Efficacy of vitamin D supplementation in major depression: A meta-analysis of randomized controlled trialsF Vellekkatt, V Menon
Department of Psychiatry, Jawaharlal Institute of Post Graduate Medical Education and Research (JIPMER), Puducherry, India
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/jpgm.JPGM_571_17
Source of Support: None, Conflict of Interest: None
Keywords: Depressive disorder, meta-analysis, randomized controlled trials, vitamin D
Worldwide, it is estimated that more than 300 million people are affected by depression. Unipolar depressive disorders were the third leading cause of disability-adjusted life years (DALY) lost based on the WHO report of 2004, and were projected to rise to the second leading cause of DALY across all age group by 2020. Despite burgeoning research on its neurobiological basis in the last decade, a significant gap exists in our understanding of the origins and progression of depressive disorders. Studies have suggested that depression may have multifactorial origins involving dysfunction of multiple brain areas and alteration in many biochemical functions such as gene expression and immune response., However, classical antidepressants mainly function by correction of monoamine imbalance. This may be a possible reason for the low rates of treatment-induced remission often noted in clinical trials on depressed patients., Therefore, there is a pressing need to look at other treatment targets that can optimize clinical management of depression.
A growing body of literature links vitamin D to the pathophysiology of depression.,,, This has mainly come from three lines of evidence; first, lower serum vitamin D levels in depressed persons compared to controls;,,, second, presence of vitamin D receptors in various parts of the brain limbic system, cerebellum, and cortex,,, which controls emotions and behavior; and third, the important modulatory role that vitamin D plays in regulating immunoinflammatory pathways that have been found to be relevant to the pathophysiology of depression.,, This evidence has spawned a series of trials that have tried to answer the question whether vitamin D supplementation among depressed patients might improve their depression scores with mixed results. Thus far, a few meta-analyses have included studies assessing the effect of vitamin D on depressive symptom scores in individuals without a clinical diagnosis of major depression.,,,, Further, as the authors of these studies point out, many of the included patients had very low depression scores to begin with which may be the reason for the marginal effects noted. Hence, there is a need to separately examine the effects of vitamin D supplementation in populations with major depression.
With this background, we aimed to combine the evidence from available randomized trials to provide a quantitative estimate of the effect of vitamin D supplementation on depression symptom ratings among patients with clinically diagnosed depression compared to a control group. We also aimed to highlight the research gaps in this area to inform future trials.
Inclusion and exclusion criteria
Using the patient, intervention, comparison, outcomes, and study design criteria, all studies that assessed the association between vitamin D and depression or vice versa were included provided they met the following a priori criteria:
Population – Patients clinically diagnosed with unipolar depression as per the standard diagnostic criteria
Intervention – Vitamin D supplementation in any dosage formulation
Comparison – With a control group of any nature
Outcomes – Primary outcome expressed in terms of depression symptom scores
We included only randomized controlled trials meeting the above criteria [Table 1]. Studies done on healthy volunteers or nonpsychiatric populations were excluded.,,,, The excluded studies with relevant details are given in [Table 2].
Search strategy and study selection
Electronic searches of database such as PubMed, ScienceDirect, and Google Scholar were carried out from inception till August 2017. Articles were generated using random combinations of the following search terms: “depression,” “depressive disorder,” “depressed individuals,” “vitamin D,” “vitamin d,” “25-hydroxyvitamin D,” “vitamin D3,” “Cholecalciferol,” or “Calcitriol.” The search was limited to articles published in English language, peer reviewed journals. The title and abstracts of the generated studies were examined by two of the authors independently (Vikas Menon and Favaz V), and a consolidated list of abstracts were drawn up after eliminating duplicates. In case of inadequate information in the abstract, the corresponding full texts of potentially relevant articles were retrieved to screen for inclusion criteria. The reference lists of included studies were also manually examined for potential articles. We did not include gray literature such as conference proceedings primarily due to concerns about study quality and inadequate reporting. The flow chart for the literature search is presented in [Figure 1].
Data extraction and selection of outcome measure
Data were abstracted from articles meeting the inclusion criteria on the following items: author and year of study, region of study, total sample size as well as sample size in each group, dosing regimen of vitamin D followed, nature and dose of antidepressant received by the comparator arm, assessment time points, total duration of study, and primary and secondary outcome measures. The outcome measure used for calculation of effect size was the depression symptom score which had to be a primary outcome measure of all the included studies. Data extraction was done by two of the authors (Vikas Menon and Favaz Vellekkatt). We relied only on published information and did not contact the authors for additional data.
Study quality assessment
We assessed the quality of individual trials using the Cochrane collaboration tool which summarizes the risk of bias under various heads. These included information about random sequence generation (selection bias), details about allocation concealment (selection bias), blinding of participants and study personnel (performance bias), blinding of outcome rater (detection bias), handling incomplete data (attrition bias), and selectively reporting originally mentioned outcomes (reporting bias). The authors, after examining the full texts of the included articles, categorized every trial on the above parameters which were reported as present, absent, or unclear [Table 3].
For every trial that assessed the efficacy of vitamin D supplementation against a comparator in major depression, effect sizes were computed as standardized mean difference (Cohen's d) with 95% confidence interval.
Pooled mean effect sizes for the intervention was calculated using both the fixed and random-effects model for the entire sample. Both the models threw up similar values for the present meta-analysis. We used the I2 statistic to assess heterogeneity among the selected studies. Its value may range from 0 to 100, with higher values suggesting greater heterogeneity.
Assessment of publication bias was done using the Funnel plot. This is essentially a scatter plot of the intervention effect estimates of each study against some measure of its size or precision indicating its overall weight in the meta-analysis. Here, we studied the scatter between the computed effect size for each study against its standard error. In the absence of a publication bias, effect estimates from smaller studies will scatter more widely towards the bottom of the graph while the larger studies aggregate more closely above. The result is the appearance of an inverted funnel. If there is a publication bias, the funnel plot tends to be proportionately asymmetrical with gaps in the bottom of the graph. Analysis was performed using STATA version 13 (StataCorp. 2013. Stata Statistical Software: Release 13. College Station, TX: StataCorp LP).
Characteristics of selected studies
A total of four trials met the inclusion criteria and were included in the present meta-analysis. Characteristics of the included studies are shown in [Table 1]. Three of the included trials were done ,, in Iran and the remaining one  was done in southeast China. Two of the trials , were done on laboratory proven vitamin D deficiency, whereas in the remaining two trials , participants were not examined for vitamin D deficiency before recruitment. The sample size of individual trials ranged from 40 to 746, and the pooled sample size was 948. Among the four trials, three were double-blind randomized control trial ,, and one was a nonblinded randomized control trial. In one of the trials, intervention group received parenteral vitamin D supplementation, whereas in all other trials,,, oral vitamin D was provided. In two trials,, the intervention group received weekly oral vitamin D supplementation, whereas in the one other trial, they received daily oral vitamin D supplementation. Duration of follow-up ranged from a minimum of 8 weeks  to a maximum of 52 weeks.
Effect sizes of individual trials ranged from 0.49 to 1.03. The effect sizes, confidence intervals, and respective weights for included trials are depicted in [Figure 2] (Forest plot). Meta-analysis of the four trials using the random effects model yielded a pooled mean effect size of 0.58 [n = 948, confidence Intervals (CI) 0.45–0.72] for vitamin D supplementation on depressive symptom ratings in major depression. The I2 value for heterogeneity of the analyzed studies was 0 in both fixed and random-effects model (P = 0.720). This indicates low levels of heterogeneity among the sample of included studies.
Quality assessment of studies
All were randomized controlled trials, of which all were double-blind trials except one which was a nonblinded study. Details regarding allocation concealment were not mentioned in two of the studies., In all four trials, the outcome data was complete and there was no reporting bias.
Evidence of publication bias
To assess publication bias, a funnel plot was derived [Figure 3]. Though the number of trials were few, it was observed that less precise studies were skewed towards the right side of the graph (favoring the intervention). To test the asymmetry of the funnel plot, Egger's test was carried out. The standard normal deviate was regressed against its precision, and the regression equation for the sample of studies was derived as follows:
Standard normal deviate (SND) =0.613 + 0.516 × (precision)
The Y-intercept (constant) for the equation was 0.613 (confidence intervals −1.558–2.784, P = 0.349, adjusted R2 = 0.95), which suggested minimal publication bias among the selected sample of studies.
The findings of the present meta-analysis suggest that vitamin D supplementation has clinical benefits in patients with syndromal depression. The pooled mean effect size was 0.58 which suggest a moderate effect size. Of the four studies included in the meta-analysis, only in one did the confidence intervals for the effect size span zero. In the remaining three studies,,, the confidence intervals were noninclusive of the null value (zero) and can be regarded as clinically effective trials. Notably, parenteral vitamin D supplementation was not found to be superior to oral vitamin D. The study by Khoraminya et al. had a relatively high effect size (standardized mean difference, 1.03), and was the only study where patients were administered adjunctive daily vitamin D along with antidepressant (fluoxetine). However, on sensitivity analysis, we observed that the pooled effect size did not change much when this study was dropped from the random-effects model. The effect size continued to be moderate in magnitude (0.57, 95% confidence interval 0.44–0.71), which indicates that vitamin D supplementation on a nondaily basis may also be as effective as a daily supplementation regimen. Visual inspection of the funnel plot indicated that there was some publication bias particularly among smaller studies though the Egger test suggested minimal publication bias.
To our knowledge, no prior meta-analysis has tried to answer the question whether supplementation of vitamin D is beneficial in clinically depressed individuals. The reason for the benefit seen with vitamin D could be due to the fact that depression has multifactorial origins including oxidative stress., Therefore, the antioxidant properties of vitamin D could be relevant , and may contribute to therapeutic effects observed. Moreover, neuroinflammation is an upcoming theory for depression ,, and is postulated to be an important factor in modulating treatment response., Certain studies have shown low levels of vitamin D to be associated with elevated levels of systemic inflammatory markers.,,,, This provides a basis for understanding the mechanisms underpinning improvement in depressive symptoms upon addition of vitamin D to the therapeutic regimen. Future studies with prospective designs must investigate whether change in levels of inflammatory markers parallel that of serum vitamin D levels in major depression to confirm and establish the directionality of these associations.
Some limitations of the current meta-analysis should be considered before drawing firm conclusions. First, only four studies met the inclusion criteria, and therefore, the results must be viewed as tentative. Second, in the current meta-analysis, one study  had disproportionately higher weightage owing to large sample size. To see how the other studies behaved as a group, we did a sensitivity analysis and noticed that the effect size showed a marginal appreciation (0.68, 95% confidence interval 0.34–1.03) after dropping the Wang et al. study from the final model. This continues to support the benefits of supplementing vitamin D in depression. Having observed this, we must add that the other three studies had small sample sizes, and therefore, group differences had to be of comparatively higher magnitude to achieve statistical significance. This implies that the therapeutic effect of vitamin D was substantial. Third, we have employed the funnel plot with Egger test to detect publication bias. However, the power of this test to detect bias is low with a small number of studies. Fourth, because patients with depression have an appreciable placebo response rate,, lack of allocation concealment in two of the studies , and lack of blinding of intervention in one trial  could have influenced the results. These were among several sources of bias noted in the current meta-analysis. The included trials differed in the dosages of adjunctive vitamin D used and this may have influenced the overall results. It is also unclear if vitamin D supplementation is beneficial only among deficient subjects or across the board in major depression and this needs to be answered through future studies comparing depressed patients with and without a concurrent diagnosis of vitamin D deficiency.
To conclude, our findings show that vitamin D supplementation favorably impacts depression ratings in major depression. The effect sizes noted were in the moderate range for this outcome. These conclusions are certainly impacted by the limited number of trials analyzed and the inherent methodological bias noted in a few of them. Clearly, more well-designed trials that address these flaws as well as concurrently investigate the possible mechanisms underpinning clinical benefits with vitamin D are required to improve our understanding and therapeutic strategy. Considering the high burden of poor response and nonresponse in depression, this will inform and expand the therapeutic armamentarium to deal with this condition better.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3]