Subclinical Hypothyroidism: When a Rising TSH Earns Treatment
A 78-year-old woman's TSH has climbed twice, mildly, and her thyroid antibodies are positive. The disagreement isn't whether her thyroid is drifting — it's whether a documented trajectory and a known risk marker earn treatment before a textbook threshold is reached.
Dolores M., a 78-year-old woman, has spent the better part of two years managing her husband's advancing Parkinson disease — medication schedules, physical therapy visits, the small daily negotiations of a marriage reorganized around someone else's tremor — and when her primary care physician asked at her annual physical how she'd been sleeping, she laughed before admitting it had been badly, for months. Her exam and routine labs were otherwise unremarkable except for one number: a TSH of 6.9 mIU/L, rechecked four months later at 7.6, with a normal free T4 of 1.2 ng/dL — persistently, mildly elevated, not a single stray value. Her LDL on the same draw was 142 mg/dL, up from 118 two years ago — a drift that is unremarkable on its own but is the change subclinical hypothyroidism is most reliably expected to produce, arriving alongside the TSH rather than at some unrelated moment. She has no personal or family history of thyroid disease, takes lisinopril for well-controlled hypertension, and has never before had an abnormal thyroid panel in the fifteen years her chart has been tracked. Her TPO antibody, sent to help settle the picture, came back positive at 180 IU/mL — a marker that diagnoses nothing on its own but meaningfully raises her odds of progressing to overt hypothyroidism over time compared with a TPO-negative patient sitting at the same TSH.
Whether that progression risk means treating her today is a different question than whether her thyroid is drifting, and the two get conflated more often than they should. TRUST, the largest randomized trial of levothyroxine in older adults with subclinical hypothyroidism, enrolled patients whose TSH ran 4.60 to 19.99 mIU/L on two occasions at least three months apart — Dolores sits squarely inside that population — and found no significant improvement in hypothyroid symptom scores or reported tiredness at one year. Her fatigue is real and is plausibly explained by eighteen months of interrupted sleep, and TRUST's own population argues against assuming it will lift on levothyroxine simply because a TSH happens to be elevated at the same time. What TRUST does not settle on its own terms is the antibody question, because it never measured antibodies at all. Lyko and colleagues later did, in stored TRUST and IEMO80+ samples, and found no greater benefit in the antibody-positive group — but that analysis still says nothing about a patient with a specifically documented upward trajectory, which is Dolores's situation rather than the average enrolled patient's. Her rising numbers and her real progression risk are a separate fact sitting alongside a null trial result, not erased by it.
At the follow-up visit, deciding whether to start now
I'd start levothyroxine today. The Whickham survey's long-term follow-up data, from Vanderpump and colleagues, found that a raised TSH with positive TPO antibodies carries a meaningfully higher annual risk of progressing to overt hypothyroidism than a raised TSH alone — in plain terms, her antibody status isn't background noise, it's telling us which direction her labs are actually headed. She isn't a stable subclinical patient we're watching out of caution; she's on a documented trajectory, and her rising LDL is a second, independent signal that inadequate thyroid hormone is already doing something measurable, not just sitting as an abstract lab value.
I won't dispute the biology of antibody-driven progression — that's real. But TRUST, Stott and colleagues' trial in the New England Journal of Medicine, is exactly the population question we're facing, not a tangential data point: the largest randomized trial of treating older adults at this TSH range, and it found no measurable improvement in symptoms or reported tiredness at one year.
You're characterizing her as outside what the randomized evidence tested because of her antibody status. TRUST itself didn't measure antibodies — but Lyko and colleagues went back to the stored TRUST and IEMO80+ samples and did, in 660 patients, 188 of them anti-TPO positive, and found no greater benefit from levothyroxine in that group. The question has since been asked directly, and the answer for symptom benefit was still no. Overtreatment in a 78-year-old woman is not a hypothetical either; the same literature that documents progression risk to overt hypothyroidism also documents that inadvertently pushing an older patient into subclinical hyperthyroidism raises real risk of atrial fibrillation and fracture. Rodondi's meta-analysis in JAMA found that subclinical hypothyroidism's own association with coronary events was concentrated at TSH of 10 or higher — she isn't there yet.
I don't think we have to choose between treating today and watching passively for a year. She has two data points moving the same direction and a real biological reason to expect a third will too. I'd hold off starting levothyroxine now, but I wouldn't leave the threshold open-ended the way routine surveillance usually does — recheck in three months with a repeat free T4, and commit now to starting treatment if that value keeps climbing, rather than waiting to see what an unplanned future visit happens to catch. And her fatigue deserves its own answer regardless of what we decide about the TSH: connecting her with caregiver respite resources addresses the thing most likely actually keeping her tired, which a thyroid pill won't touch either way.
Agreed: hold levothyroxine today, recheck TSH and free T4 in three months, and refer her for caregiver support resources in the interim rather than treating her fatigue as self-evidently thyroidal.
Not agreed: where the trigger for starting treatment actually sits if her TSH keeps rising but stays under 10. The endocrinologist would start levothyroxine at the next visit if the value clears 8.5, given her documented trajectory and antibody status; the geriatrician wants to hold to the traditional 10 mIU/L threshold regardless of trajectory, unwilling to treat a number that guideline evidence hasn't itself validated as harmful below that line. The three-month recheck was scheduled specifically because neither voice was willing to let that disagreement go unresolved for a full year.