COVID-19 studies:  C19 studies: C19:  IvermectinIVM Vitamin DV.D HC QHC Q Vitamin CV.C ZincZn PXPX BHBH FPVFPV FLVFLV PVP-IPI CICI BLBL RDRD
Vitamin D is effective for COVID-19: real-time meta analysis of 70 studies
Random effects meta-analysis of the 20 vitamin D COVID-19 treatment studies to date shows an estimated reduction of 62% in the effect measured, RR 0.38 [0.27-0.53]. 95% of the studies to date report positive effects (12 of 20 are statistically significant in isolation). There is significant heterogeneity in studies, however this overview highlights that all treatment studies show positive effects, with the exception of one very late stage cholecalciferol study.
Sufficiency studies show a strong association between vitamin D sufficiency and outcomes. Meta-analysis of the 50 sufficiency studies shows an estimated reduction of 53%, RR 0.47 [0.40-0.55].
All data to reproduce this paper and the sources are in the appendix.
ImprovementStudies AuthorsPatients
All treatment studies 62% [47‑73%] 20 206 14,808
Treatment mortality results 69% [49‑81%] 11 102 2,647
All sufficiency studies 53% [45‑60%] 50 407 11,766
A[0.01-0.11]40,000IUviral+0/287/7RRCIDose (5d)TreatmentControlAnnweiler0.11[0.03-0.48]80,000IUdeath10/575/9Annweiler0.37[0.06-2.21]80,000IUdeath2/2910/32Tau​2 = 0.84; I​2 = 69.7%Early treatment0.10[0.03-0.36]12/11422/4890% improvementTan0.20[0.04-0.93]5,000IUoxygen3/1716/26RRCIDose (5d)TreatmentControlCastillo (RCT)0.15[0.01-2.94]0.8mg (c)death0/502/26Rastogi (RCT)0.47[0.24-0.92]300,000IUviral+6/1619/24Murai (RCT)1.49[0.55-4.05]200,000IUdeath9/1196/118Ling0.20[0.08-0.48]40,000IUdeath73253Jevalikar0.18[0.02-1.70]60,000IUdeath1/1283/69Giannini0.63[0.35-1.09]400,000IUdeath/ICU14/3629/55Nogués (CLUS. RCT)0.36[0.19-0.67]0.8mg (c)death36/55157/379Lakkireddy (RCT)0.39[0.08-1.91]300,000IUdeath2/445/43Tau​2 = 0.14; I​2 = 45.1%Late treatment0.42[0.28-0.63]71/1,034137/99358% improvementAnnweiler0.07[0.01-0.61]50,000IUdeath2/2910/32RRCIDose (1m)TreatmentControlLouca0.92[0.88-0.94]n/acasesCangiano0.30[0.10-0.87]50,000IUdeath3/2039/78Vasheghani0.70[0.33-1.49]n/adeath7/8848/420Ma0.70[0.50-0.97]n/acases49/3631,329/7,934Sulli0.50[0.34-0.73]n/acases22/6643/64Meltzer0.64[0.29-1.41]n/acases6/131239/3,338Ünsal0.29[0.11-0.76]variespneumonia4/2814/28Tau​2 = 0.15; I​2 = 80.6%PrEP0.54[0.38-0.77]93/7251,722/11,89446% improvementAll studies0.38[0.27-0.53]176/1,8731,881/12,93562% improvementAll 20 vitamin D COVID-19 treatment 4/10/211 CT: study uses combined treatmentTau​2 = 0.34; I​2 = 85.6%; Z = 5.76 (p < 0.0001)Lower RiskIncreased Risk
B studiesTreatment studiesmin, Q1, median, Q3, maxLower RiskIncreased 4/10/21
100+ 75 50 25 25 50 75 100 Tanoxygen, p=0.04Espitia-Hernandezviral+, p<0.0001Castillodeath, p=0.11Annweilerdeath, p=0.002Annweilerdeath, p=0.02Annweilerdeath, p=0.28Rastogiviral+, p=0.02Muraideath, p=0.43Loucacases, p=0.0007Lingdeath, p<0.001Cangianodeath, p=0.04Jevalikardeath, p=0.12Gianninideath/ICU, p=0.13Vasheghanideath, p=0.45Noguésdeath, p=0.001Macases, p=0.03Lakkireddydeath, p=0.27Sullicases, p=0.0002Meltzercases, p=0.38Ünsalpneumonia, p=0.009Treatment studies% Lower Risk% Increased 4/10/21
100+ 75 50 25 25 50 75 100 Lau, ICU, p=0.29Mendy, death, p=0.89Panagiotou, ICU, p=0.02Faul, ventilation, p=0.03Merzon, hosp., p=0.06Carpagnano, death, p=0.05Im, cases, p=0.003Hastie, death, p=0.31Baktash, death, p=0.50Meltzer, cases, p=0.02Israel, cases, p<0.001Radujkovic, death, p=0.001Kaufman, cases, p<0.001Maghbooli, death, p=0.08Karahan, death, p<0.0001Yılmaz, severe case, p=1.00Faniyi, seropositive, p=0.003Ye, hosp., p=0.03Macaya, severe case, p=0.07Hernández, death/ICU, p<0.0001Abrishami, death, p=0.04Cereda, death, p=0.04Walk, int./death, p=1.00Luo, progression, p=0.01Jain, ICU, p<0.0001De Smet, death, p=0.02Katz, cases, p<0.001Alguwaihes, death, p=0.007Vassiliou, death, p=0.04Abdollahi, cases, p=0.001Szeto, death, p=1.00Karonova, death, p=0.07Amin, progression, p=0.20Angelidi, death, p=0.01Li, severe case, p<0.0001Bennouar, death, p=0.002Demir, severe case, p<0.001Susianti, death, p=0.32Infante, death, p=0.05Gavioli, death, p=0.83Sulli, cases, p<0.0001Ricci, death, p=0.07Mazziotti, progression, p=0.006Charoenngam, death, p=0.19Vanegas-Cedillo, death, p=0.006Meltzer, cases, p=0.11Freitas, death, p=0.02Bayramoğlu, severe case, p=0.0008Livingston, cases, p<0.0001Ünsal, death, p=0.23Sufficiency studies% Lower Risk% Increased 4/10/21
Figure 1. A. Random effects meta-analysis of treatment studies. Simplified dosages are shown for comparison, these are the total dose in the first five days for treatment, and the monthly dose for prophylaxis. Calcifediol treatment is indicated with (c). For full details see the appendix. B. Scatter plot showing the distribution of effects reported in serum level analysis (sufficiency) studies and treatment studies (the vertical lines and shaded boxes show the median and interquartile range). C and D. Chronological history of all reported effects for treatment studies and sufficiency studies. The 2 studies reporting negative effects both have very low statistical significance.
We analyze all significant studies regarding vitamin D and COVID-19. Search methods, inclusion criteria, effect extraction criteria (more serious outcomes have priority), all individual study data, PRISMA answers, and statistical methods are detailed in Appendix 1. We present random-effects meta-analysis results for studies analyzing outcomes based on sufficiency, for all treatment studies, for mortality results only, and for treatment studies within each treatment stage.
Vitamin D.
Vitamin D undergoes two conversion steps before reaching the biologically active form as shown in Figure 2. The first step is conversion to calcidiol, or 25(OH)D, in the liver. The second is conversion to calcitriol, or 1,25(OH)2D, which occurs in the kidneys, the immune system, and elsewhere. Calcitriol is the active, steroid-hormone form of vitamin D, which binds with vitamin D receptors found in most cells in the body. Vitamin D was first identified in relation to bone health, but is now known to have multiple functions, including an important role in the immune system [Martens]. There is a significant delay involved in the conversion from cholecalciferol, therefore calcidiol (calcifediol) may be preferable for treatment.
Figure 2. Simplified view of vitamin D sources and conversion.
Many vitamin D studies analyze outcomes based on serum vitamin D levels which may be maintained via sun exposure, diet, or supplementation. We refer to these studies as sufficiency studies, as they typically present outcomes based on vitamin D sufficiency. These studies do not establish a causal link between vitamin D and outcomes. In general, low vitamin D levels are correlated with many other factors that may influence COVID-19 susceptibility and severity. Therefore, beneficial effects found in these studies may be due to factors other than vitamin D. On the other hand, if vitamin D is causally linked to the observed benefits, it is possible that adjustments for correlated factors could obscure this relationship. For these reasons, we analyze sufficiency studies separately from treatment studies. We include all sufficiency studies that provide a comparison between groups having sufficient and insufficient levels.
For studies regarding treatment with vitamin D, we distinguish three stages as shown in Figure 3. Pre-Exposure Prophylaxis (PrEP) refers to regularly taking vitamin D before being infected. Early Treatment refers to treatment immediately or soon after symptoms appear, while Late Treatment refers to more delayed treatment.
Figure 3. Treatment stages.
Figure 1 shows the effects reported in sufficiency studies and treatment studies. Figure 4 and 5 show results by treatment stage. Figure 6 shows a forest plot for random effects meta-analysis of sufficiency studies, while Figure 7, 8, 9, and 10 show forest plots for all treatment studies with pooled effects, and for studies reporting mortality, case results, and viral clearance results only. Table 1 summarizes the results.
Study typeNumber of studies reporting positive results Total number of studiesPercentage of studies reporting positive results Random effects meta-analysis results
Analysis of outcomes based on sufficiency 45 50 90.0% 53% improvement
RR 0.47 [0.40‑0.55]
p < 0.0001
Early treatment 3 3 100% 90% improvement
RR 0.10 [0.03‑0.36]
p = 0.00039
Late treatment 8 9 88.9% 58% improvement
RR 0.42 [0.28‑0.63]
p < 0.0001
Pre‑Exposure Prophylaxis 8 8 100% 46% improvement
RR 0.54 [0.38‑0.77]
p = 0.00072
All treatment studies 19 20 95.0% 62% improvement
RR 0.38 [0.27‑0.53]
p < 0.0001
Table 1. Results. treatment studiesPre-Exposure ProphylaxisLate treatmentEarly treatmentmin, Q1, median, Q3, maxLower RiskIncreased 4/10/21
Figure 4. Results by treatment stage.
100+ 75 50 25 25 50 75 100 Espitia-Hernandezviral+, p<0.0001Annweilerdeath, p=0.002Annweilerdeath, p=0.28Early treatmentTanoxygen, p=0.04Castillodeath, p=0.11Rastogiviral+, p=0.02Muraideath, p=0.43Lingdeath, p<0.001Jevalikardeath, p=0.12Gianninideath/ICU, p=0.13Noguésdeath, p=0.001Lakkireddydeath, p=0.27Late treatmentAnnweilerdeath, p=0.02Loucacases, p=0.0007Cangianodeath, p=0.04Vasheghanideath, p=0.45Macases, p=0.03Sullicases, p=0.0002Meltzercases, p=0.38Ünsalpneumonia, p=0.009Pre-Exposure Prophylaxis% Lower Risk% Increased 4/10/21
Figure 5. Results by treatment stage.[0.18-1.68]ICURRCIMendy0.93[0.33-2.47]deathPanagiotou0.48[0.24-0.95]ICUFaul0.31[0.10-0.95]ventilationMerzon0.54[0.23-1.02]hosp.Carpagnano0.29[0.10-0.85]deathIm0.27[0.11-0.69]casesHastie0.83[0.57-1.20]deathBaktash0.71[0.18-2.78]deathMeltzer0.56[0.36-0.89]casesIsrael0.79[0.75-0.84]casesRadujkovic0.07[0.01-0.37]deathKaufman0.47[0.29-0.76]casesMaghbooli0.48[0.22-1.05]deathKarahan0.17[0.08-0.41]deathYılmaz0.27[0.01-5.14]severe caseFaniyi0.71[0.59-0.86]seropositiveYe0.07[0.01-0.81]hosp.Macaya0.45[0.15-1.06]severe caseHernández0.17[0.07-0.43]death/ICUAbrishami0.24[0.06-0.93]deathCereda2.20[1.01-3.22]deathWalk1.00[0.60-1.67]int./deathLuo0.37[0.17-0.81]progressionJain0.05[0.01-0.18]ICUDe Smet0.30[0.10-0.80]deathKatz0.22[0.17-0.27]casesAlguwaihes0.14[0.04-0.59]deathVassiliou0.09[0.01-1.51]deathAbdollahi0.46[0.28-0.75]casesSzeto1.06[0.49-2.26]deathKaronova0.21[0.03-1.28]deathAmin1.32[0.88-1.89]progressionAngelidi0.12[0.02-0.60]deathLi0.64[0.53-0.78]severe caseBennouar0.14[0.04-0.50]deathDemir0.11[0.03-0.44]severe caseSusianti0.09[0.01-1.34]deathInfante0.45[0.19-1.10]deathGavioli1.05[0.78-1.40]deathSulli0.49[0.35-0.69]casesRicci0.12[0.01-2.28]deathMazziotti0.63[0.40-0.90]progressionCharoenngam0.66[0.32-1.33]deathVanegas-Cedillo0.47[0.28-0.81]deathMeltzer0.65[0.39-1.10]casesFreitas0.59[0.38-0.91]deathBayramoğlu0.30[0.09-0.77]severe caseLivingston0.49[0.25-0.83]casesÜnsal0.19[0.01-3.87]deathAll studies0.47[0.40-0.55]53% improvementAll 50 vitamin D COVID-19 sufficiency 4/10/21Tau​2 = 0.17; I​2 = 81.1%; Z = 9.47 (p < 0.0001)Lower RiskIncreased Risk
Figure 6. Random effects meta-analysis for sufficiency studies.[0.01-0.11]40,000IUviral+0/287/7RRCIDose (5d)TreatmentControlAnnweiler0.11[0.03-0.48]80,000IUdeath10/575/9Annweiler0.37[0.06-2.21]80,000IUdeath2/2910/32Tau​2 = 0.84; I​2 = 69.7%Early treatment0.10[0.03-0.36]12/11422/4890% improvementTan0.20[0.04-0.93]5,000IUoxygen3/1716/26RRCIDose (5d)TreatmentControlCastillo (RCT)0.15[0.01-2.94]0.8mg (c)death0/502/26Rastogi (RCT)0.47[0.24-0.92]300,000IUviral+6/1619/24Murai (RCT)1.49[0.55-4.05]200,000IUdeath9/1196/118Ling0.20[0.08-0.48]40,000IUdeath73253Jevalikar0.18[0.02-1.70]60,000IUdeath1/1283/69Giannini0.63[0.35-1.09]400,000IUdeath/ICU14/3629/55Nogués (CLUS. RCT)0.36[0.19-0.67]0.8mg (c)death36/55157/379Lakkireddy (RCT)0.39[0.08-1.91]300,000IUdeath2/445/43Tau​2 = 0.14; I​2 = 45.1%Late treatment0.42[0.28-0.63]71/1,034137/99358% improvementAnnweiler0.07[0.01-0.61]50,000IUdeath2/2910/32RRCIDose (1m)TreatmentControlLouca0.92[0.88-0.94]n/acasesCangiano0.30[0.10-0.87]50,000IUdeath3/2039/78Vasheghani0.70[0.33-1.49]n/adeath7/8848/420Ma0.70[0.50-0.97]n/acases49/3631,329/7,934Sulli0.50[0.34-0.73]n/acases22/6643/64Meltzer0.64[0.29-1.41]n/acases6/131239/3,338Ünsal0.29[0.11-0.76]variespneumonia4/2814/28Tau​2 = 0.15; I​2 = 80.6%PrEP0.54[0.38-0.77]93/7251,722/11,89446% improvementAll studies0.38[0.27-0.53]176/1,8731,881/12,93562% improvementAll 20 vitamin D COVID-19 treatment 4/10/211 CT: study uses combined treatmentTau​2 = 0.34; I​2 = 85.6%; Z = 5.76 (p < 0.0001)Lower RiskIncreased Risk
Figure 7. Random effects meta-analysis for treatment studies.[0.03-0.48]80,000IU10/575/9RRCIDose (5d)TreatmentControlAnnweiler0.37[0.06-2.21]80,000IU2/2910/32Tau​2 = 0.38; I​2 = 52.0%Early treatment0.17[0.05-0.55]12/8615/4183% improvementCastillo (RCT)0.15[0.01-2.94]0.8mg (c)0/502/26RRCIDose (5d)TreatmentControlMurai (RCT)1.49[0.55-4.05]200,000IU9/1196/118Ling0.20[0.08-0.48]40,000IU73253Jevalikar0.18[0.02-1.70]60,000IU1/1283/69Nogués (CLUS. RCT)0.36[0.19-0.67]0.8mg (c)36/55157/379Lakkireddy (RCT)0.39[0.08-1.91]300,000IU2/445/43Tau​2 = 0.27; I​2 = 47.4%Late treatment0.40[0.21-0.76]48/96573/88860% improvementAnnweiler0.07[0.01-0.61]50,000IU2/2910/32RRCIDose (1m)TreatmentControlCangiano0.30[0.10-0.87]50,000IU3/2039/78Vasheghani0.70[0.33-1.49]n/a7/8848/420Tau​2 = 0.86; I​2 = 75.0%PrEP0.28[0.08-0.94]12/13797/53072% improvementAll studies0.31[0.19-0.51]72/1,188185/1,45969% improvementAll 11 vitamin D COVID-19 treatment mortality 4/10/21Tau​2 = 0.38; I​2 = 61.1%; Z = 4.51 (p < 0.0001)Lower RiskIncreased Risk
Figure 8. Random effects meta-analysis for mortality results only.[0.88-0.94]n/aRRCIDose (1m)TreatmentControlMa0.70[0.50-0.97]n/a49/3631,329/7,934Sulli0.50[0.34-0.73]n/a22/6643/64Meltzer0.64[0.29-1.41]n/a6/131239/3,338Tau​2 = 0.07; I​2 = 78.9%PrEP0.71[0.52-0.96]77/5601,611/11,33629% improvementAll studies0.71[0.52-0.96]77/5601,611/11,33629% improvementAll 4 vitamin D COVID-19 treatment case 4/10/21Tau​2 = 0.07; I​2 = 78.9%; Z = 2.21 (p = 0.014)Lower RiskIncreased Risk
Figure 9. Random effects meta-analysis for COVID-19 case results only.[0.01-0.11]40,000IU0/287/7RRCIDose (5d)TreatmentControlTau​2 = 0.00; I​2 = 0.0%Early treatment0.03[0.01-0.11]0/287/797% improvementRastogi (RCT)0.47[0.24-0.92]300,000IU6/1619/24RRCIDose (5d)TreatmentControlTau​2 = 0.00; I​2 = 0.0%Late treatment0.47[0.24-0.92]6/1619/2453% improvementAll studies0.12[0.01-1.97]6/4426/3188% improvementAll 2 vitamin D COVID-19 treatment viral clearance 4/10/211 CT: study uses combined treatmentTau​2 = 3.72; I​2 = 92.5%; Z = 1.48 (p = 0.069)Lower RiskIncreased Risk
Figure 10. Random effects meta-analysis for viral clearance results only.
To avoid bias in the selection of studies, we include all studies in the main analysis. Here we show the results after excluding studies with critical issues.
[Murai] is a very late stage study (mean 10 days from symptom onset, with 90% on oxygen at baseline), with poorly matched arms in terms of ethnicity, diabetes, and baseline ventilation, all of which favor the control group. Further, this study uses cholecalciferol, which may be especially poorly suited for such a late stage.
The studies excluded are as follows, and the resulting forest plot is shown in Figure 11.
[Murai], very late stage, >50% on oxygen/ventilation at baseline.[0.01-0.11]40,000IUviral+0/287/7RRCIDose (5d)TreatmentControlAnnweiler0.11[0.03-0.48]80,000IUdeath10/575/9Annweiler0.37[0.06-2.21]80,000IUdeath2/2910/32Tau​2 = 0.84; I​2 = 69.7%Early treatment0.10[0.03-0.36]12/11422/4890% improvementTan0.20[0.04-0.93]5,000IUoxygen3/1716/26RRCIDose (5d)TreatmentControlCastillo (RCT)0.15[0.01-2.94]0.8mg (c)death0/502/26Rastogi (RCT)0.47[0.24-0.92]300,000IUviral+6/1619/24Ling0.20[0.08-0.48]40,000IUdeath73253Jevalikar0.18[0.02-1.70]60,000IUdeath1/1283/69Giannini