The reference ranges for the hormones covered in this article were not designed to identify optimal function. They were designed to identify disease severe enough to warrant intervention. Those two goals produce different numbers.

The distinction matters most for hormones that decline with age. A reference range built from a tested population whose average age is 55 will have a lower bound that reflects the expected hormonal state of a 55-year-old, not the state associated with preserved function. A 42-year-old with testosterone of 330 ng/dL falls within the lab’s normal range for men, somewhere above the lower bound of 270 ng/dL. He is also in roughly the 12th percentile for men his age in population data. His lab report will not mention this.

This is the specific failure mode of hormonal reference ranges: they accommodate age-related decline rather than flagging it. For metabolic and cardiovascular markers, being within the normal range is at least directionally correct as a safety signal, even if thresholds are set too conservatively. For testosterone, DHEA-S, and IGF-1, being within the normal range may simply mean you are declining at the expected rate.

This is the third article in the blood tests series. The previous two covered metabolic health and cardiovascular risk. This one covers the hormonal panel: what the standard annual physical does not test, what declining hormone levels do to muscle mass, cognition, mood, and metabolic rate, and what optimal looks like by marker.

Why hormonal reference ranges fail differently

The first article in this series described the general problem with reference ranges: they describe the middle 95 percent of whoever got tested, not of a healthy population. For most markers, this produces a range that is directionally useful. Elevated LDL points toward real risk; the threshold is too permissive, but the direction is right.

Hormonal reference ranges introduce a different error. Testosterone, DHEA-S, and IGF-1 all follow predictable decline trajectories across adult life. When a reference range is derived from a mixed-age tested population, the lower bound reflects age-related attrition. The range tells you whether you are declining at a normal rate. It does not tell you whether that rate of decline is compatible with good function at 45, or 55, or 65.

The thyroid adds a second structural problem: TSH is not a thyroid hormone. It is a pituitary signal requesting more thyroid hormone production. Measuring TSH measures the pituitary’s request, not the thyroid’s response. A normal TSH can coexist with suboptimal thyroid hormone levels if the pituitary is compensating by signaling more loudly, and that compensation is exactly what early thyroid dysfunction looks like.

Together, these two failures mean that a standard annual panel provides almost no information about the hormonal environment governing muscle preservation, metabolic rate, and cognitive function. None of the markers in this article appear on most routine annual physicals. When they are ordered, the reference ranges tell you whether you are aging normally. They do not tell you whether you can do better than that.

Thyroid: the full axis

The thyroid gland produces two hormones: thyroxine (T4) and triiodothyronine (T3). T4 is the dominant product of thyroid secretion, accounting for roughly 80 percent of output. T3 is the biologically active form at the cellular level. T4 functions primarily as a prohormone: it circulates to peripheral tissues where it is converted to T3 by deiodinase enzymes, primarily in the liver and kidneys. That conversion is not guaranteed. Under chronic stress, caloric restriction, illness, or selenium deficiency, T4-to-T3 conversion is partially redirected toward reverse T3, an inactive form that competes with T3 at thyroid hormone receptors without activating them.

The clinical consequence is a split between what the standard panel measures and what is actually happening at the cellular level. A thyroid workup that includes only TSH, or TSH and total T4, can show normal results in someone whose peripheral conversion is impaired and whose active thyroid hormone availability is substantially below optimal.

TSH, thyroid-stimulating hormone, is produced by the pituitary in response to circulating thyroid hormone levels. When thyroid hormones fall, TSH rises to stimulate greater production. When they rise, TSH falls. The relationship is logarithmic: small changes in free T4 and T3 produce disproportionately large changes in TSH. This makes TSH a sensitive detector of significant thyroid dysfunction and a less sensitive detector of the suboptimal range where symptoms are real but the pituitary can still compensate by pushing TSH toward the upper part of the reference interval.

The standard laboratory reference range for TSH runs from approximately 0.4 to 4.0 mIU/L in most labs. The upper bound reflects the threshold above which overt thyroid disease is highly probable, not the point above which thyroid function begins to decline. A 2005 analysis by Wartofsky and Dickey examined the epidemiological data for the TSH reference range and argued that the true normal range for a population free of thyroid disease and thyroid antibodies is substantially narrower, approximately 0.5 to 2.5 mIU/L. A TSH of 3.5 mIU/L sits within most labs’ accepted range. Functionally, in a symptomatic person, it may represent the pituitary working harder than it should have to.^[1]

Free T4 and free T3 measure the unbound, biologically available fraction of each hormone. Total T4 and T3 include the protein-bound fraction, which is inactive. The free measurements are more clinically informative. A free T3 in the lower quarter of the reference range, combined with a mid-range free T4 and a TSH trending toward the upper end of normal, describes a pattern consistent with impaired conversion. Labs report it without comment. It correlates with fatigue, cold intolerance, impaired cognition, hair loss, and difficulty maintaining body composition, symptoms often attributed to normal aging.

Thyroid peroxidase antibodies (TPO-Ab) and thyroglobulin antibodies (TgAb) assess autoimmune activity directed at the thyroid. Hashimoto’s thyroiditis, in which the immune system attacks thyroid tissue, is the most common cause of hypothyroidism in developed countries and the most prevalent autoimmune condition overall, estimated to affect roughly 5 percent of the general population.^[2] Antibodies can be elevated for years before TSH becomes abnormal. They explain why a TSH trending upward year over year within the normal range is a more meaningful signal than any single reading.

A complete thyroid panel: TSH, free T4, free T3, and TPO antibodies. Reverse T3 adds context specifically when free T3 is low-normal relative to free T4. TSH alone is not adequate.

Testosterone: total, free, and SHBG

Most circulating testosterone is protein-bound and biologically inactive. Sex hormone-binding globulin (SHBG) binds testosterone tightly, making it unavailable for cellular uptake. Albumin binds it loosely; this fraction is considered bioavailable. Free testosterone, approximately 2 to 3 percent of the total, circulates unbound and is available immediately for receptor binding. Total testosterone and free testosterone can diverge substantially. A man with total testosterone of 600 ng/dL and high SHBG may have free testosterone equivalent to a man with total testosterone of 380 ng/dL and normal SHBG. The two men are in materially different hormonal environments. Total testosterone does not distinguish between them.

SHBG rises with age. In men, it increases approximately 1 to 2 percent per year after 40, meaning that even stable total testosterone represents declining free testosterone over time. Conditions that further elevate SHBG, including excess alcohol, caloric restriction, hypothyroidism, and liver disease, compound the divergence. A longitudinal analysis from the Massachusetts Male Aging Study found that total testosterone and bioavailable testosterone diverged increasingly with age: men in their 70s had substantially more of their total testosterone bound to SHBG, as a fraction of the total, than men in their 40s, even at the same total reading.^[3]

The European Male Aging Study enrolled 3,369 men aged 40 to 79 across eight European centers and is the most rigorous published analysis of testosterone and its functional consequences in aging men. The study found that the combination of three sexual symptoms — decreased morning erections, decreased sexual thoughts, and erectile dysfunction — with total testosterone below 11 nmol/L (approximately 317 ng/dL) most reliably identified symptomatic late-onset hypogonadism. It also found that free testosterone below 220 pmol/L was independently associated with adverse outcomes even at normal total testosterone, and that total testosterone alone identified only a fraction of men with significant functional impairment.^[4]

Beyond symptoms, low testosterone has measurable effects on body composition and metabolic function. Low testosterone promotes visceral fat accumulation, and visceral fat expresses aromatase, the enzyme that converts testosterone to estradiol. This creates a self-reinforcing cycle: low testosterone favors fat gain, fat gain increases aromatase activity, aromatase converts more testosterone to estradiol, which lowers testosterone further while raising estrogen. Insulin resistance and elevated adiposity lower testosterone; low testosterone promotes insulin resistance and adiposity.

A 2007 analysis of testosterone data from three cross-sectional samples of American men found that testosterone levels in the United States had declined substantially across the cohorts, independent of age, BMI, smoking, and other measured confounders. A 65-year-old man in the 2002 to 2004 cohort had testosterone levels approximately 15 percent lower than a 65-year-old man measured in 1987 to 1989. The cause of the cohort-level decline was not identified.^[5]

A complete assessment requires total testosterone, free testosterone, and SHBG. A morning draw is essential: testosterone follows a diurnal pattern, peaking in the early morning and declining through the day. A draw taken in the afternoon can be 25 to 30 percent lower than a morning draw in the same person. Add estradiol if body fat is elevated or if total testosterone is in the low-to-mid range, because the testosterone-to-estradiol ratio carries clinical implications independent of either value in isolation.

DHEA-S: the marker that peaks at twenty-five

Dehydroepiandrosterone sulfate (DHEA-S) is a steroid hormone produced primarily by the adrenal cortex. It is the most abundant steroid hormone in circulation and serves as the primary precursor to androgens and estrogens in peripheral tissues: conversion in muscle, skin, bone, and brain produces testosterone, dihydrotestosterone, and estradiol locally, supplementing what circulating hormones supply.

DHEA-S follows the most predictable decline curve in endocrinology. Production peaks in the mid-20s, then falls at approximately 2 to 3 percent per year across adult life. A 1984 analysis by Orentreich and colleagues followed participants from their 20s through their 80s and documented the trajectory precisely: the decline is continuous, roughly linear on a log scale, and essentially universal across individuals and populations studied. By age 70, most people have 20 to 30 percent of the peak level they had at 25.^[6]

The consequence for reference ranges is direct. Age-stratified normal ranges for DHEA-S reflect a level associated with expected decline, not with preserved function. Being normal for your age means your DHEA-S has declined at the expected rate. It does not tell you whether that level is still adequately supporting the peripheral hormone synthesis and cellular functions that DHEA-S contributes to.

A prospective study from the EPIC-Norfolk cohort found that lower DHEA-S was associated with significantly higher all-cause mortality in elderly men, independent of established confounders including age, BMI, smoking, and existing disease. Men in the lowest quintile of DHEA-S had substantially higher mortality than men in the upper quintiles across the follow-up period.^[7] Similar associations have been reported for muscle strength, bone density, and cognitive function in multiple large cohorts, though the evidence is less consistent than for testosterone or thyroid, partly because DHEA-S is difficult to study in isolation given its role as a precursor to multiple downstream hormones.

The challenge with DHEA-S as a monitoring target is that there is no widely accepted optimal range supported by clean intervention data. Supplementation trials have produced variable results, partly because DHEA-to-downstream-hormone conversion varies substantially between individuals and between sexes. What the marker does provide is an index of adrenal reserve and a trajectory signal. Measured longitudinally over years, it quantifies the rate of adrenal decline and identifies when that decline has moved into a range associated with adverse functional outcomes in the prospective literature.

IGF-1: the growth hormone proxy

Insulin-like growth factor 1 (IGF-1) is produced primarily in the liver in response to growth hormone stimulation. Growth hormone is released from the pituitary in pulses, particularly during sleep and exercise, making it impractical to measure directly: a single blood draw for growth hormone reflects only whether a pulse occurred in the preceding minutes. IGF-1, which integrates growth hormone output over time with a half-life of 12 to 15 hours, is the standard proxy for growth hormone status.

IGF-1 mediates many of growth hormone’s anabolic effects: it promotes protein synthesis in skeletal muscle, stimulates bone formation, supports neuronal survival and synaptic plasticity, and drives tissue repair. Production peaks in late adolescence, declines through the 20s and 30s, and continues falling across adult life. By age 70, circulating IGF-1 is substantially below the levels typical of healthy 30-year-olds.

Low IGF-1 is associated with reduced muscle mass, impaired physical recovery, and accelerated bone density loss. The relationship between IGF-1 and muscle protein synthesis is well-characterized: IGF-1 activates the PI3K-Akt-mTOR pathway in muscle, the same signaling cascade that resistance training stimulates through mechanical loading. Age-related IGF-1 decline partially explains why older adults have a blunted anabolic response to the same training stimulus that produces robust adaptation in younger people. The resistance training article on this site covers this in detail. The three primary behavioral levers that maintain IGF-1 are exercise, adequate protein intake, and sleep quality, which determines the amplitude of nocturnal growth hormone pulses.

IGF-1 carries a complication that none of the other markers in this series has. It is a growth factor, and growth factors do not restrict their proliferative action to desired targets. Several prospective studies have linked higher circulating IGF-1 to increased risk of hormone-sensitive cancers. A 1998 prospective analysis found that men in the highest quartile of plasma IGF-1 had more than four times the prostate cancer risk of men in the lowest quartile, an association that remained significant after adjustment for prostate-specific antigen and other confounders.^[8] Associations with premenopausal breast cancer and colorectal cancer have been replicated across multiple large cohorts.

This creates a U-shaped optimal range that does not exist for any other marker in this series. Very low IGF-1 is associated with poor physical function, impaired recovery, and some evidence of increased all-cause mortality. Very high IGF-1 is associated with increased cancer risk. The zone between roughly 150 and 250 ng/mL is where most longevity-oriented researchers place the target for adults, with the caveat that the evidence is not precise enough to treat any specific number as a hard cutoff. Unlike testosterone or thyroid hormones, the goal with IGF-1 is not to maximize. The goal is to understand where you sit on the curve and to calibrate the levers accordingly.

Hormonal monitoring in women

The four-marker framework above describes hormones that decline along a predictable trajectory. For men, monitoring is largely about tracking that decline against a baseline. For women, the picture is different in kind. Estrogen and progesterone fluctuate substantially across the menstrual cycle, shift erratically in the perimenopause transition that can extend for years before the final period, and then stabilize at post-menopausal levels that are categorically different from the premenopausal state. A single blood draw means almost nothing without knowing where in the cycle it was taken.

Estradiol

Estradiol (E2) is the primary estrogen in premenopausal women and the driver of downstream effects across bone, cardiovascular function, cognition, lipid metabolism, and mood. It is produced in the ovaries in response to FSH stimulation, with secondary production in adipose tissue through aromatization.

In premenopausal women, estradiol follows a cycle-specific pattern. A day-2 or day-3 draw, taken early in the follicular phase, captures the baseline used to assess ovarian reserve and cycle-independent estrogen status; values typically run 20 to 80 pg/mL at this point. Estradiol peaks around ovulation at 200 to 500 pg/mL, then falls and rises to a moderate luteal peak before declining before the next period. Interpreting a single estradiol result without knowing the cycle day it was drawn is not clinically meaningful.

In perimenopause, estradiol becomes erratic before it falls. Levels can be abnormally high in one cycle and low in the next as follicular development becomes irregular. A single normal reading in perimenopause provides false reassurance; the trajectory across multiple cycles is the relevant signal. Post-menopause, estradiol stabilizes below approximately 30 pg/mL, derived from peripheral aromatization rather than ovarian production. The cardiovascular, bone, and cognitive protective effects are substantially reduced at these levels.

FSH: the transition marker

Follicle-stimulating hormone drives ovarian follicle development. As follicular reserve declines, the pituitary compensates by increasing FSH to drive the remaining follicles harder. Rising cycle-day-3 FSH is typically the earliest measurable signal of declining ovarian function, and it rises before estradiol falls.

This is the clinical value FSH adds over estradiol alone. In early perimenopause, estradiol can appear normal because the elevated FSH is successfully stimulating remaining follicles. Persistently elevated FSH above 25 IU/L on two measurements taken in separate cycles is the standard criterion for the menopausal transition. Post-menopausal FSH typically exceeds 40 to 70 IU/L. The STRAW+10 staging system, the current international standard for classifying reproductive aging, uses FSH alongside cycle irregularity as the primary staging criteria.^[9]

Progesterone

Progesterone is produced by the corpus luteum after ovulation. A mid-luteal draw — approximately 7 days after ovulation, day 19 to 22 in a 28-day cycle — above 10 ng/mL confirms that ovulation occurred. Values below 5 ng/mL in the mid-luteal phase indicate an anovulatory cycle: no corpus luteum formed, and no progesterone was produced.

Anovulatory cycles become increasingly frequent in perimenopause, years before FSH and estradiol shift into post-menopausal ranges. A woman can have regular menstrual bleeding from anovulatory cycles while experiencing the consequences of progesterone deficiency: disrupted sleep, mood changes, and endometrial exposure to unopposed estrogen. A mid-luteal progesterone below 5 ng/mL alongside normal FSH and estradiol identifies this pattern when neither of those two markers would.

Testosterone and SHBG in women

Female total testosterone runs at approximately 5 to 10 percent of male levels, typically 15 to 70 ng/dL. Most clinical testosterone assays were designed and calibrated for male concentrations. Their precision at female levels is poor, and standard immunoassay results in women carry substantial measurement error. Accurate measurement at female testosterone levels requires liquid chromatography-tandem mass spectrometry (LC-MS/MS), a method not universally available outside academic or specialist centers. An Endocrine Society position statement on testosterone measurement identified standard immunoassays as unsuitable for routine measurement in women for this reason.^[10]

SHBG in women is further complicated by estrogen status. Estrogen stimulates SHBG production; women on combined oral contraceptives or estrogen replacement therapy can have SHBG levels two to three times higher than those off hormonal therapy at the same total testosterone. Free testosterone is therefore especially critical in women, and total testosterone especially unreliable as a standalone measure.

Post-menopause, ovarian testosterone production declines substantially, roughly halving circulating testosterone relative to the premenopausal baseline. The combination of declining testosterone, declining DHEA-S, and rising SHBG produces a compounded reduction in androgenic activity that contributes to the loss of muscle mass, libido, and energy seen after menopause, and that standard panels do not test for.

Four markers and what optimal looks like

Each marker below covers three things: what it measures, why the reference range sets the wrong target, and what optimal looks like for proactive hormonal monitoring.

TSH

TSH measures the pituitary’s demand signal to the thyroid, not the thyroid’s output. It is sensitive to significant dysfunction and less sensitive to the suboptimal range where early compensation is occurring.

Standard reference range: approximately 0.4 to 4.0 mIU/L. Optimal for most adults without known thyroid disease: 1.0 to 2.0 mIU/L. A TSH of 3.5 mIU/L is technically normal; in a person with relevant symptoms, it may represent the pituitary pushing harder than it should have to. TSH alone is not sufficient: order free T4 and free T3 alongside it. Add TPO antibodies to assess for Hashimoto’s, particularly in anyone with a family history of thyroid disease or an unexplained upward trend in TSH over time.

Free T3 and free T4

Free T4 is the biologically inactive prohormone the thyroid produces. Free T3 is the active form generated by peripheral conversion. The ratio of free T3 to free T4 reflects conversion efficiency.

Most lab reference ranges for free T3 run from approximately 2.3 to 4.2 pg/mL; optimal is toward the upper half. Free T4 reference ranges vary by assay but typically span 0.8 to 1.8 ng/dL. A free T3 in the lower quarter of its range, combined with a mid-normal free T4 and a TSH approaching the upper end of normal, describes impaired conversion. The thyroid is producing T4 adequately; the conversion to active T3 is partially blocked. Standard panels do not test for this.

Total testosterone + free testosterone + SHBG

Total testosterone measures all circulating testosterone regardless of availability. Free testosterone measures the immediately active fraction. SHBG determines how much of the total is bound and unavailable.

Standard reference range for total testosterone in men: 270 to 1070 ng/dL. Population data from aging studies place the median for healthy men in their 40s and 50s around 400 to 600 ng/dL; the lower bound of the reference range reflects clinical hypogonadism, not a low-normal functional state. Optimal for most men: above 500 ng/dL total testosterone, with free testosterone in the upper third of the lab’s reference range. SHBG above 40 nmol/L substantially reduces free testosterone even at mid-range total readings. A morning draw is required; afternoon draws are not comparable.

For women, the relevant reference ranges and optimal zones differ substantially by life stage. The principle — that free testosterone is more informative than total and that SHBG must be measured to interpret either — applies equally.

DHEA-S

DHEA-S measures adrenal production of the primary precursor to androgens and estrogens in peripheral tissues. Age-stratified reference ranges accommodate the expected decline rather than identifying a health-promoting level.

A single reading tells you where you fall relative to peers. Longitudinal tracking, measuring every one to two years, tells you whether you are declining faster or slower than expected and whether your level is moving toward the range associated with adverse functional outcomes in prospective studies. The level associated with favorable outcomes in epidemiological data is toward the upper end of the age-adjusted range, not merely within it.

IGF-1

IGF-1 measures integrated growth hormone output and reflects the anabolic hormonal environment supporting muscle, bone, and tissue repair.

Standard reference ranges for adults span approximately 100 to 310 ng/mL, varying by age and sex. The clinically relevant zone for longevity-oriented monitoring is approximately 150 to 250 ng/mL. Below 120 ng/mL suggests impaired growth hormone output with downstream consequences for muscle, recovery, and bone. Above 300 ng/mL warrants consideration of cancer risk, particularly in older men. The goal is to be in the range, and to understand what is driving the level: exercise adequacy, protein intake, and sleep quality.

What to order

None of these appear on a standard annual physical. Each requires explicit request:

Thyroid panel. Specify TSH, free T4, free T3, and TPO antibodies. Most labs default to TSH alone or TSH and total T4. The free fractions must be explicitly requested. Some clinicians are unfamiliar with the clinical rationale for free T3; noting that it assesses peripheral T4-to-T3 conversion, which TSH alone cannot detect, usually resolves the discussion.

Testosterone panel. Order total testosterone, free testosterone (or calculated free testosterone from total testosterone and SHBG), and SHBG. Specify a morning draw; afternoon draws are systematically 25 to 30 percent lower. Add estradiol in men with elevated body fat, low-normal total testosterone, or symptoms that could reflect high aromatase activity.

DHEA-S. A single blood draw, no fasting required. Available at any lab running a full hormone panel. The most useful application is longitudinal: establish a baseline and retest every one to two years.

IGF-1. A fasting draw is preferred; nutritional state influences liver IGF-1 production. Interpret against the age- and sex-specific reference range with awareness that the optimal zone is narrower than the full reference range at both ends.

For women: additional markers. The thyroid panel, DHEA-S, and IGF-1 above apply to both sexes. Replace the male testosterone panel with:

• Estradiol (E2) + FSH. Draw on cycle day 2 or 3 for a cycle-independent baseline. Note the cycle day on the requisition; an undated estradiol result is not interpretable. FSH drawn on the same day provides ovarian reserve context that estradiol alone cannot.
• Mid-luteal progesterone. Draw approximately 7 days after ovulation (day 19 to 22 in a 28-day cycle) to confirm ovulation occurred. Not relevant post-menopause.
• Total testosterone + free testosterone + SHBG. Request LC-MS/MS methodology if available; standard immunoassays are imprecise at female testosterone concentrations. SHBG is especially important in women on hormonal contraception or estrogen therapy, where it may be running two to three times higher than off-therapy baseline.

Post-menopausal women: draw estradiol and FSH on any day (no cycle timing required). Persistently elevated FSH above 25 IU/L on two separate draws confirms the menopausal transition; post-menopausal FSH typically exceeds 40 IU/L.

---

The standard annual physical was designed to detect late-stage dysfunction in organ systems. It was not designed to track the hormonal environment that governs how those organ systems function. Muscle, bone, metabolism, cognition, and mood all depend on the hormonal axes this article covers. None of them are routinely tested. All of them can be tested in a single blood draw.

The next article in this series covers organ function: liver markers beyond standard ALT and AST, kidney function beyond creatinine, and the markers that distinguish late-stage disease detection from early trajectory monitoring.

Previous: The 65 Percent

References

1. Wartofsky L, Dickey RA. (2005). The evidence for a narrower thyrotropin reference range is compelling. Journal of Clinical Endocrinology & Metabolism, 90(9), 5483–5488. https://pubmed.ncbi.nlm.nih.gov/16148345/
2. Caturegli P, De Remigis A, Rose NR. (2014). Hashimoto thyroiditis: clinical and diagnostic criteria. Autoimmunity Reviews, 13(4–5), 391–397. https://pubmed.ncbi.nlm.nih.gov/24434360/
3. Feldman HA, Longcope C, Derby CA, et al. (2002). Age trends in the level of serum testosterone and other hormones in middle-aged men: longitudinal results from the Massachusetts Male Aging Study. Journal of Clinical Endocrinology & Metabolism, 87(2), 589–598. https://pubmed.ncbi.nlm.nih.gov/11836290/
4. Wu FC, Tajar A, Beynon JM, et al. (2010). Identification of late-onset hypogonadism in middle-aged and elderly men. New England Journal of Medicine, 363(2), 123–135. https://pubmed.ncbi.nlm.nih.gov/20554979/
5. Travison TG, Araujo AB, O’Donnell AB, Kupelian V, McKinlay JB. (2007). A population-level decline in serum testosterone levels in American men. Journal of Clinical Endocrinology & Metabolism, 92(1), 196–202. https://pubmed.ncbi.nlm.nih.gov/17062768/
6. Orentreich N, Brind JL, Rizer RL, Vogelman JH. (1984). Age changes and sex differences in serum dehydroepiandrosterone sulfate concentrations throughout adulthood. Journal of Clinical Endocrinology & Metabolism, 59(3), 551–555. https://pubmed.ncbi.nlm.nih.gov/6235241/
7. Trivedi DP, Khaw KT. (2001). Dehydroepiandrosterone sulfate and mortality in elderly men and women. Journal of Clinical Endocrinology & Metabolism, 86(9), 4171–4177. https://pubmed.ncbi.nlm.nih.gov/11549649/
8. Chan JM, Stampfer MJ, Giovannucci E, et al. (1998). Plasma insulin-like growth factor-I and prostate cancer risk: a prospective study. Science, 279(5350), 563–566. https://pubmed.ncbi.nlm.nih.gov/9438850/
9. Harlow SD, Gass M, Hall JE, et al. (2012). Executive summary of the Stages of Reproductive Aging Workshop + 10: addressing the unfinished agenda of staging reproductive aging. Menopause, 19(4), 387–395. https://pubmed.ncbi.nlm.nih.gov/22343518/
10. Rosner W, Auchus RJ, Azziz R, Sluss PM, Raff H. (2007). Utility, limitations, and pitfalls in measuring testosterone: an Endocrine Society position statement. Journal of Clinical Endocrinology & Metabolism, 92(2), 405–413. https://pubmed.ncbi.nlm.nih.gov/17090633/