Aging Interventions: Calibrated Verdict Index

A continuously updated, evidence-weighted map of aging interventions. Verdicts follow a locked methodology (methodology.md) and are triangulated against five calibration anchors (CALIBRATION_ANCHORS.md).

Last index update: 2026-04-25. Total pages: 42 (38 interventions + 4 interactions).

📄 Paper: Where the Longevity Discourse is Systematically Wrong — synthesis paper covering the methodology + the eight findings the popular discourse gets wrong + what deserves more attention. HTML version · Machine-readable verdict table (CSV)

Verdicts use these bands (lowest evidence → highest):

• Insufficient evidence — too little data to form a verdict
• Mostly hype — popular intervention; tested rigorously and failed, or never had higher-tier support
• Mixed — tier-appropriate evidence exists but replication has failed or population dependence is severe
• Suggestive — T3-T4 evidence with replication; human data absent or null/early
• Probable — T3 evidence with at least suggestive human data, OR T1 evidence in defined population
• Strong — T1 / massive cohort + RCT evidence on hard endpoints; mechanism understood

If you do nothing else from this database, do exercise (aerobic + resistance), prioritize sleep (~7h), and do caloric moderation without specific timing fixation. These represent the bulk of the population-level longevity benefit.

Strong (in humans)

The interventions with the most overwhelming evidence base. These are the non-negotiables.

• Aerobic exercise — anchor; 20-40% all-cause mortality reduction at moderate doses across millions of person-years.
• Resistance training — 10-20% all-cause mortality + unique sarcopenia / functional-capacity protection.
• Sleep — Strong observational / Probable causal; ~7 hours nightly is the population optimum.
• Statins (secondary prevention) — overwhelming evidence in established CVD.

Strong (in mice — anchor band for animal evidence)

• Caloric restriction — anchor for mouse-Strong; Suggestive in humans (CALERIE-2 biomarker results, no mortality data).

Probable

Solid evidence at the relevant evidence tier and population.

• Rapamycin — anchor; T3 mouse, T2 human surrogate; the best-evidenced longevity drug candidate.
• Acarbose — Probable mice / Suggestive humans; underappreciated relative to ITP evidence quality.
• 17α-estradiol — Probable in male mice / Insufficient evidence humans; striking single-sex ITP finding.
• Canagliflozin — Probable male mice + Probable in CV/CKD/DM hard endpoints.
• GLP-1 agonists — Probable in obese / CV-risk per SELECT trial.
• Statins (primary prevention 40-75) — clear net benefit at elevated CV risk.
• Sauna / heat exposure — KIHD cohort dose-response is among strongest non-exercise lifestyle signals.
• Creatine + resistance training — Probable for sarcopenia / cognition adjunct.
• HRT in early-menopause women — Probable when started ≤age 60 / within 10 years of menopause.
• TRT in documented hypogonadism — Probable per TRAVERSE.
• Zone 2 / VO2max framework — Probable for the broader VO2max claim / Mixed for the Zone-2-is-optimal-protocol claim.

Suggestive

Real biology, replication or human translation incomplete.

• NMN / NR (NAD+ precursors) — anchor; ITP failed NR; cousin to NMN.
• Senolytics (D+Q, fisetin) — strong mouse, early human, mixed clock signals.
• Spermidine — SmartAge primary endpoint null; observational signal positive.
• GlyNAC — single-group dominance; mechanism coherent.
• Sulforaphane — best-evidenced natural Nrf2 activator.
• Plasma exchange / TPE — Conboy lab pilots; commercial layering high.
• Taurine — leaning toward downgrade; 2025 Science follow-up undermined the age-decline premise.
• Omega-3 (general) — Suggestive general / Probable in REDUCE-IT-equivalent population.
• Lithium (low-dose, observational + MCI) — drinking water signals; supplement microdosing weaker.
• CoQ10 (HF / statin myalgia) — narrow indications only.
• EGCG / green tea consumption — observational; supplement form has hepatotoxicity caveat.
• Berberine (cardiometabolic) — modest glycemic/lipid effects; longevity claim weaker.
• Cold exposure (narrow indications) — mood, BAT, recovery; longevity claim weak.

Mixed

Tier-appropriate evidence but replication failures or severe population dependence.

• Metformin — ITP null in mice; observational human data confounded; MET-PREVENT 2025 shows exercise blunting.
• Time-restricted eating — Liu 2022 NEJM showed null vs matched-calorie comparator.
• Vitamin D — VITAL primary null; modest cancer-mortality signal real; hip-fracture signal in women.
• Hyperbaric oxygen — single-group + COI; replication missing.
• Statins (>75 healthy primary) — USPSTF: insufficient evidence; STAREE pending.

Mostly hype

Popular interventions where the evidence has failed or never reached higher tiers.

• Resveratrol — anchor for Mostly hype; ITP-failed, SIRT1 mechanism not supported in humans.
• Curcumin — bioavailability ceiling; PAINS critique; modest evidence only for OA pain.
• EGCG (high-dose supplements) — hepatotoxicity caveat.
• Quercetin (standalone) — senolytic case lives in D+Q, not standalone.
• Methylene blue — TauRx Alzheimer's program failed; biohacker framing unsupported.
• Pterostilbene / PQQ / Astaxanthin — consolidated; same pattern.
• Lithium microdosing (broader claims) — supplement-form longevity claim unsupported.
• CoQ10 (general longevity) — heart failure data ≠ aging data.

Insufficient evidence

Not enough data to form a verdict, generally because the field is preclinical-only.

• Yamanaka factors / partial reprogramming — most preclinically exciting category; first FDA-cleared trial April 2026.
• Klotho upregulation — strong mouse evidence; no human supplementation pathway established.

Interactions / Combinations / Antagonisms

Synergies and important antagonisms between interventions.

• Rapamycin + Acarbose — Probable; among largest ITP combined effects (+34% males, +28% females).
• Rapamycin + Metformin — Probable but driven by rapamycin; metformin adds little.
• Metformin + Exercise (antagonism) — Probable antagonism; metformin blunts exercise adaptations.
• GLP-1 + Resistance Training (mitigation) — Probable mitigation of GLP-1-induced muscle loss.

How to use this index

• For decision-making: prioritize Strong-band interventions; treat Probable as supportable; treat Suggestive as worth tracking but not committing to; treat Mostly hype as supplement-industry signal rather than evidence.
• For staying current: verdict changes are logged in the repository commit history. Open issues on the repo for new evidence or contested verdicts.
• For the methodology behind verdicts: read methodology.md and CALIBRATION_ANCHORS.md. Both are locked; revisions require explicit changelog entries.
• For unresolved questions: open a GitHub issue.

What this database is not

• Not personal medical advice. Verdicts are population-level evidence summaries.
• Not exhaustive. New interventions and verdict updates appear via commits to the repository.
• Not free of judgment calls — every verdict is a calibration claim defensible against the methodology and anchors. Disagreements should be raised via the question-resolver process, not silent edits.