Continuous Glucose Monitoring

Continuous glucose monitoring (CGM) is a method of measuring glucose concentrations repeatedly over days or weeks, usually with a small sensor worn on the skin. Most devices do not measure blood glucose directly. A thin filament placed under the skin reads glucose in the interstitial fluid, and an enzyme-based electrochemical reaction converts that reading into an electrical signal that is then turned into a glucose value. Because readings arrive every few minutes, CGM shows the direction and speed of change, not only a single snapshot. This is the main difference from traditional fingerstick self-monitoring of blood glucose. The technology is most widely used in diabetes management, especially for people with type 1 diabetes and for some people with type 2 diabetes who take insulin.

CGM matters because low blood glucose and high blood glucose episodes are dangerous, and because the timing of these changes is hard to capture with occasional measurements. Continuous data support insulin dosing decisions, reveal patterns that a single test would miss, and feed into automated insulin delivery systems that adjust insulin in near real time. Its outputs, such as time in range and glucose management indicator, have also changed how clinicians assess glycemic control, which was traditionally summarized by a single HbA1c value. Growth in use has been accompanied by better sensor accuracy, smaller devices, longer wear times, and a shift from clinic-only use to personal use.

This entry first explains the basic measurement principle and sensor design, then describes the main system types and the metrics derived from CGM data. It next covers how accuracy is defined and regulated, the historical development of the technology, and its clinical role in diabetes care and automated insulin delivery. It then considers uses outside diabetes, access and equity issues, and unresolved questions about accuracy, data, and future non-invasive sensing.

Measurement principle and physiology

A continuous glucose monitor (CGM) most often measures glucose in the interstitial fluid (ISF) of subcutaneous fat, not in blood. The sensor's filament sits just under the skin, where glucose moves from the bloodstream into the surrounding fluid. Because glucose has to diffuse out of capillaries before it reaches the sensor, the sensor signal trails the blood value. Estimates of this delay vary. One analysis cited in the literature predicted a time constant of 5–10 min, and some reports note that CGM readings can lag behind blood glucose by as much as 20 min. A study published in Diabetes found that fibrotic encapsulation of the sensor, a scar-tissue layer that forms around the implant, contributed more to the delay than normal ISF physiology did.

The lag matters because the displayed number describes where glucose was a few minutes earlier. During a meal, a rising curve can look less steep than it is, and during insulin action or exercise, a falling glucose may reach the alarm threshold after the true value has already crossed it. Insulin decisions made on stale data can therefore push glucose too high or too low. The lag also varies between sensors and individuals, which complicates the design of automated dosing algorithms.

The sensing chemistry is electrochemical and most commercial devices use the first-generation glucose oxidase reaction. Glucose oxidase catalyzes the oxidation of β-d-glucose with oxygen to produce gluconic acid and hydrogen peroxide. A working electrode, coated with the enzyme and held at a fixed potential by a potentiostat, oxidizes the hydrogen peroxide, releasing electrons. The resulting current is proportional to the amount of glucose that diffused in and reacted, and it is typically measured in nanoamperes. A counter electrode balances the current, and a reference electrode fixes the potential. The method depends on oxygen, which is a limitation: the second generation of sensors used synthetic electron mediators in place of oxygen, and the third generation aimed to pass electrons directly to the electrode. Other electroactive substances in the tissue can also contribute to the signal, which is one source of interference.

Each reading is generated from a stream of raw signal. One design samples the sensor every 10 seconds and reports averaged values every five minutes, while other systems report every one to five minutes. Trend information comes from the change between successive readings. If the difference between two readings is divided by the time between them, the result is a rate of change, which is usually shown as an arrow or slope on the display. Because the rate is calculated from the lagged signal, the trend is also delayed, and small measurement noise becomes more visible when the difference is taken.

Self-monitoring of blood glucose (SMBG) by fingerstick is a different kind of measurement. A user pricks the fingertip with a lancet and applies a drop of capillary blood to a test strip in a meter. This yields a single snapshot, not a curve. The practical difference is large: fingerstick testing is limited by how many pricks a patient is willing to perform each day, while CGM can report up to 288 glucose values per day, which reveals temporal patterns. Fingersticks are also more accurate for a single point in time, and they measure capillary blood rather than ISF, so the two numbers can differ. Older CGM systems required fingerstick calibration, typically about twice per day, and a fingerstick is still commonly used to confirm a low or high reading before treatment.

Sensor and system design

Components and signal path

A typical CGM system is built from a few physically distinct parts. The CGM device consists of 3 basic components: a minimally invasive or non-invasive glucose sensor, the transmitter, and the receiver or display. The sensing element is a thin filament placed under the skin, where it contacts the interstitial fluid that surrounds the tissue cells. In one enzyme-based design, the sensor tip carries glucose oxidase, and the resulting electrochemical reaction produces a current that tracks ambient glucose; the transmitter then sends that signal wirelessly to a handheld receiver, which converts it into a reading in mg/dL The sensor's platinum tip is embedded with glucose oxidase enzyme, which catalyzed a reaction whereby electrical current is generated which is proportional to ambient glucose concentration. The current is then measured and translated into a glucose reading once it is wirelessly sent by the transmitter to the small hand-held receiver.

Display hardware varies. Some products use a dedicated receiver, while others forward data to a smartphone or to an insulin pump. For example, one widely used system is compatible with the iLet, Omnipod 5, Tandem Mobi, and Tandem t:slim X2. Some manufacturers merge the sensing and transmitting functions into one disposable unit; the Dexcom G7 is described as an all-in-one sensor/transmitter, whereas other designs keep a reusable transmitter that snaps onto a separate sensor base.

Membrane layers and electrode architecture

The sensing stack is built in layers on a base substrate. One description of a first-generation electrochemical design lists a base which is a ceramic or polymer having thickness <30 μm, with three electrodes (working, counter, and reference) on top of it. Above the electrodes, an interference-rejection membrane limits the passage of substances such as acetaminophen, ascorbic acid, and uric acid, which would otherwise add false signal.

The enzyme layer is the active chemistry. In enzyme-based biosensors, ambient oxygen facilitates the conversion of the reduced enzyme back to its oxidized form with concomitant production of hydrogen peroxide, and the signal can be read either from peroxide oxidation or from the loss of oxygen. This dependence creates a practical limit: the dynamic range of such sensors is limited by co-substrate (i.e., oxygen) availability. To address this, an outer diffusion-limiting membrane is added so that glucose, rather than oxygen, does not run short at high concentrations An outer diffusion-limiting membrane is thus employed to control. Membrane stability is therefore central to signal stability and linearity; membrane degradation and enzyme leaching are among the recognized causes of in vivo failure The failure of sensor components in vivo may be categorized as follows: 1) enzyme. instability and leaching; 2) membrane degradation and delamination; and, 3) electrode.

Calibration: factory versus fingerstick

Calibration links the raw current to blood glucose. In a fingerstick-calibrated system, the user measures capillary blood with a meter and enters the value, which anchors the sensor's baseline and compensates for drift. Typically, to achieve the desired level of accuracy and reliability and reduce the impact of noise and other spurious signals, the sensor data is calibrated using a known good blood glucose value. For such systems, some CGM systems must be calibrated once at the activation, or calibrated periodically with traditional blood glucose measurements. The Eversense 365 illustrates the workflow: its initialization phase involves a series of 4 fingerstick calibrations 2 to 12 hours apart, and glucose data become available after the second.

Factory calibration moves this work into manufacturing. Factory calibration typically involves batch testing and calibration algorithms. During production, sensors from each batch are tested against known glucose concentrations. The resulting correlation between the electrical current generated by the CGM sensor and the glucose concentration is embedded into the sensor's memory. Because sensors are calibrated batch by batch, sensors in each batch should have a similar deterioration rate and/or sensitivity. Workflow benefits are clear: fingerstick calibration adds burden that users find inconvenient, uncomfortable, or otherwise disfavored by patients. Accuracy is the trade-off. Factory-calibrated devices have improved, but results still depend on lot consistency and on the algorithm; one factory-calibrated Dexcom G6 trial reported an overall MARD of 11.4% in a study that also measured an average sensor lag of 4.5 minutes.

Wear duration, power, and sensor life

Wear duration is set by the sensor's chemistry and by how long the filament remains reliable in tissue. Current disposable systems typically last about between 7 and 15 days, for example Up to 14 days for FreeStyle Libre 2/3, while the Dexcom G7 is rated at a 10-day lifespan and the G7 15-Day extends wear to 15.5 days. Systems often add a grace period; the standard G7 includes a 12-hour grace period for sensor replacement. Implantable devices last far longer: the Eversense E3 is Up to 180 days, and optical systems can last up to one year, though their installation requires a minor surgery, so the device is installed by professionals in the clinic.

Power is handled differently across designs. In the integrated G7, the sensor/transmitter has a non-rechargeable, integrated battery with a 10-day wear period. Separate transmitters usually last longer than the sensor: according to one consumer summary, Transmitters will operate for 3 to 12 months, depending on how much you use them and their brand. Data are typically sent at intervals of minutes; the G7 sends the data to compatible devices including smartphones and the G7 receiver every five minutes.

Sensor life also shapes cost and convenience. Because each sensor must be replaced at the end of its approved wear, a longer-lived sensor reduces the number of insertions and consumables per month. One manufacturer reports that 73.9% of sensors lasted the full 15 days, which shows that the labeled wear time is not met in every case and that early failures remain part of routine use.

Types of systems and their operation

Real-time and intermittently scanned CGM

The two main consumer and clinical formats differ mainly in how readings reach the user. A real-time system (rtCGM) has a transmitter that sends a new glucose value to a receiver, smartphone, or smartwatch every one to five minutes, so values appear without any action by the user. Because the data are pushed continuously, rtCGM can sound alarms and, in newer designs, issue predictive alerts that warn of a trend toward a threshold, not only a crossing of it.

Intermittently scanned CGM (isCGM), often called flash glucose monitoring, works differently. The sensor stores data continuously, but a reading is shown only when the user holds a reader or a compatible phone close to the sensor. The FreeStyle Libre, introduced in 2013, popularised this format. Because isCGM has no separate reusable transmitter, the device is simpler and cheaper to use. The trade-off is that it cannot supply a continuous stream on its own, so closed-loop insulin systems depend on rtCGM. Early isCGM models had no alarms, and later generations added threshold-based alerts only.

Comparative evidence is mixed. A systematic review and meta-analysis of five randomised trials found rtCGM superior to first-generation isCGM in people with type 1 diabetes, with better time in range and less time in both hypoglycaemia and hyperglycaemia. A head-to-head study reported a time-in-range gain of about 7 percentage points (roughly 101 minutes per day) with rtCGM. However, a clinical effectiveness review found that some studies showed no difference in time in range, and that the reduction in time below range was the most consistent finding. The ALERTT1 randomised trial, which compared the two modes in adults with type 1 diabetes, has been reported at 6 and 24 months.

Retrospective and professional (blinded) CGM

Blinded, or professional, CGM records glucose in interstitial fluid but does not display values to the wearer. Data are stored and downloaded in a clinic after a wear period, typically 3, 7, or 14 days, and interpreted as a pattern review. Clinicians use it as a "checkup" when patients do not own a personal device, and it is also used in research to capture glucose data during trials. Some investigators argue that blinded data better reflect everyday behaviour, because patients are not reacting to live numbers.

The clinical value of retrospective review alone is debated. Several studies suggest it adds little to glycaemic control outside pregnancy, where real-time feedback and patient training are considered more effective. Results from individual studies vary: one prospective study in insulin-treated adults with type 2 diabetes reported a 1.3% fall in HbA1c after a year of repeated one-week blinded sessions with treatment decisions informed by the data. Because some classifications group isCGM with retrospective devices, the boundary between "retrospective" and "real-time" is not always sharp.

Standalone CGM and integration with insulin delivery

A standalone CGM provides values, trend arrows, and alarms, while the person adjusts insulin by injection or with a separate pump. In a sensor-augmented pump, the sensor and pump share data, but the user still sets basal rates and boluses, and the pump may suspend delivery when glucose approaches a low threshold. Hybrid closed-loop and fully automated systems go further: a control algorithm adjusts basal delivery at regular intervals using CGM values, and the user typically still announces meals or corrections. Since closed loop depends on continuous streaming, these systems are built around rtCGM.

Emerging optical, microneedle, multi-analyte, and non-invasive designs

Non-invasive and minimally invasive approaches aim to remove the need for a conventional subcutaneous filament. Optical methods, including mid-infrared optoacoustic time-gating, face a basic problem: interstitial glucose lags behind blood and is present at lower concentrations, and variations in pH, hydration, and temperature can affect accuracy. Sweat-based sensing has an even larger lag and has not met the error standard commonly used for clinical acceptability, which is why it is considered insufficient for closed-loop control so far.

Microneedle sensors sit between the two. They penetrate only the outer skin layers and reach interstitial fluid with less pain than a conventional needle. In laboratory tests, one enzyme-based microneedle design showed a linear response over 0 to 20 mM, a sensitivity of about 11.9 nA/mM, and a signal that stayed within ±10% over three days. Small clinical studies of microneedle patches have reported correlation with venous and capillary reference samples, along with lower pain scores. Other microneedle arrays have been built to detect lactate and glutamate, which points toward multi-analyte sensing, though multi-analyte performance has been shown mainly in vitro. Non-enzymatic designs, such as porous platinum electrodes, avoid enzyme instability but rely on protective membranes to limit interference.

The main trade-offs are accuracy against invasiveness, time lag against signal stability, and the small sample sizes of most validation studies. Several non-invasive optical systems have reported high agreement in limited clinical evaluations, but these results have not yet been confirmed in large, independent trials.

Data metrics and interpretation

Time-in-range summaries

CGM data are usually condensed into the share of time spent in each glucose band. Time in range (TIR) is the proportion of readings between 70 and 180 mg/dL (3.9–10.0 mmol/L). The 2019 international consensus uses this as its default range, with 70–140 mg/dL as a secondary option for some purposes. Time below range (TBR) is divided into level 1 (54–69 mg/dL) and level 2 (below 54 mg/dL), and level 2 is treated as clinically significant. Time above range (TAR) covers readings over 180 mg/dL, and the subset above 250 mg/dL is reported separately as "very high." A hypoglycemic event is generally counted when readings stay below the threshold for at least 15 minutes, and it ends once readings have returned to 70 mg/dL or above for 15 minutes. Level 3 hypoglycemia is defined by the need for assistance from another person rather than by a specific glucose value, which is why it is handled separately from the numerical bands. For a detailed account of the consensus origins, see the hypoglycemia entry.

Clinical targets

For most adults with type 1 or type 2 diabetes, the consensus targets are TIR above 70% (about 17 hours per day), TBR below 70 mg/dL under 4%, TBR below 54 mg/dL under 1%, TAR above 180 mg/dL under 25%, and TAR above 250 mg/dL under 5%. Separate targets have been proposed for older adults, people at high risk of hypoglycemia, and pregnant women with type 1 diabetes. The consensus also gives rough translations: each 1 percentage point of TIR corresponds to about 15 minutes per day, and a 10-point change in TIR is associated with roughly a 0.5-point change in HbA1c. TIR of 70% is linked to an HbA1c near 7%. These are population-level correspondences and do not describe any individual's exact relationship.

Glucose management indicator

The glucose management indicator (GMI) converts mean CGM glucose into an estimate of HbA1c, using the formula GMI (%) = 3.31 + 0.02392 × mean glucose (mg/dL). It is calculated from a period of at least 14 days of data and replaced the earlier term "estimated HbA1c," which was dropped because of confusion with the laboratory test. Because GMI rests on interstitial glucose averaged over days or weeks, while HbA1c reflects hemoglobin glycation over roughly two to three months, the two numbers can diverge. Several sources contribute to the gap. These include CGM measurement error, the length of monitoring, and missing data. On the HbA1c side, conditions that alter red blood cell lifespan or hemoglobin structure can distort the lab result. Discordance also varies with clinical status. In a cohort of people with chronic kidney disease, the difference exceeded 0.5 percentage points in 68.2% of participants, compared with 42.2% of those without kidney disease. In an intermittently scanned cohort, discordance increased with higher BMI and with type 2 diabetes. In a small retrospective study of 26 patients, measured HbA1c averaged 0.34 percentage points above GMI, and in a pediatric cohort, larger discrepancies were associated with greater glycemic variability. Such studies are small or single-center, so the magnitude of discordance should be read as indicative rather than general.

Glycemic variability and the ambulatory glucose profile

Glycemic variability is commonly measured by the coefficient of variation (%CV), calculated as the standard deviation of glucose divided by mean glucose, multiplied by 100. A %CV of 36% or less is considered acceptable, and values above 36% indicate unstable glucose and a need for review. Some evidence suggests that a lower threshold of 33% offers additional protection against hypoglycemia in people taking insulin or sulfonylureas. The %CV is preferred over the raw standard deviation because glucose distributions in diabetes are rarely normal, and because %CV remains comparable across periods even when average glucose changes. In one illustrative comparison, two profiles with identical GMI values of 8.3% had %CVs of 32.8% and 46.6%, showing that the average alone can conceal substantial differences in daily swings.

The ambulatory glucose profile (AGP) is the standardized one-page report used to display these patterns. It summarizes readings over a reference period, usually 10 to 14 days, as median and percentile bands plotted across a single 24-hour day, and it also includes a metrics table and individual daily graphs. The AGP can be generated from anywhere between 5 days and 3 months of data, and several consensus statements have adopted it as the default report format.

Minimum data sufficiency

A report is considered reliable only when it covers enough time and enough active sensor data. The consensus-based standard is 14 consecutive days of data with active readings at least 70% of the time, and ideally at least 90%. Studies have found that 14 days correlate strongly with 3-month data for mean glucose, TIR and hyperglycemia, which supports the shorter window. Intermittently scanned systems have their own requirement: users must scan at least every 8 hours or the gaps will appear in the report. Reports that fall below these thresholds may misrepresent typical patterns, which is why data completeness is listed first on the AGP and why GMI and TIR should be interpreted together with wear time.

Accuracy, standards, and regulation

Measuring agreement with a reference

CGM accuracy is judged by pairing each sensor value with a reference measurement taken at the same moment. A common metric for assessing CGM accuracy is the aggregate mean absolute relative difference (MARD), which compares all temporally matched sensor data with reference measurements (e.g., Yellow Springs Instrument [YSI]) across all subjects of a given study.

MARD is computed by taking each paired difference relative to the reference, treating every difference as positive regardless of its direction, and averaging the results. "Absolute" means each individual relative difference value is considered a positive value, irrespective of whether the calculated difference with respect to the comparison result is positive or negative. A smaller figure signals closer agreement, since a small MARD number indicates that the CGM readings are close to the reference glucose value. Some studies also report MARD separately for glucose ranges, and a MARD below about 10% is generally regarded as good, and most modern CGMs sit around 8–10%, although there is no official regulatory MARD "pass mark".

MARD values are not always directly comparable. Different studies vary in reference method (SMBG vs. lab analysis), sampling frequency, or even glucose range tested. Because an average can hide large individual errors, MARD is usually paired with error grid analysis. Error grids, such as the Clarke error grid and its successors, plot each device value against its reference and sort the pairs into zones according to the likely clinical consequence of acting on the reading. The consensus error grid for type 1 diabetes is one such tool, and its zones A and B are treated as clinically acceptable.

Standards and regulatory criteria

ISO 15197 was written for blood glucose monitoring (BGM) systems rather than CGM. ISO 15197:2013 specifies that at least 95% of BGMS results must be within ±15 mg/dl at glucose concentrations below 100 mg/dl and within ±15% at or above 100 mg/dl, and that at least 99% of results must fall within zones A and B of a consensus error grid analysis. The earlier 2003 edition used a ±15 mg/dL limit below 75 mg/dL, and ISO 15197:2013 introduces more stringent accuracy criteria by extending that limit to 100 mg/dL. Applying the fingerstick standard to CGM produces a mismatch: current-generation CGM systems have published accuracies meeting the 15% requirement, but are accompanied by a large reduction in percentage of samples considered accurate according to the ISO 15197 standard.

In the United States, the FDA introduced a separate category in 2018. In 2018, the US Food and Drug Administration (FDA) introduced a new regulatory category, "integrated CGM (iCGM)", and iCGM devices are approved for non-adjunctive use, allowing CGM values to be used directly for treatment decisions without confirmatory fingerstick measurements. The special controls set accuracy thresholds at the 95% lower confidence bound. For example, within 15 mg/dL below 70 mg/dL must exceed 85%, within 15% between 70 and 180 mg/dL must exceed 70%, and within 15% above 180 mg/dL must exceed 80%. Rate-of-change criteria also apply, and the clinical data must be compared with an FDA-accepted laboratory-based glucose measurement method that is precise and accurate, and that is traceable to a higher order. One 2025 review describes this category as the strictest accuracy standard defined by any regulatory agency in the world, a characterization that reflects the authors' view rather than a formal ranking.

Four dimensions of accuracy

Accuracy is often assessed along several axes rather than by one number. Analytical accuracy concerns agreement with the reference across the full glucose range. Clinical accuracy asks whether readings would lead to the same treatment decision as the reference, which matters most near hypoglycemic and hyperglycemic thresholds. Trend accuracy concerns the direction and speed of change, which the iCGM rate-of-change criteria address directly. Precision describes repeatability under the same conditions. Durability asks whether performance holds across sensor lots, the wear period, and insertion sites. The FDA requires that clinical study results must demonstrate consistent analytical and clinical performance throughout the sensor wear period. Because assessment of the accuracy of CGM has not been standardized globally, accuracy data should be considered in the context of the study design and data reporting.

Sources of error

Physiological lag is the most fundamental source of discrepancy. Interstitial glucose changes more slowly than capillary blood glucose, and the choice of reference sample matters: if the device is calibrated using capillary blood glucose, use of other samples (e.g., venous blood or arterialized venous blood) for the reference method can introduce additional bias. Calibration errors propagate, since an incorrect meter point used for calibration can falsely raise or lower calculated glucose levels. Rapid glucose excursions, such as those during exercise, enlarge the lag. Sensor compression, drug or chemical interference, and temperature effects further degrade readings, and these are discussed in more detail in the sections on sensor design and hospital use.

Historical development

Enzyme-electrode origins (1960s–1980s)

The sensing principle behind modern devices dates to 1962. The original design held a thin layer of glucose oxidase between a glucose-permeable membrane and an oxygen electrode, and it inferred glucose from the oxygen the enzyme consumed a glucose biosensor consisting of an oxygen electrode, an inner oxygen semipermeable membrane, a thin layer of GOx, and an outer dialysis membrane. Later workers measured the hydrogen peroxide produced in the reaction instead, and added a redox mediator to move electrons between the enzyme and the electrode; the addition of a redox mediator to the reaction, to facilitate the efficient transfer of electrons between the glucose oxidase enzyme and the electrode is the principle on which the current market leading sensors are based. About three decades after the 1962 work, researchers placed subcutaneous sensors in rats and obtained usable estimates of blood glucose for up to ten days. Implantable sensor research grew in the early 1980s, but this technology was commercially unavailable until the original MiniMed CGMS, CGMS Gold (Medtronic MiniMed, CA, USA) received Food and Drug Administration (FDA) approval in 1999.

Retrospective professional devices (1990s)

The first commercial system was a clinic-based device. Its sensor was worn by the patient, who could not see the readings, and the recorded data were extracted by the clinician afterwards. Because the wearer received no live values, the data served mainly to reveal patterns, and CGM readings alone were not suitable to offer the basis for therapeutic decision-making and further required a conformational finger-prick test. Early use was largely confined to research. The sensor's wear period was short, with the first device approved for three days of use in a physician's office, and the 1999 FDA approval covered that limited use pattern.

Real-time systems (2000s)

The shift to real-time display came in the following decade. Dexcom's Short Term Sensor received approval in 2006, a few weeks before Medtronic obtained clearance for its Paradigm REAL-Time system, which showed the glucose trend line and a rate-of-change arrow on the screen of a pump-linked device rather than on a separate monitor. In 2007, FDA approved MiniMed Paradigm REAL-Time system (Insulin Pump and CGM system), the first diabetes management integrated system which was a major leap of progress toward the development of a "closed-loop" insulin delivery system or artificial pancreas.

Successive sensor generations improved wear time, size, and the need for calibration. At the time, the CGM devices used in this study were all early generation; they were larger than current devices, and they required fingerstick calibrations in addition to fingerstick glucose checks for any treatment decisions. Later products moved toward factory calibration, and Medtronic MiniMed has continued to develop its Guardian sensor (GS4), which was recently approved in both Europe and the USA, that lasts for 7 days requires no calibration.

Scanned systems and consumer products (2010s–2020s)

The intermittently scanned format spread from Europe. Abbott's FreeStyle Libre first launched in Europe in 2014, with a 14-day wear time, was the first to eliminate the need for fingersticks. Users read the sensor with a handheld scanner rather than receiving a continuous stream. The FDA approved the FreeStyle Libre Flash Glucose Monitoring System in September 2017, the first continuous glucose monitoring system that can be used by adult patients to make diabetes treatment decisions without calibration using a blood sample from the fingertip. A 14-day version followed in the United States in 2018, and scanning through a smartphone app was cleared in July 2021.

The same period saw the move from clinic-based to personal devices. Professional CGMs later morphed into personal CGMs. Consumer-facing products reached the market later still: the Dexcom Stelo Glucose Biosensor System is the first over-the-counter iCGM, approved by FDA, intended for individuals aged 18 years and older who are noninsulin users, or those without diabetes.

Regulatory and reimbursement milestones

Access depended as much on payer decisions as on device approval. In the United Kingdom, the Freestyle Libre (FSL) flash glucose monitor became available on prescription (subject to local health authority approval) in all four nations of the United Kingdom from November 2017. In the United States, one review noted that CMS also recommended approval of CGM use in patients with T2D experiencing hypoglycemia without being on insulin. Such decisions widened the population of people with type 2 diabetes who could obtain devices, moving CGM beyond people treated with insulin.

Clinical use in diabetes care

Type 1 and type 2 diabetes

CGM is most firmly established for people with Type 1 Diabetes and for people with Type 2 Diabetes who take insulin. The 2026 Standards of Care from the American Diabetes Association recommend CGM for older adults with type 1 diabetes and for older adults with type 2 diabetes on insulin, citing better glycemic results, fewer low-glucose episodes, and a lighter treatment burden as the rationale. Recommend continuous glucose monitoring (CGM) for older adults with type 1 diabetes A and type 2 diabetes on insulin therapy B to improve glycemic outcomes, reduce hypoglycemia, and reduce treatment burden. The same edition extends the recommendation beyond insulin users: Use of continuous glucose monitoring (CGM) is recommended at diabetes onset and anytime thereafter for adults with diabetes who are on insulin therapy, and it also favors CGM for adults with type 2 diabetes on other glucose-lowering drugs when the device can improve management.

The hypoglycemia evidence comes mainly from randomized trials. In a US trial of adults with poorly controlled type 2 diabetes on basal insulin, the mean HbA1c fell from 9.1% to 8.0% in the CGM group, compared with a fall from 9.0% to 8.4% in the blood glucose meter group, and severe hypoglycemic events occurred in one participant in each arm. Mean HbA1c level decreased from 9.1% at baseline to 8.0% at 8 months in the CGM group and from 9.0% to 8.4% in the BGM group. Severe hypoglycemic events occurred in 1 participant (1%) in the CGM group and in 1 (2%) in the BGM group. A systematic review of mixed-population trials found that real-time CGM users recorded fewer sensor-detected hypoglycemic events than meter users, a difference driven mainly by people taking insulin or sulfonylureas. rtCGM users had fewer CGM-derived hypoglycaemia events compared to BGM users, driven by the insulin and sulphonylurea groups.

For people with type 2 diabetes who do not use insulin, the evidence is smaller. Their hypoglycemia risk is lower, which makes benefits harder to demonstrate, and they are under-represented in CGM trials. The meta-analysis of RCTs in noninsulin users included 8 studies encompassing 541 participants, and pooled results showed a modest HbA1c reduction. A 26-week trial in US primary care practices, reported in the New England Journal of Medicine, found a greater HbA1c reduction with CGM than with routine care, and no severe hypoglycemia events were reported.

Pregnancy and gestational diabetes

The largest pregnancy trial is CONCEPTT, which studied women with type 1 diabetes who were pregnant or planning pregnancy. Between March 25, 2013, and March 22, 2016, researchers randomly assigned 325 women (215 pregnant, 110 planning pregnancy) to capillary glucose monitoring with CGM or without. The HbA1c benefit was small: a mean difference of -0·19% in pregnant women using CGM. The trial's stronger findings concerned neonatal outcomes. Among pregnant participants, the incidence of large for gestational age infants was lower (odds ratio 0·51), along with fewer neonatal intensive care admissions lasting more than 24 h and fewer incidences of neonatal hypoglycaemia. The investigators linked these effects to reduced exposure to maternal hyperglycemia. They found no apparent benefit of CGM in women planning pregnancy, and severe hypoglycemia episodes were similar between groups.

Evidence for CGM in gestational diabetes is thinner than the pregnancy data for type 1 diabetes. Use in this group is usually decided case by case, often for women treated with insulin or medication, and it rests on fewer trials and smaller samples.

Guideline recommendations and routine education

Professional guidelines now treat CGM as part of routine care rather than as a tool reserved for difficult cases. The 2026 ADA update recommended use of continuous glucose monitoring at diabetes onset and anytime thereafter to improve outcomes for anyone who could benefit from its use in diabetes management. For children, CGM is recommended at diagnosis or as soon as possible for those who can use the device safely, alone or with caregivers.

In practice, the guidelines pair device prescription with structured teaching. When prescribing a continuous glucose monitoring (CGM) device, ensure that people with diabetes and caregivers are offered initial and ongoing training and education as indicated by individual circumstances. That education is expected to cover how to use the data, including uploading or sharing it so that therapy can be adjusted. Fingerstick testing also remains necessary: people using CGM must also have access to BGM at all times. Clinicians can assess glycemic status using HbA1c, CGM metrics such as time in range, or both, and fructosamine or CGM can substitute for HbA1c when HbA1c is unsuitable.

Limits of the evidence

Several limits qualify these findings. Trials have mostly followed HbA1c, time in range, and hypoglycemia over months rather than years, so the link between sensor use and long-term complications is not well established. One systematic review of real-time CGM in type 2 diabetes noted that episodic use required multiple sessions for sustained HbA1c reduction, which suggests that benefits depend on how the device is used over time. Its authors called for research on long-term efficacy and cost-effectiveness. Interpreting the data well also requires training, because raw glucose traces can prompt overcorrection if the user does not understand trends and alert settings.

Integration with insulin delivery and digital health

Glucose sensing becomes most consequential when its readings drive insulin dosing, the setting in which the artificial pancreas concept is realized. The most established form is hybrid closed-loop, or automated insulin delivery (AID), in which a subcutaneously worn glucose sensor communicates with an algorithm that adjusts basal insulin while the user still announces meals. Sensor-augmented pumps stop short of this: they display readings and may suspend delivery, but they do not calculate doses. Commercial systems differ mainly in their algorithms. The Medtronic 670G and 780G use PID with insulin feedback, whereas CamAPS FX places an adaptive model predictive controller on a smartphone and has been described as usable with different pumps and sensors (the algorithm resides on a smartphone application and has the potential to be used with any insulin pump and CGM).

Two control families dominate. A proportional-integral-derivative (PID) controller combines three signals: the current gap to target, the speed at which glucose is changing, and the accumulated error over time, each with a weight that may be fixed or adjusted (PID controllers direct insulin doses based on the difference from target glucose at the current time point (proportional component), the rate of change in sensor glucose over time (derivative component) and the area under the curve between measured and target glucose levels including memory of prior controller-induced changes (integral component). All three components are weighted with a multiplier, which may be pre-determined or adjusted over time.). Model predictive control (MPC) instead uses a model of glucose–insulin dynamics to forecast future values and select doses that keep glucose near target while avoiding lows (It uses a mathematical model of glucose-insulin dynamics to predict future glucose trajectories and compute insulin doses minimising deviations from target range while avoiding hypoglycemia.). Reviewers note that MPC has become a standard formulation partly because it handles constraints in an intuitive way (For many, model predictive control (MPC) has become the standard formulation for insulin delivery algorithms because of its intuitive treatment of constraints and ease of modification.). Any design must account for sensor inaccuracy, the slow action of insulin, and the rapid effect of carbohydrate (The algorithm must account for device accuracy and reliability, the relatively slow pharmacodynamics of insulin, and the relatively fast effects of carbohydrate ingestion.). The physiological lag of interstitial sensing therefore enters the controller directly, not only as a reporting issue.

Safety limits are a separate layer on top of the control law. Commercial designs embed bounds on dosing; the 780G, for example, keeps adaptive insulin limits alongside its PID algorithm and model-based corrections (The PID algorithm including insulin feedback with adaptive insulin limits remains embedded on the 780G pump, and includes added model-based autocorrections.). The central risk is that the controller acts on a wrong input. A basic loop sees only the glucose value and cannot check whether a change is physiologically plausible, and without bounds it could dose on erroneous data (In typical existing infusion pump systems for the automatic regulation of the delivery of insulin formulation, there may be no safety limits on the amount of insulin formulation that is delivered based on an erroneous input. In other words, the typical closed-loop algorithm may only examine the glucose level input and may have no way to verify whether the change in glucose level is a reasonable change). Typical safeguards, which vary by product, include limits on dosing rate and total insulin, sensor-signal validity checks, automatic suspension when glucose is low or falling, and fallback to basal delivery when the sensor drops out. Clinical evidence is still mostly short-term: a 2026 meta-analysis of AID in type 2 diabetes reported that AID use increases time in range, reduces time above target glucose, and maintains similar rates of hypoglycemia and serious adverse events. While AID slightly increases the risk of device deficiency, safety is still within acceptable limits. The same review judged that longer and larger trials were still needed (larger and longer-term studies are necessary to confirm the benefits and assess cost-effectiveness).

Alarm design largely determines whether users keep these systems running. Alert fatigue occurs when alerts arrive, but too often, too early or for readings that resolve on their own. Suggested remedies aim for fewer, more actionable alerts while keeping the warnings for clinically important lows (fewer, more actionable alerts, while preserving clinically important lows). Alerts can also fail to arrive at all because of connectivity or operating-system settings, which is a separate problem from over-alerting (Signal loss, Bluetooth, range, permissions, an OS update or app restrictions are dropping the alert.). Smartphone apps allow users to set alert profiles, including two distinct schedules for their alerts, and they upload data to cloud servers that can feed remote-monitoring apps used by caregivers or clinicians (The CGM app 410 can provide the insulin data and event entry data ... to the Cloud). On the clinical side, automatic uploads can overwhelm staff, since clinicians may receive hundreds of data points a day and might miss an out-of-range alarm if they are expected to review results in real time. Open-source AID systems, which combine pumps, sensors and phone apps outside the manufacturer's bundle, add further questions about support and safety (Open-source AID systems like Loop, AndroidAPS, and Trio combine pumps, CGMs, and smartphone apps to deliver customizable insulin automation.).

Data flows that connect sensors, phones and clouds also raise questions of security and ownership, discussed in data privacy and cybersecurity. Machine learning is being tested for the prediction side of the loop: one 2026 prototype used a Transformer network to forecast glucose 30 minutes ahead and fed that forecast into a PID-controlled insulin release (This system supports a proportional-integral-differential (PID)-controlled GRI release mechanism with precise 30-minute glucose predictions.). Whether such predictive layers improve safety in routine use remains to be established in larger trials.

Use beyond diabetes and in hospitals

Outside routine outpatient diabetes care, CGM is used in intensive care units, in research on people without diabetes, in over-the-counter consumer products, and in sports science. Each setting raises different questions about accuracy and interpretation.

In critical care, the main concern is whether sensors remain reliable when patients are severely ill. In an observational study of 103 ICU patients, nearly all of whom were mechanically ventilated and most of whom were receiving vasopressors, sensors showed a median bias of −0.30 mmol/L and a median MARD of about 11%, and no clinical hypoglycemia occurred while sensors were in place. MARD values in other ICU cohorts are less favorable. A multicenter study of an intermittently scanned system reported an overall MARD of 18.0%, and the error was largest in the hypoglycemic range, where it reached 33.1%. Sepsis adds rapid hemodynamic and metabolic shifts and microvascular dysfunction, which have prompted doubts about sensor accuracy in that population. Hospital practice has adapted to these limits: one safety-net hospital placed sensors on the upper arm so that patients could be repositioned, including prone positioning, and it required fingerstick or laboratory confirmation until two consecutive readings agreed with the sensor within 30 mg/dL before insulin dosing was based on CGM values.

Research on people without diabetes uses CGM to observe how blood glucose responds to meals, sleep, and exercise in daily life. Observational data from healthy adults indicate that most of the day is spent between 70 and 140 mg/dL, with brief post-meal rises above 140 mg/dL that typically resolve within about two hours. Such studies can describe individual variation in glucose responses to the same food, which is a central motivation for nutrition and prediabetes research. However, no outcome trial has yet shown that wearing a sensor improves long-term health in people without diabetes, and interpretation of normal glucose excursions remains an open research question. The sensor cannot diagnose diabetes or prediabetes, and diagnosis still depends on laboratory tests.

Consumer products began in the United States in 2024. The FDA authorized Dexcom's Stelo in March 2024 as the first CGM for over-the-counter sale, labeled for adults aged 18 and older who do not use insulin, including people without diabetes who want to understand how food and activity affect them. Abbott's Lingo was cleared in June 2024 as a wellness product for adults not using insulin, and its Libre Rio was cleared for adults with type 2 diabetes who do not use insulin. Stelo reports readings every 15 minutes through a smartphone app, with wear of up to 15 days, and it is not designed to alert users to hypoglycemia. The regulatory questions concern the boundary between wellness use and medical use, while the interpretive questions concern how users, without clinical training, should respond to normal fluctuations or to readings that cannot be verified without a fingerstick.

In sports and exercise physiology, rapid glucose changes make sensor accuracy a particular concern. In one study of athletes without diabetes, the median MARD was 7.3% during rigorous exercise but rose to 25.1% during sedentary periods. The authors attributed this pattern to interstitial fluid not being actively pumped like blood, relying instead on muscle movement to circulate and mix, so that exercise promotes faster equilibration between blood and interstitial glucose. In para-cyclists, accuracy was acceptable at rest (MARD 12%) but markedly lower during exercise (MARD 34%), and it was worst in hand-bike riders, especially when the sensor sat near highly active muscles. Lag also increases after large carbohydrate boluses, when blood glucose is changing quickly. Because interstitial glucose is not a direct measure of carbohydrate availability for working muscles, sports applications require cautious interpretation.

Access, cost, and equity

Access to CGM depends as much on payment rules as on sensor performance. Prices in the United States vary sharply by channel. One 2026 price guide puts the cost from $54 for a single two-week over-the-counter (OTC) starter sensor to $579 a month for a brand-name prescription system bought without insurance. Other estimates give a single sensor typically ranges from $70 to $150, so published figures are not directly comparable. Original Medicare covers CGM and supplies as durable medical equipment, and beneficiaries generally pay 20% of the approved amount after the Part B deductible, which was $283 in 2026 (After you meet the annual Part B deductible ($283 in 2026), you generally pay 20% of the Medicare-approved amount, as long as your supplier accepts assignment.). Medicare eligibility was widened in 2023 to include people with diabetes who do not use insulin. Private coverage is broad for type 1 diabetes, while coverage for type 2 diabetes often depends on insulin use. Medicaid coverage is set state by state. A prescription is generally needed for insurance to pay (A prescription is required for insurance to cover a continuous glucose monitor.). Over-the-counter devices sold directly to consumers have created a separate, uninsured price tier. Manufacturer-supplied figures should be read with caution: according to Abbott, more than 95% of private plans cover its FreeStyle Libre systems, a claim the company itself reports.

Disparities by income, region, and setting

Within high-income systems, use is uneven. A 2025 systematic review of children and young people with type 1 diabetes found that decreased CGM use and higher discontinuation for low SES, low education, publicly insured and minority ethnic, especially Black, CYP, and these associations were generally robust to adjustment for other sociodemographic variables. The authors concluded that there are significant SES, ethnic and education inequalities in CGM use, with the gap widest particularly when reimbursement is limited.

In low- and middle-income countries (LMICs), access is far narrower. In Africa, for example, advanced diabetes technologies including insulin pumps and CGM devices are virtually absent from public healthcare systems in Nigeria, and in more affluent countries such as Egypt, Morocco and South Africa, uptake remains largely limited to private beneficiaries on insurance. In Brazil, insulin pumps and continuous glucose monitoring (CGM) systems are expensive and are not currently provided by Brazil's public healthcare system (SUS), nor do private insurance routinely cover them. In the Philippines, barriers include unaffordability, high illiteracy rate, reimbursement constraints, lack of healthcare professionals promoting and implementing use of CGM, and inexperience of both patients and clinicians. Shortages of trained staff and limited infrastructure compound these financial constraints (see health systems and universal-health-coverage), and the resulting gap is often described as a digital divide in diabetes care.

Data privacy, ownership, and security

Continuously collected glucose data are sensitive. A CGM trace can reveal meal times, stress responses, sleep patterns, and alcohol consumption, and when combined with smartphone location data, it forms a detailed behavioral profile. The path from body to third party is often long: the data originate from a medical device on the patient's body, travel through a consumer application, sit on commercial cloud servers, and reach third parties with no connection to the patient's care.

Legal protection is uneven. Data ownership remains legally ambiguous in most jurisdictions, with patients, healthcare providers, device manufacturers, and app developers each holding competing claims. In the United States, HIPAA's entity-based scope leaves a regulatory gap for consumer-grade applications. By contrast, the GDPR covers all processing of European Union residents' health data regardless of entity type. Deidentification is another weak point: many CGM manufacturers disclose in their privacy policies that they may deidentify certain information, but they are not legally obligated to meet HIPAA's Safe Harbor standard. Re-identification is a further risk, since CGM data can be used to pseudo-reidentify individuals using ML algorithms, and reported accuracy may reach as high as 86%. Consent for research reuse is also a concern, because consent frameworks in place often fail to explain what will happen to the data, who will access them, or for what purpose. Consent and data governance are therefore linked to data-privacy, informed-consent and cybersecurity.

Domestic manufacturing and national policy

Local production and national reimbursement decisions are central to expanding access, because they affect both unit prices and whether public programmes can finance devices at all. Researchers writing about the Philippines have argued that local governments could potentially collaborate with private sectors to make CGM more accessible, and economic analyses can also spearhead the amendment of national health policies for a better reimbursement system of diabetes technology. They also observed that as diabetes technology advances, national health policies clearly lag behind. Assessment of cost-effectiveness through health technology assessment and health financing decisions thus determine how quickly these devices reach public patients.

Open questions and research frontiers

Current research is moving in several directions at once: measuring more than one analyte, replacing the skin-worn electrochemical sensor with optical or non-invasive methods, and using machine learning to process signals and predict glucose excursions. Each direction raises its own questions about reliability and evidence.

Multi-analyte sensing is the most advanced of these. Abbott's dual glucose–ketone sensor measures interstitial glucose and ketones with the same size and on-body footprint as its glucose-only sensor, and both analytes are read through enzymatic amperometric reactions that produce a current. A sensor membrane covers each sensing layer and the underlying working electrodes and forms a biocompatible interface to the interstitial fluid. This membrane allows the analyte to diffuse at a rate that ensures a linear sensor response and sufficient sensitivity. The company reported that factory calibration of two analytes on one sensor was a central technical difficulty, because each new analyte brings its own chemistry and membrane constraints. The rationale is clinical: glucose and ketone concentrations may not directly correlate with each other in a diabetic individual also exhibiting ketoacidosis, which is why concurrent measurement is of interest for diabetic ketoacidosis. A CE mark in Europe has allowed the product to be marketed there, though competitor Dexcom has also talked about dual-analyte sensing, although plans for continuous ketone monitoring are further out. Laboratory work on other designs, such as an optical hydrogel "barcode" sensor for glucose and oxygen, reports storage stability over 12 weeks and about 6% reproducibility in phosphorescence lifetime, but these are bench-scale results without clinical outcome data.

Non-invasive and minimally invasive glucose sensing remains the least settled area. A systematic review and meta-analysis of 15 studies with 733 participants found that pooled sensitivity for hypoglycaemia detection was 69.3% and mean specificity was 93.3%. The same authors found no evidence of better performance for the most recent generation of devices and concluded that these systems were not yet accurate enough for hypoglycaemia detection in routine use. Some newer prototypes report encouraging figures. A mid-infrared device reached a mean absolute relative difference (MARD) of 19.6–20.7% across two sessions, equivalent to the performance of early CGM systems cleared by the FDA for adjunctive use. A visible-light optical device presented at a scientific meeting reported a Mean Absolute Relative Deviation of 8.8%, but conference presentations of this kind have not been subjected to the same peer review as full publications. Independent testing of a photoplethysmography-based home meter found that the device does not meet the minimum criteria set by ISO or FDA, which illustrates why independent validation is essential.

Durability and biocompatibility limit how long implanted or long-wearing sensors can perform. Commercial electrochemical sensors are typically worn for only a few days to two weeks, with constant sensor-tissue friction shortening the longevity of these sensors to 3-14 days. An optical implant lasts considerably longer, with a reported lifespan of 180 days, but it requires a surgical procedure. Tissue responses are a shared problem: the foreign body reaction hampers metabolite diffusion and degrades optical chemistry, ultimately causing inaccuracies in the sensor's readings. How repeated insertion and long wear affect skin and local immune responses across diverse populations is still insufficiently characterized in the published work examined here.

Standardization is a parallel concern. Accuracy is reported with different metrics, thresholds and reference methods across manufacturers and regulators. For example, FDA criteria state that 95% of meter readings should be within ±15% of the reference reading and 99% within ±20%, while ISO-based claims are often framed through error-grid zones, so results across studies are not always directly comparable. Harmonizing these criteria internationally would make cross-device comparisons meaningful.

Artificial intelligence is being applied to signal processing, forecasting and dosing. Forecasting studies generally report promising technical performance, but a systematic review found that substantial heterogeneity in datasets, model architectures, prediction horizons, outcome definitions, and validation strategies limited direct comparisons. Clinical evidence lags further behind: relatively few studies evaluated prospective clinical or decision-support interventions and clinically meaningful outcomes such as time in range or HbA1c. In intensive care, a review concluded that routine adoption depends on prospective multicenter trials with standardized endpoints, transparent calibration, workflow integration, and safety monitoring. The evidence needed before routine clinical use therefore includes prospective, multicenter clinical trials with pre-specified outcomes, external validation in populations different from those used for training, and explicit safety monitoring for closed-loop dosing decisions.

References

1. 962-P: Continuous Glucose Monitor Readings Lag Interstitial Glucose by Several Minutesdiabetesjournals.org
2. Fibrotic Encapsulation Is the Dominant Source of Continuous Glucose Monitor Delaysdiabetesjournals.org
3. Continuous Glucose Monitoring (CGM): What It Ismy.clevelandclinic.org
4. Interstitial fluid glucose time-lag correction for real-time continuous glucose monitoring - ScienceDirectsciencedirect.com
5. A systematic review of continuous glucose monitoring sensors: principles, core technologies and performance evaluation - ScienceDirectsciencedirect.com
6. Glucose Oxidoreductases for Glucose Monitoring: A Bird’s Eye View of Glucose-Sensing Enzymes II - PMCpmc.ncbi.nlm.nih.gov
7. Glucose Sensor - an overviewsciencedirect.com
8. An Overview of Current and Emerging Biomaterials Technology for Continuous Glucose Monitoring (CGM) Devices -- Current state and future perspectives of the leading technologiesarxiv.org
9. Continuous Glucose Monitoring Versus Self-monitoring of Blood Glucose in Children with Type 1 Diabetes- Are there Pros and Cons for Both? - PMCpmc.ncbi.nlm.nih.gov
10. Continuous Glucose Monitoring Versus Self-monitoring of Blood Glucose in Type 2 Diabetes Mellitus: A Systematic Review with Meta-analysisncbi.nlm.nih.gov
11. CGM vs Finger Stick: Which Is More Accurate?healthcentral.com
12. Continuous blood glucose monitorimage-ppubs.uspto.gov
13. Blood glucose monitoringen.wikipedia.org
14. Review—Electrochemistry and Other Emerging Technologies for Continuous Glucose Monitoring Devices - IOPscienceiopscience.iop.org
15. Efficacy of Continuous Glucose Monitoring in Neonates With Hypoglycemiacdn.clinicaltrials.gov
16. Dexcom G7 and G7 15-Day CGM Systemconsumerguide.diabetes.org
17. Biocompatible Materials for Continuous Glucose Monitoring Devicescdr.lib.unc.edu
18. Manufacturing controls for sensor calibration using fabrication measurementsimage-ppubs.uspto.gov
19. Continuous glucose monitoren.wikipedia.org
20. CALIBRATIONeversensecgm.com
21. Implantable continuous glucose monitor deviceimage-ppubs.uspto.gov
22. How Long Does a CGM Sensor Last? (All Brands Compared)cgmmonitors.com
23. How Long Does A CGM Sensor Last?fitlyfast.com
24. Dexcom CGMen.wikipedia.org
25. How Long Does a CGM Last? Device Lifespanhealifynow.com
26. New Dexcom G7 15 Day CGM: Longer Wear Time, More Accuracydexcom.com
27. Factory-Calibrated Continuous Glucose Monitoring: How and Why It Works, and the Dangers of Reuse Beyond Approved Duration of Wear - PMCpmc.ncbi.nlm.nih.gov
28. Continuous Glucose Monitor (CGM) Glossaryadces.org
29. How Long Do CGM Sensors Last?aeroflowdiabetes.com
30. Real-Time CGM Shows Benefits Over Intermittently Scanned CGM in T1Dajmc.com
31. The accuracy of capillary blood glucose testing versus real time and intermittently scanned continuous glucose monitoring - El‐Abd - 2023 - Practical Diabetes - Wiley Online Librarywchh.onlinelibrary.wiley.com
32. Intermittently Scanned and Real-Time Continuous Glucose Monitoring for People With Diabetes - NCBI Bookshelfncbi.nlm.nih.gov
33. Effect of switching from intermittently scanned to real-time continuous glucose monitoring in adults with type 1 diabetes: 24-month results from the randomised ALERTT1 trial - ScienceDirectsciencedirect.com
34. Comparing real-time and intermittently scanned continuous glucose monitoring in adults with type 1 diabetes (ALERTT1): a 6-month, prospective, multicentre, randomised controlled trial - ScienceDirectsciencedirect.com
35. Comparison of flash glucose monitoring with real time continuous glucose monitoring in children and adolescents with type 1 diabetes treated with continuous subcutaneous insulin infusion - ScienceDirectsciencedirect.com
36. Flash vs Continuous Glucose Monitoringomnipod.com
37. Impact of blinded retrospective continuous glucose monitoring on clinical decision making and glycemic control in persons with type 2 diabetes on insulin therapy - ScienceDirectsciencedirect.com
38. Clinical Use of Continuous Glucose Monitoring in Adults with Type 2 Diabetes - Anders L. Carlson, Deborah M. Mullen, Richard M. Bergenstal, 2017journals.sagepub.com
39. Comparison of 3 Versus 6-Month Use of CGMS in Non-Insulin Using T2DM Patientsclinicaltrials.gov
40. Pros and cons of continuous glucose monitoring in the intensive care unit - PMCpmc.ncbi.nlm.nih.gov
41. Benefits of Blinded Continuous Glucose Monitoring during a Randomized Clinical Trial - PMCpmc.ncbi.nlm.nih.gov
42. Unblinded CGM Should Replace Blinded CGM in the Clinical Management of Diabetesresearchgate.net
43. Non-invasive measurements of blood glucose levels by time-gating mid-infrared optoacoustic signalsnature.com
44. Integrated individually addressable microneedle arrays for robust glucose monitoring and on-demand insulin releasingnature.com
45. Clinical evaluation of a polarization-based optical noninvasive glucose sensing systemnature.com
46. A comprehensive review on the integration of microneedle technologies with biosensing platforms for advancements in fabrication, biomarker detection, and therapeutic monitoring in precision medicinelink.springer.com
47. Porous Platinum Black-Coated Minimally Invasive Microneedles for Non-Enzymatic Continuous Glucose Monitoring in Interstitial Fluidncbi.nlm.nih.gov
48. Review of Non-Invasive Glucose Sensing Techniques: Optical, Electrical and Breath Acetonencbi.nlm.nih.gov
49. Sensitive Electrochemical Non-Enzymatic Detection of Glucose Based on Wireless Data Transmissionncbi.nlm.nih.gov
50. Time in range: a new parameter to evaluate blood glucose control in patients with diabetesncbi.nlm.nih.gov
51. International Consensus on Use of Continuous Glucose Monitoring - PMCpmc.ncbi.nlm.nih.gov
52. Clinical Targets for Continuous Glucose Monitoring Data - Blog - NIDDKniddk.nih.gov
53. Core Glycemic Metrics Explained. TIR, TBR, TAR, GMI, CV, and Mean Glucose for Diabetes Careadces.org
54. Clinical Targets for CGM Data Interpretationnovomedlink.com
55. Clinical guidance and targets for time in rangetirhub.com
56. Time in Range Targets & Guidelines for Diabetesnovomedlink.com
57. The Discrepancy Between Hemoglobin A1c and Glucose Management Indicators in 26 Patients Treated With Continuous Glucose Monitoring in an Internal Medicine Residency Clinicncbi.nlm.nih.gov
58. Discordance between Glucose Management Indicator and Glycated Hemoglobin in a Pediatric Cohort with Type 1 Diabetes: A Real-World Studyncbi.nlm.nih.gov
59. Discordance Between Glycated Hemoglobin A1c and the Glucose Management Indicator in People With Diabetes and Chronic Kidney Disease - Philippe Oriot, Claire Viry, Antoine Vandelaer, Sébastien Grigioni, Malanie Roy, Jean Christophe Philips, Gaëtan Prévost, 2023journals.sagepub.com
60. HbA1c and Glucose Management Indicator Discordance Associated with Obesity and Type 2 Diabetes in Intermittent Scanning Glucose Monitoring Systemncbi.nlm.nih.gov
61. Estimating Glycemia From HbA1c and CGM: Analysis of Accuracy and Sources of Discrepancydiabetesjournals.org
62. The ambulatory glucose profile and its interpretationmja.com.au
63. (PDF) The ambulatory glucose profile and its interpretationresearchgate.net
64. Glossarypro.aace.com
65. AGP Glucose Metrics Average Glucose 175 mg/dL Goal: <154 mg/dLhealthpartners.com
66. Landscape of Continuous Glucose Monitoring (CGM) and Integrated CGM: Accuracy Considerations - Timothy S. Bailey, Shridhara Alva, 2021journals.sagepub.com
67. Measures of Accuracy for Continuous Glucose Monitoring and Blood Glucose Monitoring Devices - PMCpmc.ncbi.nlm.nih.gov
68. CGM Accuracy Explained: What MARD Means & How Accurate Sensors Are (2026)glucorxcgm.in
69. MARD & CGM Accuracy: The Numbers You Trust Aren’t Always Standarddiabetech.info
70. ISO 15197: 2013 Evaluation of a Blood Glucose Monitoring System’s Measurement Accuracy - Nina Jendrike, Annette Baumstark, Ulrike Kamecke, Cornelia Haug, Guido Freckmann, 2017journals.sagepub.com
71. (PDF) Analytical Performance Requirements for Systems for Self-Monitoring of Blood Glucose With Focus on System Accuracy: Relevant Differences Among ISO 15197:2003, ISO 15197:2013, and Current FDA Recommendations · Nutrition and Dieteticsacademia.edu
72. Glucose sensor systemimage-ppubs.uspto.gov
73. Trends in accuracy management of continuous glucose monitoring systems - PMCpmc.ncbi.nlm.nih.gov
74. Importance of FDA-Integrated Continuous Glucose Monitors to Ensure Accuracy of Continuous Glucose Monitoring - David C. Klonoff, Monica Gabbay, Sun Joon Moon, Emma G. Wilmot, 2025journals.sagepub.com
75. eCFR :: 21 CFR 862.1355 -- Integrated continuous glucose monitoring system.ecfr.gov
76. (PDF) Importance of FDA-Integrated Continuous Glucose Monitors to Ensure Accuracy of Continuous Glucose Monitoringresearchgate.net
77. March 27, 2018 Dexcom, Inc. Neeta Sharma Vice President, Regulatory Affairsaccessdata.fda.gov
78. Importance of FDA-Integrated Continuous Glucose Monitors to Ensure Accuracy of Continuous Glucose Monitoringovid.com
79. Glucose Biosensor - an overviewsciencedirect.com
80. Glucose sensing technology—current practice?link.springer.com
81. The Historical Evolution of Continuous Glucose... : International Journal of Diabetes and Technologyjournals.lww.com
82. Expanding the horizon of continuous glucose monitoring into the future of pediatric medicinenature.com
83. Past, Present, and Future of Continuous Glucose Monitorsliebertpub.com
84. Abbott's FreeStyle® Libre 14 Day Flash Glucose Monitoring System Now Approved in U.S. - Jul 27, 2018abbott.mediaroom.com
85. FDA approves first continuous glucose monitoring system for adults not requiring blood sample calibrationfda.gov
86. How continuous glucose monitors have transformed the diabetes landscape - Drug Delivery Businessdrugdeliverybusiness.com
87. Flash-glucose Monitoring in Sub-optimally Controlled Type 1 Diabetes (FLASH-UK)cdn.clinicaltrials.gov
88. Glucose biosensors in clinical practice: principles, limits and perspectives of currently used devices - PMCpmc.ncbi.nlm.nih.gov
89. An Inpatient Performance Evaluation of a New Subcutaneous Glucose Sensorclinicaltrials.gov
90. MiniMed Guardian CGM - Diabetesnet.comdiabetesnet.com
91. FDA Approves FreeStyle Libre System for Patientsmedscape.com
92. All About the Abbott FreeStyle Libre Flash Glucose Monitorhealthline.com
93. 13. Older Adults: Standards of Care in Diabetes—2026diabetesjournals.org
94. What’s New in Continuous Glucose Monitoring: ADA Standards of Care 2026journals.lww.com
95. Effect of Continuous Glucose Monitoring on Glycemic Control in Patients With Type 2 Diabetes Treated With Basal Insulin: A Randomized Clinical Trial - PubMedpubmed.ncbi.nlm.nih.gov
96. Continuous glucose monitoring in noninsulin‐treated type 2 diabetes: A critical review of reported trials with an updated systematic review and meta‐analysis of randomised controlled trials - Aronson - 2025 - Diabetes, Obesity and Metabolism - Wiley Online Librarydom-pubs.onlinelibrary.wiley.com
97. Continuous Glucose Monitoring in Non–Insulin-Treated Type 2 Diabetesnejm.org
98. Continuous glucose monitoring in pregnant women with type 1 diabetes (CONCEPTT): a multicentre international randomised controlled trialncbi.nlm.nih.gov
99. Continuous glucose monitoring in pregnant women with type 1 diabetes (CONCEPTT): a multicentre international randomised controlled trial - PubMedpubmed.ncbi.nlm.nih.gov
100. Continuous glucose monitoring in pregnant women with type 1 diabetes (CONCEPTT): a multicentre international randomised controlled trial.repository.cam.ac.uk
101. News Release Contact: press@diabetes.orgdiabetes.org
102. 7. Diabetes Technology: Standards of Care in Diabetes—2026diabetesjournals.org
103. 2026 ADA Diabetes Standards of Medical Care Clinical Guideline Summary - American Diabetes Association Guidelines - Guideline Centralguidelinecentral.com
104. Efficacy of Real-Time Continuous Glucose Monitoring in Improving Glycemic Outcomes Among Adults With Type 2 Diabetes: A Systematic Review of Randomized Controlled Trialsncbi.nlm.nih.gov
105. The American Diabetes Association Releases “Standards of Care in Diabetes—2026”diabetes.org
106. Key CGM Recommendations from ADA 2026 SOC Standards of Care in Diabetes—2026provider.dexcom.com
107. The effectiveness of continuous glucose monitoring with remote telemonitoring-enabled virtual educator visits in adults with non-insulin dependent type 2 diabetes: A randomized trial - Diabetes Research and Clinical Practicediabetesresearchclinicalpractice.com
108. New closed-loop insulin systems - PMC - NIHpmc.ncbi.nlm.nih.gov
109. Closed-loop insulin delivery: update on the state of the field and emerging technologies - PMCpmc.ncbi.nlm.nih.gov
110. Recent advances in closed-loop insulin delivery - ScienceDirectsciencedirect.com
111. Closed Loop Insulin Delivery: Technology Landscape 2026patsnap.com
112. Physiological Closed-Loop Control (PCLC) Systems: Review of a Modern Frontier in Automationarxiv.org
113. Hybrid Closed-Loop Insulin Delivery in Type 1 Diabetes During Supervised Outpatient Conditions - PMCpmc.ncbi.nlm.nih.gov
114. Safety limits for closed-loop infusion pump controlimage-ppubs.uspto.gov
115. Role of automated insulin delivery in managing insulin-treated outpatients with type 2 diabetes: A systematic review and meta-analysis - PMCpmc.ncbi.nlm.nih.gov
116. Troubleshooting CGM Alerts. A Clinician's Triage Guide from danatech.orgadces.org
117. Methods for providing an alert or an alarm to a user of a mobile communications deviceimage-ppubs.uspto.gov
118. The Need for Data Standards and Implementation Policies to Integrate CGM Data into the Electronic Health Record - PMCpmc.ncbi.nlm.nih.gov
119. A Wearable, Dual Closed‐loop Insulin Delivery System for Precision Diabetes Management - He - 2026 - Advanced Materials - Wiley Online Libraryadvanced.onlinelibrary.wiley.com
120. Efficacy and Safety of Automated Insulin Delivery Systems in Patients with Type 1 Diabetes Mellitus: A Systematic Review and Meta-Analysise-dmj.org
121. Accuracy of continuous glucose monitoring in critically ill patientsncbi.nlm.nih.gov
122. Accuracy of the intermittently scanned continuous glucose monitoring system in critically ill patients: a prospective, multicenter, observational studylink.springer.com
123. Real-world Accuracy of CGM in Inpatient Critical and Noncritical Care Settings at a Safety-Net Hospitaldiabetesjournals.org
124. Clinical validation of continuous glucose monitoring in glycemic management of critically ill sepsis patients: A prospective randomized clinical study - PubMedpubmed.ncbi.nlm.nih.gov
125. Stelo & OTC CGMs: Glucose Data Without Diabetes, Explainedacibademinternational.com
126. First over-the-counter continuous glucose monitor FDA-clearedcontemporarypediatrics.com
127. FDA Clears First OTC Continuous Glucose Monitoring, Dexcom Stelo Glucose Biosensor Systempharmacytimes.com
128. CGM Without Diabetes: Why Sensors Hide Highs and Lowsmymedicineadvisor.com
129. Continuous Glucose Monitors Without a Prescription Since 2024 · Galegale.care
130. Accuracy and Performance of Continuous Glucose Monitors in Athletes - ScienceDirectsciencedirect.com
131. Continuous glucose monitoring in para cyclists: An observational studyncbi.nlm.nih.gov
132. Application potential of continuous glucose monitoring (CGM) in elite endurance athletes without diabetes: What do physiology and current evidence tell us?link.springer.com
133. Continuous Glucose Monitor Cost (2026): Cash vs Insurancelongevitybenchmark.com
134. Continuous Glucose Monitor Costs: Insurance, Subscriptions & Devicesdietitianwithoutborders.com
135. Does Medicare Cover a Continuous Glucose Monitor? (2026)brevy.com
136. Does Insurance Cover a Continuous Glucose Monitor? Medicare, Medicaid & Private Plans Explained (2026)momentarylab.com
137. Equity of Continuous Glucose Monitoring in Children and Young People With Type 1 Diabetes: A Systematic Review - Howard Dicks - 2025 - Pediatric Diabetes - Wiley Online Libraryonlinelibrary.wiley.com
138. Use of continuous glucose monitoring (CGM) in diabetes management in Africa: opportunities, challenges, and future prospects - PMCpmc.ncbi.nlm.nih.gov
139. Type 1 diabetes technology gap between high‐income and developing countries: Continuous glucose monitoring access remains a challenge in Brazilncbi.nlm.nih.gov
140. A call to action in improving access to diabetes care in lower- and middle-income countries - The Lancet Regional Healththelancet.com
141. Data Privacy, Ownership, and Secondary Use of Clinical Data Generated by Continuous Glucose Monitors and Mobile Health Applications: A Review - PMCpmc.ncbi.nlm.nih.gov
142. Data Privacy, Ownership, and Secondary Use of Clinical Data Generated by Continuous Glucose Monitors and Mobile Health Applications: A Review - Paulo Dario, 2026journals.sagepub.com
143. Privacy and Security Issues Surrounding the Protection of Data Generated by Continuous Glucose Monitors - PMCpmc.ncbi.nlm.nih.gov
144. Open Data Sharing in Clinical Research and Participants Privacy: Challenges and Opportunities in the Era of Artificial Intelligencearxiv.org
145. Medicare and Continuous Glucose Monitors: Coverage Rules for 2026farisinsurancenetwork.com
146. CGM Insurance Coverage Guide in 2026 by CGM Monitorscgmmonitors.com
147. How to Get a Continuous Glucose Monitor (CGM) Covered by Insurance in 2026: A Step-by-Step Guidehealthcosttools.com
148. How Much Does a CGM Cost in 2026? Complete Pricing Guideglucoseintel.com
149. Exploring the feasibility and acceptability of Continuous Glucose Monitoring among people with type 1 diabetes and healthcare providers in South Africa’s public sector: A qualitative studyjournals.plos.org
150. Continuous Dual Glucose–Ketone Sensing Technology - Udo Hoss, Shridhara Alva, Heather Pryor, Zoe Welsh, Benjamin Feldman, 2025doi.org
151. Analyte sensors and sensing methods featuring dual detection of glucose and ketonesimage-ppubs.uspto.gov
152. Abbott receives CE mark for dual glucose-ketone sensormedtechdive.com
153. Non-invasive and minimally invasive glucose monitoring devices: a systematic review and meta-analysis on diagnostic accuracy of hypoglycaemia detectionlink.springer.com
154. Clinical validation of noninvasive blood glucose measurements by midinfrared spectroscopynature.com
155. Noninvasive Glucose Monitor Demonstrates Accuracy in Patients With Diabetes Using Visible Light Technologyconsultant360.com
156. Accuracy of a Non-Invasive Home Glucose Monitor for Measurement of Blood Glucose - PMCpmc.ncbi.nlm.nih.gov
157. Continuous Monitoring of Glucose and Oxygen using an Insertable Biomaterial-based Multianalyte Barcode Sensorbiorxiv.org
158. Artificial Intelligence for Glucose Prediction and Glycemic Management in Diabetes: A Systematic Review of Predictive Performance and Clinical Outcomescureus.com
159. Systematic review of Artificial Intelligence-based methods for glycemic control and risk prediction in intensive care units - PubMedpubmed.ncbi.nlm.nih.gov
160. WO2020159956A1 - Analyte sensors and sensing methods featuring dual detection of glucose and ketones - Google Patentspatents.google.com
161. Insertable Biomaterial-Based Multianalyte Barcode Sensor toward Continuous Monitoring of Glucose and Oxygenncbi.nlm.nih.gov