The decision this changes

Subclinical hypothyroidism (SCH) — mildly elevated thyrotropin (TSH) with a normal free thyroxine — is common (NHANES III: ≈13 million in the US, higher in women and older people) and levothyroxine is a default prescription for it, especially when tiredness, constipation or weight gain are attributed to it. A gold meta-analysis says: in nonpregnant adults, treating SCH does not improve quality of life or symptoms. This is a de-escalation decision — a standing drug not worth starting for these outcomes in this stratum, and a lever to stop treating as a big rock. It sits in the pharmacotherapy taper (standing-drug efficacy/limitations appraisal), not the prescriber zone: the finding is stratum-level (start-or-not for the general SCH adult), not per-person titration. (inferred from Feller et al., 2018)

The effect estimate (Feller 2018 — 21 RCTs, 2192 nonpregnant adults)

SMDs coded so positive = benefit of levothyroxine (0.2/0.5/0.8 = small/moderate/large). Primary outcomes, gold SR+MA:

  • «thyroid hormone therapy … was not associated with benefit regarding general quality of life (n = 796; SMD, −0.11; 95% CI, −0.25 to 0.03; I2=66.7%) or thyroid-related symptoms (n = 858; SMD, 0.01; 95% CI, −0.12 to 0.14; I2=0.0%). Overall, risk of bias was low and the quality of evidence assessed with the GRADE tool was judged moderate to high.» (Feller et al., 2018)
  • Certainty was HIGH for both primary outcomes (and for muscle strength, blood pressure, BMI). The QoL SMD back-transforms to ≈0.02 on the EQ-5D and the symptom SMD to ≈0.18 on the ThyPRO (0-100) score — clinically nil. (Feller et al., 2018)
  • Every secondary patient-important outcome was null too: depressive symptoms (SMD −0.10, −0.34 to 0.13), cognitive function (0.09, −0.05 to 0.22), muscle strength (0.1, −0.1 to 0.2), systolic BP (−0.7 mm Hg, −2.6 to 1.2), BMI (0.2, −0.4 to 0.8). No study measured pain. (Feller et al., 2018)
  • Verdict: «These findings do not support the routine use of thyroid hormone therapy in adults with subclinical hypothyroidism.» (Feller et al., 2018)

No-meaningful-effect, not insufficient-evidence — and where the line falls

The four evidence states must be kept apart, and this source splits cleanly across two of them:

  • No-meaningful-effect for QoL, symptoms, mood, cognition, muscle strength, BP, BMI: GRADE moderate-to-high, narrow CIs tight around a point estimate of ~0, and the two largest and most recent trials (incl. TRUST/Stott 2017) drive the pool. This is a positive null, not an absence.
  • Insufficient-evidence for cardiovascular events and mortality: GRADE low, evaluated by a single trial only — «definitive evidence is lacking regarding the association of therapy for subclinical hypothy- roidism with reduced cardiovascular event rates.» (Feller et al., 2018) type-G

Conflating the two would either overclaim a proven-harmless CV profile or discount the genuinely strong QoL/symptom null. (inferred from Feller et al., 2018)

The surrogate the drug hits — TSH normalizes while the person does not move

Levothyroxine reliably drives the treated arm’s TSH into range (0.5-3.7 mIU/L) while placebo stays elevated (4.6-14.7 mIU/L) — «indicating that treatment was associated with nor- malization of thyrotropin levels» — yet no patient-important outcome follows. (Feller et al., 2018) This is a textbook marker-moved-outcome-didn’t case, sharpened because the surrogate (TSH) is also the diagnostic criterion: treating-to-target normalizes the number that defines the disease without changing how the person feels or functions. -> Surrogate Outcomes. (inferred from Feller et al., 2018)

Transportability — where the null may NOT hold (name the boundary)

The null is established for mildly elevated TSH in adults up to ~74 y with mild-to-moderate symptoms. It does not automatically transport to three strata, each a route-(b) effect-modification candidate the trials could not test:

  • TSH >10 mIU/L (more severe SCH): «only 2 RCTs examined participants with a mean baseline thyro- tropin level higher than 10 mIU/L. Therefore, the current findings may not be generalizable to people with subclinical hypothyroidism and a thyrotropin level higher than 10 mIU/L» (baseline TSH <7.0 in 11 of 21 RCTs). (Feller et al., 2018)
  • Age >80: highest mean age was 74 y, so results «may not be gener- alizable to people older than 80 years.» (Feller et al., 2018)
  • High symptom burden: «It is possible that the subgroup of people with subclini- cal hypothyroidism and a high burden of symptoms would still benefit from treatment» — severe-symptom patients are plausibly underrepresented (treated immediately, not enrolled). (Feller et al., 2018) type-G

These are stated candidates, not demonstrated benefits: no positive effect-modification evidence exists yet, so under symmetric standards they stay open gaps, not carve-outs that rescue treatment. (inferred from Feller et al., 2018)

Guidance dissonance + the overtreatment cost

The MA runs against the practical reach of guidelines: «more than 90% of persons with subclinical hypothyroidism and a thyrotropin level of less than 10 mIU/L would actually qualify for treatment. How- ever, results of this meta-analysis are not consistent with these guideline recommendations. In addition to absence of an as- sociation of thyroid hormone therapy with improved out- comes, thyroid hormone therapy is associated with adverse ef- fects when overtreatment occurs.» (Feller et al., 2018) So the net at the margin is a drug that carries overtreatment harm (iatrogenic hyperthyroidism) with no offsetting patient-important benefit for the modal SCH adult — the case for not pulling this lever, and for questioning a standing prescription. (inferred from Feller et al., 2018)

Limits

  • Single gold MA (loop open). Confidence in the no-benefit finding is high on this source’s own GRADE, but it is one meta-analysis; the wiki holds no independent SR to corroborate or contest it, and grades coherence, not validity. A second, independent gold SR/MA on levothyroxine for SCH (or the individual large RCTs — TRUST/Stott 2017, Zhao 2016) would upgrade the independent-backing state; none is held yet.
  • Scope is treatment of established SCH for QoL/symptoms; it says nothing about frank hypothyroidism, pregnancy (excluded — the correct general-adult stratum), or SCH-plus-comorbidity populations (excluded as unrepresentative). (inferred from Feller et al., 2018)

References

Feller, M., Snel, M., Moutzouri, E., Bauer, D. C., de Montmollin, M., Aujesky, D., Ford, I., Gussekloo, J., Kearney, P. M., Mooijaart, S., Quinn, T., Stott, D., Westendorp, R., Rodondi, N., & Dekkers, O. M. (2018). Association of Thyroid Hormone Therapy With Quality of Life and Thyroid-Related Symptoms in Patients With Subclinical Hypothyroidism: A Systematic Review and Meta-analysis. JAMA, 320(13), 1349. https://doi.org/10.1001/jama.2018.13770