TSH index and 24-hour mean heart rate in euthyroid adults: opposing FT3 and heart rate variability components

Central thyroid homeostasis may contribute to variation in heart rate even within the euthyroid range, but the relative contributions of endocrine and autonomic correlates remain uncertain.

To examine the association between the TSH index (TSHI) and 24-hour mean heart rate and to explore whether this association could be statistically decomposed into components involving free triiodothyronine (FT3) and heart rate variability.

This cross-sectional observational study was conducted as a secondary analysis of a previously described cohort. The cohort comprised 550 euthyroid adults who underwent 24-hour ambulatory electrocardiography and contemporaneous thyroid function testing. The present analysis focused specifically on the association between TSHI and 24-hour mean heart rate and its statistical decomposition involving FT3 and ln(SDNN). Multivariable linear regression and parallel mediation analysis, interpreted as statistical decomposition rather than causal mediation, were used after adjustment for demographic factors, body mass index, blood pressure, lifestyle factors, and major cardiometabolic comorbidities.

Higher TSHI was associated with lower mean heart rate (β -3.488 bpm per 1-unit increase; 95% CI -4.633 to -2.343; P < 0.001). Parallel mediation suggested opposite-direction indirect components: a positive association through FT3 (indirect estimate +0.251 bpm; 95% CI 0.052 to 0.520) and a negative indirect association through ln(SDNN) (indirect estimate -0.607 bpm; 95% CI -1.219 to -0.039). The net indirect association was not significant (-0.356 bpm; 95% CI -0.998 to 0.270), whereas the residual direct association remained predominant (c' -3.132 bpm; 95% CI -4.076 to -2.138).

In euthyroid adults, the association between TSHI and 24-hour mean heart rate appears to comprise FT3 and heart rate variability components in opposite directions, while the residual direct association remains predominant. These findings should be interpreted as reflecting cross-sectional statistical decomposition, rather than providing evidence for causal mediation.