A population-level decline in serum testosterone levels in American men
This longitudinal analysis compared testosterone measurements from successive sampling periods in the Massachusetts Male Aging Study. It was designed to distinguish ordinary age-related decline from a broader birth-cohort effect in the male population.
Design
Longitudinal cohort
Massachusetts Male Aging Study
Sampling
Successive waves
Same community, repeated over years
Question
Ageing or cohort?
Is a 55-year-old today the same as before?
Answer
Population-level fall
Independent of age
Structured abstract
Method
Serum testosterone was measured in successive sampling periods among men drawn from the same community. Because the same age groups were sampled again years later, ordinary age-related decline could be separated from a broader birth-cohort effect.
Result
Total and free testosterone were lower in the later sampling periods at the same ages. The difference persisted after adjustment for age and for several health and lifestyle factors that had changed in the meantime, including weight and smoking.
Conclusion
Something beyond ageing moved the whole distribution downward. The pattern points to a change at population level rather than to individual ageing alone.
Why it matters
Later sampling waves measured lower testosterone at the same ages — so today's lab «normal» describes the men around you now, not an individual target.
Study design
The same city, sampled again
A single snapshot cannot distinguish getting older from being born later. Repeating the measurement in the same population, years apart, can.
Cohort
Community sample
Men from the same community were sampled in successive periods rather than recruited fresh from different sources each time.
Comparison
Same ages, different years
Each age band could be compared with the same age band measured years earlier, which is what isolates the cohort effect.
Adjustment
Health and lifestyle
The analysis adjusted for age and for several health and lifestyle factors that had themselves changed across the period.
Hormones
Total and free testosterone
Design
Longitudinal, repeated sampling
Adjusted for
Age, weight, smoking and other factors
Finding
Decline persisted after adjustment
What the repeated sampling showed
The result is a direction rather than a single headline number, and the direction is what matters.
Total testosterone at the same ages
Lower
Later sampling periods showed lower total testosterone than earlier ones at matched ages.
Free testosterone at the same ages
Lower
The same pattern appeared in free testosterone.
After adjustment
Persisted
Adjusting for age, health and lifestyle factors did not remove the difference.
This page does not print a single percentage for this study. The cohort effect is reported as a persistent downward shift across sampling periods, and the larger cross-country estimates are covered under the pooled analyses elsewhere in this library.
What could move an entire population
The paper documents the trend and does not identify a cause. The candidates below are discussed in the wider literature and are listed as open questions, not as findings.
Body composition
Rising weight lowers testosterone, but adjusting for it did not remove the decline in this analysis.
Adjusted analysis in the same paper
Ageing versus cohort
Longitudinal work measured the true ageing slope separately, which is what allows the remaining shift to be attributed to cohort.
Harman et al., BLSA, 2001
Reproduced elsewhere
Comparable cohort effects were later documented in Finnish and Danish population surveys and in young American men.
Perheentupa 2013 · Andersson 2007 · Lokeshwar 2020
Clinical interpretation
Normal is a moving target
Observational cohort evidence identifies a trend; it cannot establish a single cause. That limitation is real and should be stated before anything else.
What it does establish is that the comparison group behind your lab report has changed. Being inside today's reference range and being at a healthy level for you are two different claims.
Only measurement together with symptoms can tell those two apart, which is why this site pairs a symptom questionnaire with bloodwork rather than treating either alone as an answer.
Related studies in this library
Journal of Clinical Endocrinology & Metabolism 86: 724–731 · 2001
Longitudinal effects of aging on serum total and free testosterone levels in healthy men
European Journal of Endocrinology 168: 227–233 · 2013
A cohort effect on serum testosterone levels in Finnish men
Journal of Clinical Endocrinology & Metabolism 92: 4696–4705 · 2007
Secular decline in male testosterone and SHBG serum levels in Danish population surveys
Your own numbers
Measure yourself, not the average.
This study is about the population you are being compared with. The useful question is where you personally sit.
- Total testosterone and SHBG by LC-MS/MS
- Free testosterone calculated from both
- Symptom questionnaire alongside the result
- A physician to interpret the two together