What are you looking to achieve and monitor with your blood glucose monitor? (AMA #3)

Attia describes a clean natural experiment: at Fenway Park, he had french fries. He had fasted all day and worked out beforehand. The fries did not produce a spike. He 'got to have the fries without the badness.' This is the calibration the device enables: you can discover exactly which indulgences are metabolically tolerable under which prior conditions (fasted + trained = buffered). Without the CGM, this learning is impossible — you are driving the racecar without a tachometer.

Attia's specific insight is that the CGM renders the HbA1c entirely redundant and inferior. The A1c gives you one number every 3 months, assumes normal red-cell lifespan, and tells you nothing about variability. The CGM gives you average AND SD at any time window you choose. Monitoring both together allows detection of a pattern that A1c completely misses: a patient who is fine on average but has massive swings driven by specific meal patterns.

This is Attia's core CGM framework: the device is not a glucose device, it is an insulin proxy device. Every decision about diet, exercise timing, and meal composition runs through the question 'will this produce a glucose spike, and therefore an insulin spike?' The goal is to keep glucose low and flat. The CGM makes that optimization loop real-time and legible where previously it was invisible. He envisions this becoming the standard of care: 'my hope is that in 10 years maybe that's ambitious I would hope that the hemoglobin a1c can't even be ordered on a lab and everyone just has a CGM.'

Attia's Fenway Park case study is a clean example of what CGM makes possible that no other tool can: he knew in real time that the fries did not produce a spike. Without the CGM, that knowledge is unavailable. The practical implication is that the CGM does not just tell you what to avoid — it tells you what you can get away with under the right prior conditions, which is metabolically valuable information. Exercise and fasting are the two levers he explicitly identifies for creating absorptive capacity before an indulgence.

Attia's beta thalassemia trait means his red blood cells live significantly longer than 120 days, causing more hemoglobin glycation to accumulate over their extended lifespan. His lab A1c reads 5.6–6.0, which would conventionally put him in the pre-diabetic range. His CGM-based average glucose, tracked continuously, imputes to a true A1c equivalent of 4.5–5.0 — a radically different clinical picture. He has also seen the reverse in patients: short-lived red cells producing falsely low A1c despite poor glycemic control.

Attia explicitly addresses the FDA regulatory bind: making CGM OTC would require the FDA to either remove real-time feedback (the core value) or dilute accuracy (also destructive). Neither is acceptable. The pragmatic path today is physician prescription. He envisions a future where CGM data replaces HbA1c even for life insurance underwriting — 'you just wear the CGM for two months and the data comes from that as opposed to actually measuring this nonsensical number.'

The mechanism is internal competition rather than guilt: Attia describes himself as 'kind of a competitive person internally much more competitive internally than externally' — he 'just can't stand to see spikes of glucose.' The CGM converts a future health consequence (insulin resistance) into an immediate, legible loss condition (a spike on the display). That immediacy is exactly what long-horizon health goals cannot provide on their own. The behavioral result: he skipped a cookie on a plane not because of any intellectual knowledge about carbohydrates, but because he could not face having to look at the spike.

Attia uses the CGM report to simultaneously track two numbers: average glucose (proxy for baseline insulin level) and the standard deviation of glucose (proxy for reactive insulin secretion). His goal is to keep both low. He can generate these reports on 7-, 14-, 30-, and 90-day windows and track trends. The SD metric is entirely invisible on standard lab work, which is why Attia considers the CGM such a leap over the standard annual HbA1c: 'the a1c is not telling you anything about the variability.'

The beta thalassemia example is Attia's own body: his red blood cells outlive their expected 90–120 day lifespan, meaning they accumulate more glycation over time than a normal cell would, which inflates the A1c reading. His CGM, tracking actual glucose continuously, reveals that his true average glucose would correspond to a 4.5–5.0 A1c — a full point or more below what the lab reports. He makes the same point in the other direction: patients with hemolytic conditions or rapid red-cell turnover will under-estimate glycemic exposure. The CGM is right in both cases; the A1c is wrong.

Glucose is measurable at point-of-care because a simple enzymatic reaction (glucose oxidase) yields an instantaneous electrical signal. Insulin requires either a radioimmunoassay or an ELISA — multi-step processes involving sequential enzyme additions and wash steps that take minutes in a lab, not seconds on a chip. The solution Attia sees coming is computational: train a regression model on simultaneous CGM data and blood draws to build a patient-specific insulin prediction curve off glucose dynamics. But that requires large individual-level training sets and has not been solved yet.

Attia's practical advice: physicians write prescriptions for far more dangerous things every day — opioids, hormones — and writing a CGM script for a health-conscious non-diabetic patient is not a stretch. He explicitly invites patients to 'just say doc I need a CGM.' In the longer arc, he envisions CGM data replacing HbA1c in life insurance underwriting — wear a CGM for two months instead of getting a blood draw.