Continuous Glucose Monitor (CGM): What Your Numbers Change About Your Treatment

A continuous glucose monitor (CGM) is a wearable sensor, usually worn on the upper arm or abdomen, that measures glucose in interstitial fluid every 1 to 5 minutes rather than at a single fingerstick moment. FDA-cleared systems in common use include real-time CGMs (Dexcom G-series, Medtronic Guardian) and intermittently scanned or "flash" CGMs (Abbott FreeStyle Libre). This article is about the metrics these devices generate, not about individual dosing instructions.

The core answer, and its boundary

Two people can have an identical A1c of 7.0% while one holds a steady 154 mg/dL and the other swings between the 50s and the 300s. A1c cannot tell them apart; CGM can. That single fact is why CGM-derived metrics (time in range, time below range, time above range, coefficient of variation, and the glucose management indicator) are now used alongside A1c, not as a replacement for it, to decide whether to add, reduce, or switch a diabetes medication. What CGM has not been shown to do, at least not on the evidence available for this article, is establish that treating glucose variability in people without diabetes changes their long-term risk of developing diabetes or its complications. That remains an open question, not a settled one.

What a CGM measures, and why the lag matters

CGM sensors read glucose in the fluid between cells, not in blood directly. Interstitial glucose lags blood glucose by roughly several minutes, which matters most when glucose is changing quickly, for example right after a meal, during exercise, or during a hypoglycemic episode where fast confirmation matters. For most day-to-day pattern recognition, that lag is a minor limitation. For an in-the-moment decision about whether someone is having a low blood sugar emergency right now, a fingerstick or symptom-based judgment still matters, and CGM should not replace emergency glucose treatment or urgent care when someone is confused, unresponsive, or unable to safely swallow.

The five metrics that clinicians actually act on

An ambulatory glucose profile (AGP) report distills weeks of CGM data into a handful of numbers. Each one tends to prompt a different kind of clinical response.

Time in range (TIR): the percentage of readings between 70 and 180 mg/dL for most adults with diabetes. A commonly cited consensus target is TIR above 70%, associated with an A1c in the neighborhood of 7%, though the precise correlation varies by individual and should not be treated as an exact conversion.

Time below range (TBR): the percentage of readings below 70 mg/dL (and, separately, below 54 mg/dL, which is considered clinically significant). Consensus targets are generally under 4% and under 1% respectively. This is the metric most likely to trigger a dose reduction rather than a new drug.

Time above range (TAR): readings above 180 mg/dL. A commonly cited target is under 25%. Persistently high TAR is what usually prompts adding a prandial insulin, a GLP-1 receptor agonist, or an SGLT2 inhibitor.

Coefficient of variation (CV): a measure of how much glucose swings around its average. A CV at or below roughly 36% is generally considered stable; values above that threshold prompt a search for causes such as erratic meal timing, gastroparesis, medication non-adherence, or other underlying conditions, rather than an automatic increase in medication dose.

Glucose management indicator (GMI): an estimate of A1c derived from CGM averages. When GMI and lab A1c disagree substantially, clinicians look for causes that distort the lab A1c specifically, such as conditions affecting red blood cell turnover, hemoglobin variants, or iron deficiency.

These thresholds trace back to an international consensus process on CGM targets published in Diabetes Care and reflected in subsequent ADA Standards of Care language recommending that time in range and time below range be used alongside A1c. The exact numeric thresholds above should be confirmed against the current ADA Standards of Care before being used to counsel an individual patient, since guideline numbers are periodically revised.

How CGM changes medication decisions in type 2 diabetes

CGM's usefulness comes from exposing patterns a quarterly A1c cannot see.

Overnight or fasting glucose rise (the "dawn phenomenon"): when the AGP shows glucose climbing in the early morning hours, the typical response is adjusting basal insulin timing or dose, or considering a GLP-1 receptor agonist with a fasting-glucose effect. The magnitude of benefit from any specific agent varies by trial and by patient, and a specific numeric claim about how much a given drug lowers fasting glucose in this pattern should be checked against the current product label or a recent trial before being quoted to a patient.

Postprandial spikes: glucose rising above roughly 180 mg/dL for an extended period after meals is a pattern CGM catches that fingersticks routinely miss. Depending on the picture, this may lead to adding a rapid-acting insulin, an alpha-glucosidase inhibitor, or adjusting the carbohydrate composition and timing of meals. Some trial evidence suggests dietary changes such as lower-glycemic-index eating can reduce CGM-measured variability, but effect sizes differ across studies and specific percentage claims need to be checked against a verified source before being cited as precise.

Nocturnal hypoglycemia: lows overnight are often asymptomatic, meaning a patient can sleep through a genuinely dangerous glucose level and have no idea it happened until CGM data shows it. This is usually the strongest single trigger for reducing basal insulin or switching away from a sulfonylurea toward an agent with a lower hypoglycemia risk, such as a DPP-4 inhibitor or GLP-1 receptor agonist. Older adults appear to be disproportionately affected by unrecognized nocturnal lows, which is one reason simplified, lower-hypoglycemia-risk regimens are often favored in that population, though exact proportions from any single study should be verified rather than repeated as a precise figure.

CGM in type 1 diabetes: dosing gets more granular

In type 1 diabetes, CGM data feeds fairly directly into day-to-day insulin decisions: recalculating a correction factor when postprandial readings consistently run high, adjusting an insulin-to-carbohydrate ratio when two-hour post-meal glucose repeatedly misses target in the same direction, and using overnight CGM tracings to check basal rates in pump users. Randomized evidence, including flash-CGM comparisons against fingerstick monitoring, has generally reported reductions in hypoglycemia exposure without worsening A1c, which is part of why professional guidelines now recommend CGM broadly for adults with type 1 diabetes rather than reserving it for people with hypoglycemia unawareness. Specific percentage reductions from individual trials vary by study population and device generation and should be pulled from the current trial report rather than treated as fixed constants.

A recurring theme among clinicians who focus on glycemic variability is that when A1c and time-below-range point in different directions, for example a reasonable A1c alongside a meaningfully elevated time below range, the hypoglycemia signal usually takes priority for the next treatment change. That is a clinical judgment reflected in guideline language about prioritizing hypoglycemia reduction, not a universal rule that overrides individualized care.

CGM in prediabetes and general metabolic health: what is and is not established

CGM use by people without diagnosed diabetes has grown well beyond its original clinical population. The rationale is plausible: postprandial glucose excursions above roughly 140 mg/dL might flag insulin resistance before fasting glucose or A1c becomes abnormal. Some observational work has found that people without diabetes show meaningfully different glucose response patterns to identical meals, suggesting glucose regulation is more individualized than A1c alone implies.

What is not established is whether identifying these excursions and acting on them (through diet, exercise, or medication) changes anyone's long-term risk of developing diabetes or its complications. The ADA does not currently recommend CGM for prediabetes screening as a matter of routine practice. Some endocrinology guidance describes CGM as a potentially useful adjunct for lifestyle coaching in people at high metabolic risk, which is a narrower claim than "corrects prediabetes." Readers using CGM for general wellness should treat the data as a behavior-feedback tool, not a diagnostic or a proven risk-reduction intervention, and specific guideline citations here should be verified against the current version of the relevant society's published recommendations before being relied upon.

Targets differ by who is wearing the sensor

Consensus targets are not one-size-fits-all.

General adults with type 1 or type 2 diabetes: TIR above roughly 70%, time below 70 mg/dL under about 4%, time below 54 mg/dL under about 1%, time above 180 mg/dL under about 25%.

Older adults or people at higher hypoglycemia risk (history of severe lows, limited life expectancy, significant comorbidity): a looser TIR target, often cited around 50%, with an even tighter ceiling on time below range, reflecting that avoiding a dangerous low matters more than chasing a lower average in this group.

Pregnancy with preexisting diabetes: a narrower target range, often cited as roughly 63 to 140 mg/dL, because both maternal and fetal outcomes are sensitive to glucose extremes in ways that differ from non-pregnant adults. Trial evidence in this population has generally favored CGM over fingerstick-only monitoring for reducing certain neonatal complications, though exact effect sizes should be confirmed from the primary trial report rather than repeated as fixed numbers.

People without diabetes: typically spend the large majority of time in a narrower range with a lower average glucose and less variability than people with diabetes, though "normal" reference ranges from CGM are not yet standardized the way blood glucose reference ranges are.

CGM alongside GLP-1s, SGLT2 inhibitors, and testosterone therapy

CGM gives faster feedback than a quarterly A1c on whether a medication is working.

GLP-1 receptor agonists (semaglutide, tirzepatide): CGM can show the timing and magnitude of postprandial glucose improvement within weeks, well before a three-month A1c would reflect it. This lets a clinician identify a poor responder and consider a dose change or a different agent sooner, though the CGM data is a supporting signal, not a substitute for a full clinical picture including weight, gastrointestinal tolerance, and cardiovascular risk factors.

SGLT2 inhibitors (empagliflozin, dapagliflozin): these lower glucose independently of insulin. In people with type 1 diabetes who use an SGLT2 inhibitor off-label alongside insulin, CGM monitoring is particularly important because the drug's glucose-lowering effect can mask an insulin dose that is now too high, and because SGLT2 inhibitors carry a meaningful diabetic ketoacidosis risk in this population that is not always accompanied by very high glucose readings. This combination should be treated with caution and close monitoring, not as routine practice.

Testosterone replacement therapy: testosterone can improve insulin sensitivity in men with hypogonadism, and some trial evidence has linked testosterone treatment to a lower incidence of progressing to type 2 diabetes in men with impaired glucose tolerance. CGM data showing improving glycemic variability during TRT can serve as a supporting biomarker in a broader treatment picture, but it does not establish that TRT is an approved or first-line diabetes-prevention therapy, and testosterone therapy carries its own risks and monitoring requirements independent of glucose control.

A pattern-to-decision framework for reading a CGM report

This is a way to organize what a CGM report is showing before jumping to a medication change. It is a starting framework for a clinical conversation, not a substitute for individualized medical judgment, and every action listed here requires a clinician to confirm against the patient's full history, kidney and cardiac status, and current medication list.

How to think about persistently high readings

A mean glucose consistently above roughly 180 mg/dL, or a TIR well under half, generally prompts a structured review rather than an isolated dose bump: checking the timing and composition of the meal producing the largest spike, reviewing medications known to worsen glucose control (corticosteroids, some atypical antipsychotics, thiazide diuretics), and confirming medication adherence before escalating therapy. When escalation is warranted, ADA/EASD-style guidance generally frames the choice between a GLP-1 receptor agonist and an SGLT2 inhibitor around comorbidities: favoring a GLP-1 receptor agonist when weight loss is a priority or atherosclerotic cardiovascular disease is present, and favoring an SGLT2 inhibitor when heart failure or chronic kidney disease is present. CGM speeds up this decision by showing inadequate control in near real time rather than waiting for the next A1c.

How to think about low readings

A CGM reading below 70 mg/dL meets the threshold for hypoglycemia; below 54 mg/dL is considered clinically significant and generally calls for prompt treatment with fast-acting carbohydrate per standard hypoglycemia protocols. Recurrent lows on CGM, meaning a meaningful share of time below range rather than an isolated event, usually call for a medication adjustment (typically a lower insulin or sulfonylurea dose, or a switch to a lower-hypoglycemia-risk agent) rather than simply eating more to compensate. If someone is confused, unable to swallow safely, or unresponsive during a low, that is an emergency requiring glucagon or emergency medical care, not a CGM-guided decision.

What is established, what is plausible, and what is not

Established: CGM produces metrics (time in range, time below range, time above range, coefficient of variation, GMI) that are recognized in ADA Standards of Care as complementary to A1c for guiding treatment in diagnosed diabetes. Consensus targets for these metrics exist from an international expert panel process published in Diabetes Care, though the exact numeric thresholds should be checked against the current Standards of Care for any patient-facing use, since guidelines are updated periodically and this article's figures should be reverified before clinical use.

Plausible but unproven: that treating CGM-detected postprandial glucose excursions in people without diagnosed diabetes reduces future diabetes risk; that testosterone therapy prescribed primarily for glycemic benefit is an accepted first-line diabetes-prevention strategy; and precise numeric effect sizes attributed to specific trials in earlier versions of material like this, which need confirmation against the primary trial publication rather than being repeated as fixed facts.

Not established from the material available here: any specific direct quotation attributed to a named individual clinician or researcher in earlier drafts of this topic could not be verified against a primary source and has been removed rather than repeated.

Sources and verification note: The clinical thresholds referenced above (time in range, time below range, time above range, coefficient of variation) trace to an international consensus process on CGM targets published in Diabetes Care and are reflected in subsequent ADA Standards of Care guidance recommending CGM metrics alongside A1c. General background on CGM device clearance can be checked directly at fda.gov. Several specific trial names, sample sizes, and numeric effect sizes that appeared in an earlier version of this article could not be verified against a confirmed primary source and have been removed or converted into general, hedged statements. Before this article is published, an editor with access to the primary literature should confirm each remaining guideline threshold against the current ADA Standards of Care and reinsert verified, specific trial citations where they strengthen a claim.