Thyroid management

The most common mistake in midlife thyroid care is treating a slightly high thyroid-stimulating hormone (TSH) reading in someone who feels perfectly well. The biggest recent shift is the recognition that most "thyroid cancers" picked up on neck ultrasound would never have caused any harm.

The classical model of thyroid disease was tidy: hypothyroid, hyperthyroid, or normal. The 2017–2026 evidence has dismantled the borders. The reference range itself is being rewritten by age, sex, and biology. The landmark RCT in older adults with subclinical hypothyroidism showed no benefit from treatment. A protective high-TSH phenotype keeps showing up in centenarian cohorts. A meaningful fraction of patients on the standard levothyroxine monotherapy still feel hypothyroid even when their bloodwork is "perfect," and there's a genetic reason. And the global thyroid cancer "epidemic" turned out to be an iatrogenic artifact of high-resolution ultrasound. This article walks the calibrated reading.

The TSH reference-range argument (Moderate)

The standard adult TSH reference range — roughly 0.45–4.5 mIU/L — comes mostly from the third US National Health and Nutrition Examination Survey (NHANES III). The statistical objection is that the population it was drawn from included people with undiagnosed autoimmune thyroid disease, which inflates the upper end. Filter out positive thyroid antibodies, goiters, and family history, and the upper limit of "truly disease-free" sits closer to 2.5–3.0 mIU/L.^[1]

The clinical objection runs the other way. Surks and colleagues estimated that universally adopting 3.0 mIU/L as the cutoff would reclassify 22–28 million Americans as having subclinical hypothyroidism overnight — and trigger a wave of levothyroxine prescriptions whose long-term mortality, cognitive, and cardiac benefits aren't there in the data.^[2] The risks on the other side — atrial fibrillation, accelerated bone loss — are real.

The newer and more useful reframe is demographic-specific intervals. A 2025 Annals of Internal Medicine analysis combining NHANES (8,308 US adults) with a 314,302-person Chinese cohort showed the 97.5th percentile of TSH rises with age and varies across racial and ethnic groups, while total thyroxine (T4) stays largely stable with age and women run slightly higher than men.^[3] Applying these intervals reclassifies 48.5% of people previously labeled subclinically hypothyroid as normal, and 31.2% of those previously labeled subclinically hyperthyroid. The French Endocrine Society's heuristic for older adults — divide age by 10 to set the upper limit of normal, so 7.0 mIU/L at age 70 — captures the same intuition without requiring a population-genetics workup.^[4]

The practical version: a TSH of 4.5 in a 70-year-old is not the same number as a TSH of 4.5 in a 30-year-old, and shouldn't trigger the same response.

Subclinical hypothyroidism — what TRUST settled (Strong)

The defining trial in geriatric thyroid care is TRUST, an RCT of 737 community-dwelling adults aged ≥65 with persistent subclinical hypothyroidism (TSH 4.60–19.99 mIU/L on two measurements; normal free T4) randomized to levothyroxine vs placebo, titrated to a normal TSH.^[5]

Levothyroxine produced no improvement in:

Hypothyroid symptoms (on ThyPRO, a thyroid-specific quality-of-life questionnaire)
Vitality / fatigue
Executive cognitive function
Handgrip strength
Activities of daily living
Carotid intima-media thickness
Fatal or non-fatal cardiovascular events

A separate longitudinal observation strengthens the case for restraint: among 1,751 people with subclinical hypothyroidism followed at a health-screening centre, 57.9% reverted to euthyroid over a median 36 months with no intervention at all.^[6] That study's main finding explains part of why: TSH swings with the seasons, running higher in winter and spring and lower in summer and autumn, and a repeat test in summer or autumn was about 1.4 times more likely to come back normal. Most of what gets labeled subclinical hypothyroidism on a single lab in an older adult isn't a disease in progress — it's noise, or a transient illness response, or an age-appropriate drift.

The corollary: don't treat a slightly elevated TSH in a well person. Reasonable exceptions exist (TSH persistently >10, positive TPO antibodies plus symptoms, pregnancy planning, fertility evaluation), but the default in midlife is recheck-and-watch, not prescribe.

The centenarian phenotype and the longevity inverse (Moderate)

A counter-intuitive finding shows up across longevity cohorts: people who live extraordinarily long lives tend to run a higher TSH, not lower. In an Ashkenazi centenarian cohort, the TSH frequency curve was shape-similar to younger controls but shifted right — 35.2% of centenarians had a TSH above 2.5 mIU/L vs 15.4% of controls.^[7] A companion study in the same cohort found the pattern extends to the centenarians' offspring, pointing to an inherited component in the set point of the hypothalamic–pituitary–thyroid (HPT) axis.^[8]

Population-level data from the Rotterdam Study agree: people in the low-normal thyroid-function quartile lived longer, with a longer life expectancy free of cardiovascular disease, than those in the high-normal quartile.^[9] Hypothyroid men gained ~0.3 years and hypothyroid women ~1.1 years of additional life expectancy versus the strictly euthyroid.

The mechanistic case sits on the "rate of living" framework. Triiodothyronine (T3) turns up the mitochondrial proteins that deliberately let energy escape as heat instead of capturing it as usable fuel — thermogenesis bought at metabolic cost.^[10] The byproduct is reactive oxygen species, the corrosive molecules thrown off as a side-effect of burning fuel; the cumulative downstream is oxidative DNA damage, telomere shortening, and accelerated cellular senescence. Slightly downregulating the axis decelerates the engine — fewer of those reactive byproducts per unit time, less compounding damage over decades.

The healthspan-vs-lifespan catch

This is exactly not an argument for chasing low thyroid function. The Pax8+/− mouse model — a stable mild genetic hypothyroid state — lives a normal lifespan but with severe insulin resistance, hepatic steatosis, and elevated rates of aggressive liver cancer.^[11] Profound metabolic suppression buys lifespan at the cost of healthspan. The implication for humans is narrow: a mildly elevated TSH in an aging adult is adaptive and probably protective; deliberately suppressing thyroid function below the normal range to mimic the centenarian phenotype isn't a strategy, it's metabolic syndrome in slow motion.

Levothyroxine monotherapy, residual symptoms, and the DIO2 polymorphism (Moderate)

Synthetic levothyroxine (LT4) has been the global standard for over half a century. It works by being converted at the cellular level — by the deiodinases DIO1 and DIO2 — into biologically active T3. The assumption is that a normal serum TSH on LT4 equals adequate T3 at the tissues.

Roughly 5–15% of LT4-treated patients with a biochemically normalized TSH report persistent residual symptoms. Among those who do, the complaints cluster tightly: energy and fatigue (reported by 80–90%), weight management (70–75%), memory (60–80%), and mood (40–50%).^[12] After ruling out the usual confounders (iron deficiency, B12, sleep apnea, depression, occult autoimmune disease), one mechanistic explanation that holds up is localized tissue hypothyroidism: enough T4 in the blood, not enough conversion to T3 where it matters.^[13]

A native thyroid secretes T4 and T3 in roughly a 14:1 ratio. Patients on LT4 alone tend to run a lower serum T3/T4 ratio than non-treated euthyroid controls, and a depressed ratio tracks with classical hypothyroid complaints.^[14] The genetic anchor is the DIO2 Thr92Ala polymorphism — a variant that materially impairs the deiodinase enzyme. Homozygous carriers are more likely to fail on LT4 monotherapy because central T3 generation is the bottleneck, particularly in the brain.

What combination therapy actually delivers

A 2024 meta-analysis of 16 randomized trials — 15 of LT4 + LT3 combination therapy against LT4 alone, plus two of desiccated thyroid — found measurable biochemical changes (free T4 down, total T3 up) and a statistically significant improvement on the General Health Questionnaire (GHQ-28) — about 3 points of a 28-item distress score, a real but modest shift.^[15] The picture is otherwise mixed: fatigue scales, lipids, resting heart rate, and generic quality-of-life metrics often don't separate. Pharmacokinetics is part of the problem — oral liothyronine has a short half-life, producing a supraphysiological serum-T3 spike a few hours after a dose and an afternoon trough. Multidaily dosing helps; sustained-release formulations would help more and aren't widely available.

Desiccated thyroid extract (DTE; e.g., Armour Thyroid, NP Thyroid) contains both T4 and T3 in an animal-derived ~4:1 ratio. Patient satisfaction often runs high (likely the morning T3 surge), and the same 2024 meta-analysis showed higher serum total T3 and slightly higher TSH compared to LT4 monotherapy — but the trial quality remains low, and DTE's ratio is non-physiological for humans.^[16]

When to consider combination therapy. Persistent, well-characterized symptoms despite a normalized TSH; ideally after ruling out other causes and after a DIO2 genotype where available. Targeting a near-physiological T3/T4 ratio, monitoring TSH closely to avoid suppression, watching for atrial fibrillation and accelerated bone loss. It is not a generic upgrade over monotherapy for the asymptomatic patient with a normal TSH.

Hashimoto's — the autoimmune layer beneath most hypothyroidism (Moderate)

In iodine-sufficient countries, the overwhelming majority of primary hypothyroidism is downstream of Hashimoto's thyroiditis — chronic autoimmune destruction of the gland, marked by antibodies against thyroid peroxidase (TPOAb) and thyroglobulin (TGAb). Women are affected 4–10 times more often than men.^[17] The standard-of-care moment for intervention has historically been after the gland fails — i.e., once levothyroxine becomes necessary. The expanding evidence base says several upstream levers move the trajectory.

Selenium — the strongest nutritional signal

The thyroid contains the highest selenium concentration per gram of any organ. Selenium is the catalytic core of the deiodinases and glutathione peroxidases that neutralize the hydrogen peroxide thyroid hormone synthesis generates. Selenium deficiency leaves follicular cells exposed to oxidative damage and accelerates autoantigen release.

The largest meta-analysis — 35 randomized trials, graded moderate-certainty — found that supplemental selenium produced a moderate-to-large reduction in thyroid peroxidase antibodies (standardized mean difference −0.96, 95% confidence interval −1.36 to −0.56) and lowered TSH in patients not already taking levothyroxine. The antibody effect held whether or not patients were on replacement. Thyroglobulin antibodies did not change significantly.^[18] (That bracketed range is a 95% confidence interval — the span where the true effect most plausibly lies. When such a range crosses zero for a difference, or 1.0 for a ratio, the finding is compatible with no real effect.) A 2025 meta-analysis agrees on direction and adds a durability caveat: thyroglobulin antibodies fell at three months but not at six.^[19]

Typical trial doses run 50–200 µg/day. The therapeutic window is narrow — chronic intake above ~400 µg/day risks selenosis (nail changes, neuropathy, digestive symptoms) — but the corrected-deficiency dose is small enough that two Brazil nuts a day cover most people. Selenium has the most robust antibody-lowering data of any nutritional intervention in this disease; what it has not shown is that lowering antibodies changes whether someone eventually needs levothyroxine.

Vitamin D — useful, but threshold- and genotype-dependent

Vitamin D promotes the regulatory T cells that hold autoimmunity in check and damps down the inflammatory arms of the immune response. Deficiency tracks with higher antibody titers, but the dose-response is non-linear: high-dose pulse regimens correct blood vitamin D and normalize TSH, but the antibody response plateaus quickly.^[20] Common variants in the vitamin D receptor gene modulate responsiveness; antibody suppression on supplementation appears most marked in people starting from clear deficiency (below 20 ng/mL, 50 nmol/L) and carrying the susceptible genotypes.

What else helps and what doesn't

Omega-3 fatty acids — higher red-blood-cell levels of alpha-linolenic acid (ALA) and eicosapentaenoic acid (EPA) associate with roughly 26–37% lower odds of Hashimoto's (odds ratios 0.63 for ALA and 0.74 for EPA).^[21] This is cross-sectional, so it establishes association, not protection. Mechanism is anti-inflammatory membrane remodeling. See Omega-3.
Lactose exclusion in lactose-intolerant Hashimoto's patients — 8-week elimination measurably lowers TSH and improves levothyroxine absorption pharmacokinetics. A small population, but a real effect when it applies.
The autoimmune protocol (AIP) and grain-free diets — mechanistically plausible (gut barrier, molecular mimicry), short-term improvement in inflammation markers, but evidence remains small-trial and self-selected; treat as adjunctive, not foundational.
Myo-inositol — a 2024 network meta-analysis showed no benefit beyond selenium alone, whether given by itself or combined. Skip.^[22]

Iodine — a U-shaped curve, not a "more is better" nutrient (Strong)

Iodine is the obligate elemental substrate for T4 and T3 synthesis. Both deficiency and excess disrupt thyroid function; the dose-response is bimodal.

The cleanest natural experiment is the Randers-Skagen 20-year cohort in Denmark, comparing two communities with vastly different drinking-water iodine: Skagen at 139 µg/L (urinary iodine ~160 µg/L, optimal), Randers at 2 µg/L (urinary ~55 µg/L, mild–moderate deficiency).^[23] At baseline (participants aged 75–80), the iodine-deficient community had a 26% prevalence of hyperthyroidism versus 6% in the replete one — the expected consequence of a gland compensating for chronic low substrate by growing nodules that secrete hormone on their own. After 20 years, mortality in the iodine-replete cohort was about 40% lower (adjusted HR 0.60, 95% CI 0.41–0.87). The cohorts were small — 428 people followed to the end — so treat the size of the effect as indicative rather than precise.

The other side of the curve matters too. Ingesting more than the ~1,100 µg/day tolerable upper limit — easy to do with kelp supplements or Lugol's solution — can break the Wolff-Chaikoff autoregulatory escape, especially in aging glands or those with silent autoimmunity, producing iatrogenic iodine-induced hypothyroidism.^[24] The American Thyroid Association recommends against chronic supplementation above 500 µg/day; the recommended daily allowance for non-pregnant adults is 150 µg/day, and 220–290 µg in pregnancy and lactation.

Iodine intake	Effect
Severe/moderate deficiency (<100 µg/d)	Goiter, hyperthyroidism via autonomous nodules, fetal neurodevelopmental harm
Sufficient (150–290 µg/d)	Normal hormone synthesis, lowest mortality in cohort data
Excess (≥1,100 µg/d chronic)	Wolff-Chaikoff failure → iodine-induced hypothyroidism, autoimmune activation

Practical version: an iodized-salt-and-fish baseline plus a multivitamin that contains ~150 µg covers most people. Kelp and Lugol's-style megadosing is one of the easier ways to cause the disease they're marketed against.

The thyroid cancer overdiagnosis epidemic (Caution)

The thyroid cancer story is now the canonical case study in screening-driven overdiagnosis. US incidence of papillary thyroid carcinoma rose from 5 per 100,000 in 1975 to ~15 per 100,000 by the 2010s. Mortality stayed flat. Microsimulation modeling estimates that 72% to 94% of all US papillary thyroid cancer diagnoses between 1991 and 2019 were overdiagnoses — clinically silent, indolent tumors that would never have caused symptoms.^[25] The burden fell disproportionately on women: an estimated 13–17 overdiagnosed cases per 100,000 women per year, against 3–5 per 100,000 men, largely because women get more neck imaging.

The driver was the proliferation of high-resolution neck ultrasonography in primary care, combined with fine-needle aspiration of subcentimeter nonpalpable nodules. The downstream cost was several hundred thousand total thyroidectomies, radioactive iodine treatments, and lifelong levothyroxine dependencies — plus the surgical risks of recurrent laryngeal nerve injury and permanent hypoparathyroidism — for a "cancer" that was never going to do harm. The crossover with cancer screening is direct: the same length-time bias logic that complicates DCIS and PSA findings applies here, and harder.

The 2025 American Thyroid Association (ATA) guidelines for differentiated thyroid cancer brought active surveillance into the guideline itself as an appropriate strategy for selected patients with cT1aN0M0 (≤1 cm, no nodal involvement, no extrathyroidal extension) papillary thyroid cancer.^[26] Worth being precise about the strength: this is a conditional recommendation resting on low-certainty evidence — a legitimate option to discuss, not an established standard that overrides surgery. The protocol around it is austere on purpose:^[27]

Monitoring is by neck ultrasound, not biochemistry — routine thyroglobulin is explicitly not indicated.
Surgical triggers are concrete: confirmed tumor growth >3 mm, new nodal or distant disease, extrathyroidal extension, or patient anxiety that precludes surveillance.
For progressive disease that no longer responds to radioactive iodine, lenvatinib at a starting dose of 24 mg once daily is the standardized first-line systemic option, with explicit dose-reduction protocols.^[28]

The framing matters beyond oncology. A patient told they have a tiny papillary cancer, and that watching it is a legitimate choice rather than an act of negligence, is being offered something a generation of patients didn't have — preservation of native thyroid function and freedom from lifelong synthetic replacement.

Practical guidance

Test TSH when there's a reason — fatigue, weight change, mood, cold intolerance, hair changes, slow recovery, dyslipidemia, infertility evaluation, in pregnancy, or as a baseline at midlife. Don't make TSH a tracking biomarker for "optimization" — see the wider midlife lab cadence for where TSH does and doesn't belong on a screening panel. Recheck a single abnormal value in 6–12 weeks before treating.
Interpret the result in context. A TSH of 4–6 mIU/L in a 65-year-old without symptoms is most likely physiological. The French "age divided by 10" upper limit captures the right calibration.
Add antibodies when the picture is unclear — TPOAb and TGAb separate truly autoimmune disease from incidental drift. Antibody-positive patients deserve closer follow-up even if TSH is borderline.
Don't treat subclinical hypothyroidism in older adults without a specific indication. TRUST is the strongest single trial in the area; the answer was no benefit. Exceptions are TSH persistently >10, symptomatic and antibody-positive, fertility planning, or pregnancy.
If you're on levothyroxine and still feel hypothyroid with a normal TSH, rule out the usual confounders (iron, B12, sleep apnea, depression, insomnia, overtraining). If they're clean, the DIO2 polymorphism and a trial of physiological-ratio LT4+LT3 combination therapy with a thyroid-aware clinician are reasonable — not first-line, but legitimate.
Cover selenium and vitamin D adequately, especially in Hashimoto's. Two Brazil nuts a day, or 100–200 µg supplemental selenium; correct vitamin D to ~30–50 ng/mL (75–125 nmol/L) (see Vitamin D).
Get iodine from food, not megadose supplements. Iodized salt, seafood, dairy. Skip kelp tablets and Lugol's drops unless prescribed.
If you have a small thyroid nodule found on imaging — ask about active surveillance. The 2025 ATA framework endorses it for selected cT1aN0M0 papillary thyroid cancer, as a conditional recommendation. It is a conversation to have, not a foregone conclusion in either direction.

What's overrated

Pushing TSH into the "optimal" 0.5–2.0 range in an asymptomatic adult. The longevity inverse runs the other way.
"Adrenal-thyroid fatigue" protocols combining desiccated thyroid, exogenous cortisol, and unvalidated saliva panels — none of which have RCT support, several of which carry real harms.
Iodine megadosing, including kelp, Lugol's solution, and "thyroid support" formulas that exceed the tolerable upper limit.
Reflex thyroidectomy for sub-centimeter papillary cancer. Since 2025, active surveillance is a guideline-endorsed alternative for selected low-risk disease — automatic surgery is no longer the only defensible path.