Journal Club with Dr. Peter Attia | Metformin for Longevity & The Power of Belief Effects

The 2014 Bannister study found that type 2 diabetics on metformin had a ~15% lower all‑cause mortality than matched non‑diabetics. Attia explains that result relied on “informative censoring”: diabetics who worsened and left metformin (often by dying or needing stronger drugs) were removed from analysis, biasing survival upward. A newer Danish registry study (Keys et al.) with ~500,000 people, discordant twins, and sensitivity analyses (with and without censoring) showed diabetics on metformin still had *higher* mortality (hazard ratios ~1.3–2.1) than non‑diabetics. Metformin may still help diabetics versus *no* treatment, but there’s no credible evidence yet that it extends life beyond normal in healthy people.

Attia walks through Table 1 of the Keys paper to show how, even after matching age, sex, BMI, etc., medication use diverged massively: far more diabetics were on lipid‑lowering, antihypertensive, antiplatelet, and other drugs. Statistical adjustments (three levels of models) reduce but do not erase these differences, and you can never fully correct for unmeasured behaviors (diet quality, exercise, health consciousness). Large sample size and long follow‑up increase power, but without randomization you can’t cleanly isolate the causal effect of metformin.

Attia emphasizes that the primary causes of insulin resistance and type 2 diabetes are energy imbalance, low physical activity, excess ectopic fat (fat in muscle, liver, pancreas), poor sleep, and chronic stress/hypercortisolemia. The earliest warning sign is *elevated insulin* with normal glucose, not glucose itself. Exercise (especially consistent, higher‑volume activity) and adequate sleep dramatically improve insulin sensitivity, sometimes cutting glucose disposal deficits in half within a week of better sleep. He still uses metformin for insulin‑resistant patients but does not see it as a blanket longevity tool for fit, insulin‑sensitive people.

Metformin weakly inhibits mitochondrial complex I, shifting fuel partitioning and raising basal lactate. When Attia started doing lactate‑based zone 2 training, his fasted resting lactate on metformin was ~1.6 mmol/L instead of <1.0 mmol/L, suggesting he was effectively slightly anaerobic at baseline. That, plus data that metformin may blunt strength and hypertrophy gains, pushed him to stop after years of use. For athletes or people prioritizing performance and muscle mass, those tradeoffs may outweigh any speculative geroprotective benefit.

In the Gu et al. nicotine study, experienced smokers vaped the *same* low nicotine dose but were told it was low, medium, or high. Their subjective ratings matched the instructions (they “felt” low/medium/high), but critically, fMRI showed that activity in thalamic–ventromedial prefrontal cortex circuits—central to attention and learning—scaled with the believed dose. The more nicotine they *thought* they got, the stronger the circuit activation, even though blood nicotine levels were essentially identical. This goes beyond classic placebo (drug vs no drug) to show true dose‑response in belief effects at the neural level.