Clinical Care

Most of the largest mortality levers on this site are behavioural: sleep, exercise, diet, stopping smoking, social connection. A smaller set of interventions only happens through a clinician — a prescription, a screening test, a controlled lab draw — and a few of those deliver effect sizes behaviour alone cannot match. Semaglutide cut major cardiovascular events by a fifth in adults with overweight or obesity and established heart disease; statins and the drugs that followed them reduce events in proportion to how far they lower atherogenic particles; treating blood pressure to an intensive target cut deaths by more than a quarter in a high-risk trial. Other interventions in this pillar are narrower than their marketing: testosterone replacement treats deficiency and has never shown a mortality benefit in a randomised trial, and menopausal hormone therapy is well supported for symptoms and bone but is not a heart-disease drug. This pillar collects the evidence on all of them, plus the labs and screening cadences that decide who should be treated at all.

The pattern that organises this pillar

A recurring observation across these interventions — the authors of one testosterone registry named it Wilder's Law of Initial Value — is that response tends to be proportional to how far a patient has already drifted from a healthy baseline. Adults seeking optimisation from a good starting point see much smaller absolute benefits, sometimes none.

Treat this as a heuristic for reading the evidence, not as a finding in its own right: the three cases usually cited for it sit at three different evidence tiers. Semaglutide's cardiovascular benefit is randomised and strong, but was only ever tested in adults who already had cardiovascular disease, so it shows the effect exists there rather than that it is largest there^[1]. The menopausal timing hypothesis is observational and graded Moderate. And the testosterone registry that named the law reports a 77% mortality reduction that its own deep dive rejects as a selection-bias artefact^[2] — see testosterone therapy.

What the evidence actually supports

Strong:

• A few cancer screens earn their place; most of the value is in which ones. Biennial mammography from 40^[3], colorectal screening from 45^[4], primary human papillomavirus (HPV) testing every 5 years from 30 to 65 — it finds 40% more high-grade lesions per round than cytology^[5], which is what justifies the longer interval^[6] — low-dose computed tomography (CT) for 50–80-year-olds with 20 or more pack-years who still smoke or quit within the last 15 years^[7], one-time hepatitis C antibody at 18–79^[8], and one-time abdominal aortic ultrasound for men 65–75 who ever smoked. A meta-analysis of 18 trials in 2.1 million people reframed everything else: averaged across six common modalities, mean life extension is measured in days, not years^[9]. Screening does not passively harm — non-cancer deaths in screened arms rose 0.2%, indistinguishable from zero^[10] — it just earns less than its marketing implies.
• Blood pressure — treat it, and treat it early. In SPRINT, 9,361 high-risk adults treated to an intensive systolic target had 25% fewer major cardiovascular events and 27% lower all-cause mortality^[11]. Two caveats the headline usually loses: the survival advantage began to fade 2–4 years after the trial ended once pressures drifted back, so this is lifelong management rather than a finite win; and the 2025 AHA/ACC guideline sets <130/80 as the target, not the threshold for medication — at stage 1 (130–139/80–89) drug therapy is gated on a 10-year risk of 7.5% or more on the PREVENT calculator (Predicting Risk of Cardiovascular Disease Events), or on existing cardiovascular, kidney, or diabetic disease^[12]. Home or ambulatory monitoring is a Class I recommendation, not a nicety, and potassium-enriched salt substitutes are now Class I too^[13]. Bedtime dosing is dead as a survival strategy — TIME (>21,000 people) and BedMed (>3,300) both found nothing^[14][15].
• Lipids — count particles, and start early. Atherosclerosis is the dose × time integral of exposure to apolipoprotein B (apoB)-carrying particles, not a snapshot reading^[16]; genetic data suggest up to a threefold greater benefit per unit of cholesterol lowered when lowering starts early^[17]. About a quarter of adults are discordant — a reassuring LDL cholesterol masking a high particle count^[18] — though that apoB predicts better is on firmer ground than that acting on it rather than LDL cholesterol changes outcomes, which rests on observational cohorts. Ezetimibe, bempedoic acid (13% fewer major events in 13,970 statin-intolerant adults, with no new-onset-diabetes signal)^[19] and PCSK9 inhibitors (median LDL cholesterol 92 → 30 mg/dL, 15–20% fewer events depending on the endpoint)^[20] all work through the same mechanism, which is why the "statins have special non-cholesterol effects" theory is dead^[21]. A coronary calcium scan is the highest-value single test for anyone in the borderline-to-intermediate risk grey zone — a score of zero buys roughly 5–10 years; any detectable calcium converts a probability into anatomy and tightens the target.
• Testosterone's cardiovascular fear is settled — but so is the diabetes hope. TRAVERSE randomised 5,204 men with documented deficiency: heart attack, stroke, and cardiovascular death occurred in 7.0% on testosterone versus 7.3% on placebo^[22]. A pre-specified substudy found no slowing of progression to type 2 diabetes, so testosterone should not be used to prevent or treat it^[23]. The FDA removed the boxed cardiovascular warning in 2025 — and in the same action added a class-wide blood-pressure warning after post-marketing studies confirmed a rise across every product^[24]. See testosterone therapy.
• Semaglutide is the first weight-loss drug with a hard cardiovascular outcome. In SELECT — about 17,600 non-diabetic adults with overweight or obesity and established cardiovascular disease — semaglutide 2.4 mg cut major cardiovascular events by roughly 20%, from 8.0% on placebo to 6.5% on drug over about 40 months^[25]. The trial was funded by the manufacturer, Novo Nordisk. See Ozempic-class drugs.
• Menopausal hormone therapy relieves hot flashes and protects bone, and the delivery details are decisive. Transdermal estradiol beats oral for clotting risk; micronised progesterone beats medroxyprogesterone acetate and norethisterone for breast risk — in the French E3N cohort, breast cancer risk was unchanged with micronised progesterone and 69% higher with the older progestogens^[26]. Current use lowers fracture odds about 25%, but in the 1–10 years after stopping, odds return to (and for combined therapy slightly above) never-user levels — roughly 14 extra fractures per 10,000 woman-years, concentrated in women who used it for under five years, so the cessation window needs a taper or a bridging drug — beyond ten years post-cessation the odds fall below never-user levels again, so the bone gains banked on treatment are durable in the long run; it is the window that is the hazard, not the outcome^[27]. For the genitourinary symptoms of menopause, local vaginal estrogen and intravaginal prasterone are the highest-leverage interventions in the whole field — minimal systemic absorption, usable in many women who cannot take systemic hormones^[28].
• Lipoprotein(a) is a causal, genetically fixed risk factor, and everyone should be tested once. It is more than 90% genetically determined, stable from about age five, and untouched by diet, exercise, or statins; roughly 20% of adults carry enough to matter^[29]. The 2026 dyslipidaemia guideline recommends universal one-time screening — while recommending apoB more narrowly, when triglycerides exceed 200 mg/dL, in diabetes, or when cholesterol targets are met but risk remains^[30].
• Don't treat a slightly high thyroid-stimulating hormone (TSH) in someone who feels well. TRUST randomised 737 adults aged 65 and over with TSH 4.6–19.99: levothyroxine improved nothing — not symptoms, vitality, cognition, grip strength, artery-wall thickness, or cardiovascular events^[31]. About 58% of cases revert on their own within a median three years, and TSH runs seasonally high, so a single elevated reading is usually noise^[32]. See thyroid management.
• First-line is lifestyle correction. Both the American Urological Association and joint endocrine guidance require optimising sleep, sleep apnoea, exercise, alcohol, and endocrine-disrupting-chemical exposure before any hormone prescription^[33]. About 25% of men start testosterone without a baseline workup — skipping the easiest reversible causes.

Moderate:

• The menopausal timing hypothesis. Estrogen started within 10 years of menopause is associated with better cardiovascular outcomes; started more than 10 years out, the signal flips. KEEPS and ELITE tested it prospectively and both showed the predicted vascular response — but they were powered on surrogate endpoints (artery-wall thickness, endothelial function) and neither reduced heart attacks, strokes, or deaths, so they cannot carry the coronary-incidence and mortality figures usually attributed to them. Those come from observational cohorts carrying a healthy-user bias no adjustment removes; a trial meta-analysis found hormone therapy mortality-neutral overall, with the ~30% reduction confined to the early-initiation subgroup^[34]. The 2024 twenty-year WHI review concludes hormone therapy should not be used to prevent chronic disease at any age^[35]. Cardiovascular protection is not, on its own, an indication.
• Blood-pressure control protects cognition. SPRINT-MIND found 19% less mild cognitive impairment in the intensive arm, with the combined impairment-plus-dementia outcome down 15%^[36]; a 7-year follow-up found the benefit durable even after participants returned to usual care^[37].
• Two blood-pressure phenotypes that a clinic visit cannot see. Masked hypertension — normal in clinic, elevated at home — affects 10–18% of adults and carries roughly double the cardiovascular mortality^[38]. Untreated white-coat hypertension carries about a third higher event risk, largely because ~40% progress to sustained hypertension within a decade^[39]. Only structured out-of-office measurement finds either.
• Isometric exercise lowers blood pressure more than aerobic training does. In a network meta-analysis of 270 trials in over 15,000 people, isometric work produced the largest reductions — about 8.2 mm Hg systolic and 4.0 diastolic, comparable to a first-line drug — with the wall squat scoring highest^[40].
• Lifelong testosterone exposure and coronary risk point opposite to replacement. A 2026 study took genetic markers of lifelong testosterone from 425,097 UK Biobank adults and tested them against coronary disease in a separate sample of 1.16 million people: men predisposed to higher lifelong testosterone had about 17% higher coronary-artery-disease risk — in absolute terms a lifetime risk of roughly 7.3% rising to about 8.5% — mediated largely by blood pressure, and null in women^[41]. Compatible with TRAVERSE, because decades of genetic exposure is not the same intervention as months of restoration in a deficient man — but a real argument against "optimising" a healthy adult toward 25-year-old levels.
• Statin muscle symptoms are mostly nocebo — in the people studied. SAMSON enrolled 60 adults with documented severe statin intolerance in an n-of-1 crossover of statin months, placebo months, and no-tablet months. Mean daily symptom scores came out at 8.0 with no tablet, 16.3 on the statin, and 15.4 on placebo — statistically indistinguishable from the statin. In other words, about 90% of the symptom burden the drug appeared to cause was also produced by an identical-looking placebo. Shown their own data, half restarted therapy^[42]. That is one small trial in a selected population, not a general law — but it is the right frame to run before abandoning a statin.
• The TSH reference range is being rewritten. The standard upper limit came from a population that included undiagnosed autoimmune disease. Applying age- and demographic-specific intervals reclassifies 48.5% of "subclinical hypothyroidism" as normal^[43]; a serviceable shortcut in older adults is an upper limit of age ÷ 10, so 7.0 at seventy^[44]. A TSH of 4.5 at seventy is not the same number as a TSH of 4.5 at thirty.
• Cognition and mood on testosterone. A 2025 meta-analysis of 14 trials in hypogonadal men reports standardised mean differences of 0.49 for executive function and 0.46 for memory — on a scale where about 0.2 is small, 0.5 moderate, and 0.8 large^[45]. The American Urological Association still rates the cognitive benefit "inconclusive," an institutional lag behind the meta-analytic data.
• A short midlife panel beats the untargeted annual physical. Randomised data on broad general health checks show no reduction in all-cause mortality, cardiovascular events, or cancer mortality^[46]. A six-marker annual draw — apoB, fasting insulin with HbA1c, high-sensitivity C-reactive protein (hs-CRP), ferritin, uric acid, and basic organ panels — plus a once-in-a-lifetime lipoprotein(a) does change decisions. hs-CRP is the cleanest long-horizon marker: over 30 years in 27,939 women, the top fifth carried about 70% higher cardiovascular risk, outperforming LDL cholesterol over the same interval^[47]. The discipline of refusal matters as much as the panel: no nuclear-magnetic-resonance (NMR) lipoprotein profiles when apoB is available^[48], no IgG food panels, no salivary "adrenal stress" panels^[49], no routine insulin-like growth factor 1 (IGF-1) in healthy adults, no provoked heavy-metal challenges^[50].
• Iodine sufficiency looks like a genuine longevity input, on a thin cohort. Over 20 years, a Danish iodine-replete community had about 40% lower adjusted mortality than a deficient one (hazard ratio 0.60, 95% confidence interval 0.41–0.87 — the bracketed range is where the true effect most plausibly lies, and because it stays below 1.0 the benefit is unlikely to be chance). Only 428 people were followed to the end, so read the size of the effect as indicative rather than precise^[51]. Iodised salt and seafood at the 150 µg/day allowance does the work; kelp and Lugol's megadosing does the opposite.

Weak / preliminary:

• Epigenetic-clock deceleration as a drug endpoint. A 32-week semaglutide trial in adults with HIV-associated fat redistribution moved several validated clocks by roughly two years — the first time a drug has moved multiple validated clocks in a randomised trial, though not the first time hormones have moved them at all (the uncontrolled nine-man TRIIM study reported 1.5–2.5 years of regression back in 2019, and one clock in the semaglutide trial did not move). Read it with the design in mind: a not-yet-peer-reviewed preprint reporting a post-hoc, exploratory analysis of 45 participants on drug and 39 on placebo, in a population selected for accelerated aging, with the profiling run by the vendor that sells the clock^[52]. Real, narrow, not generalisable. The same caution applies to reading a clock result at all: they predict mortality at population scale, but that a reading should drive an individual decision is not established^[53].
• Microdosing GLP-1 drugs for longevity in healthy adults. Mechanistically plausible — much of the systemic effect may occur below weight-loss doses — but there is zero randomised evidence, and the compounded supply chain bypasses FDA oversight on dose accuracy, sterility, and salt form. Speculative.
• Multi-cancer early-detection blood tests. The NHS-Galleri trial in over 142,000 people missed its primary endpoint and exists only as a company press release. It found more early-stage cancers — but a stage shift is a surrogate, and finding more early cancers is exactly what detecting indolent disease faster looks like. UK and US screening bodies require randomised mortality evidence first^[54]. See cancer screening cadence.

Caution:

• GLP-1 muscle and bone loss. 25–40% of the weight lost is fat-free mass, and hip bone mineral density fell about 2.6% over 52 weeks in a phase 2 trial of adults already at elevated fracture risk^[55] — a loss that shows up more in non-diabetic weight loss than in type 2 diabetes^[56]. Without 1.6–2.2 g/kg of protein^[57] and 2–4 weekly resistance sessions, the drug is a route to sarcopenic obesity, not longevity.
• The rest of the GLP-1 harm profile, which the weight-loss coverage tends to skip. Nausea affects about 44% versus 16% on placebo, and 4.3% stop the drug because of gastrointestinal effects^[58]. Gallbladder and biliary disease risk rises about 37% overall and more than doubles in weight-loss trials specifically^[59]. Pancreatitis and gastroparesis run at about 4.6 and 9.1 cases per 1,000 person-years respectively — low in absolute terms, but several-fold above the comparator weight-loss drug, on wide confidence intervals^[60]. And sudden painless loss of vision in one eye from optic-nerve infarction has been associated with semaglutide in two cohorts, the second matching 174,584 pairs^[61][62] — any new visual loss warrants same-day ophthalmology and the drug should not be restarted.
• Testosterone's secondary signals. Atrial fibrillation, acute kidney injury, pulmonary embolism, and elevated blood pressure on ambulatory monitoring, all from TRAVERSE. Erythrocytosis (haematocrit above 54%) is the single most consistent adverse effect. Separately, exogenous testosterone suppresses the reproductive axis and halts sperm production — men who want to preserve fertility need a different protocol entirely.
• Hormone therapy started late, taken orally, or continued for many years. Initiation in the late sixties and beyond raises heart attack, stroke, and clot risk in the first year or two — a different proposition from continuing established therapy. Oral estrogen carries a clotting and stroke signal that transdermal estradiol does not^[63]. And use beyond 5–10 years carries a small but real ovarian-cancer excess — roughly one additional death per 1,700–3,300 women for five years' use starting around 50 — which is a reason to revisit the indication annually rather than to refuse the drug.
• Compounded prescription medications. Whether microdosed semaglutide or fractionated female testosterone, both bypass FDA oversight on dose accuracy, sterility, and salt formulation; global consensus explicitly discourages compounded female testosterone^[64].
• Overdiagnosis is a real cost of indiscriminate screening. An estimated 72–94% of US papillary thyroid cancers diagnosed between 1991 and 2019 were overdiagnoses, driven by high-resolution ultrasound of nonpalpable nodules^[65]. Roughly 1 in 5 screen-detected cases of ductal carcinoma in situ is overdiagnosis^[66] — and among 1,780 women whose disease was not operated on, 8–14% developed invasive cancer at eight years, far below the 20–50% older models assumed^[67]. Prostate cancer detected by prostate-specific antigen (PSA) testing requires about 12 diagnoses to prevent one death, and 456 men invited to screen — a 13% relative reduction in prostate-cancer mortality that amounts to about 2 fewer deaths per 1,000 men invited over 23 years^[68]. The 2025 American Thyroid Association guidelines now endorse active surveillance for selected sub-centimetre papillary cancer — a conditional recommendation on low-certainty evidence, not a new standard of care^[69].
• The centenarian TSH inverse is not an instruction. Long-lived cohorts run higher TSH, which argues against pushing a well person's TSH down — but not for suppressing thyroid function. Mice held in a mild hypothyroid state live a normal lifespan with severe insulin resistance, fatty liver, and aggressive liver cancer^[70]. A mildly elevated TSH in an aging adult is adaptive; deliberately chasing one is metabolic syndrome in slow motion.

Topics covered in depth

Midlife labs →

What to draw every year, what to draw once, what to leave for a symptom, and the five marketed tests to refuse outright — read against optimal rather than merely "normal" ranges.

Blood pressure management →

The target, the home-measurement protocol the diagnosis actually runs on, the two phenotypes a clinic visit misses, single-pill combinations, and the wall squat.

Lipid management →

apoB-first, cumulative-exposure framing; risk calculators and where they mislead; the coronary calcium strata; statins through PCSK9 inhibitors; and the lipoprotein(a) pipeline.

Cancer screening cadence →

Which tests, not how much — the high-value cadence, the overdiagnosis table, and an honest read on multi-cancer blood tests.

Ozempic-class drugs →

Semaglutide and tirzepatide: weight loss, the cardiovascular outcome, the epigenetic-clock signal, and the muscle, bone, and tolerability costs that decide whether it is a longevity tool.

Testosterone therapy →

What TRAVERSE settled, what the lifelong-exposure genetics complicate, why the mortality registry doesn't survive scrutiny, and the diagnostic and monitoring protocol.

Menopausal hormone therapy →

Timing, route, and progestogen as three separate decisions; what the trials did and didn't show; bone, brain, cancer risk, and the non-hormonal alternatives.

Thyroid management →

Why a slightly high TSH in a well person is usually nothing, how the reference range is being rewritten, residual symptoms on levothyroxine, and the ultrasound-driven cancer epidemic.

Practical guidance

1. Correct the reversible inputs first. Sleep, untreated obstructive sleep apnoea, alcohol, chronic stress, endocrine-disrupting chemicals, and overtraining all suppress endogenous testosterone; obesity, sedentary living, and an ultra-processed diet drive the metabolic dysfunction GLP-1 drugs rescue. The clinic visit is more useful after these are addressed.
2. Get the baseline lab workup. Annually: apoB, fasting insulin, HbA1c (glycated haemoglobin, a three-month average of blood-sugar exposure — the two together give a simple insulin-resistance index), hs-CRP, ferritin, uric acid, complete blood count, comprehensive metabolic panel. Once in a lifetime: lipoprotein(a). Symptom-driven only: TSH, two morning testosterone draws, estradiol and follicle-stimulating hormone. About 25% of men start testosterone without this — the most common avoidable mistake in the pillar.
3. Measure blood pressure at home, not just in the clinic. Seven consecutive days, twice daily, two or three readings a minute apart, seated and silent, discarding day one. The home average is what you act on — and it is the only way to find masked hypertension.
4. If a statin decision is genuinely on the fence, get a coronary calcium scan once. It converts a borderline risk estimate into anatomy. Zero buys years; anything measurable moves the decision.
5. Target the middle of the range, not the top. The genetic signal on lifelong testosterone is the strongest argument here, and it is reinforced by the U-shaped mortality curves this site documents for uric acid and for insulin-like growth factor 1 (IGF-1), where both extremes carry risk. Repletion, not optimisation.
6. Scaffold prescription pharmacology with lifestyle, always. A GLP-1 drug without resistance training and 1.6–2.2 g/kg of protein is sarcopenic obesity in slow motion. Testosterone without sleep-apnoea screening misses the cause. Statins without dietary correction underdeliver.
7. Take the side effects seriously. Haematocrit above 54% on testosterone, bone-density loss and the biliary and vision signals on GLP-1 drugs, and the fracture rebound after stopping hormone therapy all warrant active monitoring, not reassurance by default.
8. Know the hard contraindications. Pregnancy, a family history of medullary thyroid carcinoma or multiple endocrine neoplasia type 2, severe pancreatitis, and active eating disorders rule out GLP-1 drugs. Active prostate cancer is no longer an absolute contraindication to testosterone after TRAVERSE, but specialist co-management is standard. Sex-hormone therapy in anyone with a history of hormone-sensitive cancer needs oncology input.

What's overrated

• Testosterone as a longevity drug. It treats documented, symptomatic deficiency. No randomised trial has shown a mortality benefit, and the registry that reported one is a textbook selection-bias artefact.
• Hormone therapy as cardiovascular or cognitive prevention. The trials that vindicated the timing hypothesis were powered on surrogate endpoints and reduced no hard events; the 2024 WHI review is explicit that prevention is not an indication. Started early, hormone therapy does not harm cognition — and does not independently preserve it either^[71].
• GLP-1 drugs as a course of treatment you finish. Two-thirds of lost weight returns within 68 weeks of stopping^[72]. Like blood-pressure and cholesterol drugs, this treats a chronic relapsing condition and the benefit lasts only while the drug is on board.
• GLP-1 drugs for established Alzheimer's disease. The phase 3 EVOKE trials in more than 3,800 people missed the primary cognitive endpoint, even though cerebrospinal-fluid biomarkers moved in the right direction^[73].
• Whole-body MRI and CT marketed to healthy adults. Aggressively sold, no randomised mortality evidence, substantial false-positive cascade. Routine annual skin checks in average-risk adults sit in the same category — a pooled review of 20 studies in over 6 million patients found no population-level melanoma mortality benefit^[74].
• Bedtime dosing of blood-pressure medication, and cuffless smartwatch readings as diagnostic. The first is dead; the second is a trend tool that the 2024 and 2025 guidelines explicitly advise against using clinically.
• Chasing "subclinical" findings. A mildly high TSH, a borderline ferritin, a single elevated hs-CRP on a stressful week. Each invites a prescription that confers no benefit.

What this category is not

• Not medical advice. The doses, thresholds, and interventions discussed here are evidence summaries, not prescribing recommendations. Decisions about a specific patient belong to a specific clinician.
• Not a "preventive optimisation" pillar. The site's hierarchy is lifestyle first. This pillar exists because some interventions only happen through a clinician — prescriptions, screening tests, controlled labs — not because pills outperform behaviour.
• Not a marketing surface. No brand promotion; the article titles use household names (Ozempic-class drugs) for plain language, not endorsement.

---

For the full ordered action list drawn from every pillar, see the healthspan long list.