Vitamin D supplementation and the outcomes of critically ill adult patients: a systematic review and meta-analysis of randomized controlled trials

This meta-analysis assessed the association between vitamin D supplementation and the outcomes of critically ill adult patients. A literature search was conducted using the PubMed, Web of Science, EBSCO, Cochrane Library, Ovid MEDLINE, and Embase databases until March 21, 2020. We only included randomized controlled trials (RCTs) comparing the efficacy of vitamin D supplementation with placebo in critically ill adult patients. The primary outcome was their 28-day mortality. Overall, 9 RCTs with 1867 patients were included. In the pooled analysis of the 9 RCTs, no significant difference was observed in 28-day mortality between the vitamin D supplementation and placebo groups (20.4% vs 21.7%, OR, 0.73; 95% CI, 0.46–1.15; I2 = 51%). This result did not change as per the method of vitamin D supplementation (enteral route only: 19.9% vs 18.2%, OR, 1.19; 95% CI, 0.88–1.57; I2 = 10%; intramuscular or intravenous injection route: 25.6% vs 40.8%, OR, 0.48; 95% CI, 0.21–1.06; I2 = 19%) or daily dose (high dose: 20.9% vs 19.8%, OR, 0.83; 95% CI, 0.51–1.36; I2 = 53%; low dose: 15.6% vs 21.3%, OR, 0.74; 95% CI, 0.32–1.68; I2 = 0%). No significant difference was observed between the vitamin D supplementation and placebo groups regarding the length of ICU stay (standard mean difference [SMD], − 0.30; 95% CI, − 0.61 to 0.01; I2 = 60%), length of hospital stay (SMD, − 0.17; 95% CI, − 041 to 0.08; I2 = 65%), and duration of mechanical ventilation (SMD, − 0.41; 95% CI, − 081 to 0.00; I2 = 72%). In conclusion, this meta-analysis suggested that the administration of vitamin D did not provide additional advantages over placebo for critically ill patients. However, additional studies are needed to confirm our findings.

www.nature.com/scientificreports/ study conducted in Europe showed that 13.0% of 55,844 European individuals showed average serum 25(OH) D concentrations of < 30 nmol/L 9 . In China, 30.6% of elderly people have vitamin D deficiency 10 .
In addition to its prevalence in the general population, vitamin D deficiency is common among critically ill patients. Lee et al. showed that 64.5% (n = 120) of critically ill surgical patients had serum 25(OH)D concentrations of < 20 nmol/L 11 , and Higgins et al. reported that 26% (50/196) of patients admitted to a medical/surgical intensive care unit (ICU) had vitamin D levels of ≤ 30 nmol/L 12 . A retrospective cohort study showed that 54% (65/121) of patients with severe sepsis or septic shock had vitamin D levels lower than 15 mg/mL 13 , and another prospective multicenter study demonstrated vitamin D deficiency in 78.8% (197/250) of patients 14 . Furthermore, several studies document that vitamin D deficiency could be associated with poor outcomes in critically ill patients 12,13,[15][16][17][18] . To improve the outcomes of critically ill patients, vitamin D supplementation was proposed for ICU patients. Several randomized controlled trials (RCTs) were conducted to investigate the effects of vitamin D supplementation on the outcomes of critically ill patients. However, their results are conflicting [19][20][21][22][23][24][25][26][27][28] . Some studies showed that vitamin D supplementation demonstrated positive effects by decreasing the length of hospital stay 23 , duration of mechanical ventilation (MV) 26,27 , and mortality rate 24,27 . However, some studies 11,20,21,25,29,30 reported no change in the outcomes of critically ill patients. Even 2 meta-analyses, the included studies of which were published before 2017 31,32 , provided inconsistent findings. Since 2017, four more RCTs 24-27 have reported their findings. Therefore, we conducted an updated meta-analysis of RCTs to assess the association between vitamin D supplementation and the outcomes of critically ill patients.

Methods
Study search and selection. We conducted a literature review using the databases of PubMed, Embase, Web of Science, EBSCO, Cochrane Library, Ovid Medline, Embase, and Proquest until March 21, 2020. The following search terms were used: "intensive care" "ICU, " "critically-ill, " "vitamin D, " "calcitriol, " "Cholecalciferol*, " "ergocalciferol*, " and "RCT. " Our meta-analysis only included RCTs that investigated the clinical efficacy of vitamin D supplementation compared with placebo for critically ill adult patients. The supplementation could be done in different ways, such as oral, enteral, or parenteral vitamin D administration as 1, 25-dihydroxyvitamin D (calcitriol) or 25-hydroxyvitamin D (cholecalciferol). Two authors (Lan SH and Chang SP) searched for related studies and examined the risk of bias in each study using the Cochrane Risk of Bias Assessment tool 33 . When they had different opinions, a third author (Lai CC) helped resolve the issue. Data, including the year of publication, study design, study location and duration, demographic characteristics of critically ill patients, regimen of vitamin D, patient outcomes, and adverse events, were extracted from each included study. This study followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines.

Definitions and outcomes.
Critically ill patients were defined as that the patients with acute respiratory failure required mechanical ventilation or the patients required ICU hospitalization. The primary outcome of the current study was the patients' 28-day mortality. If data on 28-day mortality were not available, hospital mortality was used in the meta-analysis. Secondary outcomes included the length of ICU and hospital stay and the duration of MV. Doses of ≥ 300,000 and < 300,000 IU of vitamin D daily were defined as high and low doses, respectively, as per a previous study 34 .

Statistical analysis. We used Review Manager software (The Cochrane Collaboration 2008, Copenhagen)
to develop a random-effects model and derive the pooled estimates and their associated 95% CIs. The odds ratio (OR) was used to evaluate the outcome of 28-day mortality. Standardized mean differences (SMDs) and 95% CIs were computed for continuous variables including length of ICU and hospital stay and the duration of MV.

Discussion
This meta-analysis included nine RCTs with 1867 patients to compare the efficacy and safety of vitamin D supplementation with placebo in critically ill patients. The outcome was numerically better in the vitamin D supplementation group than control group, which may suggest biologically significant trends favoring vitamin D supplementation, however, these differences did not reach statistical significance. Overall, our results suggested that vitamin D supplementation did not significantly improve the outcomes of critically ill patients, which was supported by the following evidence. First, 28-day mortality did not change with vitamin D supplementation in the pooled analysis of 9 RCTs. Second, this difference remained unchanged in the sensitivity test. Third, we also found no significant improvement in the mortality of critically ill patients with vitamin D deficiency in the  www.nature.com/scientificreports/ subgroup analysis. Fourth, compared with the placebo group, we found no significant difference in mortality in the vitamin D supplementation group with either enteral or injection administration of vitamin D and with administration of low-or high-dose vitamin D. Finally, we assessed the effect of vitamin D on the length of ICU and hospital stay and MV duration and found no significant difference between the vitamin D supplementation and placebo groups. Moreover, no difference was observed in the subgroup analysis of high and low doses of vitamin D. The aforementioned findings indicate that compared with placebo, the vitamin D supplementation is not associated with lower mortality in critically ill patients. Our findings are consistent with those of a meta-analysis by Langlois et al. 32 , in which they included six RCTs of 695 patients, and they found that vitamin D did not reduce the mortality, length of ICU and hospital stay, and period on a ventilator. However, another meta-analysis by Putzu et al. 31 including 7 studies of 716 patients between 2011 and 2016 showed that vitamin D supplementation was associated with lower mortality compared with placebo (OR, 0.70; 95% CI, 0.50-0.98, I 2 = 0). This difference could be because we included a recent largescale study of more than 1,000 patients by Ginde et al. 25 , in which the administration of high-dose vitamin D did not provide an additional benefit with respect to clinical outcomes, including mortality. Moreover, the data of clinical outcomes in the analysis by Putzu et al. 31 had been reported by only 3 of 4 studies, which may limit the generalizability of their findings. Conversely, our study included more patients, more updated studies, and more subgroup analyses than previous studies 31,32 . In addition, all of our analyses showed consistent findings. Therefore, our findings provide stronger evidence regarding the effect of vitamin D supplementation on the outcomes of critically ill patients than previous studies. However, this study had several limitations. Although this study focused on critically ill patients, their clinical characteristics are heterogeneous. Some were admitted to the ICU for traumatic injury, and some had ventilatorassociated pneumonia. The criteria of vitamin D deficiency varied across studies, and the disease severity of the study patients also differed. Therefore, potential positive effects of vitamin D supplementation on the patient outcomes could not be found in this pooled analysis. In addition, only limited studies reported the vit D3 level after treatment and their level increased after treatment. Thus, we cannot assess the association between the level of vitamin D after treatment and the clinical outcome. Further studies are warranted to discover specific populations who can benefit from vitamin D supplementation 35 .