Steroid formulation, standardized prednisolone-equivalent dose, and thyroid–adrenal axis function in 28,695 adults: a real-world cross-sectional study

Exogenous steroid misuse is an increasingly recognized public health concern with potential consequences for hypothalamic–pituitary–thyroid (HPT) axis regulation. However, dose-standardized evidence linking specific steroid formulations to thyroid function markers remains scarce. To examine associations between steroid type and prednisolone-equivalent dose with thyroid function markers (TSH and FT4) and adrenal axis markers in a large adult clinical cohort. Cross-sectional analysis of 28,695 adults attending a tertiary endocrinology center. Patients were classified into diagnostic groups: Thyroid-only (n = 21,755), Steroid-only (n = 4,634), and Thyroid+Steroid comorbidity (n = 2,282). Steroid formulations were identified from structured medications and free-text clinical notes, then standardized to prednisolone-equivalent (pred-eq) doses using an NLP-assisted pharmacological mapping pipeline and established conversion factors. Thyroid status was classified per standard TSH thresholds. Kruskal–Wallis and Spearman correlation analyses were performed. Among 6,916 steroid-exposed patients, four broad categories were identified: DHEA supplement (49.2%), Corticosteroid+DHEA (29.0%), Oral corticosteroid (11.8%), and Injectable corticosteroid (8.7%). In the 282 patients with standardized dose data (median pred-eq dose 5.0 [IQR 1.67–6.5] mg/day), thyroid abnormality rate was 16.7%. TSH and FT4 did not differ significantly across steroid categories in the standardized-dose subgroup (TSH p = 0.137; FT4 p = 0.390). After dose standardization, Spearman correlations between pred-eq dose and TSH (ρ=−0.139, p = 0.120) or FT4 (ρ=+0.039, p = 0.746) were non-significant. Adrenal suppression (cortisol < 5 µg/dL) was present in 30.5% of steroid-exposed patients with available cortisol testing (614/2,010). Steroid exposure category, including injectable corticosteroid use, showed descriptive variation in thyroid markers on visual inspection of distributions, although formal between-category comparisons in the dose-standardized subgroup did not reach statistical significance. After pharmacological dose standardization, dose–response associations with thyroid function are attenuated, suggesting steroid category and formulation may matter more than cumulative dose. Adrenal co-suppression is frequent and warrants co-monitoring. These findings provide a scalable real-world framework for endocrine surveillance in steroid-exposed adults.