Limits to the Evaluation of the Accuracy of Continuous Glucose Monitoring Systems by Clinical Trials

Accuracy of continuous glucose monitoring systems in intensive care unit patients: a scoping review

PURPOSE

Glycemic control poses a challenge in intensive care unit (ICU) patients and dysglycemia is associated with poor outcomes. Continuous glucose monitoring (CGM) has been successfully implemented in the type 1 diabetes out-patient setting and renewed interest has been directed into the transition of CGM into the ICU. This scoping review aimed to provide an overview of CGM accuracy in ICU patients to inform future research and CGM implementation.

METHODS

We systematically searched PubMed and EMBASE between 5th of December 2023 and 21st of May 2024 and reported findings in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline for scoping reviews (PRISMA-ScR). We assessed studies reporting the accuracy of CGM in the ICU and report study characteristics and accuracy outcomes.

RESULTS

We identified 2133 studies, of which 96 were included. Most studies were observational (91.7%), conducted in adult patients (74%), in mixed ICUs (47.9%), from 2014 and onward, and assessed subcutaneous CGM systems (80%) using arterial blood samples as reference test (40.6%). Half of the studies (56.3%) mention the use of a prespecified reference test protocol. The mean absolute relative difference (MARD) ranged from 6.6 to 30.5% for all subcutaneous CGM studies. For newer factory calibrated CGM, MARD ranged from 9.7 to 20.6%. MARD for intravenous CGM was 5-14.2% and 6.4-13% for intraarterial CGM.

CONCLUSIONS

In this scoping review of CGM accuracy in the ICU, we found great diversity in accuracy reporting. Accuracy varied depending on CGM and comparator, and may be better for intravascular CGM and potentially lower during hypoglycemia.

Analytical Performance of the FreeStyle Libre 2 Glucose Sensor in Healthy Male Adults

Continuous Glucose Monitoring (CGM) not only can be used for glycemic control in chronic diseases (e.g., diabetes), but is increasingly being utilized by individuals and athletes to monitor fluctuations in training and everyday life. However, it is not clear how accurately CGM reflects plasma glucose concentration in a healthy population in the absence of chronic diseases. In an oral glucose tolerance test (OGTT) with forty-four healthy male subjects (25.5 ± 4.5 years), the interstitial fluid glucose (ISFG) concentration obtained by a CGM sensor was compared against finger-prick capillary plasma glucose (CPG) concentration at fasting baseline (T0) and 30 (T30), 60 (T60), 90 (T90), and 120 (T120) min post OGTT to investigate differences in measurement accuracy. The overall mean absolute relative difference (MARD) was 12.9% (95%-CI: 11.8-14.0%). Approximately 100% of the ISFG values were within zones A and B in the Consensus Error Grid, indicating clinical accuracy. A paired t-test revealed statistically significant differences between CPG and ISFG at all time points (T0: 97.3 mg/dL vs. 89.7 mg/dL, T30: 159.9 mg/dL vs. 144.3 mg/dL, T60: 134.8 mg/dL vs. 126.2 mg/dL, T90: 113.7 mg/dL vs. 99.3 mg/dL, and T120: 91.8 mg/dL vs. 82.6 mg/dL; p < 0.001) with medium to large effect sizes (d = 0.57-1.02) and with ISFG systematically under-reporting the reference system CPG. CGM sensors provide a convenient and reliable method for monitoring blood glucose in the everyday lives of healthy adults. Nonetheless, their use in clinical settings wherein implications are drawn from CGM readings should be handled carefully.

Explainable Machine Learning for Real-Time Hypoglycemia and Hyperglycemia Prediction and Personalized Control Recommendations

BACKGROUND

The occurrences of acute complications arising from hypoglycemia and hyperglycemia peak as young adults with type 1 diabetes (T1D) take control of their own care. Continuous glucose monitoring (CGM) devices provide real-time glucose readings enabling users to manage their control proactively. Machine learning algorithms can use CGM data to make ahead-of-time risk predictions and provide insight into an individual's longer term control.

METHODS

We introduce explainable machine learning to make predictions of hypoglycemia (<70 mg/dL) and hyperglycemia (>270 mg/dL) up to 60 minutes ahead of time. We train our models using CGM data from 153 people living with T1D in the CITY (CGM Intervention in Teens and Young Adults With Type 1 Diabetes)survey totaling more than 28 000 days of usage, which we summarize into (short-term, medium-term, and long-term) glucose control features along with demographic information. We use machine learning explanations (SHAP [SHapley Additive exPlanations]) to identify which features have been most important in predicting risk per user.

RESULTS

Machine learning models (XGBoost) show excellent performance at predicting hypoglycemia (area under the receiver operating curve [AUROC]: 0.998, average precision: 0.953) and hyperglycemia (AUROC: 0.989, average precision: 0.931) in comparison with a baseline heuristic and logistic regression model.

CONCLUSIONS

Maximizing model performance for glucose risk prediction and management is crucial to reduce the burden of alarm fatigue on CGM users. Machine learning enables more precise and timely predictions in comparison with baseline models. SHAP helps identify what about a CGM user's glucose control has led to predictions of risk which can be used to reduce their long-term risk of complications.

Continuous Glucose Monitoring in Children and Adolescents with Congenital Adrenal Hyperplasia

Objective

Patients with congenital adrenal hyperplasia (CAH) require lifelong therapy with glucocorticoids to suppress androgen excess and substitute for deficient cortisol. An important aspect of care is the prevention of metabolic sequelae. In infants, potentially lethal nocturnal hypoglycaemia has been described. In adolescence, visceral obesity, hypertension, hyperinsulinism and insulin resistance are reported. To date, systematic studies of glucose profiles in this age group with CAH are lacking.

Methods

This was a monocentric, prospective, observational study to determine the glucose profiles under different treatment regimens in a cohort of young patients with CAH. The continuous glucose monitoring device used was the latest generation FreeStyle Libre 3® sensor in blinded mode. Therapeutic and auxological data were obtained.

Results

The cohort consisted of 10 children/adolescents with a mean age of 11 years. Three patients exhibited morning fasting hyperglycaemia. Overall, 6 out of 10 patients had unacceptably few total values in the desired range of 70-120 mg/dL. Tissue glucose values above 140-180 mg/dL were found in 5 of 10 patients. The mean value for glycosylated haemoglobin for the cohort was of 5.8%. All pubertal adolescents with reverse circadian regimens had significantly higher glucose levels at night. Two adolescents showed asymptomatic nocturnal hypoglycaemia.

Conclusion

Most of the patients exhibited abnormalities in glucose metabolism. Two-thirds had elevated total 24h glucose values outside the age-appropriate reference values. Thus, this aspect may need to be addressed early in life by adjusting the doses, treatment regimen or dietary measures. Consequently, reverse circadian therapy regimens should be critically indicated and closely monitored due to the potential metabolic risk.

Accuracy and Potential Interferences of Continuous Glucose Monitoring Sensors in the Hospital

For years, the standard of care for monitoring dysglycemia in hospitalized patients was capillary blood glucose (CBG) testing with point-of-care glucose meters. Recently, there has been a revolution in novel factory-calibrated continuous glucose monitoring (CGM) systems. Newer CGMs are smaller and less expensive, have improved accuracy and longer wear time, and do not require fingerstick CBG for calibration, resulting in increased utilization in ambulatory settings. Consequently, hospitals have noticed increased usability of CGMs among hospitalized patients and expect a progressive continued increase. During the COVID-19 pandemic, there was a critical need for innovative approaches to glycemic monitoring, with several pilot implementation projects using CGM in the intensive care unit and non-intensive care unit settings, further boosting the evidence in this area. Hence, recent guidelines have provided recommendations for the use of CGM in specific hospital scenarios and highlighted the potential of CGM to overcome CBG limitations for glucose monitoring in the inpatient setting. In this review, we provide the following: 1) an up-to-date review of the accuracy of the newer CGMs in hospitalized patients, 2) a discussion of standards for CGM accuracy metrics, 3) a contemporary overview of potential interferences that may cause inaccuracies or poor CGM performance, and 4) required steps for full regulatory approval of CGMs in the hospital and future research steps to advance the field forward.

Effects of Continuous Carbohydrate Intake with Gummies during the Golf Round on Interstitial Glucose, Golf Performance, and Cognitive Performance of Competitive Golfers: A Randomized Repeated-Measures Crossover Design

This study examined the effects of continuous carbohydrate intake during a golf round on interstitial glucose, golf performance, and the cognitive performance of competitive golfers. Eleven competitive golfers participated and played 18 holes of golf in this study. Participants were randomly assigned to the group indicated to consume the test food (CHO intake) or the group required to not consume it (NOT intake). Here, gummies were used as the test food, and the amount of carbohydrates was 30 g per h. Blood glucose levels were evaluated using interstitial glucose. Golf performance was measured in five tests, including scores, 2.5 m putting test, club head speed, driving distance, and accuracy. Cognitive performance was measured in three tests, including self-perceived levels of fatigue (PLF), self-perceived levels of concentration (PLC), and self-perceived levels of relaxation (PLR). Interstitial glucose (p < 0.001) and PLF (p < 0.001) were significantly reduced in the CHO intake compared with that in the NOT intake from the sixth hole. PLC was significantly higher in the CHO intake than in the NOT intake on all 18 holes (p = 0.032). These findings suggest that continuous carbohydrate intake may be effective in reducing fatigue and maintaining the performance of competitive golfers.

Impact of Two Different Reference Measurement Procedures on Apparent System Accuracy of 18 CE-Marked Current-Generation Blood Glucose Monitoring Systems

BACKGROUND

Measurement accuracy has been assessed for many different blood glucose monitoring systems (BGMS) over the years by different study groups. However, the choice of the comparison measurement procedure may impact the apparent level of accuracy found in such studies.

MATERIALS AND METHODS

Measurement accuracy of 18 different BGMS was assessed in a setting based on ISO 15197 using two different comparison methods in parallel: a glucose oxidase (GOD)-based and a hexokinase (HK)-based method. Accuracy limits of ISO 15197 were applied, and additional analyses were performed, including bias, linear regression, and mean absolute relative difference (MARD) to assess the impact of possible differences between comparison methods on the apparent level of accuracy.

RESULTS

While ≈80% of BGMS met the accuracy criteria of ISO 15197 when compared with the respective manufacturers' reference measurement procedure, only two-thirds did so against both comparison methods. The mean relative bias ranged from -6.6% to +5.7% for the analysis against the GOD-based method and from -11.1% to +1.3% for the analysis against the HK-based method, whereas MARD results ranged from 3.7% to 9.8% and from 2.3% to 10.5%, respectively. Results of regression analysis showed slopes between 0.85 and 1.08 (GOD-based method) and between 0.81 and 1.01 (HK-based method).

CONCLUSIONS

The results of this study indicate that there are systematic differences between the reference measurement procedures used for BGMS calibration as well as for system accuracy assessment. Because of the potential impact on therapy of patients with diabetes resulting from these differences, further steps toward harmonization of the measurement procedures' results are important.

Variation of Mean Absolute Relative Differences of Continuous Glucose Monitoring Systems Throughout the Day

BACKGROUND

There is an increasing use of continuous glucose monitoring (CGM) by people with diabetes. Measurement performance is often characterized by the mean absolute relative difference (MARD). However, MARD is influenced by a number of factors and little is known about whether MARD is stable throughout the day.

MATERIAL AND METHODS

A total of 24 participants with type 1 diabetes were enrolled in the study. The study was performed for seven in-patient days. Participants wore two CGM systems in parallel and performed additional frequent blood glucose (BG) measurements. On two days, glucose excursions were induced.MARD was calculated between pairs of CGM and BG values, with BG values serving as reference values. ARD values calculated from CGM-BG pairs were grouped by hour of the day. Results were analyzed separately for glucose excursion days and for regular days.

RESULTS

Total MARDs for the complete study duration were 12.5% ± 3.6% and 13.2% ± 2.4% (n = 24). Throughout the day marked variability of MARD was observed (8.0% ± 1.3%-16.3% ± 2.9% (G5); 9.1% ± 1.4%-16.3% ± 5.3% (FL), up to n = 157 each). Low(est) MARD values were observed before breakfast and dinner, when subjects were in or near a fasting state. Especially after breakfast and lunch, MARD values were higher than average.

CONCLUSIONS

Analytical performance of the two CGM systems, assessed by MARD, was found to vary markedly throughout the day. Activities of daily life likely triggered these variations. An increasing number of CGM users base therapeutic decisions on CGM values, and they should be aware of these variations of performance throughout the day.

Effect of Ethanol Consumption on the Accuracy of a Glucose Oxidase-Based Subcutaneous Glucose Sensor in Subjects with Type 1 Diabetes

Continuous glucose monitors (CGM) have improved the management of patients with type 1 diabetes (T1D), with glucose oxidase (GOx)-based sensors being the most used. However, they are potentially subject to both electrochemical and enzymatic interferences, including those related to changes of pH. The objective of this study is to investigate the effect of ethanol, given as beer along with a mixed meal, on the accuracy of a commercial GOx-CGM. Data from 12 T1D participants in a randomized crossover trial to evaluate the effect of meal composition and alcohol consumption on postprandial glucose concentration were used. Absolute error (AE) and mean absolute relative difference (MARD) were calculated. The differences between the alcohol and nonalcohol scenarios were assessed using the Mann−Whitney U and Wilcoxon signed-rank tests. The AE in the alcohol study was low, but significantly greater as compared to the study without alcohol (p-value = 0.0418). The MARD was numerically but not significantly greater. However, both variables were greater at pH < 7.36 and significantly affected by time only in the alcohol arm. In T1D, alcohol consumption affects the accuracy of a GOx-CGM. This effect could be at least partially related to the ethanol-induced changes in pH.

Validation of the Application of Solid Contact Ion-Selective Electrode for Off-Body Sweat Ion Monitoring

The solid contact ion-selective electrode (ISE) is a promising skin-interfaced monitoring system for sweat ions. Despite a growing number of on-body usages of ISE with fancy new materials and device fabrications, there are very few reports attempting to validate ISE results with a gold standard technique. For this purpose, this work uses inductively coupled plasma-optical emission spectrometry (ICP-OES) as a reference technique to conduct a direct evaluation of the sweat sodium and potassium ion levels obtained by ISE in an off-body approach. Eight healthy male subjects were recruited to collect exercise-induced sweat. It was found that sweat sodium and potassium ions present a rather wide concentration range. The sweat sodium concentration did not vary greatly in an exercise period of half an hour, while the sweat potassium concentration typically decreased with exercise. Mineral drink intake had no clear impact on the sweat sodium level, but increased the sweat potassium level. A paired t-test and mean absolute relative difference (MARD) analysis, a method typically used for evaluating the performance of glucometers, was employed to compare the results of ISE and ICP-OES. The statistical analysis validated the feasibility of ISE for measuring sweat ions, although better accuracy is required. Our data suggests that overweight subjects are likely to possess a higher sweat sodium level.

Continuous Glucose Monitoring in Healthy Adults-Possible Applications in Health Care, Wellness, and Sports

INTRODUCTION

Continuous glucose monitoring (CGM) systems were primarily developed for patients with diabetes mellitus. However, these systems are increasingly being used by individuals who do not have diabetes mellitus. This mini review describes possible applications of CGM systems in healthy adults in health care, wellness, and sports.

RESULTS

CGM systems can be used for early detection of abnormal glucose regulation. Learning from CGM data how the intake of foods with different glycemic loads and physical activity affect glucose responses can be helpful in improving nutritional and/or physical activity behavior. Furthermore, states of stress that affect glucose dynamics could be made visible. Physical performance and/or regeneration can be improved as CGM systems can provide information on glucose values and dynamics that may help optimize nutritional strategies pre-, during, and post-exercise.

CONCLUSIONS

CGM has a high potential for health benefits and self-optimization. More scientific studies are needed to improve the interpretation of CGM data. The interaction with other wearables and combined data collection and analysis in one single device would contribute to developing more precise recommendations for users.

American Medical Society for Sports Medicine Position Statement on the Care of the Athlete and Athletic Person With Diabetes

ABSTRACT

The American Medical Society for Sports Medicine (AMSSM) developed this position statement to assist physicians and other health professionals in managing athletes and active people with diabetes. The AMSSM selected the author panel through an application process to identify members with clinical and academic expertise in the care of active patients with diabetes. This article reviews the current knowledge and gaps regarding the benefits and risks of various types of exercise and management issues for athletes and physically active people with diabetes, including nutrition and rehabilitation issues. Resistance exercises seem to be beneficial for patients with type 1 diabetes, and the new medications for patients with type 2 diabetes generally do not need adjustment with exercise. In preparing this statement, the authors conducted an evidence review and received open comment from the AMSSM Board of Directors before finalizing the recommendations.

Noninvasive Glucose Monitor Using Dielectric Spectroscopy

OBJECTIVE

The Alertgy noninvasive continuous glucose monitor (ANICGM) is a novel wristband device that reports glucose levels without entailing skin puncture. This study evaluated the performance of the ANICGM compared to a Food and Drug Administration-approved glucose meter in patients with type 2 diabetes.

METHODS

The ANICGM device measures changes in the electromagnetic field generated by its sensor to produce a dielectric spectrum. The data contained within this spectrum are used in tandem with machine learning algorithms to estimate blood glucose levels. Values from the ANICGM were collected, sent to the Alertgy lab, formatted, and compared with fingerstick blood glucose levels, which were measured using the Accuchek Inform II glucometer. Fifteen patients completed three 120-minute sessions. The mean absolute relative difference (MARD) was calculated.

RESULTS

MARD values were compared between study days 2 and 3. The MARD for day 2 was 18.5% (95% CI, 12.8-42.2%), and the MARD for day 3 was 15.3% (95% CI, 12.3-18.4%). The difference in the MARD between days 2 and 3 was not statistically significant (P = .210).

CONCLUSION

The resulting MARDs suggest that further investigation into the use of dielectric spectroscopy for glucose monitoring should be explored.

Benefits of a Switch from Intermittently Scanned Continuous Glucose Monitoring (isCGM) to Real-Time (rt) CGM in Diabetes Type 1 Suboptimal Controlled Patients in Real-Life: A One-Year Prospective Study §

The switch from intermittently scanned continuous glucose monitoring (isCGM) to real-time (rt) CGM could improve glycemic management in suboptimal controlled type 1 diabetes patients, but long-term study is lacking. We evaluated retrospectively the ambulatory glucose profile (AGP) in such patients after switching from Free Style Libre 1 (FSL1) to Dexcom G4 (DG4) biosensors over 1 year. Patients (n = 21, 43 ± 15 years, BMI 25 ± 5, HbA1c 8.1 ± 1.0%) had severe hypoglycemia and/or HbA1c ≥ 8%. AGP metrics (time-in-range (TIR) 70–180 mg/dL, time-below-range (TBR) <70 mg/dL or <54 mg/dL, glucose coefficient of variation (%CV), time-above-range (TAR) >180 mg/dL or >250 mg/dL, glucose management indicator (GMI), average glucose) were collected the last 3 months of FSL1 use (M0) and of DG4 for 3, 6 (M6) and 12 (M12) months of use. Values were means ± standard deviation or medians [Q1;Q3]. At M12 versus M0, the higher TIR (50 ± 17 vs. 45 ± 16, p = 0.036), and lower TBR < 70 mg/dL (2.5 [1.6;5.5] vs. 7.0 [4.5;12.5], p = 0.0007), TBR < 54 mg/dL (0.7 [0.4;0.8] vs. 2.3 [0.8;7.0], p = 0.007) and %CV (39 ± 5 vs. 45 ± 8, p = 0.0009), evidenced a long-term effectiveness of the switch. Compared to M6, TBR < 70 mg/dL decreased, %CV remained stable, while the improvement on hyperglycemia exposure decreased (higher GMI, TAR and average glucose). This switch was a relevant therapeutic option, though a loss of benefit on hyperglycemia stressed the need for optimized management of threshold alarms. Nevertheless, few patients attained the recommended values for AGP metrics, and the reasons why some patients are “responders” vs. “non-responders” warrant to be investigated.

Tracking metabolic responses based on macronutrient consumption: A comprehensive study to continuously monitor and quantify dual markers (cortisol and glucose) in human sweat using WATCH sensor

Wearable Awareness Through Continuous Hidrosis (WATCH) sensor is a sweat based monitoring platform that tracks cortisol and glucose for the purpose of understanding metabolic responses related to macronutrient consumption. In this research article, we have demonstrated the ability of tracking these two biomarkers in passive human sweat over a workday period (8 h) for 10 human subjects in conjunction with their macronutrient consumption. The validation of the WATCH sensor performance was carried out via standard reference methods such as Luminex and ELISA This is a first demonstration of a passive sweat sensing technology that can detect interrelated dual metabolites, cortisol, and glucose, on a single sensing platform. The significance of detecting the two biomarkers simultaneously is that capturing the body's metabolic and endocrinal responses to dietary triggers can lead to improved lifestyle management. For sweat cortisol, we achieved a detection limit of 1 ng/ml (range ∼1-12.5 ng/ml) with Pearson's "r" of 0.897 in reference studies and 0.868 in WATCH studies. Similarly, for sweat glucose, we achieved a detection limit of 1 mg/dl (range ∼ 1-11 mg/dl) with Pearson's "r" of 0.968 in reference studies and 0.947 in WATCH studies, respectively. The statistical robustness of the WATCH sensor was established through the Bland-Altman analysis, whereby the sweat cortisol and sweat glucose levels are comparable to the standard reference method. The probability distribution (t-test), power analysis (power 0.82-0.87), α = 0.05. Mean absolute relative difference (MARD) outcome of ˷5.10-5.15% further confirmed the statistical robustness of the sweat sensing WATCH device output.

Review of Wearable Devices and Data Collection Considerations for Connected Health

Wearable sensor technology has gradually extended its usability into a wide range of well-known applications. Wearable sensors can typically assess and quantify the wearer’s physiology and are commonly employed for human activity detection and quantified self-assessment. Wearable sensors are increasingly utilised to monitor patient health, rapidly assist with disease diagnosis, and help predict and often improve patient outcomes. Clinicians use various self-report questionnaires and well-known tests to report patient symptoms and assess their functional ability. These assessments are time consuming and costly and depend on subjective patient recall. Moreover, measurements may not accurately demonstrate the patient’s functional ability whilst at home. Wearable sensors can be used to detect and quantify specific movements in different applications. The volume of data collected by wearable sensors during long-term assessment of ambulatory movement can become immense in tuple size. This paper discusses current techniques used to track and record various human body movements, as well as techniques used to measure activity and sleep from long-term data collected by wearable technology devices.

Touch-Based Fingertip Blood-Free Reliable Glucose Monitoring: Personalized Data Processing for Predicting Blood Glucose Concentrations

Diabetes prevalence has been rising exponentially, increasing the need for reliable noninvasive approaches for glucose monitoring. Different biofluids have been explored recently for replacing current blood finger-stick glucose strips with noninvasive painless sensing devices. While sweat has received considerable attention, there are mixed reports on correlating the sweat results with blood glucose levels. Here, we demonstrate a new rapid and reliable approach that combines a simple touch-based fingertip sweat electrochemical sensor with a new algorithm that addresses for personal variations toward the accurate estimate of blood glucose concentrations. The new painless and simple glucose self-testing protocol leverages the fast sweat rate on the fingertip for rapid assays of natural perspiration, without any sweat stimulation, along with the personalized sweat-response-to-blood concentration translation. A reliable estimate of the blood glucose sensing concentrations can thus be realized through a simple one-time personal precalibration. Such system training leads to a substantially improved accuracy with a Pearson correlation coefficient higher than 0.95, along with an overall mean absolute relative difference of 7.79%, with 100% paired points residing in the A + B region of the Clarke error grid. The speed and simplicity of the touch-based blood-free fingertip sweat assay, and the elimination of periodic blood calibrations, should lead to frequent self-testing of glucose and enhanced patient compliance toward the improved management of diabetes.

Impact of factory-calibrated Freestyle Libre System with new glucose algorithm measurement accuracy and clinical performance in children with type 1 diabetes during summer camp

BACKGROUND

Factory-calibrated intermittently-scanned Continuous Glucose Monitoring (isCGM) device FreeStyle Libre (FSL) has recently received improvements in its glucose tracking algorithm and calibration procedures, which are claimed to have improved its accuracy.

OBJECTIVE

To compare the accuracy of two generations of 14-days FSL devices (A in 2016, B in 2019) to self-monitored blood glucose measurements (SMBG) in children with type 1 diabetes in real-life conditions during a summer camp.

MATERIALS AND METHODS

Two largely independent groups of youth with type 1 diabetes took part in summer camps. In 2016 they used FSL-A, in 2019 FSL-B. On scheduled days, participants performed supervised 8-point glucose profiles with FSL and SMBG. The accuracy vs SMBG was assessed with mean absolute relative difference (MARD) and clinical surveillance error grid (SEG).

RESULTS

We collected 1655 FSL-SMBG measurement pairs from 78 FSL-A patients (age 13 ± 2.3 years old; HbA1c: 7.6 ± 0.8%) and 1796 from 58 in FSL-B group (age 13.8 ± 2.3 years old, HbA1c: 7.5 ± 1.1%)-in total 3451 measurements. FSL-B displayed lower MARD than FSL-A (11.3 ± 3.1% vs 13.7 ± 4.6%, P = .0003), lower SD of errors (20.2 ± 6.7 mg/dL vs 24.1 ± 9.6 mg/dL, P = .0090) but similar bias (-7.6 ± 11.8 mg/dL vs -6.5 ± 8 mg/dL, P = .5240). Both FSL-A and FSL-B showed significantly higher MARD when glycaemia was decreasing >2 mg/dL/min (FSL-A:22.3 ± 18.5%; FSL-B:17.9 ± 15.8%, P < .0001) compared with stable conditions (FSL-A: 11.4 ± 10.4%, FSL-B:10.1 ± 9.1%) and when the system could not define the glycaemic trend (FSL-A:16.5 ± 16.3%; FSL-B:15.2 ± 14.9%, P < .0001). Both generations demonstrated high percentage of A-class and B-class results in SEG (FSL-A: 96.4%, FSL-B: 97.6%) with a significant shift from B (decrease by 3.7%) to A category (increase by 3.9%) between generations (FSL-A: 16/80.4%; FSL-B:12.3/85.3%, P = .0012).

CONCLUSION

FSL-B demonstrated higher accuracy when compared to FSL-A However, when glycemia is decreasing or its trend is uncertain, the verification with a glucose meter is still advisable.

Time in Range from Continuous Glucose Monitoring: A Novel Metric for Glycemic Control

Glycosylated hemoglobin (HbA1c) has been the sole surrogate marker for assessing diabetic complications. However, consistently reported limitations of HbA1c are that it lacks detailed information on short-term glycemic control and can be easily interfered with by various clinical conditions such as anemia, pregnancy, or liver disease. Thus, HbA1c alone may not represent the real glycemic status of a patient. The advancement of continuous glucose monitoring (CGM) has enabled both patients and healthcare providers to monitor glucose trends for a whole single day, which is not possible with HbA1c. This has allowed for the development of core metrics such as time spent in time in range (TIR), hyperglycemia, or hypoglycemia, and glycemic variability. Among the 10 core metrics, TIR is reported to represent overall glycemic control better than HbA1c alone. Moreover, various evidence supports TIR as a predictive marker of diabetes complications as well as HbA1c, as the inverse relationship between HbA1c and TIR reveals. However, there are more complex relationships between HbA1c, TIR, and other CGM metrics. This article provides information about 10 core metrics with particular focus on TIR and the relationships between the CGM metrics for comprehensive understanding of glycemic status using CGM.

Remote Digital Monitoring for Medical Product Development

The use of digital health products has gained considerable interest as a new way to improve therapeutic research and development. Although these products are being adopted by various industries and stakeholders, their incorporation in clinical trials has been slow due to a disconnect between the promises of digital products and potential risks in using these new technologies in the absence of regulatory support. The Foundation for the National Institutes of Health (FNIH) Biomarkers Consortium hosted a public workshop to address challenges and opportunities in this field. Important characteristics of tool development were addressed in a series of presentations, case studies, and open panel sessions. The workshop participants endorsed the usefulness of an evidentiary criteria framework, highlighted the importance of early patient engagement, and emphasized the potential impact of digital monitoring tools and precompetitive collaborations. Concerns were expressed about the lack of real‐life validation examples and the limitations of legacy standards used as a benchmark for novel tool development and validation. Participants recognized the need for novel analytical and statistical approaches to accommodate analyses of these novel data types. Future directions are to harmonize definitions to build common methodologies and foster multidisciplinary collaborations; to develop approaches toward integrating digital monitoring data with the totality of the data in clinical trials, and to continue an open dialog in the community. There was a consensus that all these efforts combined may create a paradigm shift of how clinical trials are planned, conducted, and results brought to regulatory reviews.

Assessment of factors that determine the mean absolute relative difference in flash glucose monitoring with reference to plasma glucose levels in Japanese subjects without diabetes

The Abbott FreeStyle Libre flash glucose monitoring system (FGM) is a recently introduced, but widespread continuous glucose monitoring system. While its mean absolute relative difference (MARD) value indicating its accuracy is acceptable with reference to the self-monitoring of blood glucose (SMBG) levels, few reports have examined the MARD in sensor glucose values of FGM (FGM-SG) with reference to plasma glucose (PG) levels and the factors determining it. We performed oral glucose tolerance tests (OGTTs) in 25 Japanese subjects without diabetes. Parkes error grid analyses showed that FGM-SG with either SMBG or PG levels as a reference met International Organization for Standardization criteria. The MARD in FGM-SG with reference to SMBG levels was 10.9 ± 4.1% during OGTTs. Surprisingly, the MARD in FGM-SG with reference to PG levels was 20.3 ± 10.3% during OGTTs, revealing a discrepancy in the accuracy of FGM-SG compared with that of PG levels; moreover, the MARD showed negative correlations with fasting blood sugar level, homeostasis model assessment insulin resistance index, and body mass index (BMI). Multiple regression analyses revealed that BMI contributed the most to the MARD when FGM-SG and PG level were compared, as lean individuals have a greater MARD regardless of glucose levels. Inaccurate FGM data could potentially increase the risk of inappropriate treatment; consideration of such factors is critical to ensure reliable FGM values.

Accuracy and stability of an arterial sensor for glucose monitoring in a porcine model using glucose clamp technique

Intravascular glucose sensors have the potential to improve and facilitate glycemic control in critically ill patients and might overcome measurement delay and accuracy issues. This study investigated the accuracy and stability of a biosensor for arterial glucose monitoring tested in a hypo- and hyperglycemic clamp experiment in pigs. 12 sensors were tested over 5 consecutive days in 6 different pigs. Samples of sensor and reference measurement pairs were obtained every 15 minutes. 1337 pairs of glucose values (range 37–458 mg/dl) were available for analysis. The systems met ISO 15197:2013 criteria in 99.2% in total, 100% for glucose <100 mg/dl (n = 414) and 98.8% for glucose ≥100 mg/dl (n = 923). The mean absolute relative difference (MARD) during the entire glycemic range of all sensors was 4.3%. The MARDs within the hypoglycemic (<70 mg/dl), euglycemic (≥70–180 mg/dl) and hyperglycemic glucose ranges (≥180 mg/dl) were 6.1%, 3.6% and 4.7%, respectively. Sensors indicated comparable performance on all days investigated (day 1, 3 and 5). None of the systems showed premature failures. In a porcine model, the performance of the biosensor revealed a promising performance. The transfer of these results into a human setting is the logical next step.

Benefits and Limitations of MARD as a Performance Parameter for Continuous Glucose Monitoring in the Interstitial Space

High-quality performance of medical devices for glucose monitoring is important for a safe and efficient usage of this diagnostic option by patients with diabetes. The mean absolute relative difference (MARD) parameter is used most often to characterize the measurement performance of systems for continuous glucose monitoring (CGM). Calculation of this parameter is relatively easy and comparison of the MARD numbers between different CGM systems appears to be straightforward on the first glance. However, a closer look reveals that a number of complex aspects make interpretation of the MARD numbers provided by the manufacturer for their CGM systems difficult. In this review, these aspects are discussed and considerations are made for a systematic and appropriate evaluation of the MARD in clinical trials. The MARD should not be used as the sole parameter to characterize CGM systems, especially when it comes to nonadjunctive usage of such systems.

System accuracy evaluation of 18 CE-marked current-generation blood glucose monitoring systems based on EN ISO 15197:2015

OBJECTIVE

Accuracy of 18 current-generation blood glucose monitoring systems (BGMS) available in Europe was evaluated applying criteria adapted from EN ISO 15197:2015 with one reagent system lot. BGMS were selected based on market research data.

RESEARCH DESIGN AND METHODS

The BGMS ABRA, Accu-Chek Guide, AURUM, CareSens Dual, CERA-CHEK 1CODE, ContourNext One, eBsensor, FreeStyle Freedom Lite, GL50 evo, GlucoCheck GOLD, GlucoMen areo 2K, GluNEO, MyStar DoseCoach, OneTouch Verio Flex, Pic GlucoTest, Rightest GM700S, TRUEyou, and WaveSense JAZZ Wireless were tested using capillary blood from 100 different subjects and assessing the percentage of results within ±15 mg/dL (0.83 mmol/L) or 15% of comparison method results for BG concentrations below or above 100 mg/dL (5.55 mmol/L), respectively. In addition, the minimal deviation from comparison method results within which ≥95% of results of the respective BGMS were found was calculated.

RESULTS

In total, 14 BGMS had ≥95% of results within ±15 mg/dL (0.83 mmol/L) or ±15% and 3 BGMS had ≥95% of results within ±10 mg/dL (0.55 mmol/L) or ±10% of the results obtained with the comparison method. The smallest deviation from comparison method results within which ≥95% of results were found was ±7.7 mg/dL (0.43 mmol/L) or ±7.7%; the highest deviation was ±19.7 mg/dL (1.09 mmol/L) or ±19.7%.

CONCLUSIONS

This accuracy evaluation shows that not all CE-labeled BGMS fulfill accuracy requirements of ISO 15197 reliably and that there is considerable variation even among BGMS fulfilling these criteria. This safety-related information should be taken into account by patients and healthcare professionals when making therapy decisions.

TRIAL REGISTRATION NUMBER

NCT03737188.

Long-Term Accurate Continuous Glucose Biosensors via Extended Nitric Oxide Release

Analytical performance and tissue interactions of nitric oxide (NO)-releasing continuous glucose sensors were evaluated over a 28 d study in a diabetic swine model. Interstitial glucose was detected using an implanted needle-type amperometric glucose sensor. Two NO-release durations from the sensor surface were achieved by doping the membranes with nonporous (14 d release) or porous (30 d release) S-nitrosothiol-functionalized silica nanoparticles. Numerical and clinical accuracy of the sensors were assessed at time points (1, 7, 14, 21, and 28 d) following implantation. Nitric oxide-releasing sensors demonstrated accurate glucose detection over a time period directly correlated with the active release of NO. Silica particle-doped sensors that released NO for 30 d showed standard-compliant accuracy (i.e., mean absolute relative difference ≤ 15%) for >3 weeks post-implantation. Histological staining for inflammatory biomarkers suggested that the observed performance improvement was the result of decreased inflammatory cell count and a lower density collagen capsule.

Continuous noninvasive glucose monitoring; water as a relevant marker of glucose uptake in vivo

With diabetes set to become the number 3 killer in the Western hemisphere and proportionally growing in other parts of the world, the subject of noninvasive monitoring of glucose dynamics in blood remains a "hot" topic, with the involvement of many groups worldwide. There is a plethora of techniques involved in this academic push, but the so-called multisensor system with an impedance-based core seems to feature increasingly strongly. However, the symmetrical structure of the glucose molecule and its shielding by the smaller dipoles of water would suggest that this option should be less enticing. Yet there is enough phenomenological evidence to suggest that impedance-based methods are truly sensitive to the biophysical effects of glucose variations in the blood. We have been trying to answer this very fundamental conundrum: "Why is impedance or dielectric spectroscopy sensitive to glucose concentration changes in the blood and why can this be done over a very broad frequency band, including microwaves?" The vistas for medical diagnostics are very enticing. There have been a significant number of papers published that look seriously at this problem. In this review, we want to summarize this body of research and the underlying mechanisms and propose a perspective toward utilizing the phenomena. It is our impression that the current world view on the dielectric response of glucose in solution, as outlined below, will support the further evolution and implementation toward practical noninvasive glucose monitoring solutions.

Novel Single-Site Device for Conjoined Glucose Sensing and Insulin Infusion: Performance Evaluation in Diabetes Patients During Home-Use

OBJECTIVE

This study evaluated a novel diabetes treatment device that combines commercially available continuous glucose monitoring and insulin infusion technology in such a way as to perform insulin delivery and glucose sensing through a single skin insertion site (single-port device).

METHODS

Ten type 1 diabetes patients used the device for up to six days in their home/work environment for open-loop insulin delivery and glucose sensing. On an additional day, the device was used in combination with an algorithm to perform automated closed-loop glucose control under hospital settings. To assess the performance of the device, capillary blood glucose concentrations were frequently determined and a continuous glucose sensor was additionally worn by the patients.

RESULTS

The average mean absolute relative deviation from blood glucose concentrations obtained for the sensor of the device was low (median, 13.0%; interquartile range, 10.5-16.7%; n = 10) and did not differ from that of the additionally worn glucose sensor (versus 13.9%; 11.9-15.3%; P = 0.922). Furthermore, insulin delivery with the single-port device was reliable and safe during home use and, when performed in combination with the control algorithm, was adequate to achieve and maintain near normoglycemia.

CONCLUSION

Our data show the feasibility of open- and closed-loop glucose control in diabetes patients using a device that combines insulin delivery and glucose sensing at a single tissue site.

SIGNIFICANCE

The reduction in device size and invasiveness achieved by this design may largely increase patient convenience and enhance acceptance of diabetes treatment with continuous glucose monitoring and insulin delivery technology.

Differences Between Flash Glucose Monitor and Fingerprick Measurements

Freestyle Libre (FL) is a factory calibrated Flash Glucose Monitor (FGM). We investigated Mean Absolute Relative Difference (MARD) between Self Monitoring of Blood Glucose (SMBG) and FL measurements in the first day of sensor wear in 39 subjects with Type 1 diabetes. The overall MARD was 12.3%, while the individual MARDs ranged from 4% to 25%. Five participants had a MARD ≥ 20%. We estimated bias and lag between the FL and SMBG measurements. The estimated biases range from -1.8 mmol / L to 1.4 mmol / L , and lags range from 2 min to 24 min . Bias is identified as a main cause of poor individual MARDs. The biases seem to persist in days 2⁻7 of sensor usage. All cases of MARD ≥ 20% in the first day are eliminated by bias correction, and overall MARD is reduced from 12.3% to 9.2%, indicating that adding support for voluntary user-supplied bias correction in the FL could improve its performance.