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
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.
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.
| Claim | Evidence | Verdict |
|---|---|---|
| "Medicine 3.0" — proactive, personalized, prevention-first | A 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, risk | Well-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 90 | A goal-setting heuristic; the underlying specificity/reserve logic is sound. | Solid |
| Emotional health is the "fifth horseman" / foundation | A 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.
| Claim | Evidence | Verdict |
|---|---|---|
| apoB (particle number), not LDL-C, is the central causal driver | Strong — mechanism + discordance analyses; endorsed by the 2021 EAS consensus. | Solid |
| Causality proven by Mendelian randomization + RCTs | The 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 elevated | Solid; societies now endorse once-in-a-lifetime screening. His nmol/L preference is correct. | Solid |
| Statins first-line; most "intolerance" is nocebo/manageable | Among the most robust evidence bases in medicine (SAMSON for nocebo). | Solid |
| Add ezetimibe / PCSK9 / bempedoic acid to drive apoB low | Solid pharmacology and outcomes — but those trials were in higher-risk populations. | Solid |
| A CAC score of zero does NOT mean zero risk | Correct — 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 ~40 | Direction 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 plaque | The 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 people | Small 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.
| Claim | Evidence | Verdict |
|---|---|---|
| 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×/week | The 4×4 protocol is RCT-validated for raising VO₂ max. | Solid |
| Zone 2, ~3–4 h/week, lactate ~1.7–2.0 mmol | Endurance 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 mortality | Well-documented observational links (PURE, etc.); correctly framed as a marker, not a grip-training mandate. | Solid |
| Resistance train 3–4×/week with progressive overload | Interventionally 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 bodyweight | Top 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.
| Claim | Evidence | Verdict |
|---|---|---|
| Insulin resistance is an early, upstream driver of the Four Horsemen | Strong 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-insulin | Mechanistically 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 variability | Real 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 fat | Solid and mainstream (Taylor's work); accurately represented. | Solid |
| Visceral/ectopic fat matters far more than total fat; track via DEXA | Solid; annual DEXA-for-all is a practice preference, not a necessity. | Solid |
| Three levers (caloric / dietary / time restriction); no single best diet | A fair organizing framework; his critique of nutritional epidemiology's weakness is widely shared. | Solid |
| Quit prolonged fasting / aggressive TRE — it cost lean mass | A 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 leucine | The 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.
| Item | Evidence | Verdict |
|---|---|---|
| Rapamycin (~6–8 mg/week) for longevity | Best 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 back | Stopped 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) — skeptical | Human 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 |
| Item | Evidence | Verdict |
|---|---|---|
| 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 <9 | B-vitamins reliably lower homocysteine — but large RCTs show that does not cut events. Hitting the number ≠ benefit. | Overstated |
| Glycine ~2 g / curcumin for sleep & inflammation | Low-risk, lightly evidenced. | Mixed |
Hormone replacement
| Claim | Evidence | Verdict |
|---|---|---|
| TRT for symptomatic hypogonadal men — treat the symptom, not the number | Well-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 progesterone | Observationally 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.
| Claim | Evidence | Verdict |
|---|---|---|
| Colonoscopy by ~40 (vs guideline 45), shorter intervals | His most evidence-grounded screening position — colonoscopy removes precancerous polyps; guidelines have trended younger. | Solid |
| Whole-body MRI for asymptomatic average-risk adults | No 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 groups | Proven 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 pillar | Mainstream 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:
- Biograph. He co-founded and was Chief Medical Officer of a longevity-diagnostics startup whose clinics sell exactly the aggressive testing he advocates — whole-body MRI, CT angiography, DEXA, 1,000+ biomarkers (memberships ~$7,500–$15,000/yr). This is the most direct advocacy-meets-business overlap, and it sits on his weakest-evidenced positions.
- Early Medical. His concierge practice reportedly starts around $100k/year, capped at ~100 patients. Not deception, but it frames who his protocol realistically serves — the "medicine for the wealthy" critique.
- AG1. Scientific advisor + investor + past podcast sponsor — a triple overlap on a weakly-evidenced supplement.
- Broad portfolio. Equity/advisory roles across longevity, diagnostics, and health tech (disclosed on his site). Transparent, but pervasive — many recommendations touch a category he holds a stake in.
Credibility
The fair summary: well above the wellness-influencer baseline — and honest about it.
- Rigor. He cites primary literature, grades evidence ("promising / fuzzy / nonsense"), and flags uncertainty. Eric Topol called Outlive "highly informative."
- He updates. Documented reversals on fasting, metformin, and resveratrol — a genuine marker of intellectual honesty, and the trait that most separates him from Huberman.
- The consistent bias. His thumb is on the scale toward more testing and more early intervention — which (a) outruns the evidence in several places (MRI, MCED, rapamycin, ultra-low lipid targets) and (b) aligns with his businesses. Trust the mechanism; treat the screening enthusiasm as a hypothesis his clinic profits from. (As with the Huberman card, personal controversies are separate from, and not the basis of, this assessment.)
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 |
|---|---|
| apoB is causal; lower it early and aggressively | High-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) once | Already done — Lp(a) 15 nmol/L, optimal. No further action. |
| Insulin resistance is the upstream driver; reverse it | Squarely 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 levers | Adopt 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 protein | Already doing both — you track VAT (65.6 → <52 target) on DEXA and run ~170 g protein (~2 g/kg). Pure Attia. |
| Rapamycin for longevity | Not 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 tests | Apply 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 caution | Your 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.