CGM and FGM
(Continuous - Flash Glucose Monitoring)

Continuous glucose monitoring has radically transformed the way diabetes is observed because it has shifted attention from a single isolated glucose value to glucose dynamics throughout the entire day and night. For many years, diabetes management relied mainly on point capillary glucose measurements, which were useful but inevitably fragmented: each measurement provided an instantaneous snapshot without showing what had happened beforehand, where glucose was heading or how much time the patient spent in hypoglycemia, within the desired range or in hyperglycemia. Sensors have changed this scenario. It is now possible to follow the glucose profile almost in real time and observe trends, rates of change, cumulative exposure to glycemic extremes and the concrete impact of meals, exercise, insulin therapy, intercurrent illness and daily routines.

CGM, or continuous glucose monitoring, refers to a family of technologies that measure glucose in interstitial fluid through a subcutaneous sensor. This family includes both rtCGM, or real-time continuous glucose monitoring, systems that automatically transmit continuous readings and alarms, and isCGM, or intermittently scanned continuous glucose monitoring, systems historically also known as FGM or flash glucose monitoring, in which the reading is displayed by scanning the sensor. This distinction is not merely terminological because it reflects technical, clinical and educational differences that affect indications, safety, adherence and the patient’s ability to use the technology correctly.

Continuous monitoring systems do not directly measure glucose in capillary or venous blood, but the glucose present in interstitial fluid, the extracellular compartment surrounding cells in subcutaneous tissue. The sensor, inserted under the skin through a small filament, uses an enzymatic reaction—usually based on glucose oxidase or related technologies—to generate an electrical signal proportional to the glucose concentration. This signal is then converted into a readable value and transmitted to a dedicated receiver, smartphone or insulin pump.

From a physiological standpoint, this choice has a fundamental consequence: the sensor value does not coincide absolutely and simultaneously with blood glucose. A physiological lag exists between blood and interstitial fluid, and it is particularly evident when glucose is changing rapidly, such as after a meal, during intense physical activity, after an insulin bolus or while treating hypoglycemia. Under stable conditions the correlation is very good, but during rapid changes the interstitial value may “follow” the blood value with a delay of several minutes. This explains why a sensor may appear clinically excellent in steady-state conditions and less accurate during highly dynamic glucose changes.

Data quality also depends on how the device algorithm filters and processes the signal. The sensor does not merely measure a concentration; it interprets a continuous, potentially noisy biological signal that must be converted into a stable, clinically useful value. Every system therefore incorporates a biochemical, engineering and algorithmic component. Actual CGM performance depends not only on the enzyme or the subcutaneous filament, but on the entire system architecture, including data refresh rate, temporary signal loss, calibration, alarms and the user interface.

Interstitial glucose correlates with plasma glucose and, through serial measurement, makes it possible to describe the pattern, direction, and duration of glucose excursions. It is not, however, a direct measure of intracellular glucose in tissues and may differ from blood glucose, especially during rapid changes; in certain situations, an immediate clinical decision still requires capillary confirmation. Continuous monitoring should therefore not be regarded simply as a technological replacement for fingerstick glucose testing, but as a paradigm shift in how glucose metabolism is represented.

Within continuous monitoring, it is essential to distinguish between rtCGM and isCGM. Real-time systems automatically transmit glucose values at regular intervals, display the curve continuously and, in most cases, provide predictive or threshold alarms for hypoglycemia, hyperglycemia or rapid change. This makes them particularly useful when safety depends on timely warning, such as in patients with nocturnal hypoglycemia, impaired awareness of hypoglycemia, pediatric age, pregnancy or intensive insulin therapy.

isCGM, or intermittently scanned CGM, systems instead require the patient to scan the sensor actively to display the value, trend arrow and graph of the preceding hours. Historically, this technology has been described as FGM, or flash glucose monitoring. In current clinical practice, flash monitoring is classified as a form of isCGM, although commercial terminology and habitual language have preserved both definitions. The most important difference is therefore not that one system “really measures” and the other does not, but how the data are made available to the patient.

This distinction has very practical implications. A scanned system requires more active behavior: if the patient does not scan, the current value is not seen. A real-time system continues transmitting and can issue a warning even without any action by the user. rtCGM therefore tends to be more advantageous when the priority is prevention of hypoglycemia or continuous supervision, whereas isCGM can be highly effective in motivated patients who are skilled at interpreting data and prefer monitoring with fewer alarms.

An overly rigid hierarchy should nevertheless be avoided. There is no single system that is “best in absolute terms” for everyone. The choice depends on age, type of diabetes, treatment, risk of hypoglycemia, educational ability, tolerance of alarms, digital familiarity, financial availability and organizational context. In many patients, the true benefit derives not from the abstract superiority of a technology, but from its fit with the actual behavior of the person using it.

The real innovation of continuous monitoring is not simply the increased number of measurements, but the ability to describe glucose as a time-dependent phenomenon. With capillary testing, the clinician sees scattered values; with a sensor, the clinician observes the trajectory, amplitude of fluctuations, persistence of excursions, relationship with meals and insulin, nocturnal distribution and vulnerability at specific times of day. This shift from point measurements to continuity has had an enormous impact on the ability to individualize care.

In type 1 diabetes, the advantage is clear: CGM identifies asymptomatic hypoglycemia, periods of nocturnal instability, postprandial rises missed by traditional self-monitoring and variability patterns that explain how an apparently acceptable HbA1c can coexist with clinically unsatisfactory control. In insulin-treated type 2 diabetes, the sensor helps determine whether the dominant problem is fasting glucose, postprandial glucose, an excessive evening dose or the response to physical activity. Even in selected people not treated with insulin, continuous monitoring can provide a much more precise picture of actual glycemic behavior when chosen according to the clinical problem.

The change has also been epistemological. For years, glycemic control was summarized almost exclusively by HbA1c, which is useful but incomplete. Sensors introduced new metrics that describe previously invisible aspects, such as time in range, time spent within the desired range, time in hypoglycemia, time in hyperglycemia and day-to-day variability. This improved the quality of clinical decisions by distinguishing patients with similar averages but very different risk profiles.

Another decisive aspect is educational value. The patient no longer receives only a number, but a graphical narrative of their metabolism. They see the rise after a meal, observe the effect of exercise, and recognize the behavior of a late bolus or excessive correction. This improves practical understanding of diabetes and makes treatment more participatory. The sensor is therefore not only a measurement device, but also a tool for learning physiology as it applies to everyday life.

Indications for continuous monitoring have progressively expanded. Today, type 1 diabetes is the most established and best-documented field of use because sensors improve glycemic control, increase time in the target range and reduce the risk of hypoglycemia, particularly when used continuously and integrated into an appropriate educational pathway. The benefit is also increasingly clear in insulin-treated type 2 diabetes, whether using a basal-bolus regimen or less intensive but clinically unstable regimens, especially in people with out-of-range values, hypoglycemia or marked variability.

The most recent development concerns non-insulin-treated type 2 diabetes. Current recommendations do not mandate generalized use, but recognize potential value in adults treated with other glucose-lowering medications when the goal is to improve or maintain individualized glycemic targets. In this population, the sensor is particularly valuable as a tool for metabolic phenotyping, educational support and assessment of the impact of diet, exercise, medications or corticosteroids.

There are also high-priority clinical contexts, including pregnancy, pediatric age, impaired awareness of hypoglycemia, recurrent severe hypoglycemia, marked glycemic variability, high-risk occupations, competitive sport, irregular nocturnal schedules, chronic kidney disease and use of insulin pumps or automated systems. In these settings, the benefit of CGM is not merely numerical, but relates to safety. The sensor makes it possible to recognize dangerous patterns early and adjust treatment more promptly than with capillary monitoring alone.

Patient selection, however, cannot be based solely on the diagnosis. The ability to interact with the technology must also be considered. A sensor generates a large amount of data, but those data must be understood, shared and translated into behavior. Patients with limited health literacy, severe alarm anxiety, rejection of the device or low willingness to use it may derive less benefit than expected on paper. The most appropriate prescription therefore combines a pathophysiological indication with behavioral sustainability.

The introduction of continuous monitoring created a new clinical language. The most important metric is time in range, meaning the time spent within the target glucose interval, generally 70-180 mg/dL for most adults with diabetes. This measure is far more informative than an isolated average because it indicates how many hours of the day the patient actually remains within a clinically reasonable zone. A high time in range suggests not only good average control, but also less exposure to glycemic extremes.

Alongside time in range are time below range and time above range. These metrics respectively describe exposure to hypoglycemia and hyperglycemia. Their usefulness is considerable because two patients with the same mean glucose can have completely different risk distributions: one may spend many hours in hypoglycemia and many hours in hyperglycemia, while the other remains stable for most of the time. Analysis of time outside the range clarifies the metabolic cost of apparently similar control.

Another essential component is glycemic variability, often summarized by the coefficient of variation. Variability is not merely a mathematical concept, but a clinical dimension of metabolic fragility. Wide and rapid fluctuations increase the risk of hypoglycemia, make treatment titration more difficult and may impair quality of life. The sensor helps determine whether the patient’s main problem is the mean glucose, the dispersion of values or both.

These metrics are complemented by mean sensor glucose and the GMI, or glucose management indicator, a value that translates the sensor mean into an estimate analogous to an expected HbA1c. GMI is useful because it allows recent glucose behavior to be compared with laboratory-measured glycated hemoglobin. When the two values diverge substantially, the clinician may suspect biological discordance, a recent change in control or an interpretive limitation of HbA1c.

• Time in range: the proportion of time generally spent between 70 and 180 mg/dL.
• Time below range: time spent below 70 mg/dL, with particular attention to values below 54 mg/dL.
• Time above range: time spent above 180 mg/dL, distinguishing more marked levels above 250 mg/dL.
• Glycemic variability: the magnitude and instability of fluctuations over 24 hours.
• GMI: an estimate of HbA1c derived from mean sensor data.

The value of a sensor lies not only in reading a single number, but in the ability to interpret trends and patterns. Trend arrows indicate the direction and, to some extent, the rate at which glucose is moving. This profoundly changes the clinical meaning of the value. A reading of 100 mg/dL with a steady arrow is very different from the same value with a rapidly falling arrow, especially in a patient with active insulin, imminent exercise or impaired awareness of hypoglycemia. The sensor therefore supports decisions not only about the present, but about the immediate future of the glucose curve.

More advanced analysis goes beyond real-time readings and uses the ambulatory glucose profile, a graphical summary of data from several days that displays the median, dispersion and recurrent periods of instability. This tool identifies repeated patterns: consistent rises after breakfast, nocturnal hypoglycemia, afternoon instability, insufficient post-dinner coverage or excessive evening corrections. The advantage is substantial because treatment can be adjusted on the basis of reproducible phenomena rather than anecdotal values.

In type 1 diabetes, the trace helps identify incorrect bolus timing, an inappropriate insulin-to-carbohydrate ratio, the effect of exercise, rebound after hypoglycemia and nocturnal vulnerability. In type 2 diabetes, it clarifies whether fasting, postprandial or variability-related abnormalities associated with irregular meals and medications predominate. Even in patients not treated with insulin, the sensor can reveal otherwise invisible phenomena, such as large excursions after specific foods or systematic deterioration associated with evening inactivity.

Interpretation, however, requires a method. A single abnormal trace is not enough to change treatment. Decisions should be based on consistent recurrence, analyzed over an adequate number of days and interpreted together with the clinical context. The strength of continuous monitoring is not that it provides more numbers, but that it gives greater structure to clinical reasoning.

Sensor accuracy has improved enormously, but no system is perfect. One of the most widely used measures is MARD, or mean absolute relative difference, which expresses the average difference between the sensor value and a glucose reference. A lower MARD generally suggests better analytical accuracy, but its true meaning depends on the context in which it is measured and it is not sufficient by itself to define a device’s clinical value.

The best-known physiological limitation is the aforementioned lag between blood and interstitial fluid. During rapid changes, the sensor may underestimate or overestimate current capillary glucose. There are also practical limitations: pressure on the sensor during sleep causing false compression-related hypoglycemia, signal loss, skin-adhesion problems, local interference, site inflammation and differences in performance between the first and last days of sensor wear.

For these reasons, capillary blood glucose retains an essential role. It is needed when symptoms do not match the sensor, when the value appears implausible, during rapid clinical deterioration, in suspected severe hypoglycemia, in acute illness with ketosis or diabetic ketoacidosis, when the sensor reports questionable values or when the system is not functioning correctly. Even in patients using advanced CGM, fingerstick testing remains a safety reference for many critical decisions.

A common conceptual error is to assume that a sensor makes every other method unnecessary. In reality, continuous monitoring broadens the field of view for both clinician and patient, but does not eliminate the need for critical validation of the data. The best use of CGM comes from intelligent integration of sensor information, clinical assessment and capillary confirmation when required.

Continuous monitoring is beneficial only if the patient knows how to use it. Without education, the sensor may become a source of numbers, alarms and anxiety. Patients must learn to understand the difference between an absolute value and the direction of the trend, know that interstitial glucose does not always match blood glucose, recognize when the sensor may be less reliable and understand when capillary confirmation is necessary.

A central issue is management of trend arrows. Decisions about a meal, an insulin correction or carbohydrate intake to treat hypoglycemia should not depend only on the displayed number, but also on the direction of the curve. This requires practical training. Similarly, patients must learn to respond to alarms without turning them into prompts for continuous, impulsive corrections, which can lead to “chasing” glucose with excessive interventions that increase variability.

Education also concerns organizational aspects: sensor application, site rotation, skin care, management of detachment, synchronization with devices, data sharing with the care team and interpretation of periodic summaries. In pediatric patients and families, balancing useful supervision against over-monitoring becomes crucial. Too many alarms or excessive attention to minor deviations can generate decision fatigue and impair quality of life.

Adherence improves when the technology is incorporated into a realistic plan. It is not enough simply to “apply the sensor.” The objectives must be clear: reducing nocturnal hypoglycemia, increasing time in range, understanding the effect of breakfast, improving insulin titration or documenting instability in a complex patient. When the sensor answers a specific clinical question, it is much easier to use it correctly and sustain its use over time.

Continuous monitoring has acquired even greater value since it began interacting with insulin delivery. In patients using insulin pens, the sensor makes it possible to determine far more precisely whether the basal dose is adequate, whether the bolus is timely, whether the insulin-to-carbohydrate ratio is correct and whether corrections are producing dangerous stacking. Continuous data reduce the need for empirical decisions and make titration more physiological.

With insulin pumps, the relationship becomes even closer. The sensor no longer merely informs the patient, but can interact with the pump to suspend insulin in anticipation of hypoglycemia or automatically adjust delivery in more advanced systems. This integration gave rise to hybrid and automated insulin-delivery systems, in which CGM is the algorithm’s true physiological input.

When used correctly, flash monitoring can also support insulin therapy very effectively, particularly when the patient scans frequently, interprets trends accurately and uses the information to anticipate better decisions. However, the advantages of rtCGM with alarms are especially evident in people at high risk of hypoglycemia or with marked instability because the timeliness of the warning adds a level of protection that intermittent scanning alone cannot provide to the same extent.

This integration of monitoring and treatment has also changed the patient’s relationship with insulin. Decisions are no longer based only on experience, symptoms and a few glucose readings, but on a continuous data stream that makes treatment more adaptive. The trade-off is that the quality of care increasingly depends on competence in interpreting the data and on the reliability of the technology used.

In pediatric patients, continuous monitoring has taken on a central role because it reduces the burden of fingerstick tests, facilitates family supervision, improves recognition of hypoglycemia and integrates well with insulin pumps and automated systems. In children and adolescents, however, the technical benefit must always be accompanied by educational support, school management, data sharing with caregivers and attention to the psychological tolerability of devices.

During pregnancy, a sensor can be particularly useful because it detects postprandial excursions with great precision and reduces time spent outside target, which is highly relevant to maternal and fetal outcomes. Expert interpretation remains essential, however, because targets are tighter, insulin response changes rapidly and the physiology of pregnancy requires far more dynamic management than usual.

During sports and exercise, CGM provides a valuable window into the individual response to movement. Patients can observe whether glucose falls during prolonged aerobic exercise, rises during intense anaerobic activity, how long delayed effects last and how to adjust carbohydrates and insulin. Precisely during exercise, however, the interstitial lag may become clinically important, so in some situations the sensor must be interpreted cautiously and, if necessary, checked with capillary blood glucose.

During acute illness, the sensor may show loss of glycemic control early, but it does not replace ketone assessment or capillary or laboratory confirmation when diabetic ketoacidosis or severe instability is suspected. In these conditions, clinical safety takes priority over technological convenience.

In non-insulin-treated type 2 diabetes, a sensor is not yet a tool to prescribe indiscriminately to everyone, but it can be very useful when used selectively: for nutritional education, understanding the impact of newer treatments, assessing the effect of corticosteroids, documenting unexpected hypoglycemia or showing patients how strongly daily behavior affects the glucose profile. This is where CGM is evolving from a technology used exclusively in insulin therapy into a tool for individualized metabolic assessment.

Despite major progress, continuous monitoring has real limitations. The first is unequal access, which depends on healthcare systems, reimbursement, age, type of diabetes and local organization. This creates a paradox: a clinically valuable technology may be inconsistently available precisely to the groups that need it most. Guidelines have progressively broadened indications, but practical implementation remains heterogeneous.

A second limitation is cognitive burden. More data do not automatically mean better care. Some patients develop alarm fatigue, compulsive checking, excessive corrections or frustration with unavoidable fluctuations. Without adequate support, technology can increase the psychological burden of diabetes rather than reduce it. The care team must also be prepared: correct interpretation of reports requires time, specific expertise and the ability to translate data into realistic decisions.

Technical and skin-related problems also occur, including adhesive allergies, premature sensor detachment, difficult application sites, false compression alarms and the need to carry a device at all times. For some individuals, these issues outweigh the metabolic benefit, which is why CGM adoption must always respect the principle of individual sustainability.

Future developments are aimed at sensors that are more accurate, longer-lasting, less invasive and better integrated with connected pens, insulin pumps and predictive algorithms. Diabetology is moving toward a model in which glucose is observed as a continuous flow interpreted by intelligent systems, rather than as a sequence of isolated checks. Even with the most advanced technology, however, clinical understanding of the data will remain essential. The sensor can display the curve, but the meaning of that curve still depends on physiology, treatment and the ability of the clinician and patient to interpret it correctly.