Continuous Glucose Monitoring Data Research Articles

Continuous Glucose Monitoring Captures Glycemic Variability After Roux-en-Y Gastric Bypass in Patients with and Without Type 2 Diabetes Mellitus: A Prospective Cohort Study.

To evaluate glycemic variability (GV) using continuous glucose monitoring (CGM) in individuals with and without type 2 diabetes mellitus (T2DM) undergoing Roux-en-Y gastric bypass (RYGB). This prospective cohort study compared the CGM data of fourteen patients with T2DM (n = 7) and without T2DM (n = 7) undergoing RYGB. After 6months, these patients were compared to a non-operative control group (n = 7) matched by BMI, sex, and age to the T2DM group. Fourteen patients underwent RYGB, with a mean BMI of 46.9 ± 5.3kg/m2 and an average age of 47.9 ± 8.9years; 85% were female. After 6months post-surgery, the total weight loss (TWL) was 27.1 ± 6.3%, with no significant differences between the groups. Patients without diabetes had lower mean interstitial glucose levels (81 vs. 94 and 98mg/dl, p < 0.01) and lower glucose management indicator (GMI) (5.2 vs. 5.6 and 5.65%, p = 0.01) compared to the control and T2DM groups, respectively. The coefficient of variation (CV) significantly increased only in patients with diabetes (17% vs. 26.7%, p < 0.01). Both groups with (0% vs. 2%, p = 0.03) and without (3% vs. 22%, p = 0.03) T2DM experienced an increased time below range with low glucose (54-69mg/dL). However, patients without T2DM had significantly less time in rage (70-180mg/dL) (97% vs. 78%, p = 0.04). Significant differences in CGM metrics among RYGB patients suggest an increase in glycemic variability after surgery, with a longer duration of hypoglycemia, especially in patients without T2DM.

Continuous Glucose Monitoring-Derived Differences in Pregnancies With and Without Adverse Perinatal Outcomes.

To evaluate whether continuous glucose monitoring (CGM)-derived glycemic patterns observed throughout pregnancy were associated with adverse perinatal outcomes, specifically fetal growth disorders and hypertensive disorders of pregnancy (HDP). We conducted a prospective observational study of individuals with viable singleton pregnancies and screening hemoglobin A 1c levels less than 6.5%. Those with preexisting diabetes were excluded. Enrollment occurred at the earliest gestational age before 17 weeks. Participants wore blinded continuous glucose monitors consecutively as willing until delivery. Those with at least 14 days of CGM data were included in analysis. Rates of large-for-gestational-age (LGA) neonates, small-for-gestational age (SGA) neonates, and HDP were assessed. Continuous glucose monitoring-derived glycemic metrics were calculated, including mean glucose level and percent time above and below thresholds. Two-sample t tests were used to compare glycemic metrics between participants with and without adverse perinatal outcomes. Of 937 participants enrolled, 760 met inclusion criteria. Those delivering LGA neonates or who were diagnosed with HDP had higher mean glucose levels (102±9 vs 100±8, P =.01 and 103±8 vs 99±8, P <.001) and spent more time above 120 mg/dL (median 16% vs 12%, P =.006, and 16% vs 12%, P <.001, respectively) and above 140 mg/dL (median 3.9% vs 2.8%, P =.006, and 3.5% vs 2.8%, P <.001, respectively) throughout gestation than those without these outcomes. These findings were present regardless of gestational diabetes mellitus status. Participants with SGA neonates had lower mean glucose levels (97±7 vs 101±8, P =.01) and spent less time above 140 mg/dL (median 1.6% vs 2.3%, P =.01) and more time below 63 mg/dL (median 3.0% vs 2.3%, P =.02) than those without SGA neonates. Individuals with LGA neonates or HDP exhibit a slightly higher mean glucose levels and spend more time hyperglycemic in early pregnancy than those who do not experience these outcomes.

Personalized Nutrition Therapy without Weight Loss Counseling Produces Weight Loss in Individuals with Prediabetes Who Are Overweight/Obese: A Randomized Controlled Trial.

Obesity stands out as a primary risk factor for diabetes. Attaining healthy weight loss, especially reducing body fat, is important in managing prediabetes and preventing progression to full diabetes and its co-morbidities. This study examined the effects of personalized nutrition therapy (PNT) combined with continuous glucose monitoring (CGM) on body weight and composition in individuals with prediabetes. A total of 30 individuals with prediabetes who were overweight or obese were assigned randomly to either the treatment, observed CGM data plus PNT, or the control group which was blinded to their blood glucose results throughout the study. Both groups were provided with dietary recommendations for calorie intake and macronutrient distribution, coupled with personalized goal setting for glucose control and healthy eating, without any specific emphasis on weight reduction or changes in physical activity. Regular visits were scheduled every 10 days to perform measurements and replace CGMs. Data were analyzed using General Linear Model with repeated measures. Over the 30-day follow-up period, both groups experienced significant reductions in weight and fat mass. The treatment group exhibited two-fold greater reductions in both weight and fat mass, a significant decrease in carbohydrate intake, and a significant increase in time spent on physical activitycompared to the control group. In addition, compliance was notably higher in the treatment group. These findings indicate that overweight or obese individuals with prediabetes can achieve weight loss and improved body composition through personalized education for glucose control, without exclusively emphasizing weight loss as the primary objective. Additionally, the real-time feedback provided by CGM enhances these improvements.

Continuous Glucose Monitoring in Intensive Care Unit Patients with Diabetes: Insights and Outcomes from a Single-center Study in India

Abstract Introduction: The use of continuous glucose monitoring (CGM) for in-hospital glucose control has been widely studied. However, its application in intensive care unit (ICU) settings is debatable owing to its perceived limitations. In ICU patients with diabetes, glycemic excursions, including frequent hypoglycemic episodes, are more prevalent. This retrospective study was undertaken to evaluate the utility of CGM in the management of diabetes patients admitted to the ICU. Methods: We collected data of patients with diabetes mellitus who were admitted to the ICU at our institution and were advised CGM by the intensivist. A retrospective analysis was conducted, examining demographic, clinical, and laboratory data, along with glucose meter readings and CGM data. The results are presented as numbers, percentages, or median values, as appropriate. Results: Out of 65 patients in the ICU, the median values of percentage time in range, time below range, and time above range were 59%, 2%, and 36%, respectively. The median HbA1C and capillary glucose on admission were 8.8% and 246 mg/dl, respectively. During the ICU stay, the CGM derived average glucose level was 167 mg/dl (percent coefficient of variation [%CV] =25.96) and the glucose management indicator was 7%. In total, 65 episodes of hypoglycemia were detected, all confirmed by capillary glucose tests. A total of 27 patients experienced hypoglycemia, of which 12 had nocturnal hypoglycemia. Conclusion: The use of CGM in the ICU improved glycemic control and helped to detect hypoglycemia, particularly nocturnal with acceptably low glycemic variability as measured by percentage CV. Further studies are required to corroborate and confirm these findings.

Role for DPP4 inhibitor therapy in cystic fibrosis related diabetes: A single centre experience

IntroductionInsulin remains the only recommended medical treatment for cystic fibrosis related diabetes (CFRD) Whilst there is an established role for orally bioavailable incretin mimetic agents such as the dipeptidyl peptidase-4 inhibitors (DPP4-I) in Type 2 diabetes mellitus, there exists little data on their utility in CFRD. AimTo examine the use of DPP4-I therapy in patients with CFRD at a single large adult cystic fibrosis center. MethodPeople with CFRD prescribed a DPP4-I were identified from our specialist CFRD clinic and records were retrospectively examined for indication for therapy, tolerability and effectiveness. Analysis of continuous glucose monitoring data Libre 2 was done for these patients (CGM) pre and at least 3 months post therapy was performed. Results23 people with CF (PwCF) with a mean (SD) age of 35.0 ± 2.4 years were included in this analysis . In 21 patients DPP4-I was prescribed as a monotherapy and it was given in combination with insulin in 2 others. Indications for therapy included reactive hypoglycaemia (n = 10) post prandial hyperglycaemia (8), insulin avoidance (3), metformin intolerance (1) and unclear (1). Therapy was well tolerated with no discontinuations due to adverse effects. Significant improvements were noted in Time in Range- Pre vs Post: 78.0 [67.5 – 84.0] vs 89.0 [79.8 – 96.0]%, p = 0.005, Time above Range -Pre vs Post: 19.5 [12.5 – 30.8] vs 6.0 [2.5 – 16.5]%, p = 0.006 and glucose variability Pre versus Post: 28.3 [25.4 – 31.1] vs 26.9 [23.1 – 31.3], p = 0.021, Of the 10 subjects who initiated therapy for hypoglycaemia, 7 reported an improvement in symptoms. No significant difference was found in weight pre and post: 61.5 ± 15.0 kg vs 62.5 ± 15.3 kg, p = 0.326 or Hba1c pre vs post: 41.0 [36.0 – 53.3] mmol/mol versus 40.5 [36.8 – 47.3], p = 0.727. ConclusionDPP4-I is well tolerated in CFRD and can lead to an improved glycaemic control in these patients with significant improvement in validated CGM metrics

Continuous glucose monitoring with FreeStyle Libre PRO sensor in patients with type 2 diabetes and end-stage renal failure on haemoDIALysis (FSLPRO-DIAL pilot study).

For end-stage renal disease (ESRD) patients with diabetes on haemodialysis, diabetes control is difficult to achieve. Hypoglycaemia is a major problem in these frailty subjects. Continuous glucose monitoring (CGM) devices appear therefore to be a good tool to help patients monitor their glycaemic control and to help practitioners optimize treatment. We aimed to compare the laboratory value of Hba1c with the sensor-estimated value of Hba1c (= glucose management indicator, GMI) in ESRD patients with type 2 diabetes (T2D) (with or without insulin treatment) on haemodialysis. Secondly, we aimed to identify CGM-derived monitoring parameters [time in range, time in hypo/hyperglycaemia, glycaemic variability (coefficient of variation, CV)] to identify patients at risk of frequent hypo- or hyperglycaemia. The FSLPRO-DIAL pilot study (NCT04641650) was a prospective monocentric cohort study including 29 subjects with T2D who achieve the protocol. Inclusion criteria were: age ≥ 18years, haemodialysis duration for at least 3months, type 2 diabetes with no change in treatment for at least 3months. Demographic data and blood sample were collected at the day of inclusion. Freestyle Libre pro IQ sensor (blinded CGM) was inserted for 14days. After this period, all CGMs data were collected and analysed. Data were available for 27 patients. Mean age was 73 ± 10, mean BMI 27.2kg/m2, mean duration of diabetes 16.9years and mean dialysis duration 2.9years. Twenty-four subjects were treated with insulin. Mean HbA1c was 6.6% (SD 1.2), and mean GMI was 6.7% (SD 0.9) (no significant difference, p = 0.3). Twelve subjects (44.4%) had a discordance between HbA1c and GMI of < 0.5%, 11 (40.8%) had a discordance between 0.5 and 1%, and only 4 (14.8%) had a discordance of > 1%. Mean time in range (70-180mg/dl) was 71.9%, mean time below range (< 70mg/dl) was 5.6%, and mean time above range (> 180mg/dl) was 22.1%. Mean CV was 31.8%. For 13 out of 27 patients, we reduced antidiabetic treatment by stopping treatments or reducing insulin doses. In this pilot study, there was no global significant difference between HbA1c and GMI in this particular cohort with very well-controlled diabetes. However, the use of the sensor enabled us to identify an excessive time in hypoglycemia in this fragile population and to adapt their treatment.

Decreased complexity of glucose time series index associated with adverse pregnancy outcomes in gestational diabetes mellitus.

We investigated the relationship between the complexity of the glucose time series index (CGI) during pregnancy and adverse pregnancy outcomes in women with gestational diabetes mellitus (GDM). In this retrospective cohort study, 388 singleton pregnant women with GDM underwent continuous glucose monitoring (CGM) at a median of 26.86 gestational weeks. CGI was calculated using refined composite multiscale entropy based on CGM data. The participants were categorized into tertiles according to their baseline CGI (CGI <2.32, 2.32-3.10, ≥3.10). Logistic regression was used to assess the association between CGI and composite adverse outcomes or large for gestational age (LGA). The discrimination performance of CGI was estimated using receiver operating characteristic analysis. Of the 388 participants, 71 (18.3%) had LGA infants and 63 (16.2%) had composite adverse outcomes. After adjustments were made for confounders, compared with those with a high CGI (CGI ≥3.10), participants with a low CGI (CGI <2.32) had a higher risk of composite adverse outcomes (odds ratio: 12.10, 95% confidence interval: 4.41-33.18) and LGA (odds ratio: 12.68, 95% confidence interval: 4.04-39.75). According to the receiver operating characteristic analysis, CGI was significantly better than glycated haemoglobin and conventional CGM indicators for the prediction of adverse pregnancy outcomes (all p < .05). A lower CGI during pregnancy was associated with composite adverse outcomes and LGA. CGI, a novel glucose homeostasis predictor, seems to be superior to conventional glucose indicators for the prediction of adverse pregnancy outcomes in women with GDM.

Perceived stress and associations between physical activity, sedentary time, and interstitial glucose in healthy adolescents

BackgroundLess moderate to vigorous physical activity (MVPA), more sedentary time (ST), and higher perceived stress are related to type 2 diabetes mellitus (T2DM) occurrence, but observational evidence addressing the interaction of these factors is lacking in youth. This pilot study investigated momentary stress as a moderator in the acute associations of MVPA and ST with subsequent glucose in healthy adolescents. MethodsParticipants (N=15, Mage=13.1±1.0 years, 10 girls, 5 with overweight/obesity) simultaneously wore a continuous glucose monitor (CGM), thigh-mounted accelerometer, and reported momentary stress via random ecological momentary assessments (EMA; Time T) for 7–14 days. MVPA and ST (min) were calculated for 60- and 120-minute time windows before each EMA prompt (Time T-1). Mean CGM-measured interstitial glucose (mg/dL) was calculated after each prompt (Mmin=120.0±25.4; Time T+1). Multilevel models assessed whether within-subject MVPA and ST (Time T-1) predicted mean glucose (Time T+1), with momentary stress as a moderator (Time T). ResultsThere were 513 time-matched EMA reports of stress, accelerometer, and CGM data. Momentary stress moderated the effects of MVPA 60 (β=-0.22, p=.001) and 120 min (β=-0.08, p=.02) before the prompt on subsequent glucose levels. When youth spent more time in MVPA than their average and when momentary stress was higher than their average, subsequent glucose was lower. Stress did not moderate associations of ST with glucose (p>.05). ConclusionsHigher momentary stress may interact with higher MVPA to lower subsequent glucose in youth. Accelerometers, EMA, and CGMs can be integrated in future studies to further understand these associations in free-living environments.

145-OR: Assessing the Current State of Diabetes Care in Long-Term Facilities Using Continuous Glucose Monitoring

Introduction: Recommendations for diabetes care in long term care facilities (LTCF) focuses on avoidance of hypoglycemia and symptomatic hyperglycemia. Using continuous glucose monitoring (CGM), we evaluated the current state of glycemia in LTCF residents with multiple comorbidities. Methods: We conducted a cross-sectional study using CGM in 8 LTC facilities in the United States. A masked Dexcom G6 pro CGM was placed for 10 days on residents with diagnosis of diabetes mellitus (DM) who were on any oral or injectable hypoglycemic medications. Clincial and demographic information were collected from medical records. Results: We evaluated 65 residents (mean age 65 years, 50% female). Fourteen percent of the cohort was on sulfonylurea medications and 68% were on insulin. CGM data showed 26% of the cohort with &amp;gt;1% time spent in hypoglycemia (time &amp;lt;70 mg/dL). A larger burden of hyperglycemia was seen with 54% of the cohort spending &amp;gt;10% time &amp;gt;250mg/dL, 37% spending &amp;gt;25% time &amp;gt;250 mg/dL, and 14% spending &amp;gt;90% time &amp;gt;250 mg/dL. Only 44% of the cohort had &amp;gt;50% Time-in-Range (70-180 mg/dL). The latest A1C was &amp;lt;7% in 45% of the cohort, 7-8% in 27%, 8.1-9% in 12%, and &amp;gt;9% in 16%. Fingerstick reading frequency was 2 or fewer times/day in 74% of the cohort, 3 times/day in 13%, and 4 or more times/day in 13%. On average, this cohort had 13 comorbidities, with 85% having functional disabilities. There was 14% with dementia and 42% with obesity (BMI&amp;gt;30). All 100% of the participants had polypharmacy(≥5 medications/day). Conclusions: In this multi-morbid cohort of residents with diabetes living in LTCF, we identified a high burden of both hypoglycemia and severe hyperglycemia, despite fair control of A1C. More consistent use of CGM could help identify glycemic excursions to improve therapeutic decision-making. Disclosure C. Slyne: None. K. Roberts: None. C.D. Conery: None. H. Brabant: None. N. Krakoff: None. E. Toschi: Consultant; Vertex Pharmaceuticals Incorporated, Sanofi. M. Munshi: Consultant; Sanofi. Funding Supported by an investigator initiated grant from Dexcom

39-PUB: A Comprehensive Tool for Analyzing Continuous Glucose Monitoring Data

39-PUB: A Comprehensive Tool for Analyzing Continuous Glucose Monitoring Data

Assessing Patterns of Continuous Glucose Monitoring Use and Metrics of Glycemic Control in Type 1 Diabetes and Type 2 Diabetes Patients in the Veterans Health Care System: Integrating Continuous Glucose Monitoring Device Data with Electronic Health Records Data.

Objective: To integrate long-term daily continuous glucose monitoring (CGM) device data with electronic health records (EHR) for patients with type 1 and type 2 diabetes (T1D and T2D) in the national Veterans Affairs Healthcare System to assess real-world patterns of CGM use and the reliability of EHR-based CGM information. Research Design and Methods: This observational study used Dexcom CGM device data linked with EHR (from 2015 to 2020) for a large national cohort of patients with diabetes. We tracked the initiation and consistency of CGM use, assessed concordance of CGM use and measures of glucose control between CGM device data and EHR records, and examined results by age, ethnicity, and diabetes type. Results: The time from pharmacy release of CGM to patients to initiation of uploading CGM data to Dexcom servers averaged 3 weeks but demonstrated wide variation among individuals; importantly, this delay decreased markedly over the later years. The average daily wear time of CGM exceeded 22 h over nearly 3 years of follow-up. Patterns of CGM use were generally consistent across age, race/ethnicity groups, and diabetes type. There was strong concordance between EHR-based estimates of CGM use and Dexcom CGM wear time and between estimates of glucose control from both sources. Conclusions: The study demonstrates our ability to reliably integrate CGM devices and EHR data to provide valuable insights into CGM use patterns. The results indicate in the real-world environment that CGM is worn consistently over many years for both patients with T1D and T2D within the Veterans Affairs Healthcare System and is similar across major race/ethnic groups and age-groups.

Continuous glucose monitoring (CGM) for 308 older-age participants in an English birth cohort: variability and correlates

Introduction &amp; BackgroundEpochs of hyperglycaemia and hypoglycaemia may each increase risk of common chronic diseases and impair both cognitive and physical function even in people without diabetes. Older people may have greater frequency of adverse glycaemic excursions, partly due to disordered autonomic function and sleep quality. Data for older, non-diabetic people are however scant. Objectives &amp; Approach1) To describe blood glucose variability (completed) and 2) its socio-demographic and lifestyle correlates in a predominantly non-diabetic cohort of older adults (planned). Participants were recruited during 2021-2023 from an English birth cohort (the 1946 National Survey for Health and Development Study). They wore a continuous glucose monitor (Freestyle libre Abbott), which measured circulating glucose four times/hour, for seven days. Summary statistics and time outside range (4.4-7.8mmol/L) were calculated. Further information on glycaemic excursions and day-to-day variability will be gleaned using the R “iglu” package. For all CGM summary and excursion measures, future analyses will investigate: associations with HbA1c, socio-demographics, body composition, physical activity, diet and alcohol use. Results will be stratified by sleep/ wake time periods estimated from simultaneous actigraphy (Philips Actiwatch Spectrum Plus). Sensitivity analyses will exclude people taking hypo/ hyperglycaemic medications and those with diabetes. Relevance to Digital FootprintsDerived summary measures can be used by future studies to give insights into glycaemic variability as a population-level risk factor. This work will bring together multiple data sources, i.e. from CGM, actigraphy and baseline cohort data. ResultsParticipants were aged 75-76 years, 45% female and 10% had diagnosed diabetes; median (IQR) BMI was 26.8 (24.6-29.2) kg/m2. CGM data from 308 participants was collected, for a median (IQR) of 6.9 (6.7-7.6) days. Average glucose over the recording period was 5.7mmol/L (5.3-6.2mmol/L), standard deviation was 1.0mmol/L (0.8-1.3mmol/L), time outside range was 12.8% (6.2-24.7%) and 16% of participants spent ≥1 hour/day above and ≥1 hour/day below range. Conclusions &amp; ImplicationsCGM was feasible for this cohort of older adults, and demonstrated high levels of time outside range for a predominantly non-diabetic group. Future analysis will determine whether enhanced characterisation of glycaemic variability is a potentially more accurate tool for predicting future disease risk than isolated glucose measurements.

Optimal Coefficient of Variance Threshold to Minimize Hypoglycemia Risk in Individuals with Well-Controlled Type 1 Diabetes Mellitus.

This study investigated the optimal coefficient of variance (%CV) for preventing hypoglycemia based on real-time continuous glucose monitoring (rt-CGM) data in people with type 1 diabetes mellitus (T1DM) already achieving their mean glucose (MG) target. Data from 172 subjects who underwent rt-CGM for at least 90 days and for whom 439 90-day glycemic profiles were available were analyzed. Receiver operator characteristic analysis was conducted to determine the cut-off value of %CV to achieve time below range (%TBR)<54 mg/dL <1 and =0. Overall mean glycosylated hemoglobin was 6.8% and median %TBR<54 mg/dL was 0.2%. MG was significantly higher and %CV significantly lower in profiles achieving %TBR<54 mg/dL <1 compared to %TBR<54 mg/dL ≥1 (all P<0.001). The cut-off value of %CV for achieving %TBR<54 mg/dL <1 was 37.5%, 37.3%, and 31.0%, in the whole population, MG >135 mg/dL, and ≤135 mg/dL, respectively. The cut-off value for %TBR<54 mg/dL=0% was 29.2% in MG ≤135 mg/dL. In profiles with MG ≤135 mg/dL, 94.2% of profiles with a %CV <31 achieved the target of %TBR<54 mg/dL <1, and 97.3% with a %CV <29.2 achieved the target of %TBR<54 mg/ dL=0%. When MG was >135 mg/dL, 99.4% of profiles with a %CV <37.3 achieved %TBR<54 mg/dL <1. In well-controlled T1DM with MG ≤135 mg/dL, we suggest a %CV <31% to achieve the %TBR<54 mg/dL <1 target. Furthermore, we suggest a %CV <29.2% to achieve the target of %TBR<54 mg/dL =0 for people at high risk of hypoglycemia.

Novel Detection and Progression Markers for Diabetes Based on Continuous Glucose Monitoring Data Dynamics.

Develop a diabetes diagnostic tool based on two markers of continuous glucose monitoring (CGM) dynamics: CGM entropy rate (ER) and Poincaré plot (PP) ellipse area (S). 5,754 daily CGM profiles from 843 individuals with type 1, type 2 diabetes, or healthy individuals with or without islet autoantibody status were used to compute two individual dynamic markers: ER (in bits per transition; BPT) of daily probability matrices describing CGM transitions between eight glycemic states, and the area S (mg2/dL2) of individual CGM PP ellipses using standard PP descriptors. The Youden's index was used to determine "optimal" cut-points for ER and S for health vs. diabetes (case 1); type 1 vs. type 2 (case 2); and low vs. high type 1 immunological risk (case 3). The markers' discriminative power was assessed through the area under the receiver operating characteristics curves (AUC). Optimal cut-off points were determined for ER and S for each of the three cases. ER and S discriminated case 1 with AUC = 0.98 (95% CI: 0.97-0.99) and AUC = 0.99 (95% CI: 0.99-1.00), respectively, (cut-offs ERcase1 = 0.76 BPT, Scase1 = 1993.91 mg2/dL2), case 2 with AUC = 0.81 (95% CI, 0.77-0.84) and AUC = 0.76 (95% CI, 0.72-0.81), respectively (ERcase2 = 1.00 BPT, Scase2 = 5112.98 mg2/dL2), and case 3 with AUC = 0.81 (95% CI, 0.77-0.84) and AUC = 0.76 (95% CI, 0.72-0.81), respectively (ERcase3 = 0.52 BPT, Scase3 = 923.65 mg2/dL2). CGM dynamics markers can be an alternative to fasting plasma glucose or glucose tolerance testing and identifying individuals at higher immunological risk of progressing to type 1 diabetes.

The Association of High and Low Glycation With Incident Diabetic Retinopathy in Adults With Type 1 Diabetes.

We investigated the risk of incident diabetic retinopathy (DR) among high glycator compared to low glycator patients based on the hemoglobin glycation index (HGI). Visit-to-visit variations in HGI also were assessed. Glycated hemoglobin (HbA1c) and continuous glucose monitoring data were collected up to 7 years prior to the date of eye examination defining incident DR or no retinopathy (control). Hemoglobin glycation index was calculated as difference in measured HbA1c and an estimated A1c from sensor glucose (eA1c) to define high (HbA1c - eA1c >0%) or low (HbA1c - eA1c <0%) glycator. Stable glycators were defined as ≥75% of visits with same HGI category. Logistic regression was used to assess the association between glycation category and incident DR. Of 119 adults with type 1 diabetes (T1D), 49 (41%) were stable low glycator (HbA1c - eA1c <0%), 36 (30%) were stable high glycator (HbA1c - eA1c >0%), and 34 (29%) were unstable glycator. Using alternate criteria to define high vs low glycator (consistent difference in HbA1c - eA1c of > 0.4% or <0.4%, respectively), 53% of the adults were characterized as unstable glycator. Compared to low glycators, high glycators did not have a significantly higher risk for incident DR over time when adjusted for age, T1D duration and continuous glucose monitoring (CGM) sensor type (odds ratio [OR] = 1.31, 95% confidence interval [CI] = 0.48-3.62, P = .15). The risk of diabetic retinopathy was not found to differ significantly comparing high glycators to low glycators in adults with T1D. Moreover, HbA1c - eA1c relationship was not stable in nearly 30% to 50% adults with T1D, suggesting that discordance in HbA1c and eA1c are mostly related either HbA1c measurements or estimation of A1c from sensor glucose rather than physiological reasons.

Generalization of a Deep Learning Model for Continuous Glucose Monitoring-Based Hypoglycemia Prediction: Algorithm Development and Validation Study.

Predicting hypoglycemia while maintaining a low false alarm rate is a challenge for the wide adoption of continuous glucose monitoring (CGM) devices in diabetes management. One small study suggested that a deep learning model based on the long short-term memory (LSTM) network had better performance in hypoglycemia prediction than traditional machine learning algorithms in European patients with type 1 diabetes. However, given that many well-recognized deep learning models perform poorly outside the training setting, it remains unclear whether the LSTM model could be generalized to different populations or patients with other diabetes subtypes. The aim of this study was to validate LSTM hypoglycemia prediction models in more diverse populations and across a wide spectrum of patients with different subtypes of diabetes. We assembled two large data sets of patients with type 1 and type 2 diabetes. The primary data set including CGM data from 192 Chinese patients with diabetes was used to develop the LSTM, support vector machine (SVM), and random forest (RF) models for hypoglycemia prediction with a prediction horizon of 30 minutes. Hypoglycemia was categorized into mild (glucose=54-70 mg/dL) and severe (glucose<54 mg/dL) levels. The validation data set of 427 patients of European-American ancestry in the United States was used to validate the models and examine their generalizations. The predictive performance of the models was evaluated according to the sensitivity, specificity, and area under the receiver operating characteristic curve (AUC). For the difficult-to-predict mild hypoglycemia events, the LSTM model consistently achieved AUC values greater than 97% in the primary data set, with a less than 3% AUC reduction in the validation data set, indicating that the model was robust and generalizable across populations. AUC values above 93% were also achieved when the LSTM model was applied to both type 1 and type 2 diabetes in the validation data set, further strengthening the generalizability of the model. Under different satisfactory levels of sensitivity for mild and severe hypoglycemia prediction, the LSTM model achieved higher specificity than the SVM and RF models, thereby reducing false alarms. Our results demonstrate that the LSTM model is robust for hypoglycemia prediction and is generalizable across populations or diabetes subtypes. Given its additional advantage of false-alarm reduction, the LSTM model is a strong candidate to be widely implemented in future CGM devices for hypoglycemia prediction.

Continuous glucose monitoring data for artificial intelligence-based predictive glycemic event: A potential aspect for diabetic care

Continuous glucose monitoring data for artificial intelligence-based predictive glycemic event: A potential aspect for diabetic care

Cross-sectional association of continuous glucose monitoring-derived metrics with cerebral small vessel disease in older adults with type 2 diabetes.

To examine cross-sectional associations between continuous glucose monitoring (CGM)-derived metrics and cerebral small vessel disease (SVD) in older adults with type 2 diabetes. In total, 80 patients with type 2 diabetes aged ≥70 years were analysed. Participants underwent CGM for 14 days. From the CGM data, we derived mean sensor glucose, percentage glucose coefficient of variation, mean amplitude of glucose excursion, time in range (TIR, 70-180 mg/dl), time above range (TAR) and time below range metrics, glycaemia risk index and high/low blood glucose index. The presence of cerebral SVD, including lacunes, microbleeds, enlarged perivascular spaces and white matter hyperintensities, was assessed, and the total number of these findings comprised the total cerebral SVD score (0-4). Ordinal logistic regression analyses were performed to examine the association of CGM-derived metrics with the total SVD score. The median SVD score was 1 (interquartile range 0-2). Higher hyperglycaemic metrics, including mean sensor glucose, TAR >180 mg/dl, TAR >250 mg/dl, and high blood glucose index and glycaemia risk index, were associated with a higher total SVD score. In contrast, a higher TIR (per 10% increase) was associated with a lower total SVD score (odds ratio 0.73, 95% confidence interval 0.56-0.95). Glycated haemoglobin, percentage glucose coefficient of variation, mean amplitude of glucose excursions, time below range and low blood glucose index were not associated with total cerebral SVD scores. The hyperglycaemia metrics and TIR, derived from CGM, were associated with cerebral SVD in older adults with type 2 diabetes.

Glycemic control assessed by continuous glucose monitoring during pregnancy in women with type 1 diabetes and its association with preeclampsia, an observational Swedish cohort study.

Women with type 1 diabetes have an increased risk of preeclampsia (PE), but it is not fully understood if degree of glycemic control is associated with this risk. The aim of this study was to assess glycemic control during pregnancy analyzed by continuous glucose monitoring (CGM) in women with and without PE and to investigate if glycemic control is associated with increased risk of PE. A total of 120 pregnant Swedish women with type 1 diabetes using CGM were included. Background factors and pregnancy outcomes were collected from medical records. CGM data were collected via the internet-based platform Diasend. Mean glucose, standard deviation of mean glucose, percentage of time in target, time below target, and time above target were presented for each trimester in women who did or did not develop PE. Associations between CGM-derived metrics and PE were analyzed with logistic regression and adjusted for confounders. Twenty-two women (18.3%) developed PE. There were no significant differences in maternal characteristics between women with and without PE. Glycemic control improved in each trimester but was suboptimal in both groups. Time in target increased from 59% in the non-PE group and 54% in the PE group in the first trimester to 65% in both groups in the third trimester. There were no significant associations between glycemic control and PE after adjustment for confounders. Degree of glycemic control during pregnancy assessed by CGM was not associated with development of PE in women with type 1 diabetes. However, more research is needed to understand the role of glycemic control in relation to development of PE.

Ultra rapid lispro improves postprandial glucose control versus lispro in combination with basal insulin: a study based on CGM in type 2 diabetes in China.

To evaluate the efficacy and safety of URLi (ultra rapid lispro insulin) compared to insulin lispro as bolus insulin with basal insulin using CGM in the individuals with type 2 diabetes(T2D) in China. This was a double-blind, randomized, parallel, prospective, phase 3 study. Subjects with uncontrolled T2D were recruited and randomized 1:2 into the insulin lispro and URLi groups. Subjects received a consistent basal insulin regimen during the study and self-administered insulin lispro or URLi before each meal throughout the treatment period. Subjects underwent a 3-day continuous glucose monitoring (CGM) at the baseline and endpoint respectively, and then CGM data were analyzed. The primary endpoint was to compare the difference in postprandial glucose (PPG) control using CGM between the two groups. A total of 57 subjects with T2D completed the study. Our CGM data showed that postprandial glucose excursions after breakfast (BPPGE) in the URLi group was lower than that in the insulin lispro group (1.59 ± 1.57 mmol/L vs 2.51 ± 1.73 mmol/L, p = 0.046). 1-hour PPG was observed to decrease more in the URLi group than that in the insulin lispro group (-1.37 ± 3.28 mmol/L vs 0.24 ± 2.58 mmol/L, p = 0.047). 2-hour PPG was observed to decrease more in the URLi group than that in the insulin lispro group (-1.12 ± 4.00 mmol/L vs 1.22 ± 2.90 mmol/L, p = 0.021). The mean HbA1c level decreased by 1.1% in the URLi group and 0.99% in the insulin lispro group, with no treatment difference (p = 0.642). In the CGM profile, TBR was not significantly different between the two groups (p = 0.743). The weight gain also did not differ between the two groups (p = 0.303). URLi can control breakfast PPG better than insulin lispro in adults with T2D in China, while it is non-inferior in improving HbA1c. The incidence of hypoglycemic and weight gain were similar between the two groups.