Association of Vitamin D Levels, Race/Ethnicity, and Clinical Characteristics With COVID-19 Test Results | Health Disparities | JAMA Network Open | JAMA Network
[Skip to Navigation]
Sign In
Figure.  Estimated COVID-19 Rates by Most Recent Vitamin D Level
Estimated COVID-19 Rates by Most Recent Vitamin D Level

To convert vitamin D to nanomoles per liter, multiply by 2.496.

Table 1.  Characteristics of the Sample
Characteristics of the Sample
Table 2.  Multivariable Association of Vitamin D Levels and Treatment With Test Results Positive for COVID-19
Multivariable Association of Vitamin D Levels and Treatment With Test Results Positive for COVID-19
Table 3.  Multivariable Association of Vitamin D Levels and Treatment With Testing Positive for COVID-19 Among Individuals Reporting Black Race
Multivariable Association of Vitamin D Levels and Treatment With Testing Positive for COVID-19 Among Individuals Reporting Black Race
Table 4.  COVID-19 Positivity Rate by Most Recent Vitamin D Level and by Treatment
COVID-19 Positivity Rate by Most Recent Vitamin D Level and by Treatment
1.
Forrest  KY, Stuhldreher  WL.  Prevalence and correlates of vitamin D deficiency in US adults.   Nutr Res. 2011;31(1):48-54. doi:10.1016/j.nutres.2010.12.001 PubMedGoogle ScholarCrossref
2.
Holick  MF, Binkley  NC, Bischoff-Ferrari  HA,  et al; Endocrine Society.  Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline.   J Clin Endocrinol Metab. 2011;96(7):1911-1930. doi:10.1210/jc.2011-0385 PubMedGoogle ScholarCrossref
3.
Charoenngam  N, Holick  MF.  Immunologic effects of vitamin D on human health and disease.   Nutrients. 2020;12(7):2097. doi:10.3390/nu12072097 PubMedGoogle ScholarCrossref
4.
Del Valle  HB, Yaktine  AL, Taylor  CL, Ross  AC, eds.  Dietary Reference Intakes for Calcium and Vitamin D. National Academies Press; 2011.
5.
Brown  LL, Cohen  B, Tabor  D, Zappalà  G, Maruvada  P, Coates  PM.  The vitamin D paradox in Black Americans: a systems-based approach to investigating clinical practice, research, and public health—expert panel meeting report.   BMC Proc. 2018;12(suppl 6):6. doi:10.1186/s12919-018-0102-4 PubMedGoogle ScholarCrossref
6.
Martineau  AR, Jolliffe  DA, Hooper  RL,  et al.  Vitamin D supplementation to prevent acute respiratory tract infections: systematic review and meta-analysis of individual participant data.   BMJ. 2017;356:i6583. doi:10.1136/bmj.i6583 PubMedGoogle ScholarCrossref
7.
Benskin  LL.  A basic review of the preliminary evidence that COVID-19 risk and severity is increased in vitamin D deficiency.   Front Public Health. 2020;8:513. doi:10.3389/fpubh.2020.00513 PubMedGoogle ScholarCrossref
8.
Maghbooli  Z, Sahraian  MA, Ebrahimi  M,  et al.  Vitamin D sufficiency, a serum 25-hydroxyvitamin D at least 30 ng/mL reduced risk for adverse clinical outcomes in patients with COVID-19 infection.   PLoS One. 2020;15(9):e0239799. doi:10.1371/journal.pone.0239799 PubMedGoogle Scholar
9.
Baktash  V, Hosack  T, Patel  N,  et al.  Vitamin D status and outcomes for hospitalised older patients with COVID-19.   Postgrad Med J. Published online August 27, 2020. doi:10.1136/postgradmedj-2020-138712 PubMedGoogle Scholar
10.
Mardani  R, Alamdary  A, Mousavi Nasab  SD, Gholami  R, Ahmadi  N, Gholami  A.  Association of vitamin D with the modulation of the disease severity in COVID-19.   Virus Res. 2020;289:198148. doi:10.1016/j.virusres.2020.198148 PubMedGoogle Scholar
11.
Meltzer  DO, Best  TJ, Zhang  H, Vokes  T, Arora  V, Solway  J.  Association of vitamin D status and other clinical characteristics with COVID-19 test results.   JAMA Netw Open. 2020;3(9):e2019722. doi:10.1001/jamanetworkopen.2020.19722 PubMedGoogle Scholar
12.
Merzon  E, Tworowski  D, Gorohovski  A,  et al.  Low plasma 25(OH) vitamin D level is associated with increased risk of COVID-19 infection: an Israeli population-based study.   FEBS J. 2020;287(17):3693-3702. doi:10.1111/febs.15495 PubMedGoogle ScholarCrossref
13.
Kaufman  HW, Niles  JK, Kroll  MH, Bi  C, Holick  MF.  SARS-CoV-2 positivity rates associated with circulating 25-hydroxyvitamin D levels.   PLoS One. 2020;15(9):e0239252. doi:10.1371/journal.pone.0239252 PubMedGoogle Scholar
14.
Israel  A, Assi Cicurel  AA, Feldhamer  I, Dror  Y, Giveon  SM, Gillis  D, Strich  D, Lavie  G.  The link between vitamin D deficiency and COVID-19 in a large population   medRxiv. doi:10.1101/2020.09.04.20188268Google Scholar
15.
Grant  WB, Lahore  H, McDonnell  SL,  et al.  Evidence that vitamin D supplementation could reduce risk of influenza and COVID-19 infections and deaths.   Nutrients. 2020;12(4):988. doi:10.3390/nu12040988 PubMedGoogle ScholarCrossref
16.
Boucher  BJ.  Vitamin D status as a predictor of COVID-19 risk in Black, Asian and other ethnic minority groups in the UK.   Diabetes Metab Res Rev. 2020;36(8):e3375. doi:10.1002/dmrr.3375 PubMedGoogle Scholar
17.
Centers for Disease Control and Prevention. CDC Diagnostic Tests for COVID-19. Accessed July 13, 2020. https://www.cdc.gov/coronavirus/2019-ncov/lab/testing.html
18.
Eurofins Viracor. Coronavirus (COVID-19) SARS-CoV-2 RT-PCR. Accessed July 13, 2020. https://www.viracor-eurofins.com/test-menu/8300-coronavirus-covid-19-sars-cov-2-rt-pcr/
19.
Roche Diagnostics. Cobas SARS-CoV-2 Test (for the COVID-19 coronavirus). Accessed July 13 2020. https://diagnostics.roche.com/us/en/products/params/cobas-sars-cov-2-test.html
20.
Centers for Disease Control and Prevention. International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM). Accessed February 15, 2021. https://www.cdc.gov/nchs/icd/icd10cm.htm
21.
Zhang  JX, Iwashyna  TJ, Christakis  NA.  The performance of different lookback periods and sources of information for Charlson comorbidity adjustment in Medicare claims.   Med Care. 1999;37(11):1128-1139. doi:10.1097/00005650-199911000-00005 PubMedGoogle ScholarCrossref
22.
McCullagh  P,, Nelder  J.  Generalized Linear Models. 2nd ed. Chapman and Hall/CRC; 1989. doi:10.1007/978-1-4899-3242-6
23.
McNutt  LA, Wu  C, Xue  X, Hafner  JP.  Estimating the relative risk in cohort studies and clinical trials of common outcomes.   Am J Epidemiol. 2003;157(10):940-943. doi:10.1093/aje/kwg074 PubMedGoogle ScholarCrossref
24.
Pregibon  D. Data Analytic Methods for Generalized Linear Models. Dissertation. University of Toronto; 1979.
25.
Blizzard  L, Hosmer  DW.  Parameter estimation and goodness-of-fit in log binomial regression.   Biom J. 2006;48(1):5-22. doi:10.1002/bimj.200410165 PubMedGoogle ScholarCrossref
26.
Zhang  Y, Fang  F, Tang  J,  et al.  Association between vitamin D supplementation and mortality: systematic review and meta-analysis.   BMJ. 2019;366:l4673. doi:10.1136/bmj.l4673 PubMedGoogle ScholarCrossref
27.
Freedman  BI, Register  TC.  Effect of race and genetics on vitamin D metabolism, bone and vascular health.   Nat Rev Nephrol. 2012;8(8):459-466. doi:10.1038/nrneph.2012.112 PubMedGoogle ScholarCrossref
28.
Hastie  CE, Mackay  DF, Ho  F,  et al.  Vitamin D concentrations and COVID-19 infection in UK Biobank.   Diabetes Metab Syndr. 2020;14(4):561-565. doi:10.1016/j.dsx.2020.04.050 PubMedGoogle ScholarCrossref
29.
Yancy  CW.  COVID-19 and African Americans.   JAMA. 2020;323(19):1891-1892. doi:10.1001/jama.2020.6548 PubMedGoogle ScholarCrossref
30.
Libon  F, Cavalier  E, Nikkels  AF.  Skin color is relevant to vitamin D synthesis.   Dermatology. 2013;227(3):250-254. doi:10.1159/000354750 PubMedGoogle ScholarCrossref
31.
Bouillon  R, Schuit  F, Antonio  L, Rastinejad  F.  Vitamin D binding protein: a historic overview.   Front Endocrinol (Lausanne). 2020;10:910. doi:10.3389/fendo.2019.00910 PubMedGoogle ScholarCrossref
32.
Oleröd  G, Hultén  LM, Hammarsten  O, Klingberg  E.  The variation in free 25-hydroxy vitamin D and vitamin D-binding protein with season and vitamin D status.   Endocr Connect. 2017;6(2):111-120. doi:10.1530/EC-16-0078 PubMedGoogle ScholarCrossref
33.
Bikle  DD, Schwartz  J. Vitamin  D  Binding protein, total and free vitamin D levels in different physiological and pathophysiological conditions front.   Endocrinol. 2019;10:317. doi:10.3389/fendo.2019.00317Google Scholar
34.
Bergman  P, Lindh  AU, Björkhem-Bergman  L, Lindh  JD.  Vitamin D and respiratory tract infections: a systematic review and meta-analysis of randomized controlled trials.   PLoS One. 2013;8(6):e65835. doi:10.1371/journal.pone.0065835 PubMedGoogle Scholar
35.
Looker  AC, Johnson  CL, Lacher  DA,  et al.  Vitamin D Status: United States 2001–2006. NCHS Data Brief, No 59. National Center for Health Statistics. 2011.
36.
McCullough  PJ, Lehrer  DS, Amend  J.  Daily oral dosing of vitamin D3 using 5000 to 50,000 international units a day in long-term hospitalized patients: insights from a seven year experience.   J Steroid Biochem Mol Biol. 2019;189:228-239. doi:10.1016/j.jsbmb.2018.12.010 PubMedGoogle ScholarCrossref
37.
Dhaliwal  R, Mikhail  M, Feuerman  M, Aloia  JF.  The vitamin d dose response in obesity.   Endocr Pract. 2014;20(12):1258-1264. doi:10.4158/EP13518.OR PubMedGoogle ScholarCrossref
38.
Billington  EO, Burt  LA, Rose  MS,  et al.  Safety of high-dose vitamin D supplementation: secondary analysis of a randomized controlled trial.   J Clin Endocrinol Metab. 2020;105(4):dgz212. doi:10.1210/clinem/dgz212 PubMedGoogle Scholar
39.
Garg  S, Kim  L, Whitaker  M,  et al.  Hospitalization rates and characteristics of patients hospitalized with laboratory-confirmed coronavirus disease 2019—COVID-NET, 14 states, March 1-30, 2020.   MMWR Morb Mortal Wkly Rep. 2020;69(15):458-464. doi:10.15585/mmwr.mm6915e3 PubMedGoogle ScholarCrossref
40.
Xue  KS, Moncla  LH, Bedford  T, Bloom  JD.  Within-host evolution of human influenza virus.   Trends Microbiol. 2018;26(9):781-793. doi:10.1016/j.tim.2018.02.007 PubMedGoogle ScholarCrossref
41.
Prietl  B, Treiber  G, Pieber  TR, Amrein  K.  Vitamin D and immune function.   Nutrients. 2013;5(7):2502-2521. doi:10.3390/nu5072502 PubMedGoogle ScholarCrossref
42.
Mohammad  S, Mishra  A, Ashraf  MZ.  Emerging role of vitamin D and its associated molecules in pathways related to pathogenesis of thrombosis.   Biomolecules. 2019;9(11):649. doi:10.3390/biom9110649 PubMedGoogle ScholarCrossref
43.
National Center for Immunization and Respiratory Diseases (NCIRD), Division of Viral Diseases. Preparing for COVID-19 in nursing homes. Accessed April 27, 2020. https://www.cdc.gov/coronavirus/2019-ncov/hcp/long-term-care.html
44.
CDC COVID-19 Response Team.  Characteristics of Health Care Personnel with COVID-19—United States, February 12-April 9, 2020.   MMWR Morb Mortal Wkly Rep. 2020;69(15):477-481. doi:10.15585/mmwr.mm6915e6 PubMedGoogle ScholarCrossref
45.
Huotari  A, Herzig  KH.  Vitamin D and living in northern latitudes—an endemic risk area for vitamin D deficiency.   Int J Circumpolar Health. 2008;67(2-3):164-178. doi:10.3402/ijch.v67i2-3.18258 PubMedGoogle ScholarCrossref
46.
Elliott  ME, Binkley  NC, Carnes  M,  et al.  Fracture risks for women in long-term care: high prevalence of calcaneal osteoporosis and hypovitaminosis D.   Pharmacotherapy. 2003;23(6):702-710. doi:10.1592/phco.23.6.702.32182 PubMedGoogle ScholarCrossref
47.
Sowah  D, Fan  X, Dennett  L, Hagtvedt  R, Straube  S.  Vitamin D levels and deficiency with different occupations: a systematic review.   BMC Public Health. 2017;17(1):519. doi:10.1186/s12889-017-4436-z PubMedGoogle ScholarCrossref
48.
Frieden  T. Former CDC Chief Dr. Tom Frieden: coronavirus infection risk may be reduced by vitamin D. Fox News. March 23, 2020. Accessed April 13, 2020. https://www.foxnews.com/opinion/former-cdc-chief-tom-frieden-coronavirus-risk-may-be-reduced-with-vitamin-d
49.
McCall  B. Medical societies advise on vitamin D in midst of COVID-19. Medscape Medical News. July 10, 2020. Accessed July 10, 2020. https://www.medscape.com/viewarticle/933715
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Views 27,261
    Citations 0
    Original Investigation
    Infectious Diseases
    March 19, 2021

    Association of Vitamin D Levels, Race/Ethnicity, and Clinical Characteristics With COVID-19 Test Results

    Author Affiliations
    • 1The University of Chicago, Chicago, Illinois
    JAMA Netw Open. 2021;4(3):e214117. doi:10.1001/jamanetworkopen.2021.4117
    Key Points

    Question  Are differences in vitamin D levels greater than levels traditionally considered sufficient (30 ng/mL) associated with having test results positive for coronavirus disease 2019 (COVID-19) in White and in Black individuals?

    Findings  In this cohort study of 4638 individuals with a measured vitamin D level in the year before undergoing COVID-19 testing, the risk of having positive results in Black individuals was 2.64-fold greater with a vitamin D level of 30 to 39.9 ng/mL than a level of 40 ng/mL or greater and decreased by 5% per 1-ng/mL increase in level among individuals with a level of 30 ng/mL or greater. There were no statistically significant associations of vitamin D levels with COVID-19 positivity rates in White individuals.

    Meaning  These findings suggest that randomized clinical trials to determine whether increasing vitamin D levels to greater than 30 to 40 ng/mL affect COVID-19 risk are warranted, especially in Black individuals.

    Abstract

    Importance  Deficient (ie, <20 ng/mL) or insufficient (ie, 20 to <30 ng/mL) 25-hydroxyvitamin D (also known as calcifediol) levels are more common in Black individuals than White individuals and are associated with increased coronavirus disease 2019 (COVID-19) risk. Whether COVID-19 risk is associated with differences in vitamin D levels of 30 ng/mL or greater is not known.

    Objective  To examine whether COVID-19 test results are associated with differences in vitamin D levels of 30 ng/mL or greater, including for White individuals and for Black individuals.

    Design, Setting, and Participants  This retrospective cohort study was conducted at an academic medical center in Chicago, Illinois. Participants included individuals with data on vitamin D level within 365 days before COVID-19 testing, which was conducted from March 3 to December 30, 2020. Data were analyzed from September 11, 2020, to February 5, 2021.

    Exposures  The last vitamin D level before COVID-19 testing was categorized as less than 20 ng/mL (ie, deficient), 20 to less than 30 ng/mL (ie, insufficient), 30 to less than 40 ng/mL, or 40 ng/mL or greater. Treatment was defined by vitamin D type and dose 14 days before COVID-19 testing and treatment changes after last vitamin D level.

    Main Outcomes and Measures  The main outcome was a positive result for COVID-19 in polymerase chain reaction testing. Multivariable analyses tested whether previously measured vitamin D level was associated with having test results positive for COVID-19 in White individuals and in Black individuals, controlling for months and treatment changes since the vitamin D level was measured, as well as demographic characteristics and comorbidity indicators.

    Results  A total of 4638 individuals (mean [SD] age 52.8 [19.5] years; 3205 [69%] women) had data for a vitamin D level within 1 year before COVID-19 testing, including 2288 (49%) Black individuals, 1999 (43%) White individuals, and 351 individuals (8%) who were another race/ethnicity (eg, Asian, Mideast Indian, >1 race). Stratified by vitamin D level, 1251 individuals (27%) had less than 20 ng/mL, 1267 individuals (27%) had 20 to less than 30 ng/mL, 1023 individuals (22%) had 30 to less than 40 ng/mL, and 1097 individuals (24%) had 40 ng/mL or greater. Lower vitamin D levels were more common in Black individuals (<20 ng/mL: 829 of 2288 Black individuals [36%]) than White individuals (<20 ng/mL: 315 of 1999 White individuals [16%]). A total of 333 individuals (7%) had test results positive for COVID-19, including 102 White individuals (5%) and 211 Black individuals (9%). Multivariate analysis controlling for time since last vitamin D level measurement was used to estimate the outcomes associated with levels 14 days before COVID-19 testing. A positive test result for COVID-19 was not significantly associated with vitamin D levels in White individuals but was associated with vitamin D levels in Black individuals (compared with ≥40 ng/mL: <20 ng/mL incidence rate ratio [IRR], 2.55 [95% CI, 1.26-5.15]; P = .009; 20 to <30 ng/mL IRR, 1.69 [95% CI, 0.75-3.84]; P = .21; 30 to <40 ng/mL IRR, 2.64 [95% CI, 1.24-5.66]; P = .01). Stratified by vitamin D level, estimated COVID-19 positivity rates in Black individuals were 9.72% (95% CI, 6.74%-13.41%) for individuals with a vitamin D level less than 20 ng/mL, 6.47% (95% CI, 3.33%-10.28%) for individuals with a vitamin D level of 20 to less than 30 ng/mL, 10.10% (95% CI, 6.00%-15.47%) for individuals with a vitamin D level of 30 to less than 40 ng/mL, and 3.82% (95% CI, 1.78%-6.68%) for individuals with a vitamin D level of 40 ng/mL or higher. Multivariate analysis in individuals with a vitamin D level of 30 ng/mL or greater found that the IRR of a positive COVID-19 test result was 0.97 (95% CI, 0.94-0.99; P = .008) per 1-ng/mL increase in vitamin D overall and 0.95 (95% CI, 0.91-0.98; P = .003) per 1-ng/mL increase in vitamin D in Black individuals.

    Conclusions and Relevance  In this single-center retrospective cohort study, COVID-19 risk increased among Black individuals with vitamin D level less than 40 ng/mL compared with those with 40 ng/mL or greater and decreased with increasing levels among individuals with levels greater than 30 ng/mL. No significant associations were noted for White individuals. Randomized clinical trials should examine whether increasing vitamin D level to greater than 40 ng/mL affects COVID-19 risk.

    Introduction

    Vitamin D has diverse physiological effects, including on calcium regulation, bone density, and immune function. Deficient levels, typically defined as 25-hydroxyvitamin D (also known as calcifediol) level less than 20 ng/mL (to convert to nanomoles per liter, multiply by 2.496), are common, especially in Black individuals, and some expert opinions define sufficient levels as 30 ng/mL or greater or 40 ng/mL or greater.1-3 Definitions of vitamin D deficiency and supplementation guidelines have been largely informed by effects on bone health from mostly White populations,4 and Black individuals have been reported to preserve bone health even with lower vitamin D levels,5 calling into question the notion that functional vitamin D deficiency is more common in Black individuals than White individuals. However, vitamin D is also important for immune function, and a meta-analysis of randomized clinical trials using daily or weekly dosing6 has found vitamin D supplementation was associated with substantially decreased viral respiratory infections, especially in individuals who were deficient in vitamin D, but also individuals without a deficient level. There is no evidence that immune function is better preserved with low vitamin D levels in Black individuals than in White individuals. This is especially important currently because deficient vitamin D levels have been associated with increased COVID-19 incidence and worse outcomes, especially in Black, Hispanic, and other non-White populations, who have also borne a disproportionate burden of COVID-19.7-14 Evidence on whether vitamin D levels above the deficient range are associated with COVID-19 risk, and whether such associations differ between White individuals and individuals of other races, could inform the design of randomized clinical trials to test whether vitamin D supplementation reduces COVID-19 risk15 and clinical decision-making before completion of such trials.16

    We used data from the electronic health record at the University of Chicago Medicine (UCM) in Chicago, Illinois, to examine whether the likelihood of a positive COVID-19 test was associated a person’s most recent vitamin D level within 365 days before COVID-19 testing, and whether such associations were present for White individuals and for Black individuals. Because more distant level may be less informative of level at the time of COVID-19 testing, analyses controlled for the timing of the most recent measurement and changes in vitamin D treatment after that level and before COVID-19 testing.

    Methods

    This cohort study was approved by the University of Chicago Biological Sciences Division institutional review board, which granted a waiver of participant consent because the data were deidentified except for elements of dates, and the institutional review board determined that the risk to the privacy of participants was minimal. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.

    Participants

    We obtained data for all 106 635 individuals tested for COVID-19 using polymerase chain reaction testing at UCM from March 3, 2020, to December 30, 2020. This sample included approximately 4314 persons tested for COVID-19 at UCM from March 3, 2020 to April 10, 2020, on whom we reported in a prior study of vitamin D deficiency and COVID-19 risk.10 We obtained electronic health record data for demographic, comorbidity, laboratory, and medication information within 365 days before participants’ first COVID-19 test. Vitamin D levels and treatments within 14 days of COVID-19 testing were excluded from analyses to avoid confounding by potential early manifestations of COVID-19, for example, presenting for care with symptoms that could lead to vitamin D testing and treatment. A total of 491 individuals were excluded because the only measured vitamin D levels or treatments were within 14 days of COVID-19 testing.

    Measurements

    All variables were obtained from the UCM electronic health record, Epic (Epic Systems). COVID-19 status was determined by the Centers for Disease Control and Prevention17 or Viacor18 polymerase chain reaction(PCR) test used until in-house testing with the Cobas SARS-CoV-2 RT-PCR (Roche) began March 15, 2020.19 Testing initially focused on individuals presenting with potential COVID-19 symptoms admitted to the hospital or health care workers with COVID-19 symptoms and exposure and was gradually broadened. Individuals’ most recent vitamin D level within 365 days before their first COVID-19 test was used to classify their level as less than 20 ng/mL (ie, deficient), 20 to less than 30 ng/mL (ie, insufficient), 30 to less than 40 ng/mL, or 40 ng/mL or greater. An additional analysis controlled for months from most recent vitamin D level measurement to 14 days before COVID-19 test result. The vitamin D dosing regimen was defined by the type of vitamin D (D2, D3, or calcitriol) and mean daily dose of vitamin D3. Changes in individuals’ recorded vitamin D treatment between the date of their most recent vitamin D level measurement and 14 days before COVID-19 testing were categorized as increased, unchanged, or decreased based on movement between the following regimens, from highest to lowest: 4001 IU/d or more vitamin D3, 2001 to 4000 IU/d vitamin D3, 1001 to 2000 IU/d vitamin D3, D2, 1 to 1000 IU/d vitamin D3 or a daily multivitamin, and no vitamin D.

    Age, sex, race (classified as White, Black, or other), and Hispanic/Latino ethnicity were collected based on participant self-selection among institutionally selected options. We used the most recent data from 365 to 14 days before COVID-19 testing to calculate body mass index. The Elixhauser comorbidity clusters potentially related to COVID-19 or vitamin D status were calculated using International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM)20 codes during 2 years, including hypertension, immunosuppression, diabetes, renal failure, chronic pulmonary disease, depression, liver disease, any psychosis, any pulmonary circulation disorder, and dementia.21

    Statistical Analysis

    Descriptive statistics were reviewed and compared using χ2 test or analysis of variance, including rates of positive test results for COVID-19, stratified by vitamin D level. Multivariable analyses were performed for the full population and separately for White and Black individuals. A multivariable generalized linear model with Poisson residuals and logit link function22,23 was estimated with the aforementioned covariates except months from last measured vitamin D level to 14 days before COVID-19 testing. Another model controlled for months from the last measured vitamin D level to 14 days before COVID-19 testing and interaction of the months variable with the last measured vitamin D level. This specification allows the association of vitamin D level with the outcome to vary by time since COVID-19 testing. The coefficient on the level variable (main term) indicates the estimated association with the vitamin D level 14 days before COVID-19 testing. Estimated probabilities of a positive COVID-19 test result were compared across level categories using bootstrapped bias-corrected 95% CIs to adjust for potential nonnormality in the distribution of the estimated probabilities. Additional analyses examined individuals with a vitamin D level of 30 ng/mL or greater, including analysis of vitamin D as a continuous variable. Multivariable model selection was informed by information criteria (Akaike or Bayesian), examination of residuals vs estimated values, and goodness-of-link and Hosmer-Lemeshow tests,24,25 resulting in inclusion of interactions of age with month of testing. We also examined rates of positive test results stratified by most recent vitamin D level and recorded vitamin D3 dosing (including none) 14 days before COVID-19 testing after excluding individuals using vitamin D2 or 1,25-dihydroxyvitamin D supplements at that time. Sensitivity analyses repeated multivariable analyses in persons with last vitamin D level measurement within 90 days of COVID-19 testing, stratified by whether persons were recorded as using vitamin D supplements at 14 days before COVID testing, and combining all non-White individuals, including Black, Mideast Indian, Asian, and other or multiple races. Statistical analysis was performed in Stata statistical software (StataCorp). P values were 2-sided, and statistical significance was set at P = .05. Data were analyzed from September 11, 2020, to February 5, 2021.

    Results
    Characteristics of the Sample

    Of 106 635 individuals tested for COVID-19 at UCM from March 3, 2020, to December 30, 2020, 4941 individuals had data on vitamin D level from within 365 days before COVID-19 testing, and 4638 individuals had complete data and were included in our analytic sample. The mean (SD) age was 52.8 (19.5) years, 3205 (69%) were women, and 1999 individuals (43%) were White, 2288 individuals (49%) were Black, and 351 (8%) individuals were another race (Table 1). Mean (SD) time since last measured vitamin D level was 167 (103) days and did not vary significantly by vitamin D level. The most recent vitamin D level was less than 20 ng/mL for 1251 individuals (27%), 20 to less than 30 ng/mL for 1267 individuals (27%), 30 to less than 40 ng/mL for 1023 individuals (22%), and greater than 40 ng/mL for 1097 individuals (24%).

    Table 1 also provides descriptive statistics for demographic and comorbidity measures, vitamin D deficiency, treatments, and rates of positive COVID-19 test results stratified by categories of vitamin D level. Vitamin D levels were lower among individuals who were younger, male, Black, Hispanic, University of Chicago employees, tested for COVID-19 earlier in the calendar year, or obese and among individuals who had a treatment dose increase after the most recent vitamin D level measurement, chronic pulmonary disease, or depression. Recorded vitamin D treatments 14 days before COVID-19 testing were similar across vitamin D categories, except that individuals who were deficient in vitamin D were recorded more frequently using vitamin D2 and less frequently using vitamin D3 at a dosage of 1000 IU/d or less, and individuals with vitamin D level 40 ng/mL or greater were more often recorded as using 4001 IU/d or more vitamin D3.

    Follow-up and Outcomes

    Overall, 333 individuals (7%) had test results positive for COVID-19, including 118 of 1251 individuals (9%) with vitamin D level less than 20 ng/mL (P < .001 vs individuals with vitamin D level ≥40 ng/mL), 76 of 1267 individuals (6%) with a vitamin D level of 20 to less than 30 ng/mL (P = .65 vs individuals with vitamin D level ≥40 ng/mL), 78 of 1023 individuals (8%) with a vitamin D level of 30 to less than 40 ng/mL (P = .06 vs individuals with vitamin D level ≥40 ng/mL), and 61 of 1097 individuals (6%) with vitamin D level 40 ng/mL or higher (P < .001 overall). Among Black individuals, 211 (9%) had positive COVID-19 test results, compared with 102 White individuals (5%) (P < .001). The eFigure in the Supplement reports COVID-19 positivity rates by race and vitamin D level.

    Table 2 and Table 3 report the multivariable analysis of COVID-19 testing results for the full sample and in Black individuals. In model 1, the model not controlling for time since the most recent vitamin D level measurement for the full sample, the incidence rate ratio (IRR) of having test results positive for COVID-19 was 1.41 (95% CI, 1.04-1.92; P = .03) for individuals with deficient vitamin D level, 0.95 (95% CI, 0.69-1.32; P = .78) for individuals with insufficient vitamin D level, and 1.27 (95% CI, 0.93-1.75; P = .14) for individuals with a vitamin D level of 20 to less than 30 ng/mL, compared with individuals with a vitamin D level of 40 ng/mL or greater. In model 2, the model controlling for time since last vitamin D level to estimate the association of vitamin D levels 14 days before COVID-19 testing with COVID-19 test results, the IRR of positive results for COVID-19 was 1.53 (95% CI, 0.91-2.56; P = .11) for individuals with vitamin D level less than 20 ng/mL, 1.21 (95% CI, 0.68-2.13; P = .52) for individuals with a vitamin D level of 20 to less than 30 ng/mL, and 1.42 (95% CI, 0.82-2.48; P = .21) for individuals with a vitamin D level of 30 to less than 40 ng/mL compared with individuals with a vitamin D level of 40 ng/mL or greater. Time since most recent measurement of vitamin D level was not associated with the risk of having positive results for COVID-19. In the models limited to individuals with a vitamin D level of 30 ng/mL or greater, a vitamin D level of 40 ng/mL or greater was not significantly associated with increased risk (model 3), but a 1-ng/mL increase in vitamin D level was associated with a 3% decrease in risk of positive COVID-19 test results (model 4) (IRR, 0.97 [95% CI, 0.94-0.99]; P = .008). Having a positive test result for COVID-19 was also associated with dementia, and age interacted with testing in March or April in all models, Black race and testing in April or November interacted in models with all individuals, and employment at the University of Chicago and age interacted with testing in November in models limited to individuals with a vitamin D level of 30 ng/mL or greater. Having negative test results for COVID-19 was associated with immunosuppressed conditions and testing in June through September in all models, female sex in models with all individuals, and age in models limited to individuals with a vitamin D level of 30 ng/mL or greater.

    For Black individuals, the IRR of test results positive for COVID-19 in the model not controlling for time since the most recent measurement of vitamin D level was 1.34 (95% CI 0.88-2.05; P = .18) for individuals with vitamin D level less than 20 ng/mL, 0.97 (95% CI, 0.62-1.52; P = .90) for individuals with a vitamin D level of 20 to less than 30 ng/mL, and 1.57 (95% CI, 1.06-2.33; P = .02) for individuals with a vitamin D level of 30 to less than 40 ng/mL, compared with those with a vitamin D level of 40 ng/mL or greater. In the model controlling for time since last vitamin D level measurement to estimate the association of vitamin D level 14 days before COVID-19 testing with test results, the IRR was 2.64 (95% CI, 1.24-5.66; P = .01) for individuals with vitamin D level less than 20 ng/mL, 1.69 (95% CI, 0.75-3.84; P = .21) for individuals with a vitamin D level of 20 to less than 30 ng/mL, and 2.55 (95% CI, 1.26-5.15; P = .009) for individuals with a vitamin D level of 30 to less than 40 ng/mL, compared with individuals with a vitamin D level of 40 ng/mL or greater. Time since most recent measurement of vitamin D level was significantly associated with increased risk of positive test results. However, for individuals with vitamin D level less than 40 ng/mL, this was offset by a decreased risk of positive test results with increased time since the vitamin D level was measured. In the models limited to individuals with a vitamin D level of 30 ng/mL or greater, the IRR was 2.49 (95% CI, 1.17-5.28; P = .02) for individuals with a vitamin D level of 40 ng/mL, and a 1-ng/mL increase in vitamin D level decreased the risk of positive test results by 5% (IRR, 0.95 [95% CI, 0.91-0.98]; P = .003). None of these analyses were statistically significant for White individuals, although 95% CIs were too wide to exclude the presence of the associations seen for Black individuals (eTable 1 in the Supplement). Notably, White individuals who decreased their vitamin D treatment between the time of vitamin D measurement and 14 days before COVID-19 testing were at increased risk of testing positive for COVID-19 (model 1: IRR, 2.44 [95% CI, 1.01-5.90]; P = .05; model 2: IRR, 2.50 [95% CI, 1.03-6.09]; P = .04).

    The Figure depicts estimated COVID-19 positivity rates based on the multivariable models accounting for time since vitamin D testing stratified by most recent vitamin D level for the full population and separately for Black and White individuals. Estimated COVID-19 positivity rates were lowest for the highest vitamin D levels in Black individuals but not in White individuals. Estimated COVID-19 positivity rates in Black individuals were 9.72% (95% CI, 6.74%-13.41%) for those with a vitamin D level less than 20 ng/mL, 6.47% (95% CI, 3.33%-10.28%) for individuals with a vitamin D level of 20 to less than 30 ng/mL, 10.10% (95% CI, 6.00%-15.47%) for individuals with a vitamin D level of 30 to less than 40 ng/mL, and 3.82% (95% CI, 1.78%-6.68%) for individuals with a vitamin D level of 40 ng/mL or greater.

    Given the overall trend for a vitamin D level of 20 to less than 30 ng/mL to be less strongly associated with COVID-19 risk than a level of 30 to less than 40 ng/mL, we tested whether treatments were more likely to be increased if levels were lower. Indeed, for individuals reported as using vitamin D3 supplements on the date of their most recently measured vitamin D level, only 5 of 188 individuals (3%) with a vitamin D level of 30 to less than 40 ng/mL had a dose increase within 60 days of vitamin D measurement vs 15 of 189 individuals (8%) with a vitamin D level of 20 to less than 30 ng/mL (P = .04).

    Results of Sensitivity Analysis

    eTable 2 in the Supplement reports rates of positive COVID-19 test results stratified by most recent vitamin D level and days since that level was measured. The sensitivity analysis reinforces our main finding that COVID-19 positivity rates were lowest for individuals whose vitamin D level was measured most recently and was 40 ng/mL or greater and highlights that positivity rates increased for individuals whose vitamin D level was measured most recently or 1 year prior and was found to be less than 20 ng/mL. Stratifying this analysis by race, this outcome was evident only among Black individuals. eTable 3 and eTable 4 in the Supplement report results for the full sample stratified by whether the person was using a vitamin D supplement 14 days before COVID-19 testing. Vitamin D levels were associated with positive test results only in individuals using supplements, and Hispanic ethnicity was associated with increased risk only in individuals not using supplements.

    Table 4 reports rates of positive COVID-19 test results stratified by most recent vitamin D level and vitamin D3 dose 14 days before COVID-19 testing. COVID-19 positivity rates remained lowest for individuals with a vitamin D level of 40 ng/mL or greater. Among 1046 individuals with a vitamin D level of 40 or greater, 788 individuals (75%) did not use vitamin D supplements. Overall, only 920 of 4258 individuals (22%) used vitamin D3 supplements, 304 individuals (7%) used more than 1000 IU/d vitamin D3, and 131 individuals (3%) used more than 2000 IU/d. Among 131 individuals using 2001 IU/d vitamin D3 or more, 4 of 24 individuals (17%) with vitamin D level less than 20 ng/mL had test results positive for COVID-19, while 2 of 107 individuals (2%) with a vitamin D level of 20 ng/mL or higher had test results positive for COVID-19 (P = .01). Among persons with a vitamin D level of 20 or greater, 2 of 107 individuals (2%) using 2001 IU/d or more vitamin D3 had test results positive for COVID-19, compared with 197 of 3075 individuals (6%) using no vitamin D or lower doses having test results positive for COVID-19 (P = .06). Among 24 individuals with vitamin D level less than 20 ng/mL who reported using 2001 IU/d or more vitamin D3, 4 (17%) had test results positive for COVID-19; while among 1052 individuals who reported using 2000 IU/d or less vitamin D3, 96 (9%) had test results positive for COVID-19 (P = .27).

    As only 239 individuals (5%) in total had a vitamin D level of 60 ng/mL or greater, we do not report multivariable analysis with this category. However, among 858 individuals with a vitamin D level of 40 to less than 60 ng/mL, 56 (7%) had test results positive for COVID-19, and among 239 individuals with a vitamin D level of 60 ng/mL or greater, 5 (2%) had test results positive for COVID-19 (P = .008). Of 11 individuals with a vitamin D level of 70 ng/mL or higher, none had positive COVID-19 test results, compared with 5 individuals (4%) with positive COVID-19 test results among 128 individuals with a vitamin D level of 60 to less than 70 ng/mL (P = .04).

    Discussion

    This findings of this cohort study reinforce prior findings by ourselves and others11-13 that lower vitamin D levels (eg, <20 ng/mL) are associated with increased risk of having test results positive for COVID-19 and provides evidence that COVID-19 risk is also increased for Black individuals with a vitamin D level of 30 to less than 40 ng/mL compared with individuals with a vitamin D level of 40 ng/mL or greater. Risk of positive COVID-19 test results decreased significantly with increased vitamin D level of 30 ng/mL or greater when measured as a continuous variable. Our findings that controlling for time since last vitamin D level measurement increased the reduction in the odds of positive test results with increased vitamin D levels and that the risk of positive test results with lower vitamin D levels increased when those levels were more recent relative to the timing of COVID-19 testing support the idea that vitamin D level at the time of testing is most strongly associated with COVID-19 risk. Our finding that COVID-19 positivity rates were increased among individuals with a vitamin D level of less than 20 ng/mL measured 275 to 365 days before COVID-testing may reflect seasonal vitamin D variation in this group.

    These findings add to the evolving literature demonstrating associations between vitamin D levels and COVID-19 risks and outcomes. As noted, a meta-analysis by Martineau et al6 reported vitamin D supplementation was associated with reducing other viral respiratory infections, even in people with vitamin D levels that were sufficient by current standards based largely on needs for bone health. All-cause mortality and other outcomes may be better with vitamin D levels of 40 to 60 ng/mL, at least in some racial groups.3,26,27 Our findings that vitamin D levels measured a longer period from COVID-19 testing were less associated with COVID-19 risk and that the association of vitamin D levels with risk were strongest in Black individuals may help explain why a 2020 study by Hastie et al28 using the UK Biobank did not find a statistically significant association between vitamin D levels 10 to 14 years before the COVID-19 testing and the risk of having positive test results.

    Our finding that vitamin D dose increases were more likely when the level was 20 to less than 30 ng/mL compared with a level of 30 to less than 40 ng/mL may explain the otherwise surprising result that a level of 20 to less than 30 ng/mL was less strongly associated with increased COVID-19 risk than a level of 30 to less than 40 ng/mL. This possible explanation would be particularly important if treatment is not fully captured in the electronic health record, which is likely given the availability of vitamin D over the counter.

    The significant association of vitamin D levels with COVID-19 risk in Black individuals that was not found in White individuals could reflect their higher COVID-19 risk, to which socioeconomic factors and structural inequities clearly contribute.29 Biological susceptibility to vitamin D deficiency may also be less frequent in White than Black individuals, since lighter skin increases vitamin D production in response to sunlight30 and vitamin D binding proteins may vary by race and affect vitamin D bioavailability.31,32 Many persons in our other race category, consisting larging of Asian or Mideast Indian individuals and individuals with multiple races, come from racial groups with higher incidence of vitamin D deficiency, but this group was too small to support multivariable analysis in our sample. Nevertheless, Hispanic ethnicity was associated with increased risk in some multivariable analyses (eTable 3 in the Supplement). We note also that our analysis for White individuals lacked statistical power to exclude large associations of vitamin D levels with COVID-19 risk.

    That vitamin D levels were associated with COVID-19 positivity rates, especially in individuals using vitamin D supplements, supports the idea that supplementation might decrease COVID-19 risk by increasing vitamin D levels. Our finding of lower rates of positive COVID-19 test results in individuals with vitamin D levels of 20 ng/mL or greater using at least 2001 IU/d vitamin D3 supplements compared with individuals with levels of 20 ng/mL or greater using less than 2001 IU/d vitamin D3 supplements could reflect the effect of those higher doses on levels and therefore COVID-19 risk. The findings of increased risk for White individuals who decreased treatment after their vitamin D levels were tested also supports the potential effectiveness of vitamin D supplementation. It has also been hypothesized that free vitamin D levels may be more important to immune function than vitamin D levels and that free levels may be more affected by vitamin D intake or synthesis in some individuals since vitamin D may not be easily mobilized from vitamin D binding protein in some circumstances.33 This may explain our findings that vitamin D supplement use independent of vitamin D levels or potential proxies for unmeasured supplementation (eg, crossing into the 20 to <30 ng/mL insufficient range) are associated with decreased COVID-19 risk, and published findings that more frequent vitamin D dosing is more effective than intermittent dosing in improving immune function.34

    These results increase the urgency to consider whether increased sun exposure or vitamin D supplementation could reduce COVID-19 risk. It is likely that less than 5% to 10% of US adults currently have vitamin D levels greater than 40 ng/mL.13,35 When increased sun exposure is impractical, achieving vitamin D levels of 40 ng/mL or greater typically requires greater supplementation than currently recommended for most individuals of 600-800 IU/d vitamin D3. The National Academy of Medicine’s highest recommended dose of 4000 IU/d vitamin D3 would substantially increase the fraction of people eventually reaching such levels, but more than 25% of individuals might not reach such levels,36 especially if they are obese or vitamin D deficient.37 Sustained use of 10 000 IU/d vitamin D3 would likely decrease the fraction of people with vitamin D levels less than 40 ng/mL to less than 5%. Such doses may increase the risk of hypercalcemia, but the risk of hypercalcemia with 10 000 IU/d vitamin D3 is low,36,38 can be mitigated by monitoring, and should be weighed against potential benefits. These arguments suggest that randomized clinical trials to understand if vitamin D can reduce COVID-19 risk should include doses likely to increase vitamin D to at least 40 ng/mL. Although limited by sample size that precludes multivariable analysis, our findings of lower COVID-19 positivity rates for individuals with vitamin D levels of 60 ng/mL or higher compared with lower levels suggest that randomized clinical trials are needed to study interventions to increase vitamin D levels to greater than 60 or even 70 ng/mL, which would increase the importance of considering safety concerns. Studies should also consider the role of vitamin D testing, loading doses, dose adjustments for individuals who are obese or overweight, risks for hypercalcemia, and strategies to monitor for and mitigate hypercalcemia, and that non-White populations, such as Black individuals, may have greater needs for supplementation, especially given their greater rates of vitamin D deficiency and disproportionate burden of COVID-19 morbidity and mortality. Cost and the potential for rapid scalability may also be important concerns, especially since vitamin D may cost less than $10 per year while screening may cost more than $50 and carry COVID-19 exposure risk for individuals not otherwise undergoing phlebotomy or unable to complete home testing. Since Medicare and Medicaid often do not cover the costs of screening for vitamin D deficiency, the increased rates of vitamin D deficiency in older adults and in racial minority groups in the United States, especially Black and Hispanic persons,1 and increased rates of economic disadvantage and COVID-19 in these groups39 suggest the need to reexamine these payment policies in the current context.

    While the increasing availability of COVID-19 vaccines is likely to reduce the spread of COVID-19 and hence any potential benefits of vitamin D supplementation, the presence of new strains resistant to existing vaccines may increase the potential benefits of vitamin D supplementation, and vitamin D supplementation may be useful in populations not receiving the vaccine. Weakened host responses may also enhance the conditions for development of viral mutation,40 so supplementation might have benefits at the population level in reducing the risk of mutant strains.

    Limitations

    This study has limitations. Despite basic scientific evidence of effects of vitamin D on innate and adaptive immunity, immunomodulation, and thrombotic regulation3,41,42 and the clinical trial evidence of effects of vitamin D supplementation on viral respiratory tract infections, the observed associations could be affected by omitted confounders so that they may not reflect causal effects of vitamin D deficiency on COVID-19 risk. However, the results are robust to including demographic characteristics and comorbidity indicators that have either physiological reasons for consideration or have been suggested to be associated with COVID-19 outcomes. Since our data are limited to the UCM electronic health record, patterns of vitamin D screening, treatment, or COVID-19 testing might have somehow selected for participants in a manner that induced an association between observed vitamin D status and testing positive for COVID-19. However, studies in many different clinical settings have now reported low vitamin D levels to be associated with increased COVID-19 risk.6-8,10 Our analyses are also limited by the possibility that vitamin D supplement use recorded in the electronic health record might not capture nonprescription vitamin D supplements and that individuals may not use supplements as reported in the electronic health record. Our analysis of the association of dosing with rates of COVID-19 positivity was also limited to individuals not using vitamin D supplements or vitamin D3 alone, excluding individuals using vitamin D2 or calcitriol, which are both frequently given in subgroups (eg, individuals with chronic kidney disease, hypoparathyroidism) deserving of separate analysis. Our study population from a northern city contained many Black individuals, older adults with chronic illness, and health care workers, all of which are risk factors for vitamin D deficiency1,43-47 and hence may not represent other populations. Related to this, larger sample sizes may be needed to establish if vitamin D levels affects COVID-19 risk in populations with lower risks of vitamin D deficiency or COVID-19, including White individuals and those in warmer climates where sun-related vitamin D production may be greater.

    Conclusions

    The findings of this cohort study support a role of vitamin D levels in COVID-19 risk. Randomized clinical trials of interventions to raise vitamin D levels into ranges at or above levels currently considered sufficient are needed to determine if those interventions could reduce COVID-19 incidence, perhaps especially in Black and other populations known to be at increased risk of vitamin D deficiency. Because such levels exceed levels recommended for other reasons, individual and policy decisions about higher supplement dosing and vitamin D testing to achieve such levels should be even more carefully considered than dosing to avoid vitamin D deficiency as currently defined, which some have argued should be pursued given current evidence that vitamin D might reduce the risk of COVID-19.48,49

    Back to top
    Article Information

    Accepted for Publication: February 10, 2021.

    Published: March 19, 2021. doi:10.1001/jamanetworkopen.2021.4117

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Meltzer DO et al. JAMA Network Open.

    Corresponding Author: David O. Meltzer, MD, PhD, University of Chicago, 5841 S Maryland, MC 5000, Chicago, IL 60637 (dmeltzer@medicine.bsd.uchicago.edu).

    Author Contributions: Drs Best and Zhang had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Meltzer, Arora.

    Acquisition, analysis, or interpretation of data: Meltzer, Best, Zhang, Vokes, Solway.

    Drafting of the manuscript: Meltzer.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Meltzer, Best, Zhang.

    Obtained funding: Meltzer.

    Administrative, technical, or material support: Meltzer.

    Supervision: Meltzer.

    Conflict of Interest Disclosures: None reported.

    Funding/Support: Drs Meltzer, Best and Zhang were supported by the Learning Health Care System Core of the University of Chicago/Rush University Institute for Translational Medicine (ITM) Clinical and Translational Science Award (ITM 2.0: Advancing Translational Science in Metropolitan Chicago; grant No. UL1TR002389, principal investigator: Dr Solway) and the African American Cardiovascular pharmacogenetic Consortium (grant No. U54-MD010723, Drs Perera and Meltzer).

    Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Additional Contributions: Data from this study were provided by the Clinical Research Data Warehouse (CRDW) maintained by the Center for Research Informatics (CRI) at University of Chicago. The CRI is funded by Clinical and Translational Sciences Award from the Biological Sciences Division of the Institute for Translational Medicine. Thomas Sutton, MS, and Julie Johnson, PhD, MPH, RN (UChicago CRI), appended additional data to the CRDW for this analysis; they were not compensated for this contribution.

    References
    1.
    Forrest  KY, Stuhldreher  WL.  Prevalence and correlates of vitamin D deficiency in US adults.   Nutr Res. 2011;31(1):48-54. doi:10.1016/j.nutres.2010.12.001 PubMedGoogle ScholarCrossref
    2.
    Holick  MF, Binkley  NC, Bischoff-Ferrari  HA,  et al; Endocrine Society.  Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline.   J Clin Endocrinol Metab. 2011;96(7):1911-1930. doi:10.1210/jc.2011-0385 PubMedGoogle ScholarCrossref
    3.
    Charoenngam  N, Holick  MF.  Immunologic effects of vitamin D on human health and disease.   Nutrients. 2020;12(7):2097. doi:10.3390/nu12072097 PubMedGoogle ScholarCrossref
    4.
    Del Valle  HB, Yaktine  AL, Taylor  CL, Ross  AC, eds.  Dietary Reference Intakes for Calcium and Vitamin D. National Academies Press; 2011.
    5.
    Brown  LL, Cohen  B, Tabor  D, Zappalà  G, Maruvada  P, Coates  PM.  The vitamin D paradox in Black Americans: a systems-based approach to investigating clinical practice, research, and public health—expert panel meeting report.   BMC Proc. 2018;12(suppl 6):6. doi:10.1186/s12919-018-0102-4 PubMedGoogle ScholarCrossref
    6.
    Martineau  AR, Jolliffe  DA, Hooper  RL,  et al.  Vitamin D supplementation to prevent acute respiratory tract infections: systematic review and meta-analysis of individual participant data.   BMJ. 2017;356:i6583. doi:10.1136/bmj.i6583 PubMedGoogle ScholarCrossref
    7.
    Benskin  LL.  A basic review of the preliminary evidence that COVID-19 risk and severity is increased in vitamin D deficiency.   Front Public Health. 2020;8:513. doi:10.3389/fpubh.2020.00513 PubMedGoogle ScholarCrossref
    8.
    Maghbooli  Z, Sahraian  MA, Ebrahimi  M,  et al.  Vitamin D sufficiency, a serum 25-hydroxyvitamin D at least 30 ng/mL reduced risk for adverse clinical outcomes in patients with COVID-19 infection.   PLoS One. 2020;15(9):e0239799. doi:10.1371/journal.pone.0239799 PubMedGoogle Scholar
    9.
    Baktash  V, Hosack  T, Patel  N,  et al.  Vitamin D status and outcomes for hospitalised older patients with COVID-19.   Postgrad Med J. Published online August 27, 2020. doi:10.1136/postgradmedj-2020-138712 PubMedGoogle Scholar
    10.
    Mardani  R, Alamdary  A, Mousavi Nasab  SD, Gholami  R, Ahmadi  N, Gholami  A.  Association of vitamin D with the modulation of the disease severity in COVID-19.   Virus Res. 2020;289:198148. doi:10.1016/j.virusres.2020.198148 PubMedGoogle Scholar
    11.
    Meltzer  DO, Best  TJ, Zhang  H, Vokes  T, Arora  V, Solway  J.  Association of vitamin D status and other clinical characteristics with COVID-19 test results.   JAMA Netw Open. 2020;3(9):e2019722. doi:10.1001/jamanetworkopen.2020.19722 PubMedGoogle Scholar
    12.
    Merzon  E, Tworowski  D, Gorohovski  A,  et al.  Low plasma 25(OH) vitamin D level is associated with increased risk of COVID-19 infection: an Israeli population-based study.   FEBS J. 2020;287(17):3693-3702. doi:10.1111/febs.15495 PubMedGoogle ScholarCrossref
    13.
    Kaufman  HW, Niles  JK, Kroll  MH, Bi  C, Holick  MF.  SARS-CoV-2 positivity rates associated with circulating 25-hydroxyvitamin D levels.   PLoS One. 2020;15(9):e0239252. doi:10.1371/journal.pone.0239252 PubMedGoogle Scholar
    14.
    Israel  A, Assi Cicurel  AA, Feldhamer  I, Dror  Y, Giveon  SM, Gillis  D, Strich  D, Lavie  G.  The link between vitamin D deficiency and COVID-19 in a large population   medRxiv. doi:10.1101/2020.09.04.20188268Google Scholar
    15.
    Grant  WB, Lahore  H, McDonnell  SL,  et al.  Evidence that vitamin D supplementation could reduce risk of influenza and COVID-19 infections and deaths.   Nutrients. 2020;12(4):988. doi:10.3390/nu12040988 PubMedGoogle ScholarCrossref
    16.
    Boucher  BJ.  Vitamin D status as a predictor of COVID-19 risk in Black, Asian and other ethnic minority groups in the UK.   Diabetes Metab Res Rev. 2020;36(8):e3375. doi:10.1002/dmrr.3375 PubMedGoogle Scholar
    17.
    Centers for Disease Control and Prevention. CDC Diagnostic Tests for COVID-19. Accessed July 13, 2020. https://www.cdc.gov/coronavirus/2019-ncov/lab/testing.html
    18.
    Eurofins Viracor. Coronavirus (COVID-19) SARS-CoV-2 RT-PCR. Accessed July 13, 2020. https://www.viracor-eurofins.com/test-menu/8300-coronavirus-covid-19-sars-cov-2-rt-pcr/
    19.
    Roche Diagnostics. Cobas SARS-CoV-2 Test (for the COVID-19 coronavirus). Accessed July 13 2020. https://diagnostics.roche.com/us/en/products/params/cobas-sars-cov-2-test.html
    20.
    Centers for Disease Control and Prevention. International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM). Accessed February 15, 2021. https://www.cdc.gov/nchs/icd/icd10cm.htm
    21.
    Zhang  JX, Iwashyna  TJ, Christakis  NA.  The performance of different lookback periods and sources of information for Charlson comorbidity adjustment in Medicare claims.   Med Care. 1999;37(11):1128-1139. doi:10.1097/00005650-199911000-00005 PubMedGoogle ScholarCrossref
    22.
    McCullagh  P,, Nelder  J.  Generalized Linear Models. 2nd ed. Chapman and Hall/CRC; 1989. doi:10.1007/978-1-4899-3242-6
    23.
    McNutt  LA, Wu  C, Xue  X, Hafner  JP.  Estimating the relative risk in cohort studies and clinical trials of common outcomes.   Am J Epidemiol. 2003;157(10):940-943. doi:10.1093/aje/kwg074 PubMedGoogle ScholarCrossref
    24.
    Pregibon  D. Data Analytic Methods for Generalized Linear Models. Dissertation. University of Toronto; 1979.
    25.
    Blizzard  L, Hosmer  DW.  Parameter estimation and goodness-of-fit in log binomial regression.   Biom J. 2006;48(1):5-22. doi:10.1002/bimj.200410165 PubMedGoogle ScholarCrossref
    26.
    Zhang  Y, Fang  F, Tang  J,  et al.  Association between vitamin D supplementation and mortality: systematic review and meta-analysis.   BMJ. 2019;366:l4673. doi:10.1136/bmj.l4673 PubMedGoogle ScholarCrossref
    27.
    Freedman  BI, Register  TC.  Effect of race and genetics on vitamin D metabolism, bone and vascular health.   Nat Rev Nephrol. 2012;8(8):459-466. doi:10.1038/nrneph.2012.112 PubMedGoogle ScholarCrossref
    28.
    Hastie  CE, Mackay  DF, Ho  F,  et al.  Vitamin D concentrations and COVID-19 infection in UK Biobank.   Diabetes Metab Syndr. 2020;14(4):561-565. doi:10.1016/j.dsx.2020.04.050 PubMedGoogle ScholarCrossref
    29.
    Yancy  CW.  COVID-19 and African Americans.   JAMA. 2020;323(19):1891-1892. doi:10.1001/jama.2020.6548 PubMedGoogle ScholarCrossref
    30.
    Libon  F, Cavalier  E, Nikkels  AF.  Skin color is relevant to vitamin D synthesis.   Dermatology. 2013;227(3):250-254. doi:10.1159/000354750 PubMedGoogle ScholarCrossref
    31.
    Bouillon  R, Schuit  F, Antonio  L, Rastinejad  F.  Vitamin D binding protein: a historic overview.   Front Endocrinol (Lausanne). 2020;10:910. doi:10.3389/fendo.2019.00910 PubMedGoogle ScholarCrossref
    32.
    Oleröd  G, Hultén  LM, Hammarsten  O, Klingberg  E.  The variation in free 25-hydroxy vitamin D and vitamin D-binding protein with season and vitamin D status.   Endocr Connect. 2017;6(2):111-120. doi:10.1530/EC-16-0078 PubMedGoogle ScholarCrossref
    33.
    Bikle  DD, Schwartz  J. Vitamin  D  Binding protein, total and free vitamin D levels in different physiological and pathophysiological conditions front.   Endocrinol. 2019;10:317. doi:10.3389/fendo.2019.00317Google Scholar
    34.
    Bergman  P, Lindh  AU, Björkhem-Bergman  L, Lindh  JD.  Vitamin D and respiratory tract infections: a systematic review and meta-analysis of randomized controlled trials.   PLoS One. 2013;8(6):e65835. doi:10.1371/journal.pone.0065835 PubMedGoogle Scholar
    35.
    Looker  AC, Johnson  CL, Lacher  DA,  et al.  Vitamin D Status: United States 2001–2006. NCHS Data Brief, No 59. National Center for Health Statistics. 2011.
    36.
    McCullough  PJ, Lehrer  DS, Amend  J.  Daily oral dosing of vitamin D3 using 5000 to 50,000 international units a day in long-term hospitalized patients: insights from a seven year experience.   J Steroid Biochem Mol Biol. 2019;189:228-239. doi:10.1016/j.jsbmb.2018.12.010 PubMedGoogle ScholarCrossref
    37.
    Dhaliwal  R, Mikhail  M, Feuerman  M, Aloia  JF.  The vitamin d dose response in obesity.   Endocr Pract. 2014;20(12):1258-1264. doi:10.4158/EP13518.OR PubMedGoogle ScholarCrossref
    38.
    Billington  EO, Burt  LA, Rose  MS,  et al.  Safety of high-dose vitamin D supplementation: secondary analysis of a randomized controlled trial.   J Clin Endocrinol Metab. 2020;105(4):dgz212. doi:10.1210/clinem/dgz212 PubMedGoogle Scholar
    39.
    Garg  S, Kim  L, Whitaker  M,  et al.  Hospitalization rates and characteristics of patients hospitalized with laboratory-confirmed coronavirus disease 2019—COVID-NET, 14 states, March 1-30, 2020.   MMWR Morb Mortal Wkly Rep. 2020;69(15):458-464. doi:10.15585/mmwr.mm6915e3 PubMedGoogle ScholarCrossref
    40.
    Xue  KS, Moncla  LH, Bedford  T, Bloom  JD.  Within-host evolution of human influenza virus.   Trends Microbiol. 2018;26(9):781-793. doi:10.1016/j.tim.2018.02.007 PubMedGoogle ScholarCrossref
    41.
    Prietl  B, Treiber  G, Pieber  TR, Amrein  K.  Vitamin D and immune function.   Nutrients. 2013;5(7):2502-2521. doi:10.3390/nu5072502 PubMedGoogle ScholarCrossref
    42.
    Mohammad  S, Mishra  A, Ashraf  MZ.  Emerging role of vitamin D and its associated molecules in pathways related to pathogenesis of thrombosis.   Biomolecules. 2019;9(11):649. doi:10.3390/biom9110649 PubMedGoogle ScholarCrossref
    43.
    National Center for Immunization and Respiratory Diseases (NCIRD), Division of Viral Diseases. Preparing for COVID-19 in nursing homes. Accessed April 27, 2020. https://www.cdc.gov/coronavirus/2019-ncov/hcp/long-term-care.html
    44.
    CDC COVID-19 Response Team.  Characteristics of Health Care Personnel with COVID-19—United States, February 12-April 9, 2020.   MMWR Morb Mortal Wkly Rep. 2020;69(15):477-481. doi:10.15585/mmwr.mm6915e6 PubMedGoogle ScholarCrossref
    45.
    Huotari  A, Herzig  KH.  Vitamin D and living in northern latitudes—an endemic risk area for vitamin D deficiency.   Int J Circumpolar Health. 2008;67(2-3):164-178. doi:10.3402/ijch.v67i2-3.18258 PubMedGoogle ScholarCrossref
    46.
    Elliott  ME, Binkley  NC, Carnes  M,  et al.  Fracture risks for women in long-term care: high prevalence of calcaneal osteoporosis and hypovitaminosis D.   Pharmacotherapy. 2003;23(6):702-710. doi:10.1592/phco.23.6.702.32182 PubMedGoogle ScholarCrossref
    47.
    Sowah  D, Fan  X, Dennett  L, Hagtvedt  R, Straube  S.  Vitamin D levels and deficiency with different occupations: a systematic review.   BMC Public Health. 2017;17(1):519. doi:10.1186/s12889-017-4436-z PubMedGoogle ScholarCrossref
    48.
    Frieden  T. Former CDC Chief Dr. Tom Frieden: coronavirus infection risk may be reduced by vitamin D. Fox News. March 23, 2020. Accessed April 13, 2020. https://www.foxnews.com/opinion/former-cdc-chief-tom-frieden-coronavirus-risk-may-be-reduced-with-vitamin-d
    49.
    McCall  B. Medical societies advise on vitamin D in midst of COVID-19. Medscape Medical News. July 10, 2020. Accessed July 10, 2020. https://www.medscape.com/viewarticle/933715
    ×