The flow diagram for study selection is displayed in Fig. 1. A total of 446 records through searching the electronic database and 27 in the reference list articles identified. After removing duplicate publications, 128 studies were retrieved for title and abstract review. A total of 20 studies went further for full-text evaluation. 11 studies were excluded due to inconsistent study and grouping methods;2 were excluded because the report form cannot be used to calculate the results, 2 because repetition of research; 1 because we could not reach the author. We eliminated two overlapping data, including only the most recent or comprehensive data. Lai's study contains two control groups, age- and gender-matched healthy subjects(h) and critically ill patients without AKI(i), so it is divided into 2 cohorts. Finally, four studies with five cohorts fulfilled the inclusion criteria and were suitable for the analysis.
The characteristics of the included studies are summarized in Table 1. Total sample size of four studies was 413. Methodological quality of the included studies using the NOS are presented in table 2. All the enrolled articles showed moderate to high quality.
Vitamin D values in AKI patients and non-AKI controls
To assess the quantitative difference of vitamin D status between AKI patients and non-AKI controls, five cohorts provided data on serum 25(OH)D level, and four on serum 1,25(OH)2D level. The average 25(OH)D levels in AKI patients was not significant different compared to non-AKI controls (MD=-7.04, 95% CI =-15.57 to 1.49, P = 0.11) (Fig. 2a), which showed significant heterogeneity (I2=96%, p＜0.0001). The average 1,25(OH)2D levels were significant lower in AKI patients than that in non-AKI controls (MD=-17.79, 95% CI =-32.73 to -2.85, P = 0.02) (Fig. 2b), which also exhibited great heterogeneity (I2=95%, p＜0.0001). After sensitivity analysis, removing Vijiyuan et al.significantly reduced this heterogeneity (I2 = 0, p = 0.45), without much impact on the outcome of 1,25(OH)2D levels (MD=-11.04, 95% CI =-15.57 to -6.50, P＜0.01) (Fig. 2c).
We further performed a subgroup analysis based on two different control populations, which were divided into healthy subjects(h) subgroup and critically ill patients without AKI(i) subgroup. The serum 25(OH)D levels remained no significant difference in both subgroups. (Fig. 3a). The serum 1,25(OH)2D levels tend to be lower in AKI patients than in healthy subjects (MD= -22.91, 95% CI = -45.44 to -0.38, P = 0.05) (Fig. 3b), but did not reach statistical difference. The serum 1,25 (OH)2 D levels were significant lower in AKI patients than critically ill patients without AKI (MD=-11.59, 95% CI = -18.89 to -4.30, P = 0.02) (Fig. 3b).
Relationship between vitamin D levels and prognosis of AKI
Of three included studies reporting AKI patients’ clinical outcome, only Vijayan provided quantitative data on vitamin D levels between the non-survivors and survivors among the AKI patients. Their study found that 1,25(OH)2D levels were significant higher in non-survivors than survivors (62±41.4 pg/mL vs. 33.7±24.2 pg/mL, P = 0.046), while 25(OH)D levels did not differ between two groups. However, higher levels of 1,25(OH)2D was not associated with mortality in AKI patients on multivariate regression analysis t after adjusting for age and APACHE II. Laigrouped the patients with AKI according to median serum vitamin D concentrations. They come to the conclusion there was no association between the serum vitamin D levels, both 25(OH)D and 1,25(OH)2D, and all-cause mortality in AKI patients. Similarly, Leaf reported that hospital mortality of AKI patients was not associated with 25(OH)D and 1,25(OH)2D after adjusting for age and serum creatinine.
Therefore, the relationship between vitamin D levels and mortality in AKI patients remains controversial. We are unable to give a consistent conclusion for this problem based on current limited and conflict study results. More comprehensive trials with sufficient power and longer follow-up duration are needed to draw a clear conclusion.