Mean absolute relative difference (MARD) summarizes the relative differences between matched sensor and reference glucose values. It is useful, but it does not tell the whole story of a sensor’s performance or make unrelated studies directly comparable.
A simple arithmetic example
Fictional example: a reference value is 100 mg/dL and a matched sensor value is 110 mg/dL. The absolute relative difference is |110 − 100| ÷ 100 × 100 = 10%. MARD averages those absolute relative differences across the matched pairs in a study. This example is arithmetic only, not a test result for any product.
A MARD of a given percentage is not the percentage of readings that are accurate. It is also not a guaranteed plus-or-minus band for an individual result.
Before comparing two published numbers
| Study detail | Why it changes the comparison |
|---|---|
| Population and conditions | Age, clinical setting and glucose patterns influence the matched observations. |
| Reference method | The comparator and sampling method affect the calculated differences. |
| Glucose ranges and change | Low values and fast-changing periods may be represented differently. |
| Sensor days and calibration | Performance over the wear period and calibration procedures may differ. |
| Missing data and exclusions | A neat average can leave interruptions, excluded pairs or outliers out of view. |
| Study design | A well-designed head-to-head comparison answers a different question from separate trials. |
What else belongs beside MARD
Review range-specific performance, outliers, data availability, alert behavior and usability when the evidence reports them. Keep manufacturer claims separate from independently assessed results. The cited methodological review explains the metric’s limitations; it is not a current product ranking or a new validation trial.
Does the lowest advertised MARD automatically mean the best CGM?
No. First check whether the studies are comparable and what the intended use requires. An average metric alone cannot establish the best system for an individual, clinic or market.
This is general educational information. Use the instructions for the exact device and discuss personal glucose results or treatment questions with your healthcare professional.
Original sources
- Heinemann et al. · Benefits and Limitations of MARD, 2019/2020
- American Diabetes Association · Diabetes Technology, Standards of Care 2026
Sources reviewed 18 October 2026. Model features and local access should be checked against the current instructions and market record.