Medicine 2.0 is the system most adults already inhabit: an annual physical, statins after the first cardiac event, a colonoscopy once you turn 50, and a slow accumulation of pill bottles labeled with the names of organs that have already started failing. Medicine 3.0, the term Peter Attia popularized, swaps the trigger. You start measuring upstream risk in your 30s, you treat the biomarkers (not the symptoms) in your 40s, and the goal is not "live to 90" but "lift, walk, think, and recognize your grandchildren at 90."
The framework is right. What is sold around it (concierge clinics with 5-figure annual fees, branded supplement bundles, the implicit suggestion that buying enough Brain Octane Oil grants exemption from biology) is mostly not. This article separates the two.
What is the difference between healthspan and lifespan?
Lifespan is total years lived. Healthspan is total years lived in good function: walking unaided, lifting your own bodyweight, not on dialysis, not in cognitive decline. The two have decoupled. Olshansky 2018 (JAMA viewpoint) noted that US life expectancy at 65 has gained roughly 5 years since 1970, but disability-free life expectancy has gained closer to 1-2 years over the same period ( Olshansky 2018 ). Most of the lifespan gain has been spent inside hospitals.
The MacArthur Foundation Research Network on Successful Aging, summarized in Rowe & Kahn's 1987 Science paper, was the first to formally split aging into "usual" (typical decline trajectory) and "successful" (low disease burden, preserved cognitive and physical function) ( Rowe & Kahn 1987 ). Their cohort work showed individual variation within any age band is enormous; a 75-year-old can have the cardiovascular profile of a typical 55-year-old or a typical 85-year-old depending on a small number of measurable variables.
The Medicine 3.0 thesis sits on top of that empirical observation. If between-person variance at 75 is wide, and that variance is driven mostly by lifestyle and metabolic load accumulated over decades, then the optimal time to intervene is not at 75. It is at 35.
What are the four horsemen of premature death?
After age 50, four disease clusters account for roughly 80% of mortality in high-income countries:
- Atherosclerotic cardiovascular disease. The single largest killer. Driven by ApoB-particle exposure over decades, hypertension, smoking, and insulin resistance. CTT 2019 meta-analysis (n=186,854) showed each 1 mmol/L LDL-C reduction cuts major vascular events ~22% ( Cholesterol Treatment Trialists Collaboration 2019, n=186854 ). The implication: starting earlier and going lower compounds.
- Cancer. Roughly 30-35% of premature deaths, with breast, colorectal, lung, prostate, and pancreatic dominating. Most early detection is screening-based (colonoscopy, mammography, low-dose chest CT for smokers), and most lifestyle leverage runs through metabolic health.
- Neurodegeneration. Alzheimer disease, vascular dementia, Parkinson disease. Risk windows open in the 60s; the molecular runway starts decades earlier. Sleep architecture, cardiometabolic control, and APOE-genotype-aware planning are the actionable inputs.
- Type 2 diabetes and accelerated metabolic aging. Not a horseman of acute death so much as a fuel injector for the other three. Each year spent insulin-resistant raises the all-cause mortality hazard.
The Medicine 3.0 move is to track the upstream markers of all four (ApoB, fasting insulin, hsCRP, eventually CAC scores and methylation panels) and to treat what is trending wrong in your 30s and 40s, not what has finished going wrong in your 60s.
Two operational notes worth flagging. First, the four horsemen are not independent. Insulin resistance is a multiplier on all three of the others: it raises atherosclerosis hazard via small-dense LDL particles and chronic vascular inflammation, it raises cancer hazard via insulin and IGF-1 signaling, and it raises dementia hazard via the cerebrovascular and Alzheimer pathways collectively sometimes called "type 3 diabetes." Fixing the metabolic axis tends to bend the other three trajectories at the same time, which is why most rational longevity programs put cardiometabolic management at or near the top of the priority list rather than treating it as a fourth equal item. Second, screening is asymmetric across the four. Cancer screening (colonoscopy, mammography, low-dose chest CT for heavy smokers) has the firmest randomized-trial evidence of any preventive step in this list. Cardiac CT-based CAC scoring has strong observational support: Detrano 2008 MESA (n=6,722) showed CAC scores above 300 raised coronary event hazard roughly 10x versus CAC of 0 across racial and ethnic subgroups ( Detrano et al. (MESA) 2008, n=6722 ). Neurodegeneration screening is the youngest of the four and the most uncertain; APOE genotyping plus cognitive baselining at 50 is reasonable, but no validated dementia-prevention RCT pipeline yet exists at the population level.
What does proactive medicine actually look like in practice?
Three concrete practices, in order of leverage.
1. Quarterly-to-annual biomarker tracking, not just a "physical." A typical insurance-paid annual visit measures total cholesterol, LDL-C (often calculated, not direct), HbA1c, and a chemistry panel. That panel will miss elevated ApoB at borderline LDL, miss elevated Lp(a) (a genetic risk factor present in ~20% of the population), and miss early insulin resistance. The minimum proactive panel: ApoB, Lp(a) once in your life, hsCRP, fasting insulin, HbA1c, comprehensive metabolic panel, and full thyroid (TSH plus free T4 plus free T3). Add testosterone and SHBG for men, full menstrual-cycle hormonal panel for women. See What Your Doctor Isn't Testing.
2. Lifestyle intervention as first-line therapy. Mandsager 2018 (n=122,007) found each 1-MET higher cardiorespiratory fitness associated with ~11% lower all-cause mortality ( Mandsager et al. 2018, n=122007 ). The hazard ratio for low CRF (bottom 25%) versus high CRF (top 2.5%) was 5.04. No pill achieves a 5x mortality hazard reduction.
3. Evidence-tiered supplementation, not enthusiasm-tiered. The shelf gets stacked from Tier 1 down: compounds with replicated outcome trials; then mechanistic-plus-correlational support; then mouse data plus a podcast. Most adults stop at Tier 1.
The honest framing on lifestyle as therapy: the median dose of structured exercise needed to move a typical adult from the bottom CRF quartile (where 5-year all-cause mortality hazards are roughly 5x baseline) into the middle two quartiles is 3-5 hours per week of intentional aerobic plus 2-3 hours per week of resistance training. That is achievable on a 40-hour-per-week working schedule with no exotic equipment. The Mandsager 2018 hazard ratio of 5.04 between bottom and top fitness quintiles dwarfs the 0.78 hazard ratio (a 22% reduction) for statins per mmol/L LDL drop seen in CTT 2019; lifestyle is not a "soft" intervention compared with pharmacotherapy, it is the largest-effect intervention in the toolbox. Anyone selling you a Tier 3 protocol while you skip the gym has the priority order inverted.
The data-gathering case
Two patterns make biomarker tracking work. First, repeated measurement turns one ambiguous number into a slope. A single LDL-C of 130 mg/dL is a question. Three quarterly measurements at 130, 135, 142 is an answer. Second, slope detection in your 40s gives you 10-20 years of lead time on most of the diseases that will kill you, which is enough time for low-risk interventions (diet, exercise, statin if needed) to bend the trajectory before it requires high-risk ones (surgery, oncology, intensive care).
Concrete tools:
- The biological-age estimator at /tools/biological-age/ implements the Levine 2018 PhenoAge clock from 9 routine blood markers, giving you a single number that compresses cardiometabolic and inflammatory signal into something trendable across visits ( Levine et al. 2018 ).
- The bloodwork tracker at /tools/bloodwork-tracker/ lets you log per-marker results over time and flags trajectories outside the optimal range. The optimal-range framing matters: "in normal range" can hide a 3-year trend toward the upper bound.
- The biomarker dictionary articles in the research index document why each marker matters, what the optimal range is, and how to interpret movement.
The honest framing: this is observational data analysis on yourself. It is not an RCT of your life. It does not tell you whether the intervention you tried in Q1 caused the Q3 result. It does tell you whether your trajectory is moving in a direction that 10-year mortality cohorts associate with worse outcomes. That is enough leverage to act on.
The specific compound layer
Stacked from highest evidence floor to lowest. Lifestyle dominates each tier; supplements amplify margins.
Tier 1: replicated outcome benefit, low side-effect profile.
- Creatine monohydrate, 5 g/day. The most-studied supplement in sports science. Kreider 2017 ISSN position stand summarizes ~700 trials documenting 5-15% strength and power gains, modest lean-mass gains, and a smaller but real cognitive case under stress ( Kreider et al. 2017 ). See the creatine compound entry.
- Omega-3 (EPA-dominant), 2-4 g/day if you have elevated triglycerides or established cardiovascular risk. Bhatt 2019 REDUCE-IT (n=8,179) showed icosapent ethyl (4 g/day pure EPA) cut MACE 25% over 4.9 years in statin-treated patients with elevated triglycerides ( Bhatt et al. (REDUCE-IT) 2019, n=8179 ). The mixed-formulation contrast (Nicholls 2020 STRENGTH null result) is a reminder that "fish oil" is not one drug. See /compounds/omega-3/.
- Vitamin D, only to correct a measured deficit. Manson 2019 VITAL (n=25,871) tested 2,000 IU/day vs placebo for 5.3 years and found null primary endpoints for cardiovascular events and total cancer ( Manson et al. (VITAL) 2019, n=25871 ). Subgroup analyses showed cancer-mortality signal in those with BMI under 25. The evidence does not support dosing in already-replete adults; it does support replacement to a 25-OH D level of 40-60 ng/mL. See /compounds/vitamin-d3-k2/.
Tier 2: solid mechanism, moderate human data.
- Magnesium glycinate or threonate, 200-400 mg elemental at night, especially if your serum magnesium is low-normal or your diet is processed-food-heavy.
- Adequate protein (1.6-2.2 g/kg/day) is not a supplement, but most adults under-eat it. Morton 2018 meta plateaus around 1.62 g/kg/day for active adults ( Morton et al. 2017, n=1863 ).
Tier 3: speculative, off-label, or under-studied in humans.
- Rapamycin, weekly cycled at 5-6 mg, off-label for longevity. Mannick 2018 (n=264) showed TORC1 inhibition improved post-vaccine antibody titers and reduced respiratory infections in healthy elderly ( Mannick et al. 2018, n=264 ). Lifespan-in-humans data does not exist. Side effects include mouth sores and dyslipidemia. See /posts/rapamycin-cycling-protocols/ and /compounds/rapamycin/.
- NAD+ precursors (NMN, NR), senolytics, peptides. Mechanism-strong, RCT-thin in humans. Effect sizes on hard outcomes are not yet documented.
The asymmetric framing: Tier 1 has a strong floor and a low ceiling. Tier 3 has an unknown floor and a possibly-high ceiling. Most adults belong on Tier 1 plus 1-2 Tier 2 entries. The cost of running a 4-supplement Tier 1 stack is roughly $25-40/month in 2026 US prices.
Where Attia and others go too far
The Medicine 3.0 framework, as a system, is sound. The cultural product around it has three honest weaknesses that the loudest voices rarely flag.
1. The longevity-as-religion problem. When the goal becomes "maximize lifespan" rather than "maximize the years you can live well," the calculus stops making sense. Attia himself has written about this: spending 3 hours/day on training and biomarker management at age 45 to gain a probabilistic 2 years at age 90 is a bet most people, on reflection, would not take. The framework is a tool. It is not an identity.
2. The expensive-test trap. Whole-body MRI screening, full-genome sequencing for routine clinical use, and quarterly methylation-clock panels are mostly not yet supported by outcome evidence. They produce data. Whether the data changes your decision (versus just generating anxiety and incidentalomas) is the question almost no marketing material answers honestly. Use them when the marginal information would change a treatment decision; skip them when they would not.
3. The signal-to-noise problem in supplement stacks. Bryan Johnson's Blueprint protocol lists ~70 compounds. There is no plausible world in which all 70 have additive benefit; most are at best neutral, some interact, and the cost-per-marginal-benefit collapses past about 5-7 well-chosen compounds. The framework is right that supplementation has a place. It is also right (when stated honestly) that the place is small.
A specific dissenter: Matt Kaeberlein, one of the central rapamycin researchers, has been publicly skeptical of pharmaceutical longevity branding. His repeated argument: the demonstrated effect sizes from sleep, cardiovascular fitness, and resistance training are larger and more reliable than anything the supplement aisle currently sells, and the population-scale gain from getting the bottom-quartile-fitness population off the bottom would dwarf any pharmaceutical longevity intervention now in trials.
A practical Medicine 3.0 starter checklist
The 90-day on-ramp, in order. No one needs all of it on day 1.
Weeks 1-2: data acquisition.
- Order a comprehensive blood panel: lipid panel with ApoB, Lp(a) (once-in-life), hsCRP, fasting insulin, HbA1c, comprehensive metabolic panel, full thyroid, 25-OH vitamin D, ferritin, CBC. Add testosterone + SHBG for men.
- Compute an estimated VO2 max (treadmill submaximal, or wearable proxy). Log it.
- Take baseline body composition: weight, waist circumference, ideally a DEXA scan.
Weeks 3-12: Tier 1 lifestyle build.
- 3-4 Zone-2 cardio sessions per week, 45-60 min each, at 65-75% HRmax.
- 2-3 resistance training sessions per week, 30-45 min each, full-body compound lifts.
- Protein at 1.6-2.0 g/kg/day, distributed across 3-4 meals.
- Sleep: 7-9 hours, consistent timing, bedroom 16-19°C.
- Weight stable in optimal-BMI range (men 22-25, women 21-24 BMI as a starting heuristic).
Months 4-6: Tier 1 supplementation, retest.
- Add creatine 5 g/day, omega-3 if measured deficient or if you have CV risk, vitamin D only to correct a measured low 25-OH D.
- Re-run the blood panel at month 6. Compare to baseline. Look for ApoB trajectory, hsCRP, fasting insulin.
Months 7-12: targeted layering.
- If ApoB still above 80 mg/dL despite lifestyle, talk to a clinician about a statin or ezetimibe.
- If sleep architecture is the issue, audit hygiene before pharmacology. See /posts/sleep-hygiene-ranked/.
- If you remain interested in Tier 3 interventions (rapamycin, off-label), this is the earliest reasonable time to discuss them with a willing clinician. Not before.
Annual.
- Repeat the panel. Add a CAC score in your 40s. Consider a methylation-clock panel if budget allows; treat it as one signal among many, not a verdict.
- Re-audit the stack. Drop anything that has not produced a measurable change in the marker it was supposed to move.
The whole thing is dull on purpose. Medicine 3.0 done well looks less like a podcast appearance and more like 90% adherence to a list a 1990s exercise physiologist would recognize, with a quarterly blood panel, a stable supplement stack of 3-5 things, and one or two off-label experiments where the evidence justifies the risk profile.
