow-up when compared by the AA dose (− 1.02 to − 0.16); p = 0.008; I 2 = 0.0%, Fig. 6) was proposed. A fixed Mantel-Haenszel model was employed for this analysis.

Publication bias
A Begg test was performed to assess the publication bias of the 11 included studies (p = 0.640). The analysis suggested that minimal publication bias occurred.

Main Findings IV AA administration and ICU/hospital mortality
Compared to other MAs [49,50,52], the major finding of this study was that IV medium doses (3-10 g/d) of AA were associated with decreased mortality, with neither low doses (< 3 g/d) nor high doses (≥ 10 g/d) having a significant impact. The MA conducted by Langlois and coworkers [52] included enteral and IV supplementation and suggested no association of AA with reduced mortality; subgroup analysis revealed that oral-enteral or parenteral, low or high administration of doses of AA (1.5 g/d as a boundary) did not significantly influence mortality.
Both dosing and bio-distribution data in humans suggest that pharmacological concentrations of AA are only attainable through IV administration due to the saturation of intestinal transporters (sodium-vitamin C transporter-1) [51]. Previous studies have shown that critically ill patients often have low AA plasma levels [4][5][6][7]. Furthermore, during the post-injury period, 2 days of 3000 mg/day AA significantly increase plasma AA concentrations [4]. Other studies [48] have shown that 10 g/d AA is associated with supranormal plasma concentrations. Considerable adverse effects have not been reported at high AA doses (66 mg/kg/h) [17][18][19]. In this study, we defined a dose lower than 3 g/d (not inclusive of 3 g/d) as low, higher than 10 g/d (inclusive of 10 g/d) as high, and 3-10 g/d as medium. Our analysis revealed that low-dose AA had little effect on mortality partly due to patients receiving IV AA ≤ 2 g/d (Table 1). Studies have shown that IV AA 2 g/d leads to only normal plasma concentrations [48]. In addition, AA is depleted by free iron, free radical scavengers in the plasma, and the destruction of oxidized AA and dehydroascorbic acid [43]. Thus, low AA doses do not influence mortality. For medium doses, improved patient prognosis was observed. Low   levels of plasma AA in septic patients inversely correlated with the incidence of multiple organ failure [7], while medium doses restored the AA concentrations to normal plasma levels [48]. Additionally, studies performed by Straaten and coworkers [68] suggested that highdose AA (3-6 g/d) decreases the formation of superoxide and peroxynitrite, bidirectly scavenges superoxides, augments antibacterial defenses, and protects against oxidative stress in critically ill patients. AA can counteract lipid peroxidative damage through the scavenging of oxygen-derived free radicals and the restoration of vascular function [53]. Counteracting oxidative stress represents a likely mechanism by which moderate AA reduces mortality. For high doses of AA, no loss of mortality was observed. Among the three trials [15][16][17][18][19]21] participants suffering from burn injuries were recruited and the majority of deaths did not occur during resuscitation, but from subsequent infections [54]. In addition, patients in group B of the study by Razmkon and coworkers [21] received 10 g on the first day of admission which was repeated on day 4, followed by 4 g/d for the remaining 3 days. The final follow-up of mortality in these trials differed, partly accounting for the outcomes.
Through subgroup analysis of the patient characteristics, AA had no effect on sepsis or burns. For sepsis, heterogeneity was observed, mainly due to the Galley et al. [24] study. The Galley trial was performed in 1997, and the study protocols drastically differed from those of the more recent studies. When the Galley et al. trial was excluded, AA had a positive effect on mortality. For burn patients, AA had little influence. Further studies are warranted to dissect this relationship given that burn deaths mainly occur after resuscitation [54].

Fluid requirement, urine output, and patients suffering from AKI
The results demonstrate that AA has little effect on fluid requirements or urine output during the initial 24 h, or  [16], although a significant increase in urine output during the first 24 h did not occur, a tendency toward increasing urine output over time was noted, which may have influenced the overall significance of the analysis. AA is hyperosmolar and is a risk factor for osmotic diuresis [17]. Despite the trial of Tanaka et al. [18], AA did not cause osmotic diuresis and no differences in the patients' urine and serum osmolality were observed. In the trial of Kahn and colleagues [17], several patients showed signs of hypovolemia in the absence of decreased urine output, which was noted as a possible sign of osmotic diuresis. Other trials lacked data on urine output and osmotic diuresis and future studies should consider these parameters. Previous studies suggested that following a severe burn injury, potent oxygen-free radicals (OFRs) are produced from the ischemia and reperfusion of burnt skin [60]. Animal studies reveal that antioxidant therapy through the administration of high-dose AA, an ORF scavenger [61], reduces post-burn lipid peroxidation [9], decreases vascular permeability [57], decreases burn and no-burn tissue edema [58], and reduces the requirement for resuscitation fluid [10,18,59]. However, this was not observed in this study. Possible reasons for these discrepancies include the varied initiation of treatment between the studies. In the study by Kahn et al. [17], treatment was initiated at 52 ± 26 min post-admission, while in study by Lin et al. [19] treatment was initiated at 4.01 ± 15 h. Zabet and coworkers [16] did not provide a specific treatment time. All the included studies were single-center, and the fluid levels received by the patients were variable. In the study by Lin and colleagues [19], although a reduction in fluid requirement was noted, both groups had higher weights and had a higher % of total body surface area (TBSA) of burn injuries than would be estimated based on the Parkland formula. The trials lacked data on fluid requirement for longer periods and future studies should consider these parameters.
It has been reported that AA promotes acute renal failure [62,63]. In patients with renal impairment who received high AA doses, increased levels of oxalate (typically excreted by the kidney) were observed in conjunction with crystallization, leading to impaired kidney function [64,65]. In this study, effects on patients suffering from AKI were not observed partly due to the AA dose. In studies by Marik and Nathens [20,36], AA was administrated at doses of 6 g/d and 3 g/d, respectively, which was considered medium doses, while Lin and colleagues [19] assessed the effects of 66 mg/kg/h AA. These trials included lacked data on time span during which AKI was measured. This may account for the overall outcome observed.

Duration of vasopressor requirement
From our studies, AA significantly reduced the duration of vasopressor requirement. AA is an essential cofactor for the copper-containing enzyme dopamine β-hydroxylase during catecholamine (dopamine, norepinephrine, and epinephrine) synthesis [55]. It was found that IV AA improved cardiovascular function and decreased the requirement for catecholamine in a patient with septic shock [56]. Daniel and colleagues [69] stated that AA was a cofactor during collagen synthesis that was required to support cardiovascular functions. Additionally, Carr et al. [13] found that AA increases vasopressor sensitivity.

Duration of mechanical ventilation
A decreased duration of mechanical ventilation upon AA administration was observed, but whether this reflected improved pulmonary function was uncertain. Tanaka et al. [18] commented that fewer days of mechanical ventilation and improved early respiratory function were associated with fluid reduction. The opposite scenario of fluid requirement was observed in this study. Animal models of sepsis suggest that IV AA attenuates proinflammatory and procoagulant states, reducing lung vascular injury [11] and oxidative stress, induced histopathological alterations, thus improving pulmonary function [66]. Only the Tanaka et al. [18] study provided pulmonary function measurements. Future studies should consider these parameters.

Comparison with other studies
The major strength of this MA was the investigation of how different AA doses contribute to the clinical outcomes of patients with critical illness. We compared the effects of AA alone or in combination with other agents for its effects on mortality and compared the effects of AA on different patient characteristics (sepsis, burn, or others). Such studies have not been previously performed. Furthermore, a varied mix of medical, surgical, and burn injury patients was included, and each trial equally contributed to the final study outcomes.

Limitations
This MA had several weaknesses that should be noted. Firstly, 12 trials were included of which 8 were RCTs and