Dexcom G6 for real-time continuous glucose monitoring

Study size, design and location

A randomised, controlled crossover trial at 2 UK hospitals of 31 people with type 1 diabetes aged 16 to 24 years old being treated with multiple daily injections or insulin pump therapy. The study ran for 8 weeks and compared time in target blood glucose range (haemoglobin A1c [HbA1c] levels between 3.9 mmol/litre and 10 mmol/litre) using Dexcom G6 compared with SMBG. Measuring HbA1c is a commonly used way to understand a person's blood glucose levels over time. Elevated HbA1c levels are associated with diabetes-related complications.

Intervention and comparator

Intervention: Dexcom G6 (n=31).

Comparator: routine SMBG (n=31). People in the study acted as their own controls.

Reference standard: laboratory blood glucose test.

Key outcomes

Time in target blood glucose range (defined as 70 mg/dl to 180 mg/dl) was significantly higher during Dexcom G6 use compared with control (35.7% plus or minus 13.5% compared with 24.6% plus or minus 9.3%; p<0.001). Times spent below range (below 70 mg/dl and below 54 mg/dl) were low and not significantly different during both study periods. The Dexcom G6 was worn 84% of the study period. People reported significantly higher satisfaction levels using Dexcom G6 compared with SMBG.

Strengths and limitations

This was a UK RCT that was adequately powered (at the 80% level) to detect the primary outcome.

The sample size of the study limited any subgroup analyses such as glycemic outcomes in multiple daily injections and pump users. The amount of sensor data available for analysis was not equal between the Dexcom G6 and control periods. The study was funded by the company.

Study size, design and location

A multicentre, randomised controlled trial at 24 endocrinology centres in the US of 158 adults with type 1 diabetes and elevated HbA1c treated with multiple daily insulin injections. The study ran for 6 months and assessed if using the Dexcom G5 reduced HbA1c levels compared with standard methods.

Intervention and comparator

Intervention: Dexcom G5 (n=105).

Comparator: routine SMBG (n=53).

Reference standard: laboratory blood glucose test.

Key outcomes

HbA1c levels were significantly reduced in the Dexcom G5 group compared with SMBG (1.1% at 12 weeks and 1.0% at 24 weeks compared with 0.5% and 0.4%, respectively, p <0.001). Median duration of hypoglycaemia (defined as below 70 mg/dl) was 43 minutes per day in the Dexcom G5 group compared with 80 minutes per day in the SMBG group (p=0.002). There were severe hypoglycaemic events in 2 people in each group (p=0.67). In 102 people in the Dexcom G5 group who completed the trial, median sensor use was 7 days per week at 4, 12, and 24 weeks.

Strengths and limitations

This multicentre RCT was adequately powered to detect a difference in mean blood glucose level between treatment groups (randomised 2:1). The study had a low dropout rate (3 people).

People in the study were unblinded to the group to which they had been assigned (but this is expected with interventions such as Dexcom, which are designed to encourage behaviour change). The study was funded by the company.

Study size, design and location

A multicentre, randomised controlled trial at 25 endocrinology centres in the US of 158 adults with type 2 diabetes treated with multiple daily insulin injections. The study ran for 6 months and assessed if using the Dexcom G5 reduced HbA1c levels compared with standard methods.

Intervention and comparator

Intervention: Dexcom G5 (n=79).

Comparator: routine SMBG (n=79).

Reference standard: laboratory blood glucose test.

Key outcomes

Mean HbA1c levels, which at baseline were 8.5% in both groups. HbA1c levels were significantly reduced in the Dexcom G5 group compared with SMBG (1.0% at 12 weeks and 0.8% at 24 weeks compared with 0.6% and 0.5%, respectively, p<0.05). The amount of hypoglycemia was extremely low at baseline (defined as below 70 mg/dl), limiting any assessment of differences in the 2 groups. In 77 people in the Dexcom G5 group who completed the trial, median sensor use was 6.9, 6.7, 6.7 days per week at 4, 12, and 24 weeks, respectively.

Strengths and limitations

This multicentre RCT was adequately powered to detect a difference in mean blood glucose level between treatment groups. The study had a low dropout rate (6 people).

People in the study were unblinded to the group to which they had been assigned (but, as noted above, this is expected with interventions such as Dexcom that are designed to encourage behaviour change). The study was funded by the company.

Study size, design and location

A multicentre, randomised controlled trial at 12 diabetes centres in Germany of 149 adults with type 1 diabetes treated with multiple daily insulin injections. People who were eligible for the study had a history of impaired hypoglycaemia awareness or severe hypoglycaemia during the previous year. The study duration was 6 months.

Intervention and comparator

Intervention: Dexcom G5.

Comparator: routine SMBG.

Reference standard: laboratory blood glucose test.

Key outcomes

A hypoglycaemic event was defined as glucose values of 3.0 mmol/litre (54 mg/dl) or lower for at least 20 minutes, preceded by a minimum of 30 minutes with glucose values greater than 3.0 mmol/litre (54 mg/dl). The mean number of hypoglycaemic events in the Dexcom G5 group was significantly reduced over 28 days (10.8 at 4-week baseline to 3.5 events at 4‑week follow up). There was no significant reduction in the SMBG group. The incidence of all severe hypoglycaemia events among control group participants during follow up was about twice the incidence seen in the Dexcom G5 group (standardised as 1.18 compared with 0.64 events per patient-year). Variability in blood glucose levels fell in the Dexcom G5 group to a greater extent than in the SMBG group. Severe hypoglycaemia events needing third-party assistance without medical assistance for recovery were also less frequent in the Dexcom G5 group than in the control group (19 events compared with 36 events).

Strengths and limitations

This multicentre RCT was adequately powered for the primary outcome of assessing the effect of Dexcom G5 on the number of hypoglycaemic events compared with SMBG. The glucose value in this study is defined as a severe hypoglycaemic event according to experts (below 4.0 mmol/litre defined as hypoglycaemia or a severe hypoglycaemic event).

People were unblinded to the group to which they had been assigned. The study was funded by the company.

Study size, design and location

A prospective multicentre study of 262 people 6 years or older with type 1 diabetes or insulin-treated type 2 diabetes from 11 sites in the US. The study duration was 10 days.

Intervention and comparator

Intervention: Dexcom G6.

No comparator.

Reference standard: laboratory blood glucose test (Yellow Springs Instrument, YSI).

Key outcomes

There was 10% mean absolute relative difference (MARD) between Dexcom G6 and reference blood glucose measurements. Matched pairs from 134 adults and 128 children and young people aged 6 to 17 years were similar. Dexcom G6 values were within 20% of paired blood glucose values in 92.4% and 91.9% of instances in adults and children and young people. Similarly, MARD was 9.9% and 10.1% respectively. The hypoglycaemia alert was correctly activated in 84.4% of instances within 30 minutes of a hypoglycaemic event (defined as below 70 mg/dl). The corresponding false alert rate was 15.6% and missed detection rate was 15.0%. The 10-day sensor survival rate was 87%.

Strengths and limitations

This is a prospective multicentre site study. There were 28 people excluded from an initial 290 people. This was primarily because of a lack of corresponding blood glucose test data. This was study with a short duration (10 days) in a chronic condition. Larger long-term studies could provide more generalisable results. The study was funded by the company.

Study size, design and location

A study retrospectively comparing the accuracy of Dexcom G5 (n=50) compared with Dexcom G6 (n=159) in people with type 1 or type 2 diabetes from 3 previous separate prospective studies (compared with laboratory tested blood glucose values). The study also compared the clinical outcomes in 10,000 people who switched from the G5 to the G6 system. The location was not reported.

Intervention and comparator

Intervention: Dexcom G6.

Comparator: Dexcom G5.

Reference standard: laboratory blood glucose test.

Key outcomes

The G5 system showed slightly better accuracy than the G6 system in terms of MARD compared with the laboratory blood test, but the statistical significance of this result was not tested. The G6 system had a higher utilisation rates over 30 days compared with the G5 system (95.3% compared with 93.8% respectively, p<0.001). Using G6 system was associated with fewer recorded hypoglycaemic glucose values, defined as below 55 mg/dl (3.1 mmol/litre; 0.7% compared with 1.1%, p<0.001).

Strengths and limitations

The accuracy of the Dexcom G6 was assessed using data from previous studies (including Wadwa et al. 2018), so there is population overlap. The patient groups were from different studies, so the performance differences may be because of differences in study design. Propensity score matching was used to adjust for differences between G5 and G6 study populations, but it is unclear whether the method was appropriate. The study did not collect data about changes in diabetes treatment, diet, or exercise patterns. The extent to which the switch from G5 to G6 accounted for the changes is therefore uncertain. The glucose value in this study is defined as a severe hypoglycaemic event. The study was funded by the company.