Sociodemographic, Lifestyle and Medical Influences on Serum Testosterone and Sex Hormone–Binding Globulin in Men From UK Biobank

Bu B. Yeap; Ross J. Marriott; Leen Antonio; Shalender Bhasin; Adrian S. Dobs; Girish Dwivedi; Leon Flicker; Alvin M. Matsumoto; Claes Ohlsson; Eric S. Orwoll; Suchitra Raj; Christopher M. Reid; Dirk Vanderschueren; Gary A. Wittert; Frederick C. W. Wu; Kevin Murray


Clin Endocrinol. 2021;94(2):290-302. 

In This Article

Abstract and Introduction


Objective: Serum testosterone concentrations are affected by factors unrelated to hypothalamo-pituitary-testicular axis pathology. We evaluated the impact of sociodemographic, lifestyle and medical factors, on serum testosterone and sex hormone–binding globulin (SHBG) in men aged 40–69 years.

Design: Cross-sectional analysis of 208,677 community-dwelling men from the UK Biobank.

Measurements: We analysed associations of different factors with serum testosterone and SHBG (immunoassays) and calculated free testosterone (cFT), using smoothed centile plots, linear mixed models and effect size estimates.

Results: Median (interquartile range) for serum testosterone was 11.6 (9.4–14.1) nmol/L, SHBG 36.9 (27.9–48.1) nmol/L and cFT 213 (178–255) pmol/L. Age and BMI were inversely associated with testosterone and cFT, while SHBG was associated with age and inversely with BMI (all P < .001). Living with a partner, (South) Asian ethnicity, never or previous smoker and some medical conditions were associated with lower testosterone. Poultry or fish eater, and higher physical activity were associated with higher testosterone (all P < .001). Testosterone was lowered by ~0.5 nmol/L across ages, ~1.5 nmol/L for BMI 30 vs 25 kg/m2, ~2 nmol/L for (South) Asian ethnicity, living with partner, college/university qualifications, low red meat eater, insufficient physical activity and 0.3–1.0 nmol/L with cardiovascular disease or diabetes. Different combinations of these factors varied serum testosterone by ~4 nmol/L, SHBG by ~30 nmol/L and cFT by ~60 pmol/L.

Conclusions: The identified modifiable risk factors support lifestyle-based interventions in men with low testosterone concentrations. Considering sociodemographic, lifestyle and medical factors facilitates more personalized interpretation of testosterone testing results with respect to existing reference ranges.


Circulating testosterone concentrations decline with increasing age,[1–3] and this decline is substantially contributed to obesity and medical comorbidities.[4–6] Loss of spouse, ethnicity, smoking and physical activity also have been associated with testosterone concentrations.[4,7–9] However, the extent to which obesity, sociodemographic and lifestyle factors, as well as medical conditions, alone and in combination, contribute to circulating testosterone in middle- and older-aged men remains unclear.

Testosterone in the circulation is bound to sex hormone–binding globulin (SHBG), to a lesser extent albumin, with a small proportion regarded as unbound or free.[10] Total testosterone and SHBG concentrations are correlated, but SHBG increases with age while testosterone concentrations fall.[1,4,9,11] The age-associated increase in SHBG results in steeper declines in calculated free testosterone (cFT) values (derived from total testosterone and SHBG using formulae based on mass action equations) with increasing age, compared with total testosterone.[1,4,11] While cFT has been advocated as a marker of androgen and health status in men,[12] its applicability is debated.[13]

Reference ranges based on results from selected healthy men have been used to define normal testosterone concentrations.[14–18] Sociodemographic, lifestyle and medical factors can affect testosterone and SHBG concentrations, and the full extent to which they influence the distributions of these measurements on a population scale has not been determined. If the independent and combined influences of these factors on testosterone and SHBG are accurately characterized, the interpretation of sex steroid measurement results can be refined. A sufficiently large data set enables more precise estimates of effect sizes.

The aims of this study were to elucidate the independent and combined influences of physical, sociodemographic, lifestyle and medical factors, on serum testosterone, SHBG and cFT concentrations in men.