Weight-based levothyroxine dosing and long-term cardiovascular, bone, and mortality outcomes in older adults: an emulated target trial using UK primary care data

Background Levothyroxine is widely prescribed in older adults with hypothyroidism, yet evidence guiding safe initial weight-based dosing is scarce. Both under-replacement and over-replacement of levothyroxine might adversely affect cardiovascular health, bone integrity, and survival. Given the impracticality of conducting a long-term randomised trial in this population, we aimed to emulate a target trial to assess associations between initial levothyroxine dose per bodyweight and long-term clinical outcomes in older adults using routinely collected primary care data. Methods This emulated target trial was based on observational data from The Health Improvement Network, a UK primary care database, from Jan 1, 2006, to Dec 31, 2021. Eligible participants were adults older than 50 years diagnosed with hypothyroidism with at least one recorded thyroid-stimulating hormone measurement exceeding 4·0 mU/L who initiated levothyroxine during this period. The target trial compared initiation of levothyroxine at greater than 1·0 μg/kg per day (high dose) versus 1·0 μg/kg per day or less (low dose) using total bodyweight and lean bodyweight definitions. Participants were followed up for up to 10 years from levothyroxine initiation until outcome occurrence, death, or study end. Inverse probability of treatment weighting adjusted for baseline confounders. Associations with cardiovascular events, bone outcomes, and all-cause mortality were estimated as primary outcomes using weighted Cox proportional hazards models, and 10-year absolute risk differences were calculated in the pseudo-weighted population. Outcomes were analysed according to baseline prescribed dose group. Each of the primary outcomes was analysed separately, using bespoke cohorts. Findings Overall, 19 027 records were eligible for inclusion in this study (14 296 [75%] were for female individuals and 4731 [25%] were for male individuals). 15 463 adults were included in the primary cardiovascular analysis, 11 762 (76%) of whom initiated levothyroxine at a dose of 1·0 μg/kg per day or less and 3701 (24%) at a dose of more than 1·0 μg/kg per day. 14 947 adults were included in the bone health analysis, 11 988 (80%) of whom initiated levothyroxine at a dose of 1·0 μg/kg per day or less and 2959 (20%) at a dose of more than 1·0 μg/kg per day. All 19 027 individuals were included in the all-cause mortality analysis, 14 503 (76%) of whom initiated levothyroxine at a dose of 1·0 μg/kg per day or less and 4524 (24%) at a dose of more than 1·0 μg/kg per day. Unweighted event rates were higher in the high-dose group than the low-dose group for cardiovascular events (728 [20%] of 3701 vs 2079 [18%] of 11 762), bone outcomes (387 [13%] of 2959 vs 1089 [9%] of 11 988), and all-cause mortality (836 [18%] of 4524 vs 2532 [17%] of 14 503). High levothyroxine dose was associated with increased risk of cardiovascular events (adjusted hazard ratio 1·18, 95% CI 1·08–1·29), bone outcomes (1·29, 1·16–1·43), and all-cause mortality (1·17, 1·09–1·25). The weighted 10-year absolute risk difference between the high-dose group and the low-dose group was 1·3% (95% CI 0·3–3·0) for cardiovascular events, 3·0% (1·6–4·5) for bone outcomes, and 3·4% (1·7–5·1) for all-cause mortality for total bodyweight dosing. Similar associations were observed across outcomes and analyses using lean bodyweight-based dosing. Interpretation Our findings suggest that high initial weight-based levothyroxine dosing in older adults is associated with less favourable long-term outcomes, supporting cautious initiation and titration in this population. Our study is based on observational data and findings are associations, so should be interpreted with caution and validated with prospective studies. Funding National Institute for Health and Care Research Applied Research Collaboration North East and North Cumbria.