A CGM can help you explore glucose patterns, but it cannot diagnose diabetes or prediabetes. Its most useful role here is generating questions, not delivering a metabolic health verdict. 1

The aim is not to make every curve flat. It is to decide whether a repeated observation suggests a useful habit change or deserves a conversation with your GP.

What a CGM actually measures in a non-diabetic adult

A CGM estimates glucose in interstitial fluid beneath the skin, rather than directly measuring blood glucose. Readings can differ from a finger-prick, especially during rapid changes. 2

There is no universal 15% error margin or fixed delay that converts a sensor result into a blood result. Check the particular device’s guidance. 2

Diagnosis uses HbA1c or laboratory plasma glucose tests, depending on the situation. Fasting insulin is not a standard diagnostic test for diabetes or prediabetes. 1

Before buying in NZ, check the supplier’s current price, intended use, phone compatibility and labelled wear period. Decide what question you want the sensor to help answer before committing to ongoing purchases.

The four numbers worth watching

Use these four summaries to describe patterns, not to set your own diagnostic cut-offs. No universal meal-rise or variability target has been established for adults without diabetes. 1 3

  • Waking glucose: note the reading before breakfast alongside the overnight trace. Do not label it a laboratory fasting glucose result. 1 2
  • Post-meal peak: record the starting level, peak and meal timing. A rise alone does not establish that the food is harmful. 3
  • Time towards baseline: note whether the curve is settling, but do not grade glucose disposal against a universal two-hour deadline. Meal composition and context matter. 4
  • Variability: standard deviation describes the spread of readings around their average. It is not a diagnosis or a validated explanation for afternoon fatigue. 1 3

For context, a study of 153 healthy participants found a median 96% of sensor time between 3.9 and 7.8 mmol/L. Some readings fell outside that interval. 3

That is a description of a selected study population, not a personal pass mark. It also means a brief reading above 7.8 mmol/L is not, by itself, evidence of disease. 3

Average glucose still helps summarise the record. Just do not let it replace the trace: the same average can conceal quite different patterns. 3

Meal patterns worth investigating

Treat these as questions to test, not patterns guaranteed to occur in every person.

Carbohydrate alone versus a mixed meal: protein, fat and fibre can alter the response to carbohydrate. PREDICT found associations with lower glucose responses, not a universal one-third reduction. 4

Try toast with eggs, rice with tofu and vegetables, or a banana with plain yoghurt. Record hunger separately rather than assuming a visible peak predicts a crash.

The second-meal effect: breakfast and the interval before lunch are part of lunch’s context. Do not assume a protein-led breakfast will always improve the response to the next meal. 4

A rise without food: stress hormones such as cortisol and adrenaline can affect glucose regulation. The effect is not a predictable number, and the curve does not reveal hormone concentrations. 2 5

A rise during a difficult meeting is therefore a possible association, not proof that stress caused it. Do not automatically dismiss repeated unexplained elevations as normal stress physiology. 5

The poor-sleep pattern: in a study of 953 adults, poorer sleep efficiency and later sleep timing were associated with larger breakfast responses. Sleep was not randomly assigned. 6

Log sleep duration and quality, but do not assign a fixed glucose penalty to a short night or assume every meal response will worsen. 6

The later-meal response: timing and previous meals can influence a curve. If kūmara looks different at lunch and dinner, compare portion size, accompanying foods and activity before blaming the clock. 4

These short-term findings do not prove that lowering a particular peak will prevent disease or improve long-term health. 4 6

Where CGMs add to a finger-prick

A finger-prick is a snapshot. Repeated sensor readings provide a timeline, including periods when taking frequent finger-pricks would be inconvenient. 2 3

Use that timeline to ask specific questions:

  • Overnight: is an apparent dip isolated or repeated? Does it coincide with sleeping on the sensor?
  • Exercise: what happens around a morning run, weights session or Sunday hike?
  • Real-world meals: does the same flat white and muffin at a Christchurch café produce a similar pattern on another day?

These are observation prompts, not reasons to conclude that a workout, late dinner or particular food is damaging.

The tools are complementary. A finger-prick can help check an unexpected sensor result, but it is not an error-free laboratory measurement either. 2

Where CGMs mislead

A meal is not a glucose tolerance test. Diagnostic testing uses defined laboratory methods. You cannot apply its thresholds directly to a mixed-meal sensor peak. 1

An overnight low is not automatically false. Pressure can distort readings. Check unexpected results with a meter, using clean, dry hands and your device’s instructions. Do not rely on feeling well alone. 2

One comparison is not a controlled trial. Repeating similar meals helps distinguish a pattern from ordinary variation, but differences in sleep, activity and portions still complicate interpretation. 4

If elective tracking starts driving food fear or compulsive checking, pause the experiment and discuss a less intrusive approach with a health professional.

What to do this week

  1. Choose one question. Start with a familiar breakfast or dinner rather than trying to score everything you eat.
  2. Record the context. Note meal time, approximate portions, a photo, sleep, stress, activity and symptoms. An incomplete log can still be useful; it simply limits interpretation.
  3. Change one thing. Compare your usual breakfast with a balanced version, or repeat the same dinner with and without a comfortable 15-minute walk afterwards.
  4. Repeat each version. Keep portions and timing reasonably similar. Avoid changing breakfast, exercise and sleep routines together if you want an interpretable comparison.
  5. Review, then decide. Consider hunger, enjoyment, affordability and practicality alongside the glucose trace. Keep a useful habit, not necessarily the sensor subscription.

The walking suggestion has some experimental support: ten older adults at increased metabolic risk improved short-term glucose control with 15-minute walks after meals. That small study cannot guarantee your response. 7

For a broader tracking framework, see building a personal health dashboard.

When to involve your GP

Take repeated unexplained readings or symptoms to your GP rather than trying to diagnose yourself from the app. Bring the trace, meal notes and any confirmatory meter readings. Laboratory testing can clarify the concern. 1

Do not wait until the sensor finishes if you feel unwell. Confusion, fainting or severe illness warrants urgent medical help, not an extended sensor experiment.

If you already have diabetes or use glucose-lowering medication, follow your clinician’s monitoring plan. This article is not a basis for changing medication or replacing prescribed monitoring.