// learn.shawon.ch / evidence-desk / attia SCORECARD
← Evidence desk

Evidence desk · Source scorecard

Peter Attia

The verdict

A physician, and the most evidence-disciplined of the big longevity voices — the useful contrast to Huberman. He cites primary literature, grades his own confidence, keeps a public disclosures page, and — rare in this world — publicly reverses himself when the data turn (he dropped prolonged fasting and metformin, and calls resveratrol "nonsense"). His core science is strong: apoB as the causal driver of heart disease, insulin resistance as an early target, and exercise — VO₂ max, strength, muscle — as the highest-leverage longevity levers.

Where he runs ahead of the evidence is consistent and worth knowing: aggressive cancer screening (whole-body MRI, multi-cancer blood tests), ultra-low lipid targets for low-risk young people, and rapamycin — bets extrapolated past the trials, several of them aligned with a diagnostics clinic he co-founded. So the rule here is different from Huberman's: usually follow his reasoning, question the aggressiveness — and trust that he'll tell you his confidence level and update it.

What's inside

  1. The pattern at a glance
  2. The framework
  3. Cardiovascular & lipids
  4. Exercise
  5. Metabolic health & nutrition
  6. Longevity drugs & supplements
  7. Hormone replacement
  8. Cancer screening & sleep
  9. The conflicts
  10. Credibility
  11. For you

The pattern at a glance

Where Huberman's claims spread across the whole axis, Attia's cluster on the right — mechanism and exercise are strong — with a tight knot of overreach on the left around aggressive screening and unproven longevity drugs. His single best trait isn't on the chart: he downgrades his own claims when the evidence does.

outruns the evidence tracks the evidence apoB is causal · lower it Strength · muscle · exercise Insulin resistance upstream Walks back his own claims Protein 1 g/lb · VO₂max as causal apoB <20–30 for everyone Rapamycin for longevity Whole-body MRI · Galleri
Mechanism and exercise sit right; aggressive screening and rapamycin drift left.

Verdicts: Solid (mainstream-backed), Mixed (real but weaker/contested than implied), Overstated (more confidence than the data support), Safety flag (a caution worth heeding). A recurring caveat below: many of his strongest numbers (VO₂ max, strength, sauna) come from observational data — robust associations, but causation and the transfer to trained gains aren't proven.

The framework

Most of this is philosophy, not testable claims — and as philosophy it's sound and well-argued. The diagnosis (medicine is reactive and weak on slow chronic disease) is widely shared; the packaging is his.

Medicine 3.0 & the longevity model
ClaimEvidenceVerdict
"Medicine 3.0" — proactive, personalized, prevention-firstA framing, not a tested model. The critique of reactive care is uncontroversial.Solid
Healthspan > lifespan; preserve the "marginal decade"Shared geroscience goal (compression of morbidity), predates him. Directionally sound.Solid
The "Four Horsemen" (heart disease, cancer, neurodegeneration, metabolic dysfunction)Matches mainstream mortality epidemiology; a useful organizing device.Solid
Prevention must start decades early; use lifetime, not 10-year, riskWell-supported in principle (risk tracks cumulative apoB exposure). The leap to drugging low-short-term-risk 30-year-olds lacks direct RCT data.Solid
The "Centenarian Decathlon" — backcast training from what you want at 90A goal-setting heuristic; the underlying specificity/reserve logic is sound.Solid
Emotional health is the "fifth horseman" / foundationA values/personal argument, candidly non-clinical. He presents it as lived experience, not science.Solid

Cardiovascular & lipids

His flagship territory and his strongest. The apoB-is-causal claim is genuinely settled science (Mendelian randomization + a stack of drug trials all agree). The split: the mechanism and metrics are solid; the aggressiveness of his targets for low-risk people outruns the trials and draws fair overtreatment pushback.

Lipids & atherosclerosis
ClaimEvidenceVerdict
apoB (particle number), not LDL-C, is the central causal driverStrong — mechanism + discordance analyses; endorsed by the 2021 EAS consensus.Solid
Causality proven by Mendelian randomization + RCTsThe strongest causal evidence in the field — genetics and statin/ezetimibe/PCSK9 trials all converge.Solid
Risk is driven by cumulative apoB exposure ("area under the curve")Well-supported concept; the exact integral is a model, not a clinical measurement.Solid
Measure Lp(a) once — genetic, causal, ~10–20% of people elevatedSolid; societies now endorse once-in-a-lifetime screening. His nmol/L preference is correct.Solid
Statins first-line; most "intolerance" is nocebo/manageableAmong the most robust evidence bases in medicine (SAMSON for nocebo).Solid
Add ezetimibe / PCSK9 / bempedoic acid to drive apoB lowSolid pharmacology and outcomes — but those trials were in higher-risk populations.Solid
A CAC score of zero does NOT mean zero riskCorrect — soft plaque is invisible to CAC; CAC=0 is low but non-zero risk.Solid
apoB "never above 60, ideally 20–30" for everyone, starting ~40Direction supported; the specific ultra-low targets for low-risk young people have no RCT and unknown decades-long safety. "Eliminate ASCVD" is a thought experiment.Overstated
Routine CT angiography to find non-calcified plaqueThe CAC-limits point is solid; routine CTA screening of the asymptomatic isn't guideline-backed (cost/contrast/radiation).Mixed
Aggressive multi-drug lowering for low-risk peopleSmall absolute benefit at real cost (PCSK9i ~$500/mo) and unproven long-term; a legitimate overtreatment concern at the low-risk end.Safety flag

Exercise

Mainstream sports medicine, conservatively prescribed — no safety flags anywhere. The one caveat is causal framing: the VO₂ max / strength / muscle–mortality links are robust but observational, and it's unproven that training-acquired gains capture the full benefit seen when you compare naturally-fit to unfit people.

Exercise & longevity
ClaimEvidenceVerdict
VO₂ max is among the strongest mortality predictors; train it, target "elite for age"Real and large (Mandsager 2018: low vs elite ≈5× mortality) — but observational. The "train and capture the full benefit" leap outruns the data.Mixed — causal framing
Raise VO₂ max with 4×4 intervals (Zone 5), 1–2×/weekThe 4×4 protocol is RCT-validated for raising VO₂ max.Solid
Zone 2, ~3–4 h/week, lactate ~1.7–2.0 mmolEndurance volume clearly helps mitochondria; "Zone 2 is uniquely special" and the exact dose are extrapolations (some argue equal-volume higher intensity matches it).Mixed
Muscle mass is a "longevity savings account"; strength & grip predict mortalityWell-documented observational links (PURE, etc.); correctly framed as a marker, not a grip-training mandate.Solid
Resistance train 3–4×/week with progressive overloadInterventionally well-supported and uncontroversial.Solid
"Stability" as the fourth pillar (DNS breathing, foot/toe work)Sound fall-prevention logic; the specific DNS protocols are practitioner wisdom, not RCT-backed. No safety concern.Mixed
~1 g protein per lb bodyweightTop of the supported range; meta-analyses plateau muscle gains ~1.6 g/kg. Defensible, mildly aggressive, safe.Mixed

Metabolic health & nutrition

Strong on the core (insulin resistance as an early, central driver), measured on diet (he's anti-dogma and admits the data are weak), and — to his credit — home to his cleanest public reversal: he quit prolonged fasting after concluding it cost him muscle.

Metabolic & nutrition
ClaimEvidenceVerdict
Insulin resistance is an early, upstream driver of the Four HorsemenStrong core (T2D, CVD); more associative for cancer/Alzheimer's. The exact risk multipliers he cites are presented more confidently than the data warrant.Mixed
Measure early with fasting insulin / HOMA-IR / OGTT-with-insulinMechanistically sound — insulin rises before glucose. Ahead of standard-of-care (which he acknowledges); assays aren't well standardized.Solid
CGM for non-diabetics reveals hidden glucose variabilityReal signal, but weak correlation with HbA1c in healthy people and no outcome trial showing benefit. A reasonable teaching tool, unproven payoff.Mixed
"Personal fat threshold" / spillover into visceral & ectopic fatSolid and mainstream (Taylor's work); accurately represented.Solid
Visceral/ectopic fat matters far more than total fat; track via DEXASolid; annual DEXA-for-all is a practice preference, not a necessity.Solid
Three levers (caloric / dietary / time restriction); no single best dietA fair organizing framework; his critique of nutritional epidemiology's weakness is widely shared.Solid
Quit prolonged fasting / aggressive TRE — it cost lean massA documented, evidence-driven reversal of his earlier keto/fasting enthusiasm. The 10-lb figure is personal; the direction is supported.Solid — updated
Distribute protein ~30–50 g/meal at ~2–3 g leucineThe leucine threshold is real, but a recent RCT found even vs skewed distribution didn't change muscle synthesis — weaker than stated.Mixed

Longevity drugs & supplements

This is where his discipline shows most clearly: he's skeptical of the hyped molecules (NMN, resveratrol) and transparent that rapamycin is a bet. His supplement stack is biomarker-target-driven — reasonable, but several targets are "hit the number" strategies whose hard-outcome payoff is unproven.

Drugs
ItemEvidenceVerdict
Rapamycin (~6–8 mg/week) for longevityBest animal data in the field (NIA ITP mice); zero human longevity RCTs, and recent human trials (PEARL) underwhelmed. He's explicit it's an unproven personal bet — credit that — but off-label use carries real immunosuppression/metabolic risk.Safety flag
Metformin for longevity — walked backStopped recommending it for healthy/fit people after RCT evidence it blunts exercise/mitochondrial adaptations. The updated skeptical stance is correct.Solid — updated
NMN / NR (NAD precursors) — skepticalHuman RCTs raise blood NAD but show null/modest hard outcomes. His skepticism matches the evidence.Solid
Resveratrol — "nonsense"Failed to extend lifespan in the NIA mouse program; weak human data. His dismissal is well-supported.Solid
Supplements
ItemEvidenceVerdict
Omega-3 (EPA+DHA) dosed to the Omega-3 Index (~8–12%)Index is a validated biomarker; EPA/DHA have reasonable lipid/CV data. Marginal benefit when already replete is uncertain. Biomarker-guided dosing is sensible.Solid
Vitamin D to a serum target (~40–60 ng/mL)Correcting deficiency is justified; VITAL found little benefit of routine supplementation in replete people. The elevated target is opinion.Mixed
Magnesium (multiple forms, incl. threonate)Correcting a common shortfall is solid; form-specific (threonate-for-brain) claims are thin.Mixed
Methylated B-vitamins to lower homocysteine <9B-vitamins reliably lower homocysteine — but large RCTs show that does not cut events. Hitting the number ≠ benefit.Overstated
Glycine ~2 g / curcumin for sleep & inflammationLow-risk, lightly evidenced.Mixed

Hormone replacement

TRT & women's HRT
ClaimEvidenceVerdict
TRT for symptomatic hypogonadal men — treat the symptom, not the numberWell-aligned with evidence; TRAVERSE was broadly reassuring on CV safety; he correctly notes no prostate-cancer signal in modern data. Appropriately monitored.Solid
Menopausal HRT was wrongly abandoned after the 2002 WHI; start early (timing hypothesis)His core "wrongly abandoned" critique has largely become the mainstream position (NAMS now endorses HRT for symptomatic women <60 / within 10 years). The strong cardioprotection framing is the contested edge.Mixed — leaning solid
Prefer transdermal estradiol + micronized progesteroneObservationally supported (lower clot/breast risk); not from large head-to-head RCTs.Solid

Cancer screening & sleep

His weakest area, and the one that most overlaps his business. The thesis "early detection beats late treatment" is true for a few cancers with good tools and runs into lead-time bias and overdiagnosis for the broad whole-body approach. In 2026 the first big multi-cancer-blood-test RCT (NHS-Galleri) missed its primary endpoint — exactly the caution flagged here.

Screening & sleep
ClaimEvidenceVerdict
Colonoscopy by ~40 (vs guideline 45), shorter intervalsHis most evidence-grounded screening position — colonoscopy removes precancerous polyps; guidelines have trended younger.Solid
Whole-body MRI for asymptomatic average-risk adultsNo mortality-benefit RCT; ~30–40% incidental findings driving biopsy cascades. ACR doesn't recommend it. He admits the "incidentaloma rabbit hole."Overstated
Multi-cancer blood tests (Galleri/MCED)The 2026 NHS-Galleri RCT (~140k) missed its endpoint; ~4× more cancers found (an overdiagnosis-shaped signal); ASCO won't recommend it. Stage-shift is a surrogate, not benefit.Overstated
Low-dose CT / CTA beyond high-risk groupsProven only in heavy smokers; extending to average-risk adults lacks RCT support and adds radiation/incidentalomas.Mixed
"Early detection beats late treatment for the Four Horsemen"True for a few screenable cancers; for broad screening it collides with lead-time/length bias. Earlier ≠ better without a mortality RCT.Mixed
Sleep as a foundational metabolic/cognitive/CV pillarMainstream and uncontroversial.Solid

The conflicts

To his real credit, he keeps a public disclosures page — more transparency than almost anyone in this space. The concern is breadth, and one specific overlap:

Credibility

The fair summary: well above the wellness-influencer baseline — and honest about it.

The pattern — how to use him

Two tiers. Mechanism and direction — apoB causality, insulin resistance, exercise and strength as longevity levers, lower-is-better lipids, measure-early — are reliable; act on them. The aggressive edge — whole-body MRI, multi-cancer blood tests, rapamycin, neonatal-level lipid targets for low-risk people — treat as hypotheses, weigh the cost and overdiagnosis, and run them through is it strong enough to act? and the funding filter (he co-owns a testing clinic). When he says "I'm not sure," believe him — that's the tell that he's worth listening to.

For you

Of all the sources on this desk, Attia maps most closely onto your Life OS — much of your roadmap already is his framework (apoB and VAT targets, HOMA-IR tracking, Zone 2, strength, DEXA, the four-horsemen lens). So the job here is mostly "you're already doing it," with the brake applied to his aggressive edge.

His position → your call
His positionYour call
apoB is causal; lower it early and aggressivelyHigh-value and genuinely yours. Your ApoB 89 is "watch," you're South Asian + insulin-resistant, and your target is already <70 — exactly the higher-risk profile where aggressive lowering has real absolute benefit. The "ultra-low for everyone" overreach doesn't apply to you because you're not low-risk.
Measure Lp(a) onceAlready done — Lp(a) 15 nmol/L, optimal. No further action.
Insulin resistance is the upstream driver; reverse itSquarely your situation (HOMA-IR 3.65, TCF7L2) — and your whole plan (Zepbound cut, fasted Zone 1, protein, muscle) is the Attia-prescribed response. Aligned.
VO₂ max, Zone 2, strength, muscle as longevity leversAdopt the structure (you do) — just hold the causal framing loosely; the mortality data are observational. Direction is right.
Track visceral fat via DEXA; ~1 g/lb proteinAlready doing both — you track VAT (65.6 → <52 target) on DEXA and run ~170 g protein (~2 g/kg). Pure Attia.
Rapamycin for longevityNot now. Animal-data bet, immunosuppressive — wrong trade mid-cut with other priorities. Park it as "watch," exactly where your own filters would put it.
Whole-body MRI / multi-cancer blood testsApply the brake. Unproven for mortality, overdiagnosis-prone (Galleri's RCT just failed), and it's where his clinic profits. Colonoscopy by ~45 is the reasonable, evidence-backed piece for you at 41.
TRT (treat symptoms, not numbers)Says no TRT for you — your total T ~597 with optimal Free T isn't a symptomatic deficiency. Matches your own deprioritization.
Omega-3 to an index target; metformin cautionYour omega-3 index is already optimal → maintain, don't escalate. You're on tirzepatide, not metformin — but note his point that some glucose drugs can blunt training adaptations is a question worth asking your prescriber about GLP-1s and your Zone 2.

Check it yourself

Re-run any row through the evidence ladder, correlation vs causation, observational studies, surrogate vs hard outcomes, base rates, and is it strong enough to act?

Sources · peterattiamd.com essays/AMAs (apoB, 10-year risk, VO₂ max, metformin, rapamycin, HRT, disclosures) · VO₂max & mortality: Mandsager 2018 · Galleri RCT (2026): Medscape · Whole-body MRI evidence: Diagnostic Imaging · Lipid-target critique: Sensible Medicine, Skeptical Cardiologist · Outlive reviews: Eric Topol, LessWrong · Biograph: Radiology Business. An evidence assessment of public claims, not a personal judgment — and not medical advice; confirm anything touching your medications with your prescriber.

Learn · Shawon Chowdhury · a source scorecard, kept rough on purpose · not medical advice