Thyroid function and risk of sepsis: a population-based prospective cohort study with traditional and genetic epidemiological analyses.

Introduction: A recent two-sample Mendelian randomization study suggested a possible causal association between hypothyroidism and selected sepsis types. We address this knowledge gap by examining the association between thyroid-stimulating hormone (TSH) levels and risk of sepsis and severe infectious diseases, using triangulation between traditional observational and Mendelian randomization (MR) analyses. Methods: Baseline characteristics and TSH-measurements were collected from adults (>20 years) at the time of participation in the prospective, population-based Trøndelag Health Study (The HUNT Study), and linked to hospital records for ascertainment of infectious diseases. Time-to-event analyses with Cox regression was used to assess the association between TSH levels and risk of sepsis, adjusting for confounders. Next, we extracted uncorrelated (R2 < 0.01) single-nucleotide polymorphisms strongly associated (p-value < 5e-8) with TSH levels from genome-wide association studies of European ancestry participants in the ThyroidOmics Consortium. Genetic associations with risk of sepsis were extracted from European ancestry participants in the UK Biobank. Secondary genetic analyses examined other measures of thyroid function (FT4, FT3, autoimmune thyroid disease and deiodinase activity) with sepsis risk, along with genetic and observational analyses of risk of lower respiratory tract infections (LRTI) and upper urinary tract infections (UUTI). Results: In the observational analyses of 45,364 subjects in HUNT there was no association between baseline normal-range TSH and sepsis risk [HR 0.98 (95% CI 0.93-1.04) per mU/L unit increase]. TSH levels <0.5 mU/L was associated with higher sepsis risk (HR 1.50, 95% CI 1.19-1.90). In the MR analyses (271,040 subjects with TSH-measurements, and 10,154 cases with sepsis), there was no association between normal-range TSH and sepsis risk [OR 1.04 (95% CI 0.98-1.10), per SD increase]. Secondary analyses supported no link between thyroid function and risk of sepsis, LRTI or UUTI. Conclusion: Variation in baseline thyroid function in the general adult population does not causally influence the risk of sepsis, LRTI or UUTI. In contrast to earlier MR work based on genetic liability to overt hypothyroidism in clinical populations, our findings indicate that mild deviations in thyroid function within and around the reference range are unlikely to be useful targets for sepsis prevention or risk stratification.