Dr. Mark Gordon on covid-19: what the evidence says · JRE #1589
SUBJECT: COVID-19
The short answer
The speaker claims a study found 85% of people with vitamin D below 30 ng/mL got COVID-19 or its illness, versus 90% of people above 60 ng/mL having no symptoms. Gordon appears to be referencing the type of retrospective analysis published by Kaufman et al. (2020) in PLOS ONE, which examined over 190,000 US patients and found SARS-CoV-2 test positivity was inversely associated with circulating 25-hydroxyvitamin D levels.
Not a true/false call. Every claim is logged with its sources; read the exhibits below.
It's like 85% of the people who have low, less than 30, have it. And as you move up the scale to the higher level, greater than 60, you see 90% of the people have nothing.
What the evidence says 01 / RECORD
Gordon appears to be referencing the type of retrospective analysis published by Kaufman et al. (2020) in PLOS ONE, which examined over 190,000 US patients and found SARS-CoV-2 test positivity was inversely associated with circulating 25-hydroxyvitamin D levels. However, the actual figures in that study are far smaller than Gordon's stated 85%-vs-90% split: positivity was 12.5% among patients with deficient levels (below 20 ng/mL), 8.1% among those with adequate levels (30-34 ng/mL), and 5.9% among those with levels of 55 ng/mL or higher. No published study shows 85% of people with vitamin D below 30 ng/mL testing positive for or contracting COVID-19, nor 90% of people above 60 ng/mL being asymptomatic or unaffected. The vitamin D-COVID literature broadly shows a modest, observational, dose-dependent inverse association between vitamin D status and infection risk, not a threshold effect of the magnitude described. Because these are observational data, the association could partly reflect confounding (age, obesity, comorbidities, indoor lifestyle) rather than a direct causal effect of vitamin D itself, and no large randomized controlled trial has confirmed that vitamin D supplementation prevents COVID-19 infection at the rate implied. The specific 85%/90% figures as stated do not match any known published dataset and substantially overstate the strength of the real association.