A French Study to Evaluate the Usefulness of an Implantable Continuous Glucose Monitoring (CGM) Sensor to Improve Glycemic Control in Participants With Diabetes Mellitus

An adverse event (AE) is any untoward medical occurrence, unintended disease or injury, or untoward clinical signs (including abnormal laboratory findings) in participants, whether or not related to the investigational medical device. AEs were coded using the MedDRA dictionary version 20.1. All AEs were assessed by the investigator for relation to the device, seriousness (according to serious AE criteria), and severity (i.e., intensity of the event: mild, moderate, or severe). AEs of hypoglycemia were identified using a Standardized MedDRA Query (SMQ) for a list of Preferred Terms; all terms for hypoglycemia reported in the AE forms were considered symptomatic hypoglycemia. 'Severe hypoglycemia' included all serious and severe/life-threatening events of the SMQ for hypoglycemia. Serious diabetic ketoacidosis was identified in the AE forms using a set list of MedDRA terms. Adverse device effects (ADEs) included all AEs that were pre-specified in the electronic Case Report Form (eCRF).

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohorts 1 and 2 spent in euglycemia with a blood glucose level ≥70 mg/dL to ≤180 mg/dL between the Day 90 and Day 120 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm and centre (and diabetes type for Cohort 1) as fixed classification effects and the time in euglycemia at baseline (i.e. percentage of time spent in euglycemia between the Day 0 and Day 30 visits) as baseline covariates. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohorts 1 and 2 spent in hyperglycemia with a blood glucose level >250 mg/dL between the Day 90 and Day 120 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm and center (and diabetes type for Cohort 1) as fixed classification effects and the level of hyperglycemia at baseline (i.e. percentage of time spent in hyperglycemia >250 mg/dL between the Day 0 and Day 30 visits) as baseline covariates. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohorts 1 and 2 spent in hyperglycemia with a blood glucose level >180 mg/dL between the Day 90 and Day 120 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm and center (and diabetes type for Cohort 1) as fixed classification effects and the level of hyperglycemia at baseline (i.e. percentage of time spent in hyperglycemia >180 mg/dL between the Day 0 and Day 30 visits) as baseline covariate. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohort 1 spent in hypoglycemia with a blood glucose level <70 mg/dL between the Day 90 and Day 120 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm and center (and diabetes type for Cohort 1) as fixed classification effects and the level of hypoglycemia at baseline (i.e. percentage of time spent in hypoglycemia <70 mg/dL between the Day 0 and Day 30 visits) as baseline covariate. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohort 1 spent in hypoglycemia with a blood glucose level <54 mg/dL between the Day 90 and Day 120 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm, center, and diabetes type as fixed classification effects and the level of hypoglycemia at baseline (i.e. percentage of time spent in hypoglycemia <54 mg/dL between the Day 0 and Day 30 visits) as baseline covariate. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohorts 1 and 2 spent in euglycemia with a blood glucose level ≥70 mg/dL to ≤180 mg/dL between the Day 150 and Day 180 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm and center (and diabetes type for Cohort 1) as fixed classification effects and the level of hypoglycemia at baseline (i.e. percentage of time spent in euglycemia between the Day 0 and Day 30 visits) as baseline covariate. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohorts 1 and 2 spent in hyperglycemia with a blood glucose level >250 mg/dL between the Day 150 and Day 180 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm and center (and diabetes type for Cohort 1) as fixed classification effects and the level of hypoglycemia at baseline (i.e. percentage of time spent in hyperglycemia >250 mg/dL between the Day 0 and Day 30 visits) as baseline covariates. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohorts 1 and 2 spent in hyperglycemia with a blood glucose level >180 mg/dL between the Day 150 and Day 180 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm and center (and diabetes type for Cohort 1) as fixed classification effects and the level of hypoglycemia at baseline (i.e. percentage of time spent in hyperglycemia >180 mg/dL between the Day 0 and Day 30 visits) as baseline covariate. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohorts 1 and 2 spent in hypoglycemia with a blood glucose level <70 mg/dL between the Day 150 and Day 180 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm and center (and diabetes type for Cohort 1) as fixed classification effects and the level of hypoglycemia at baseline (i.e. percentage of time spent in hypoglycemia <70 mg/dL between the Day 0 and Day 30 visits) as baseline covariate. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in each treatment arm of Cohorts 1 and 2 spent in hypoglycemia with a blood glucose level <54 mg/dL between the Day 150 and Day 180 visits, based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the randomization arm and center (and diabetes type for Cohort 1) as fixed classification effects and the level of hypoglycemia at baseline (i.e. percentage of time spent in hypoglycemia <54 mg/dL between the Day 0 and Day 30 visits) as baseline covariate. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in the Enabled arm of Cohort 2 spent in hypoglycemia with a blood glucose level <54 mg/dL over two different time periods (between the Days 90-120 and Days 150-180), based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the time period and center as fixed classification effects and the level of hypoglycemia at baseline (i.e. percentage of time spent in hypoglycemia <54 mg/dL between the Day 0 and Day 30 visits) as baseline covariate. Adjusted means with their 95% confidence intervals are provided.

The results show an analysis of covariance (ANCOVA) comparing the percentage of time participants in the Cohort 2, Control arm spent in hypoglycemia with a blood glucose level <54 mg/dL with a different intervention over each of the two time periods (the control period, using their usual glucose monitoring system: Days 90-120, and the enabled period, using the Eversense XL CGM system: Days 150-180), based on the total duration of time with available CGM data (i.e., excluding temporary discontinuations). The statistical model included the time period and center as fixed classification effects and the level of hypoglycemia at baseline (i.e. percentage of time spent in hypoglycemia <54 mg/dL between the Day 0 and Day 30 visits) as baseline covariate. Adjusted means with their 95% confidence intervals are provided.

Glucose variability was estimated with a unitless coefficient of variation, expressed as a percentage, which was computed using CGM data with the following formula: the sum of the ratios of standard deviation of daily glycemia to the mean of daily glycemia divided by the number of days with available data for the period of interest (Days 90 to 120). According to Monnier et al. Diabetes Care 2017, a coefficient of variation threshold set to 36% allows to distinguish between stable and unstable glycemia.

Glucose variability was estimated with a unitless coefficient of variation, expressed as a percentage, which was computed using CGM data with the following formula: the sum of the ratios of standard deviation of daily glycemia to the mean of daily glycemia divided by the number of days with available data for the period of interest (Days 150 to 180). According to Monnier et al. Diabetes Care 2017, a coefficient of variation threshold set to 36% allows to distinguish between stable and unstable glycemia.

The results show an analysis of covariance (ANCOVA) comparing the HbA1c (%), defined as the percentage of hemoglobin proteins that are glycated (i.e., chemically linked to a sugar), at the Day 120 visit between the Enabled and Control arms for Cohorts 1 and 2. The statistical model included the randomization arm and center (and diabetes type for Cohort 1) as fixed classification effects, and HbA1c (%) at Day 0 as baseline covariates. Adjusted means with their 95% confidence interval are provided.

The analysis of this secondary outcome measure for Cohort 2 was an analysis of covariance (ANCOVA) comparing the HbA1c (%), defined as the percentage of hemoglobin proteins that are glycated (i.e., chemically linked to a sugar), at the Day 180 visit between the Enabled and Control arms. The statistical model included the randomization arm and center as fixed classification effects, and HbA1c (%) at Day 0 as baseline covariates. Adjusted means with their 95% confidence interval are provided.

This measure assesses sensor life by counting the number of participants with their first inserted CGM sensor still operating at 150 days and at 180 days post-insertion. The assessment was based on the lifespan of the first sensor: for example, if lifespan of the first sensor was ≥150 days and ≥180 days, then the participant was counted as having their first sensor operating at 150 days and 180 days, respectively. The lifespan of the first inserted CGM sensor for each participant, measured in days, was calculated as the date of the last glucose measurement by the sensor minus the date of the first glucose measurement by the sensor (+1 day).

The lifespan of the first inserted CGM sensor for each participant, measured in days, was calculated as the date of the last glucose measurement by the sensor minus the date of the first glucose measurement by the sensor (+1 day).

This measure assesses transmitter wear time by counting the number of participants with their first CGM transmitter still operating at 150 days and at 180 days after first use. The assessment was based on the lifespan of the first transmitter: for example, if lifespan of the first transmitter was ≥150 days and ≥180 days, then the participant was counted as having their first transmitter operating at 150 days and 180 days, respectively. The lifespan of the first CGM transmitter for each participant, measured in days, was calculated as the date of the last glucose measurement with the transmitter minus the date of the first glucose measurement with the transmitter (+1 day).

The lifespan of the first CGM transmitter for each participant, measured in days, was calculated as the date of the last glucose measurement with the transmitter minus the date of the first glucose measurement with the transmitter (+1 day).

This was a descriptive analysis of the number of participants in the Eversense XL CGM System Enabled groups who accessed the various CGM application (app) pages/functions at study Days 60, 120, and 180. The data were collected in the electronic Case Report Form.

This outcome measure is a descriptive analysis of the number of participants who were issued various types of alarms and alerts by the Eversense XL CGM System through the mobile medical application. The alarms listed in the results table are a higher level classification for the different types of alerts that immediately follow it, until the next type of alarm is listed (for example, the 'Calibration' alarm included both 'Calibrate Now' and 'Calibration Past Due' alerts). Only participants in the Enabled groups of both cohorts actually received these alarms/alerts because the CGM remained in blinded mode for participants in the Control groups.

The DTSQs questionnaire contains eight items scored on a 7-point scale (i.e. each item is scored from 0 to 6). Six items measure Treatment Satisfaction, dealing with: satisfaction with current treatment; convenience of the treatment; flexibility; satisfaction with own understanding of participant's diabetes; how likely to recommend their present treatment; and how satisfied to continue with their present treatment. These are summed to produce a total Treatment Satisfaction score, with a minimum value of 0 and a maximum value of 36; a higher score indicates greater treatment satisfaction. In case of missing items, the overall score was calculated as the mean of the available items.

The DTSQs questionnaire contains eight items scored on a 7-point scale (i.e. each item is scored from 0 to 6). Six items measure Treatment Satisfaction (dealing with: satisfaction with current treatment; convenience of the treatment; flexibility; satisfaction with own understanding of participant's diabetes; how likely to recommend their present treatment; and how satisfied to continue with their present treatment). These are summed to produce a total Treatment Satisfaction score, with a minimum value of 0 and a maximum value of 36; a higher score indicates greater treatment satisfaction. For the change from baseline analysis, a positive value indicates an improvement in treatment satisfaction. In case of missing items, the overall score was calculated as the mean of the available items.

The DTSQc questionnaire contains eight items scored on a 7-point scale (i.e. each item is scored from -3 to +3). Six items measure Treatment Satisfaction (dealing with: satisfaction with current treatment; convenience of the treatment; flexibility; satisfaction with own understanding of participant's diabetes; how likely to recommend their present treatment; and how satisfied to continue with their present treatment). These are summed to produce a total Treatment Satisfaction score, with a minimum value of -18 and a maximum value of +18; a higher score indicates greater treatment satisfaction. In case of missing items, the overall score was calculated as the mean of the available items.

The ADDQoL measures the impact of diabetes and its treatment on 19 specific aspects of life. The scale ranges from -3 to +1 for 19 life domains (impact rating) and from 0 to +3 in attributed importance (importance rating). A weighted score for each domain is calculated as a multiplier of impact rating and importance rating (ranging from -9 to +3). Finally, a mean weighted impact score (ADDQOL score) is calculated for the entire scale across all applicable domains; a higher score indicates greater quality of life.

The ADDQoL measures the impact of diabetes and its treatment on 19 specific aspects of life. The scale ranges from -3 to +1 for 19 life domains (impact rating) and from 0 to +3 in attributed importance (importance rating). A weighted score for each domain is calculated as a multiplier of impact rating and importance rating (ranging from -9 to +3). Finally, a mean weighted impact score (ADDQOL score) is calculated for the entire scale across all applicable domains; a higher score indicates greater quality of life. For the change from baseline analysis, a positive value indicates an improvement in quality of life.

The CGM-SAT questionnaire comprises 44 item scales assessing experiences with CGM over the previous 6 months. The scale is designed to measure the impact of CGM on diabetes management and family relationships, plus on satisfaction with emotional, behavioral and cognitive effects of CGM use. The responses were rated on a 5-point scale from '1 = strongly agree' to '5 = strongly disagree'. The overall score corresponds to the mean of the 44 items of potential positive or negative effects of using the CGM device; a higher score reflects more favorable impact of, and satisfaction with CGM. In case of missing items, the overall score was calculated as the mean of the available items.

The CGM-SAT questionnaire comprises 44 item scales assessing experiences with CGM over the previous 6 months. The scale is designed to measure the impact of CGM on diabetes management and family relationships, plus on satisfaction with emotional, behavioral and cognitive effects of CGM use. The responses were rated on a 5-point scale from '1 = strongly agree' to '5 = strongly disagree'. The overall score corresponds to the mean of the 44 items of potential positive or negative effects of using the CGM device; a higher score (or positive change from baseline score) reflects more favorable impact of, and satisfaction with CGM. In case of missing items, the overall score was calculated as the mean of the available items.

The DDS2 is a 2-item diabetes distress screening instrument asking respondents to rate on a 6-point scale (from 1 = 'Not a problem' to 6 = 'A very serious problem') the degree to which the following items caused distress: (1) feeling overwhelmed by the demands of living with diabetes, and (2) feeling that I am often failing with my diabetes regimen. The DDS2 score is the mean of the two items (ranging from 1 to 6). A lower DDS2 score (or a negative change from baseline score) indicates a lower level of distress. If one item was missing, then no score was computed.

The DDS2 is a 2-item diabetes distress screening instrument asking respondents to rate on a 6-point scale (from 1 = 'Not a problem' to 6 = 'A very serious problem') the degree to which the following items caused distress: (1) feeling overwhelmed by the demands of living with diabetes, and (2) feeling that I am often failing with my diabetes regimen. The DDS2 score is the mean of the two items (ranging from 1 to 6). A lower DDS2 score (or a negative change from baseline score) indicates a lower level of distress. If one item was missing, then no score was computed.

HFS2 Worry items describe specific concerns that patients may have about their hypoglycemic episodes (e.g., being alone, episodes occurring during sleep, or having an accident). The 5-point Likert scale for each item ranges from 0 (never) to 4 (almost always), and the HFS2 Worry subscale score is the sum of all 18 items (ranging from 0 to 72). A lower HFS2 Worry subscale score (or a negative change from baseline score) indicates a lower level of fear of hypoglycemia. There was no plan to replace missing items: if one item was missing, the score was not computed.

HFS2 Worry items describe specific concerns that patients may have about their hypoglycemic episodes (e.g., being alone, episodes occurring during sleep, or having an accident). The 5-point Likert scale for each item ranges from 0 (never) to 4 (almost always), and the HFS2 Worry subscale score is the sum of all 18 items (ranging from 0 to 72). A lower HFS2 Worry subscale score (or a negative change from baseline score) indicates a lower level of fear of hypoglycemia. There was no plan to replace missing items: if one item was missing, the score was not computed.

The Partner-DDS is a 21-item self-report scale that highlights four critical dimensions of partner-related distress: "my partner's diabetes management", "how best to help", "diabetes and me", and "hypoglycemia". The scale ranges from 0 (not at all) to 4 (a great deal). The Partner-DDS yields a total diabetes distress score plus 4 subscale scores: Total Partner-DDS Score = Mean of the 21 items; My partner's diabetes management = Mean of item 3, 4, 10, 12, 14, 15 and 20; How best to help = Mean of item 2, 6, 7, 11 and 13; Diabetes and me = Mean of items 5, 8, 9, 16 and 21; Hypoglycemia = Mean of items 1, 17, 18 and 19. A lower Partner-DDS score (or a negative change from baseline score) indicates a lower level of distress.