Selenium: Toraih et al., 2026
A 5-year retrospective cohort study of 4,694 Hashimoto's thyroiditis patients found selenium supplementation was associated with worse thyroid antibody levels, higher rates of autoimmune comorbidities (particularly Sjögren's syndrome and psoriasis), and increased all-cause mortality compared to no supplementation — though the study could not measure baseline selenium levels, meaning it cannot distinguish between harm from supplementation versus confounding by indication.
Key takeaways
- Selenium-supplemented patients had 58% higher rates of new autoimmune diseases over 5 years (RR 1.33) and 38% higher all-cause mortality (RR 1.38) compared to matched controls.
- TPOAb levels — the antibody that attacks thyroid tissue in Hashimoto's — remained elevated at higher rates in the selenium group at 3 years (6.9% vs. 5.0%) and 5 years (7.9% vs. 5.3%).
- This is a large retrospective cohort study, not a randomized controlled trial — the selenium group may have been sicker at baseline in ways the propensity matching didn't capture.
- The study has no baseline selenium measurements, so it cannot determine whether patients were deficient, sufficient, or excessive at enrollment — a critical gap when evaluating a nutrient intervention.
- The findings conflict with smaller RCTs showing antibody reduction with selenium — the dose, formulation, and baseline selenium status likely matter.
The study
Toraih et al. published this retrospective cohort analysis in Endocrinology, Diabetes & Metabolism in 2026 (PubMed). They used TriNetX, a global federated health research network aggregating electronic health records from 92 healthcare organizations, to identify 2,347 matched pairs of Hashimoto's thyroiditis patients — one group supplementing selenium after diagnosis, one group not supplementing. Propensity score matching balanced baseline age, sex, BMI, comorbidities, and thyroid medications. The study followed outcomes for 5 years, tracking longitudinal TSH, free T4, and TPOAb levels, as well as incidence of new autoimmune diseases, thyroid cancer, need for thyroidectomy, and all-cause mortality. Selenium dose and formulation were not standardized — the study captured "selenium supplementation" as coded in medical records, regardless of type or amount.
How to read this study
What this paper does well: The sample size is large — 4,694 total patients followed for 5 years. Propensity score matching is the strongest tool available in observational research to simulate randomization by balancing known confounders like age, sex, BMI, diabetes, hypertension, thyroid medications, and prior autoimmune disease. The outcomes are concrete and clinically relevant — not just antibody levels but actual disease diagnoses, surgical interventions, and mortality. The 5-year follow-up window is long enough to capture chronic autoimmune progression and serious adverse events. The authors acknowledge the limitation openly: no baseline selenium measurements.
What this paper is missing: The absence of baseline selenium levels is the fatal flaw. Selenium behaves as a U-shaped curve — deficiency is harmful, sufficiency is protective, excess is toxic. Without knowing who was deficient, this study cannot distinguish between (1) selenium harming people who didn't need it, (2) selenium failing to help people who were already sufficient, or (3) confounding by indication — sicker patients being more likely to seek out selenium supplementation. Propensity matching can only balance variables the researchers measure. If selenium-supplementing patients had worse disease severity, higher symptom burden, or lower selenium status at baseline — and those weren't captured in the matching variables — the groups aren't truly comparable. The study also doesn't control for dose or formulation (selenomethionine vs. sodium selenite vs. selenocysteine), which prior RCTs suggest matters.
How I'd weight this paper: I treat this as a significant red flag worth investigating, not as definitive evidence that selenium is harmful. The mortality signal is large enough (38% increase) that it cannot be dismissed, but the mechanistic implausibility — selenium is required for thyroid peroxidase and deiodinase enzymes — suggests confounding. I weight this heavily as a caution against blanket selenium supplementation in non-deficient populations, and I weight it as weak evidence for selenium toxicity in the absence of dose and baseline status data. The conflict with prior RCTs means the truth likely depends on context: who benefits, at what dose, with what baseline selenium status.
What they found
At 3 years, 6.9% of selenium-supplemented patients had elevated TPOAb levels (>35 IU/mL) compared to 5.0% of controls (p = 0.01). At 5 years, the gap widened: 7.9% vs. 5.3% (p < 0.01). Mean TSH, free T4, and TPOAb levels were all persistently higher in the selenium group across the entire follow-up period, though the paper does not provide absolute values or confidence intervals for these means.
The incidence of new autoimmune diseases was 33% higher in the selenium group (RR 1.33, 95% CI 1.14-1.56). Sjögren's syndrome — an autoimmune attack on salivary and tear glands — was 61% more common (RR 1.61, 95% CI 1.03-2.52). Psoriasis was 169% more common (RR 2.69, 95% CI 1.52-4.76). Rates of thyroid cancer (RR 0.91, 95% CI 0.58-1.42) and thyroidectomy (RR 1.18, 95% CI 0.87-1.60) were statistically similar between groups.
All-cause mortality was 38% higher in the selenium group (RR 1.38, 95% CI 1.09-2.03). The paper does not report cause of death or whether mortality was driven by autoimmune complications, cardiovascular events, or other factors.
What it means for the average man
If you have Hashimoto's thyroiditis and normal selenium status, this study suggests supplementation may not help and could potentially worsen antibody levels and autoimmune disease risk. The mortality increase is concerning, but without baseline selenium measurements or cause-of-death data, it's impossible to know whether selenium itself is harmful or whether sicker patients were more likely to supplement.
The actionable takeaway: measure serum selenium before supplementing. Normal range is 70-150 mcg/L. If you're deficient, supplementation at 200 mcg/day of selenomethionine has shown antibody reduction in prior RCTs. If you're already sufficient, there's no evidence supplementation helps — and this study raises the possibility it could harm. Selenium toxicity (selenosis) begins above 400 mcg/day and presents as hair loss, nail brittleness, garlic breath, nausea, and neurological symptoms. The dose matters as much as the decision to supplement.
The caveats
This is a retrospective cohort study, not a randomized trial. Propensity matching cannot account for unmeasured confounders — particularly disease severity and baseline selenium status, both of which are likely correlated with the decision to supplement. The study captured "selenium supplementation" as a binary yes/no from medical records, with no data on dose, formulation, adherence, or duration of use. It's possible some patients took 50 mcg/day of selenomethionline while others took 400 mcg/day of sodium selenite — vastly different interventions.
Prior randomized controlled trials in Hashimoto's patients — including Turker (2006), Gartner (2002), and Nacamulli (2010) — found 200 mcg/day selenomethionine reduced TPOAb levels over 3-12 months. Those trials were smaller (30-70 patients) but better controlled. The conflict suggests context matters: baseline selenium status, dose, and formulation likely determine whether supplementation helps, does nothing, or harms.
Frequently asked questions
Should I trust a large observational study over smaller RCTs?
Not automatically. Large observational studies have statistical power, but randomized controlled trials have causal control. RCTs assign treatment randomly, eliminating confounding by indication — the possibility that sicker patients seek out the intervention. This study's size is a strength, but the lack of randomization and the absence of baseline selenium measurements are critical weaknesses. When observational studies conflict with RCTs, the disagreement usually points to unmeasured confounding. In this case, the most likely confounder is baseline selenium status — deficient patients may benefit, sufficient patients may not, and excessive intake may harm.
What is propensity score matching and does it make this study as good as an RCT?
Propensity score matching is a statistical technique that balances baseline characteristics between groups in observational research — in this case, matching selenium-supplementing patients to non-supplementing patients based on age, sex, BMI, comorbidities, and medications. It's the best tool available for retrospective data, but it only controls for variables the researchers measure. If selenium-supplementing patients had worse disease severity, higher symptom burden, or lower baseline selenium levels — and those weren't captured in the matching — the groups still aren't comparable. An RCT would randomize before exposure, eliminating both measured and unmeasured confounding.
Could the increased mortality be from selenium toxicity?
Possibly, but unlikely at typical supplement doses. Chronic selenium toxicity (selenosis) occurs above 400 mcg/day and presents with hair loss, brittle nails, garlic breath, nausea, fatigue, and neurological damage. Most over-the-counter selenium supplements contain 50-200 mcg. The study doesn't report dose, so it's possible some patients were taking excessive amounts, but the more likely explanation is confounding by indication — patients who were sicker at baseline (and therefore at higher mortality risk) were more likely to seek out selenium supplementation. Without baseline selenium measurements or cause-of-death data, this study cannot distinguish between direct harm, indirect harm, or confounding.
Does this mean selenium is bad for thyroid health?
No. Selenium is essential for thyroid function — it's required for the enzymes that convert T4 to T3 and for glutathione peroxidase, which protects the thyroid from oxidative damage. Selenium deficiency worsens Hashimoto's. The question this study raises is whether supplementation helps people who are already selenium-sufficient. The U-shaped curve is well-established for selenium: too little is harmful, optimal is protective, too much is toxic. This study cannot answer where patients fell on that curve at baseline, which makes it impossible to conclude whether supplementation itself is the problem or whether non-deficient people are supplementing unnecessarily.
Sources
- Toraih E., Yaghi S., Ardis C., Tran L., Abdelmaksoud A., Elshazli R., et al. Clinical Outcomes of Selenium Supplementation in Hashimoto's Thyroiditis Without Selenium Deficiency: A Large-Scale Retrospective Cohort Study. Endocrinology, Diabetes & Metabolism. 2026. PubMed.
- Turker O., Kumanlioglu K., Karapolat I., Dogan I. Selenium treatment in autoimmune thyroiditis: 9-month follow-up with variable doses. J Endocrinol. 2006;190(1):151-6.
- Gartner R., Gasnier BC., Dietrich JW., Krebs B., Angstwurm MW. Selenium supplementation in patients with autoimmune thyroiditis decreases thyroid peroxidase antibodies concentrations. J Clin Endocrinol Metab. 2002;87(4):1687-91.
These statements have not been evaluated by the Food and Drug Administration. This information is not intended to diagnose, treat, cure, or prevent any disease. Always consult with a qualified healthcare professional before starting any supplementation protocol.
