CGM finds cardiometabolic risk that HbA1c and fasting glucose miss

Design

cross-sectional analysis within a prospective cohort, blinded CGM

Population

1,356 Framingham participants without diabetes or cardiovascular disease, mean age 59.3 years

Primary outcome

association of CGM measures with 10-year PREVENT cardiovascular risk estimates and individual risk factors

Effect

1 SD more time >140 mg/dL: +2% 10-year risk estimate, 21–26% higher odds of hypertension and dyslipidaemia after adjusting for fasting glucose

In 1,356 Framingham Heart Study participants without diabetes or cardiovascular disease, a blinded Dexcom G6 Pro sensor was worn for up to 10 days and the resulting glycaemic measures compared with 10-year cardiovascular risk estimated by the PREVENT equations. Mean age was 59.3 years and 55.8% were women; 92.6% were non-Hispanic White.

CGM measures tracked risk even after adjusting for fasting plasma glucose. One standard deviation more time above 140 mg/dL — 15.4 percentage points of the wear period — was associated with a 2% higher 10-year cardiovascular risk estimate and 21 to 26% higher odds of hypertension and dyslipidaemia. Clustering the sensor data produced glycaemic profiles that disagreed with conventional classification: between 48.1% and 84.2% of people labelled prediabetic by fasting glucose or HbA1c fell into the higher-dysglycaemia profiles, and those profiles carried 6 to 7% higher estimated risk.

This is a cross-sectional analysis against a risk estimate, not against observed events, so it argues that continuous glucose dynamics carry information rather than that anyone should be screened with a sensor. It is worth knowing because the discordance is the point: the person whose HbA1c reassures you may be spending a good part of the day above 140 mg/dL.

• Do not start ordering CGM for people without diabetes — this is not a screening study
• Where a sensor already exists, time above 140 mg/dL is worth reading alongside HbA1c
• Reassess blood pressure and lipids in anyone whose sensor shows substantial time above range
• The cohort was 92.6% non-Hispanic White, with no Indian data — generalise cautiously

Why it matters

It raises the question of what a reassuring HbA1c actually rules out.

Don't overread it

Cross-sectional and against a predicted risk score — it cannot show that glycaemic variability causes cardiovascular disease.

The statistics, in plain English

The outcome here is an estimated 10-year risk score, not events that happened, so a 2% higher estimate means the model predicts more risk — not that more people had heart attacks. Cross-sectional design means glucose and risk factors were measured at the same time, so it cannot say which came first. The odds ratios for hypertension and dyslipidaemia are associations after adjustment, and residual confounding by adiposity is likely.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.