The Dose She No Longer Needs: Deprescribing Levothyroxine in Older Adults

                                                        Dr Neeraj Manikath DNB

 

A Clinician–Educator Review for Postgraduate Trainees and Practising Internists

 

 

 

The 60-second summary (for those who have a clinic in 5 minutes)

● Levothyroxine is among the most prescribed drugs on earth, and a meaningful minority of older adults taking it never had a durable biochemical indication, or have drifted into iatrogenic subclinical thyrotoxicosis.

● In older adults, a suppressed TSH is more dangerous than a mildly elevated one: it roughly triples atrial fibrillation risk and materially increases fracture risk — while treating mild subclinical hypothyroidism at age ≥65 has never been shown to help (TRUST trial).

● The DINE trial showed that simply halving the dose normalised TSH in two-thirds of suppressed elderly nursing-home residents within 8 weeks, with nobody tipping into hypothyroidism.

● Deprescribing is not withdrawal of therapy — it is right-sizing. The correct move ranges from a 25–50% dose reduction to a monitored full stop, chosen by auditing why the drug was started.

● Remember one mnemonic: when the patient is T.I.R.E.D. of pills — Trace the trigger, Interrogate the trend, Risk-stratify, Eliminate or reduce, Diarise the rechecks.

 

 

 

1. Introduction — The Patient Who Changed My Practice

 

Let me introduce you to the patient who quietly rearranged how I think about thyroid disease in older people.

 

Eileen was 82 when she landed in my falls clinic: four falls in six months, a new diagnosis of atrial fibrillation, a DXA scan showing a hip T-score of −2.9 with a vertebral wedge fracture, 4 kg of weight loss, insomnia, and a fine tremor her family doctor had attributed to "anxiety." Her medication list was unremarkable except for one entry she was fiercely proud of: levothyroxine 125 mcg daily, for nineteen years, never missed a dose.

 

Her TSH that morning was 0.04 mIU/L.

 

Here is the part that should stop you mid-ward-round. When I hunted down the records from 2004, I found a single TSH of 6.9 mIU/L drawn during a workup for fatigue after an influenza-like illness, no confirmatory repeat, no thyroid antibodies, no goitre, no surgery — followed by a GP note reading "borderline thyroid — start thyroxine 50 mcg and titrate to normalise." The dose was titrated. The TSH was normalised. And over nineteen years, "normalised" drifted into suppressed — because dose requirements fall with age, and nobody ever reassessed whether the drug was needed in the first place. Her atrial fibrillation, her fracture, her tremor, and quite possibly her falls all had a plausible common contributor: a perfectly adherent patient taking a dose her body stopped needing years ago.

 

Eileen is not an outlier. Levothyroxine is one of the most prescribed medicines in the world, with well over 100 million prescriptions dispensed annually in the United States alone [1]. Subclinical hypothyroidism — the entity responsible for most "borderline" starts — affects roughly 3–4% of the general European population and up to one in ten older women [2]. Yet the threshold at which we initiate treatment has crept relentlessly downward over the past two decades: Tayside data showed the TSH level at which clinicians started levothyroxine falling substantially over ten years, with the treated population growing several-fold and a meaningful fraction of the newly treated later developing suppressed TSH [3]. Natural-history data from the Whickham cohort remind us why that matters: only a minority of mildly elevated TSH values progress to overt disease — roughly 4% per year in antibody-positive women, and far less in everyone else — while a substantial proportion spontaneously normalise [4].

 

Meanwhile, the benefit side of the ledger has all but evaporated for the age group we're discussing: the TRUST trial randomised 737 adults aged ≥65 with subclinical hypothyroidism (mean TSH ~6.4, three-quarters under 10) to levothyroxine or placebo and found no difference in hypothyroid symptoms or tiredness at one year, extended follow-up across multiple domains showing nothing convincing either [5].

 

🩺 The thesis of this review: In older adults, the most dangerous thyroid number is not the one that's slightly high — it's the one that's low. Deprescribing levothyroxine is not a fringe "less-is-more" fashion; it is an evidence-aligned correction of one of modern medicine's most quietly iatrogenic habits. Done with an audit of the original indication, a clear plan, and scheduled rechecks, it is safe, reversible, and frequently transformative.

 

 

 

2. Pathophysiology — Only the Five Facts That Change Decisions

 

You do not need a thyroid biochemistry lecture at the bedside. You need these five facts, because every clinical decision in this article flows from them.

 

Fact 1 — TSH drifts up with age, and in the oldest old much of that is not disease. Population cohorts show that mild TSH elevation in adults over 80 is frequently thyroid-antibody-negative, progresses slowly, and associates with neutral — even favourable — health outcomes [4,10]. The Leiden 85+ study followed 85-year-olds for years and found that a mildly elevated TSH did not predict disability, cognitive decline, or worse survival [10]. Actionable meaning: a TSH of 5–6 in a robust 88-year-old is a finding, not a diagnosis. Chasing it with dose escalation is how suppression begins.

 

Fact 2 — Older bodies need less levothyroxine, and the requirement keeps falling. Full replacement in a healthy younger adult is roughly 1.6 mcg/kg/day; in adults over 65 it falls to approximately 1.0–1.2 mcg/kg/day, driven by lower lean mass and slower clearance [14]. A dose that was exactly right at age 55 is, at 80, a standing overdose. Actionable meaning: weight loss, sarcopenia, and ageing itself are dose-reduction indications — nobody charts them as such, but physiology does.

 

Fact 3 — A suppressed TSH is end-organ-active, not cosmetic. Thyroid hormone excess shortens atrial action potentials, upregulates Na⁺/K⁺-ATPase and β-adrenergic signalling — electrophysiology that converts into atrial fibrillation. In the Framingham cohort aged ≥60, a TSH ≤0.1 mIU/L carried a 10-year AF incidence of 28% versus 11% — roughly a threefold excess [6]; the Cardiovascular Health Study found subclinical hyperthyroidism approximately doubled AF risk [7]. Bone is the second target: TSH receptors sit on osteoblasts and osteoclasts, and the suppressed-TSH/high-T4 state accelerates bone turnover — the Study of Osteoporotic Fractures linked low TSH in women ≥65 to two- to four-fold increases in hip and vertebral fracture risk [8], confirmed in pooled prospective data showing hip fracture risk roughly 2.4-fold higher with endogenous subclinical hyperthyroidism [9]. Actionable meaning: suppressed TSH in an older adult is a cardiology and bone clinic problem wearing an endocrine lab coat.

 

Fact 4 — Levothyroxine's 7-day half-life makes deprescribing forgiving and slow. Because T4 clearance is leisurely, dose changes take 6–8 weeks to fully register in TSH; levels after a dose reduction fall gradually — there is no crash. The same half-life is why alternate-day dosing with the same tablet is pharmacologically smooth (daily exposure averages out). Actionable meaning: never judge a dose change at 3 weeks, and don't fear the halved dose — physiology is on your side.

 

Fact 5 — Permanent and transient hypothyroidism look identical on day 400 of treatment. Post-thyroidectomy, post-radioiodine, and advanced autoimmune hypothyroidism are permanent; thyroiditis-related, drug-induced (amiodarone, lithium, interferon, kinase inhibitors, checkpoint inhibitors, iodinated contrast), and non-thyroidal-illness-related hypothyroidism are frequently transient. Once both are on levothyroxine 75 mcg, you cannot tell them apart by looking — only by interrogating the origin. This is the fork in the deprescribing road, and we will walk it explicitly.

 

 

 

3. The Five Faces of Levothyroxine in Later Life

 

Every older patient on levothyroxine is one of five phenotypes. Your first job at any review is to identify which one is sitting in front of you.

 

Phenotype	Signature clue	The right move	Watch for
1. True, permanent hypothyroidism	Thyroidectomy, radioiodine, neck radiation, or overt Hashimoto (high-titre TPOAb, TSH >10 at diagnosis)	Continue lifelong; right-size the dose; never let TSH suppress	Dose creep with weight loss
2. True hypothyroidism, currently over-replaced	TSH <0.4 on genuine replacement	Reduce 25–50%, recheck 6–8 weeks	Recurrent suppression years later
3. Treated mild subclinical hypothyroidism, now euthyroid	Small dose (25–50 mcg) started for TSH 5–8 in a ≥65-year-old; TSH now normal	Monitored trial of discontinuation	Mild TSH elevation on follow-up — usually the answer, not a failure
4. Transient cause, long resolved	Started after viral illness/subacute thyroiditis, amiodarone, lithium, contrast, or hospitalisation	Stop, with a scheduled recheck	Late relapse (rare beyond 12 months)
5. Never indicated	TSH was normal at initiation, or a single borderline value never confirmed	Stop — the prescription was treating a lab-report artefact	Nothing thyroid-specific; watch the nocebo effect

 

 

⚠️ The critical reframing: phenotypes 2–5 are not "non-compliant with the plan." The prescription is not the diagnosis — the indication is the diagnosis. If the indication was transient, weak, or absent, continuing the drug is the deviation from correct practice, not stopping it.

 

 

 

4. 🪙 Clinical Pearls — Counterintuitive Bedside Observations

 

🪙 Pearl 1 — In geriatrics, the dangerous number is the low one. Every clinician is trained to flinch at a TSH of 8 and relax at a TSH of 0.05. Invert that reflex. The elevated TSH in an asymptomatic 80-year-old has a slow, largely benign natural history [4,5]; the suppressed TSH carries measurable AF and fracture risk this year [6–9]. A TSH of 0.05 is not "well-controlled hypothyroidism" — it is iatrogenic subclinical thyrotoxicosis.

 

🪙 Pearl 2 — Older adults present thyrotoxicosis without the textbook. Heat intolerance, hyperdefecation, and anxious agitation are the symptoms of the young. In the elderly, excess thyroid hormone presents as atrial fibrillation, falls, proximal weakness, weight loss, insomnia, or a "confusion" label — the so-called apathetic thyrotoxicosis [15]. When a suppressed-TSH elder looks quiet and tired, be more worried, not less.

 

🪙 Pearl 3 — A rising TSH on a stable dose is usually a new gut or drug problem, not thyroid progression. Calcium carbonate, ferrous sulphate, PPIs, sucralfate, cholestyramine, and even a new espresso habit reduce absorption; rifampicin and anticonvulsants accelerate clearance. Before increasing the dose, audit the breakfast table — you will fix the "hypothyroidism" with a 4-hour spacing instruction instead of a prescription.

 

🪙 Pearl 4 — Escalating levothyroxine to chase persistent fatigue is how suppression is born. Symptom scores in treated hypothyroidism correlate poorly with TSH once it's in range. If the TSH is normal and the fatigue persists, the fatigue is not thyroid — it's sleep, mood, deconditioning, anaemia, or cancer. Dose escalation is not a symptom management tool; it is a biochemical intervention.

 

🪙 Pearl 5 — New AF in an older adult: check TSH before the second cardiology opinion. Rate control and rhythm strategies underperform while the patient is thyrotoxic. Finding TSH 0.03 changes the entire management conversation — and it costs less than an echocardiogram.

 

🪙 Pearl 6 — The osteoporosis workup that omits TSH is incomplete. A suppressed TSH is a modifiable, high-yield secondary cause of bone loss [8,9]. If DXA is declining "despite adequate therapy," the missing lab is often thyroid, not vitamin D round two.

 

🪙 Pearl 7 — Anticoagulants and digoxin are thyroid thermometers. Thyroid status changes warfarin sensitivity (thyrotoxicosis increases it) and digoxin handling (thyrotoxicosis lowers digoxin levels and effect). After any meaningful levothyroxine dose change in a warfarin or digoxin patient, recheck INR/levels within 2–4 weeks — the thyroid change will recalibrate these drugs before your next routine review.

 

 

 

5. 🦪 Oysters — The Hidden Gems Most Clinicians Miss

 

🦪 Oyster 1 — The repeat test that rewrites the story. TSH has meaningful within-person biological and assay variability, and a substantial minority — in some series approaching half — of mildly elevated values normalise on repeat within weeks to months. Never start — and never anchor a deprescribing decision — on a single borderline TSH. Two concordant morning samples, six to eight weeks apart, is the minimum evidentiary standard.

 

🦪 Oyster 2 — The reverse-interaction landmine (this one detonates after discharge). Stop a PPI at discharge, or advise the patient to discontinue her calcium supplements after a nephrolithiasis episode, and her levothyroxine absorption jumps — because calcium and acid suppression were blunting it. Six to eight weeks later, precisely when she has left your service, the TSH plummets. Whenever you deprescribe a drug that was interfering with levothyroxine, you have effectively increased the thyroid dose. Write the 8-week TSH reminder the same day.

 

🦪 Oyster 3 — The espresso effect. Concurrent coffee (especially with breakfast dosing) can meaningfully cut levothyroxine absorption. The patient who "started taking it with breakfast instead of fasting" or developed a latte habit has decreased her effective dose; the one who switched from coffee-with-tablet to fasting has increased it. Likewise, brand switches, tablet-source changes (generic/brand, liquid, gel formulations) shift delivered dose — and the TSH moves 6–8 weeks later. Formulation and routine changes are stealth dose changes.

 

🦪 Oyster 4 — Central hypothyroidism is invisible to TSH logic. In pituitary disease, TSH can be low, normal, or mildly raised while free T4 is genuinely low. If you apply the "low TSH = overtreatment, reduce!" rule to a central hypothyroid, you will manufacture myxoema. In central hypothyroidism, never deprescribe by TSH — titrate to a mid-reference free T4, and refer.

 

🪙 → 🦪 Oyster 5 — The biotin mirage. High-dose biotin supplements ("hair, skin, and nails" gummies) interfere with streptavidin-based immunoassays, producing a falsely suppressed TSH and falsely elevated free T4 — a perfect imitation of iatrogenic thyrotoxicosis in a patient who feels entirely well. Ask about supplements before you stop a decade of therapy, and retest after a biotin-free interval.

 

🦪 Oyster 6 — The deprescription paradox. The patients who were started without indication are paradoxically the ones who stay on levothyroxine the longest. Their dose was never anchored by true pathology, every repeat prescription re-ritualises the habit, and their identity fuses with "my thyroid problem." The bedside test is beautiful: ask "What symptoms did you have when it was started?" — most phenotype-5 patients cannot recall any. That blank stare is your green light to investigate further.

 

🦪 Oyster 7 — The initiation-year interrogation is the highest-yield history you'll take. Subacute thyroiditis (post-viral neck pain, malaise), post-iodinated-contrast CTs, amiodarone, lithium, interferon, kinase inhibitors, checkpoint inhibitors, and hospitalisation with non-thyroidal illness all produce transient hypothyroid phases that resolve — often permanently. If the "hypothyroidism" began in the same year as one of these, you are likely looking at a phenotype-4 patient on a permanent prescription for a temporary event.

 

🦪 Oyster 8 — The oldest-old reference range runs higher than the textbook's. In robust adults over 80, a TSH of 5–6 with negative antibodies frequently represents the aged set point rather than disease [10]. The iatrogenic tragedy is that clinicians "correct" this toward younger norms and then discover suppression, AF, and fracture two years later. Normal-for-age and normal-for-a-textbook are not the same thing.

 

 

 

6. ⚡ Clinical Hacks & Tips — The Master Clinician's Shortcuts

 

⚡ Hack 1 — The 6–8 week rule. Half-life ~7 days; full TSH equilibration after any change takes 6–8 weeks. Write the follow-up bloods date the same day you change the dose — in the chart and in the patient's diary. This single habit prevents 80% of deprescribing chaos, including premature panic and forgotten follow-up.

 

⚡ Hack 2 — Halve, don't fine-tune. In suppressed elderly patients, a 25–50% dose reduction captures the physiology in one step — micro-adjustments of 12.5 mcg burn months of calendar time. The DINE randomised trial used a straight 50% reduction and normalised TSH in 69% of nursing-home residents within 8 weeks, with no participant crossing into biochemical hypothyroidism during follow-up [11]. Dose arithmetic with available tablet strengths: 125→75, 100→50, 75→25 (or alternate-day dosing, Hack 3).

 

⚡ Hack 3 — The odd-number trick. Alternating two tablet strengths on consecutive days (e.g., 100 mcg and 75 mcg alternating = average 87.5 mcg/day) is pharmacologically smooth because of the 7-day half-life, avoids new prescriptions, and lets patients keep "their tablets." Similarly, reducing to five days per week (e.g., skipping weekends) delivers a 29% dose cut with zero pharmacy involvement.

 

⚡ Hack 4 — Draw the TSH fan chart. Plot every TSH value from the last decade against time on one page. The trend is the diagnosis. A flat line at 0.15 for five years is a stronger argument for intervention than any single number; a sawtooth drifting downward tells you interactions or adherence are at play; a stable 4.7 for a decade in an 85-year-old tells you to leave it alone.

 

⚡ Hack 5 — The five-question origin audit (90 seconds with the old notes):

1. What was the TSH before the first dose?

2. Was it confirmed on repeat?

3. Were TPO antibodies ever positive?

4. Any thyroid surgery, radioiodine, or neck radiation — ever?

5. What else was happening that year (drugs, illness, contrast, thyroiditis)?

 

Three "no"s in questions 1, 2, and 3, and you are almost certainly holding a phenotype-4 or -5 prescription.

 

⚡ Hack 6 — Prescribe the plan, not just the change. The highest-leverage document is the chart entry itself: "Deprescribing trial: levothyroxine stopped [date]. Recheck TSH/FT4 in 6–8 weeks. Restart 25–50 mcg ONLY if symptomatic AND TSH confirmed >10. Patient given symptom card. GP informed." A covering clinician reading that at 2 a.m. will not reflexively restart 125 mcg because a week-6 TSH read 9.

 

⚡ Hack 7 — The weight-based sanity check. Expected replacement ≈ 1.6 mcg/kg/day in the young; ≈ 1.0–1.2 in older adults [14]. If a 55-kg 84-year-old is on 150 mcg with suppressed TSH, the arithmetic already told you the answer before the lab did. Conversely, if she's on 150 mcg with a normal-to-high TSH, stop blaming the thyroid and start hunting the gut — calcium, iron, PPI, coffee timing, coeliac screen.

 

⚡ Hack 8 — Never interpret inpatient TSH during acute illness. Non-thyroidal illness, glucocorticoids, and dopamine infusions distort TSH in both directions. Electively deprescribe in clinic, not on the ward; and wait ~6 weeks after discharge before acting on any peri-admission value.

 

 

 

7. State-of-the-Art Updates — What Has Changed Practice

 

TRUST (2017) removed the benefit assumption. In 737 adults ≥65 with subclinical hypothyroidism (TSH 4.6–19.7 mIU/L, mean ~6.4), levothyroxine titrated to normalise TSH produced no improvement in hypothyroid symptoms or tiredness at one year, with nothing compelling in extended follow-up [5]. For the first time, "don't start" in older adults became evidence-respectable — and by symmetry, "consider stopping" became a defensible, guideline-compatible conversation.

 

DINE (2021) added the randomised proof-of-mechanism for deprescribing itself. Fifty-one nursing-home residents (mean age ~79) with suppressive TSH levels (<0.55 mIU/L) on levothyroxine were randomised to halve the dose or continue. At 8 weeks, 69.2% versus 6.7% had TSH within the reference range, and no participant in the reduction arm developed biochemical hypothyroidism during the trial window [11]. Small, short, surrogate-endpoint — yes. But it is the first randomised evidence that the dose-halving move is safe and effective in exactly the population we fear touching.

 

The harm evidence hardened. Framingham's threefold AF excess with TSH ≤0.1 [6], the CHS findings [7], the fracture cohorts [8], and the pooled meta-analysis showing hip fracture risk roughly 2.4-fold higher with endogenous subclinical hyperthyroidism [9] collectively establish that the harms of over-replacement in elders are better evidenced than the benefits of treating mild subclinical hypothyroidism. When the harm side of the ledger is firmer than the benefit side, tolerating the current dose becomes the experimental position.

 

Guidelines have quietly converged. The ETA recommends against routine treatment of mild subclinical hypothyroidism (TSH <10) in adults ≥65, reserving consideration for persistent TSH >10 [13]; the ATA counsels observation over treatment for mild elevations in older adults, and explicitly warns against over-treatment and TSH suppression in this age group [14]. The benefit–risk calculus for continuing borderline therapy in elders has therefore shifted — deprescribing trials are not heterodoxy; they are the guidelines' logic taken to its conclusion.

 

What has not changed — honest gaps. There is still no large, long-duration randomised trial of full discontinuation with clinical (not biochemical) endpoints; persistence data (roughly 60–80% of well-selected patients remain euthyroid off-drug at 6–12 months in observational cohorts) are encouraging but imperfect; and cognitive associations of subclinical hyperthyroidism remain debated [15]. Deprescribe with a plan and a safety net, not with certainty.

 

 

 

8. Diagnostic Nuances — What Separates Good from Great

 

History — reconstruct the thyroid biography. Great clinicians spend three minutes on questions nobody else asks: "Who started this medicine, and what did the blood test show? Were you unwell that year? Any neck pain after a flu? Any CT scans with contrast that month? Any amiodarone, lithium, or new cancer treatment? Any thyroid operation or radioactive iodine — ever, even decades ago?" Patients often know precisely and no one has ever asked. Then corroborate with old records and pharmacy dispensing data — refill gaps reveal the true adherence that explains otherwise paradoxical TSH values.

 

Examination — thyrotoxicosis in elders hides in function, not signs. Check the pulse for irregularity (paroxysmal AF may be silent); test proximal power — rising from a chair without armrests — because thyrotoxic myopathy masquerades as sarcopenia and presents to the falls clinic; look for lid lag and a fine finger tremor, but expect their absence in the apathetic elder [15]; note height loss and kyphosis as vertebral fracture markers. In a suppressed-TSH patient with any of these, a rhythm strip or ambulatory patch is a reasonable diagnostic extension.

 

Investigations — four rules.

● TSH with free T4 whenever TSH is abnormal (low TSH + low-normal FT4? think central or non-thyroidal illness before congratulating yourself on tight control).

● TPO antibodies once, ideally at baseline: a strongly positive titre predicts relapse after deprescribing; a negative titre in a mild-SCH patient predicts the stop will stick.

● Confirm before acting — two concordant morning samples, drawn ≥6–8 weeks apart, away from acute illness, without recent biotin.

● Plot the trajectory (Hack 4) — the decade-long trend outperforms any single value, and it is free.

 

Interpretation traps. Hospital-acquired values are confounded (Hack 8); biotin mimics thyrotoxicosis (Oyster 5); recovery-phase non-thyroidal illness can transiently raise TSH; and in central hypothyroidism, TSH-guided logic fails entirely (Oyster 4). The great clinician's reflex when the labs and the story disagree is to disbelieve the number before disbelieving the patient.

 

 

 

9. Management Intricacies — Drug, Dose, Timing, Sequencing, and Pitfalls

 

9.1 Who gets which move

 

✅ The decision fork in one box:

● No durable indication (phenotypes 4–5): full stop, with monitoring.

● Questionable indication or small-dose treated mild SCH, now euthyroid (phenotype 3): stop or step down to stop — your choice guided by patient preference.

● Genuine hypothyroidism with TSH <0.4 (phenotype 2): reduce 25–50%, continue for life.

● Genuine hypothyroidism, TSH in range (phenotype 1): continue unchanged.

 

⛔ Do not deprescribe — full stop — if any of these apply:

● Post-thyroidectomy, post-radioiodine, or neck-irradiated patients (lifelong replacement)

● Central hypothyroidism (TSH is not your guide; titrate free T4, refer)

● History of differentiated thyroid cancer on TSH suppression (any target relaxation belongs with the cancer team)

● Pregnancy or the possibility of pregnancy — even subclinical hypothyroidism matters in early pregnancy

● Evolving overt hypothyroidism (low free T4): treat, don't trial

 

9.2 The protocol, step by step

 

Step 1 — Baseline (week 0). Confirm TSH (± free T4) off acute illness; TPOAb once; draw the fan chart; ECG if TSH <0.4; document the indication audit in the chart.

 

Step 2 — Execute (day 1). Phenotype 2: halve the dose [11]. Phenotypes 3–5: stop, or if the patient is anxious, step down (e.g., 100→75→50→25 over 6–10 weeks, or alternate-day dosing from the current tablet). Slow tapers are unproven but psychologically useful; the physiology doesn't require them.

 

Step 3 — Expect the rebound, don't fear it (weeks 2–6). After a full stop, TSH may transiently overshoot before settling by 8–12 weeks. A week-6 TSH of 9.2 in a well patient is a data point, not an emergency — this is the single commonest point of failure, when a covering clinician restarts the full old dose. Your Hack 6 chart note is the defence.

 

Step 4 — The verdict (weeks 6–8). TSH/FT4:

● Normal → the answer; recheck at 4–6 months, then 12 months, then on-symptoms.

● Mild elevation (4.5–10), asymptomatic, ≥65 → this is the underlying truth; observe, recheck 6–12 monthly [13,14] — remember TRUST found nothing to treat [5].

● >10, or overt, or symptomatic → restart, but low: 25–50 mcg, titrate at 6–8-week intervals, aiming for a TSH in the reference range (upper half is a reasonable target in the very old). Never reflexively resume the previous dose.

 

Step 5 — Close the loop. Letter to the GP and pharmacist; patient symptom card (fatigue, cold intolerance, constipation, weight gain) with a review trigger; document restart criteria. Deprescribing is a protocol, not an event.

 

9.3 Timing and formulation details that matter mid-trial

 

Keep the patient's dosing routine exactly constant during the trial — same time, same relation to breakfast, same brand. Do not simultaneously change calcium/iron schedules, coffee habits, or PPIs; each is a stealth dose change (Oysters 2–3) that will contaminate your experiment. If interacting drugs must change, that's fine — just recheck TSH 6–8 weeks later and say so in the plan.

 

9.4 Pitfalls, in the order they ambush you

 

1. The week-6 panic (rebound overshoot → reflex full-dose restart).

2. The post-discharge landmine (PPI/calcium stopped → absorption jumps → TSH falls after discharge).

3. Nocebo symptoms — tell patients in advance: "Most people notice nothing; symptoms in the first weeks are usually not thyroid, and we'll check bloods before changing anything."

4. Treating the number, not the person — a TSH of 5.2 at month 3 in a well 83-year-old is a finding, not a failure [10,13].

5. Forgetting warfarin/digoxin recalibration after dose changes (Pearl 7).

6. Deprescribing central hypothyroidism by TSH (Oyster 4) — a genuine disaster.

7. Stopping during acute illness — confounded TSH, confounded decision.

 

🗣️ The 20-second patient script that makes all of this work:
"Your thyroid blood test has been running low — that can quietly strain the heart rhythm and the bones, and it may even be contributing to your palpitations and weight loss. I'd like us to [halve/stop] your thyroid tablet for eight weeks and recheck. It's safe, it's reversible, and we'll decide together based on how you feel and the blood test — not guesswork."

 

 

 

10. When to Escalate vs When to Watch — Thresholds and the Reasoning Behind Them

 

Watch — deliberately, documented, with a date. An asymptomatic ≥65-year-old with TSH 4.5–10 and no indication for treatment: the reasoning is that annual progression risk in antibody-negative elders is small [4], treatment benefit at this level is unproven [5], and guidelines explicitly endorse observation [13,14]. Recheck 6–12 months. Likewise, a successfully deprescribed patient with a settling mild TSH is not a relapse — she is a diagnosis.

 

Escalate the deprescribing itself (endocrinology referral) when:

● central hypothyroidism or any pituitary context;

● differentiated thyroid cancer on suppression therapy;

● amiodarone-associated thyroid disease (complex, bimodal, genuinely hard);

● liothyronine or desiccated thyroid regimens;

● unstable angina/recent MI with a suppressed TSH (correct slowly, with cardiology in the room);

● pregnancy possible;

● suspected assay interference you cannot resolve;

● TSH instability after two adjustment cycles — stop tinkering and refer.

 

Act urgently when: suppressed TSH coexists with new AF, heart failure, or a fracture — this is thyrotoxic end-organ disease; dose reduction plus rhythm/anticoagulation management in parallel (watching warfarin and digoxin recalibration, Pearl 7), and cardiology/endocrine involvement.

 

Why the specific numbers? The 0.1 mIU/L threshold comes from Framingham, where AF risk tripled at TSH ≤0.1, with a weaker gradient for 0.1–0.4 [6] — depth of suppression tracks risk, so dose-reduction urgency scales with it. The 10 mIU/L threshold reflects the inflection in natural history: progression accelerates, symptom probability rises, and the cardiovascular signal (confined largely to TSH ≥10 or to younger patients) emerges [12]. And the 6–8 week interval is pure pharmacokinetics [14]. Every threshold in this article reduces to one of those three rationales — which means you can defend them at the bedside, not just recite them.

 

 

 

11. The One-Minute Recall — Mnemonic and Master Table

 

T.I.R.E.D. of pills — the five-step deprescribing algorithm

 

● T — Trace the trigger. Run the five-question origin audit before touching the dose.

● I — Interrogate the trend. TSH fan chart, free T4 when abnormal, TPO antibodies once, morning samples, no acute illness, no biotin.

● R — Risk-stratify. AF, falls, fractures, cognition, weight loss versus the strength and permanence of the original indication.

● E — Eliminate or reduce. No indication → stop. Suppressed on true replacement → halve [11].

● D — Diarise the rechecks. 6–8 weeks, 4–6 months, 12 months — with documented restart criteria and a patient symptom card.

 

The master table — what you see, what it means, what you do

 

TSH (mIU/L), on treatment	What it actually is	What the master clinician does
< 0.1	Iatrogenic thyrotoxicosis	Reduce ≥50% (or stop if no indication); ECG; urgent if AF; recheck 6–8 wk
0.1 – 0.4	Over-replacement	Reduce 25–50%; recheck 6–8 wk
0.4 – 4.0	Right-sized	Continue unchanged; annual check; recalculate the dose after major weight loss
4.1 – 10 (≥65, asymptomatic)	Possible underlying mild SCH or age-shifted set point	Confirm; if on treatment with a weak indication, consider the stop trial; if off treatment, watch [10,13,14]
> 10 or overt (low free T4)	True hypothyroidism	Treat/restart low (25–50 mcg) and titrate

 

 

Restart criteria, verbatim for your notes: restart 25–50 mcg only if symptoms consistent with hypothyroidism AND TSH >10 confirmed on repeat; TSH 4.5–10 with rising trend plus high-titre TPOAb — individualise; otherwise observe.

 

 

 

12. Parting Thoughts

 

Eileen stopped her levothyroxine on a Tuesday. At eight weeks her TSH was 4.8 with a normal free T4; she was sleeping through the night. At six months, 5.1; at a year, 4.6 — the mild, stable, antibody-negative elevation of a woman in her ninth decade, which two decades of well-intentioned prescribing had been "correcting" into a state that helped fracture her spine and fibrillate her atria. Her falls stopped being monthly events. Her AF has stayed rate-controlled. And the tremor attributed to anxiety disappeared with the dose that was causing it.

 

The lesson is not that levothyroxine is a bad drug — it is a superb one for the right patient, and phenotype 1 patients should never be talked out of it. The lesson is that in older adults, a prescription without a living indication is not neutral; it is a slow-motion adverse drug event with a cardiology and fracture clinic presentation. The audacity required here is small: ninety seconds of curiosity about why a drug was started, the humility to ask whether time has changed the answer, and the discipline to book the recheck before you make the change.

 

💡 If you remember one sentence, remember this one: the most valuable prescription you will ever write for an older patient is sometimes the one you choose not to refill — traced, planned, documented, and reviewed.

 

 

 

Key learning points

● Audit the indication, not just the dose: transient, unconfirmed, and never-indicated starts justify monitored discontinuation.

● Suppressed TSH in elders = AF and fracture risk, better evidenced than the benefit of treating mild subclinical hypothyroidism (TRUST).

● Halving the dose is the evidence-based default in suppressed older adults (DINE: 69% normalised at 8 weeks).

● Respect the 6–8 week equilibration and the rebound overshoot; document restart criteria so nobody else undoes your plan.

● Never deprescribe by TSH in central hypothyroidism, thyroid cancer suppression, or possible pregnancy.

● T.I.R.E.D.: Trace, Interrogate, Risk-stratify, Eliminate or reduce, Diarise.

 

 

 

References

 

1. Chaker L, Osman A, den Heijer T, Peeters RP. Hypothyroidism. Lancet. 2017;390(10108):1550-1562.

2. Garmendia Madariaga A, Santos Palacios S, Guillén-Grima F, Galofré JC. The incidence and prevalence of thyroid dysfunction in Europe: a meta-analysis. J Clin Endocrinol Metab. 2014;99(3):923-931.

3. Taylor PN, Iqbal A, Waldron A, et al. Falling threshold for treatment of a borderline elevated thyrotropin: consequences for action. Clin Endocrinol (Oxf). 2014;81(5):724-730.

4. Vanderpump MPJ, Tunbridge WMG, French JM, et al. The incidence of thyroid disorders in the community: a twenty-year follow-up of the Whickham Survey. Clin Endocrinol (Oxf). 1995;43(1):55-68.

5. Stott DJ, Rodondi N, Kearney PM, et al. Thyroid hormone therapy for older adults with subclinical hypothyroidism. N Engl J Med. 2017;376(26):2534-2544.

6. Sawin CT, Geller A, Wolf PA, et al. Low serum thyrotropin concentrations as a risk factor for atrial fibrillation in older persons. N Engl J Med. 1994;331(19):1249-1252.

7. Cappola AR, Fried LP, Arnold AM, et al. Thyroid status, cardiovascular risk, and mortality in older adults. JAMA. 2006;295(9):1033-1041.

8. Bauer DC, Ettinger B, Nevitt MC, Stone KL; Study of Osteoporotic Fractures Research Group. Risk for fracture in women with low serum levels of thyroid-stimulating hormone. Ann Intern Med. 2001;134(7):561-568.

9. Blum MR, Bauer DC, Collet TH, et al; Thyroid Studies Collaboration. Subclinical thyroid dysfunction and fracture risk: a meta-analysis of prospective cohort studies. J Clin Endocrinol Metab. 2015;100(9):3417-3425.

10. Gussekloo J, van Exel E, de Craen AJM, et al. Thyroid status, disability and cognitive function, and survival in old age. JAMA. 2004;292(21):2591-2599.

11. Al Rasyid A, Harahap WA, Sutanto AH, Nafrialdi N, Setiati S. Effect of reducing vs continuing levothyroxine dosage on suppressive thyrotropin levels in older adults: a randomized clinical trial. JAMA Intern Med. 2021;181(2):252-253.

12. Rodondi N, den Elzen WPJ, Bauer DC, et al. Subclinical hypothyroidism and the risk of coronary heart disease and mortality. JAMA. 2010;304(12):1365-1374.

13. Pearce SHS, Brabant G, Duntas LH, et al. 2013 ETA guideline: management of subclinical hypothyroidism. Eur Thyroid J. 2013;2(4):215-228.

14. Jonklaas J, Bianco AC, Bauer AJ, et al; American Thyroid Association Task Force on Thyroid Hormone Replacement. Guidelines for the treatment of hypothyroidism: prepared by the American Thyroid Association task force on thyroid hormone replacement. Thyroid. 2014;24(12):1670-1751.

15. Biondi B, Cappola AR, Cooper DS. Subclinical hyperthyroidism: a review of its clinical implications. Nat Rev Endocrinol. 2019;15(12):705-719.

 

 

 

This article is intended for postgraduate education. Deprescribing decisions must be individualised — verify local assays, formulations, and policies, and involve patients in shared decision-making at every step.