Mean Glucose as a Digital Biomarker

Mean glucose is the arithmetic average of all sensor glucose readings across a recording period, typically fourteen days. It is the most direct summary a continuous glucose monitor produces, and it occupies a specific position in diabetes measurement: it is the quantity that glycated haemoglobin has always been used to approximate.

That relationship was formalised twice. First by work translating the glycated haemoglobin assay into estimated average glucose values, and later by the glucose management indicator, which converts mean sensor glucose from a continuous monitor into an approximate laboratory equivalent. The second formulation exists because the two frequently disagree in individuals, since glycated haemoglobin also depends on red cell lifespan and other factors that have nothing to do with glucose.

A continuous glucose sensor sits in interstitial fluid, usually on the upper arm or abdomen, and reports a value every few minutes for up to fourteen days. Mean glucose is computed across all available readings in the period, which makes data sufficiency part of the measurement: consensus guidance recommends at least fourteen days of data with a high proportion of the period captured before the summary metrics are considered representative.

Interstitial glucose lags blood glucose by several minutes, which matters when the value is changing quickly and matters much less for an average across two weeks. Sensor accuracy is well characterised and is best in the mid range, which is where most readings in a typical recording sit, so mean glucose is among the more robust outputs the technology produces.

Mean glucose is reported as part of the standard continuous glucose monitoring summary alongside time in range, time below range and variability, and international consensus defines that set explicitly so it can function in clinical trials. Its particular value is in explaining discordance: when glycated haemoglobin and sensor data disagree, mean glucose and the glucose management indicator make the disagreement visible rather than leaving the clinician to choose between two numbers.

In research it appears as an efficacy endpoint in diabetes trials, as a physiological outcome in nutrition and behavioural studies, and as a covariate elsewhere. It is reported alongside instruments such as the Diabetes Distress Scale or the Diabetes Self-Management Questionnaire, because a regimen that lowers average glucose while increasing burden has not straightforwardly succeeded.

An average conceals the pattern that produced it. Two people with the same mean glucose can have entirely different days, one stable and one swinging between hyperglycaemia and hypoglycaemia, and only the second is at immediate risk. This is precisely why mean glucose is never reported alone and why time in range, time below range and variability exist alongside it.

The measure also depends on adequate data capture. A recording with substantial gaps, or one shorter than the recommended period, produces an average that may not represent the person's usual state. And a sensor derived average is not interchangeable with a laboratory glycated haemoglobin value, even when converted, which is the whole reason the glucose management indicator was introduced.

• Bergenstal RM, et al. Glucose management indicator (GMI): a new term for estimating A1C from continuous glucose monitoring. Diabetes Care. 2018. pubmed.ncbi.nlm.nih.gov
• Nathan DM, et al. Translating the A1C assay into estimated average glucose values. Diabetes Care. 2008. pubmed.ncbi.nlm.nih.gov
• Battelino T, et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the International Consensus on Time in Range. Diabetes Care. 2019. pubmed.ncbi.nlm.nih.gov
• Danne T, et al. International consensus on use of continuous glucose monitoring. Diabetes Care. 2017. pubmed.ncbi.nlm.nih.gov