Endocrine Abstracts

JOINT399

Background and Objective: Levothyroxine (LT4) is among the most frequently used medications in the Western world. In Belgium, where health insurance is mandatory, 5.85% of the population was prescribed LT4 in 2023, raising concerns regarding overdiagnosis and overtreatment. We investigated prevalence and incidence of LT4 use in relation to TSH screening intensity, treatment threshold, and risks of overtreatment.

Method: Retrospective registry-based cohort analysis in 2003-2020 based on aggregated data from the largest Belgian health insurance provider representing 42% of the Belgian population (±4.5 million individuals).

Findings: Prevalence of LT4 use steadily increased from 2.59% in 2003 to 5.29% in 2020 (slope over time +0.153;P < 0.0001) whereas incidence of LT4 use was stable around 0.40%/year (slope +0.003;P = 0.0628) (adjusted for age category, sex, region, socioeconomic status). In the 80+ age category, prevalence rose from 5.36% in 2003 to 11.63% in 2020 (slope +0.418;P < 0.0001), but incidence decreased from 0.70%/year to 0.53%/year (slope -0.005;P = 0.046). Among non-LT4 users, the proportion with high TSH testing rate (≥1 TSH test/year) steadily rose from 26.05% to 38.53% over time. Prescription (first and following) of exclusively the lowest dose (25µg) rose from 8.33% to 19.83%. Incidence of thyroid surgery was stable at around 0.05%/year. First-time LT4 use was associated with high TSH testing rate in the preceding year (Rho 0.86;P < 0.0001), prescription of exclusively the lowest dose (Rho 0.736;P = 0.0007), and incidence of thyroid surgery (Rho 0.552;P = 0.0252). Among LT4 starters, a low subsequent TSH testing rate (≤1 TSH test within two years after start) was present in 16.52% in 2003 and 14.19% in 2018. The proportion of low TSH testing rate was negatively associated with calendar year (Rho -0.915;P < 0.0001). Compared to non-LT4 users, those using LT4 for ≥2 years had a higher risk of subsequent initiation of antiarrhythmic (RR 1.113;P = 0.0284) and antiresorptive drugs (RR 1.129;P = 0.0305), but lower mortality risk (RR 0.885;P = 0.0082). Additionally, we found a higher risk of statin use (RR 1.483;P < 0.0001), flu vaccination (RR 1.273;P < 0.0001), and ≥2 GP visits/year (RR 1.165;P < 0.0001).

Discussion: These findings suggest that increased LT4 use may partially be attributed to increased detection (TSH testing rate) of subclinical hypothyroidism (prescription of only the lowest LT4 dose), and highlight the need for better TSH monitoring after starting LT4. Whether the higher use of antiarrhythmic and antiresorptive medications in LT4 users is causally related or explained by other factors such as more intensive medical follow-up needs to be further explored on a subject-level base.