Appendix

Appendix

Contents

Data tables

Table 1: Overview of the Choudhary et al. (2015b) poster

Table 2: Overview of the Danne et al. (2014) study

Table 3: Overview of the De Bock et al. (2014) abstract

Table 1 Overview of the Choudhary et al. (2015b) poster

Study component

Description

Objectives/hypotheses

To evaluate the MiniMed 640G system for usability and hypoglycaemia avoidance.

Study design

Prospective case series study.

Patients used the system for 4 weeks, during which they had 4 phone contacts to assess compliance. Pump and sensor data were collected at baseline and at the end of the study, and treatment satisfaction questionnaires were collected at the end of the study.

Setting

Three centres in Europe (UK, Denmark and Spain).

Inclusion/exclusion criteria

Patients with type 1 diabetes and previous experience with sensor‑augmented pump therapy.

Primary outcomes

Rate of hypoglycaemia prevention.

Statistical methods

Not stated.

Patients included

Twenty-four adults (aged 43.8±12.0 [19–65] years) and 16 children (aged 13.4±2.5 [9–17] years) with type 1 diabetes and previous experience with sensor‑augmented pump therapy.

The mean±SD (range) baseline parameters for all patients were: diabetes duration, 17.2±13.1 years; HbA1c, 7.6±0.9%; and BMI, 23.5±3.9 kg/m2. Female 45%.

Results

There were 2402 evaluable pump suspension events; 2.1 per patient‑day. Of the 2402 suspended events, 2322 were suspend‑before‑low events. In 83.1% of the 2322 suspend‑before‑low events, the SG value did not subsequently reach the pre‑set low limit.

The overall mean±SD duration of an automatic pump suspension event was 56.3±9.5 minutes (median 58.0, IQR 49.2 to 64.2).

The mean±SD lowest SG level after pump suspensions was 70.3±7.1 mg/dl (median 70.8, IQR, 67.3–75.0).

Most evaluable pump suspension events (1645/2402=68.5%) happened between 08:00 to 22:00; 31.5% happened at night.

Patients found the MiniMed 640G system and its automated features easy to use, and felt it helped diabetes management.

There were 4 mild adverse events (2 skin reactions, 1 common cold, and 1 urinary tract infection).

Conclusions

Automatic insulin pump suspension as implemented in the MiniMed 640G system can help patients avoid hypoglycaemia, without increasing hyperglycaemia.

Abbreviations: BMI, body mass index; HbA1c, glycated haemoglobin; IQR, inter‑quartile range; SD, standard deviation; SG, sensor glucose.

Table 2 Overview of the Danne et al. (2014) study

Study component

Description

Objectives/ hypotheses

To test the potential benefits of the predictive algorithm.

Study design

Prospective case series study.

Participants were told to avoid rigorous physical activity (for example, gym activity, swimming, or running) in the 24‑hour interval before hypoglycaemia induction. A variable overnight intravenous infusion of human insulin (0.5 unit/kg of body weight in 48 ml of 0.9% NaCl) was started after dinner to stabilise the fasting glucose level at 110 mg/dl (that is, stable between 90 and 140 mg/dl for at least 1 hour before the beginning of the experiment). The overnight intravenous fluid infusion rate (half isotonic glucose solution if the blood glucose level was below 300 mg/dl; 0.9% NaCl if above 300 mg/dl; both at 60–80 ml/kg/24 hours) and insulin dosing scheme (0.1 iU/kg/hour for glucose above 200 mg/dl; 0.05 iU/kg/hour for glucose of 150–200 mg/dl; 0.025 iU/kg/hour for glucose of 100–150 mg/dl; or no insulin for glucose below 100 mg/dl) depended on capillary blood glucose levels. Each person was given a predictive low glucose management system comprising a Paradigm Veo insulin pump and an Enlite glucose sensor (calibrated during the night and before the onset of exercise) connected to a MiniLink transmitter, with the PLGM algorithm installed on a Blackberry phone (Canada). The pump was pre‑programmed to personal settings without adjustment. A 30 minute predictive horizon with a sensor threshold of 70 or 80 mg/dl was used. The threshold was set at 70 mg/dl for 2 of the experiments (exercise sessions) when hypoglycaemia occurred because of rapidly falling glucose levels during exercise. The remainder had a sensor threshold of 80 mg/dl. Insulin was suspended for up to 120 minutes when sensor glucose was predicted to be at or below the threshold within 30 minutes. Study pump insulin delivery was started 1 hour before the planned exercise was started. After a stable fasting blood glucose level was reached, the intravenous glucose/insulin was stopped in the morning between 06:00 and 10:00. Patients were on an exercise ergometer for a maximum of 4.5 hours or until reference HemoCue blood glucose, checked at least every 15 minutes, reached 80 mg/dl. Under specific circumstances, the exercise session was continued even if the blood glucose value reached <80 mg/dl when the patient had no symptoms and the ''upward trend arrows'' appeared in the display of the insulin pump, indicating the glucose level was increasing. The exercise session consisted of up to 6 cycles of exercises on a stationary bicycle or treadmill, each lasting between 15 and 30 minutes. Each cycle was followed by a rest period of between 5 and 15 minutes. This was followed by an observation period while fasting until 1 of the following occurred: the patient needed a glucose rescue (based on symptoms or a HemoCue blood glucose level of <40 mg/dl); the suspended insulin infusion was restarted after successful PLGM triggering; or the maximum observation time of 4.5 hours was reached. If the person's glucose level did not reach the target range of <80 mg/dl during exercise, they were invited to repeat the experiment after at least a 14‑day interval or withdraw from the study.

Setting

Single-centre in Germany (Diabetes Centre for Children and Adolescents, Hanover).

Inclusion/ exclusion criteria

Eligible patients were aged 12–21 years (inclusive), who had been on CSII for at least 3 months before screening, had type 1 diabetes for more than 12 months, had a total daily insulin requirement of 0.6–1.2 U/kg/day, and had HbA1c levels of >5.8% and ≤12.0% at screening. People could not participate if they had experienced an episode of severe hypoglycaemia within the last 3 months before the experiment.

Patients included

Patients with type 1 diabetes having suspension of insulin on CSII, n=22.

Primary outcomes

Rate of hypoglycaemia prevention.

Statistical methods

Descriptive statistics.

Results

Data from only 16 patients were evaluated because the hypoglycaemic threshold during exercise was not reached in 6 people. In 1 of the patients, the pump did not suspend insulin delivery. Hypoglycaemia was prevented in 12 of the 15 successful experiments when insulin suspension was triggered using the PLGM system. The mean (±SD) SG level at predictive suspension was 92±7 mg/dl, resulting in a post-suspension lowest glucose level (by HemoCue) of 77–22 mg/dl. The suspension lasted for 90±35 minutes (range 30–120), resulting in an SG level at insulin resumption of 97±19 mg/dl.

Conclusions

The PLGM threshold setting of 80 mg/dl with a 30‑minute predictive horizon resulted in successful hypoglycaemia prevention in nearly all of the valid experiments. The fact that no reactive negative effects on glycaemia occurred at the end of suspension emphasises the safety and feasibility of this approach.

Abbreviations: CSII, continuous subcutaneous insulin infusion; HbA1c, glycated haemoglobin; iU, international units; kg, kilogram; mg/dl, milligram/decilitre; ml, millilitre; n, number of patients; NaCl, sodium chloride; PLGM, predictive low glucose management; SD, standard deviation; SG, sensor glucose.

Table 3 Overview of the De Bock et al. (2014) abstract

Study component

Description

Objectives/hypotheses

To investigate the performance of SmartGuard during increased overnight basal insulin delivery.

Study design

Randomised, controlled, crossover trial.

Patients were randomised to intervention or control nights with SmartGuard switched on or off respectively, before crossing over to the opposite arm at least 1 week later. Overnight hypoglycaemia was induced by increasing basal insulin delivery rates by 180%. SmartGuard parameters were set so that the pump would suspend insulin delivery when a sensor glucose of <4.4 mmol/l was predicted to occur within 30 minutes.

Setting

Not specified.

Inclusion/exclusion criteria

Inclusion/exclusion criteria were not specified.

Patients included

Patients with type 1 diabetes, aged 13–40 years (n=10).

Primary outcomes

Lowest sensor glucose level (the intervention was stopped if glucose was <2.8 mmol/l).

Statistical methods

Descriptive statistics.

Results

When in the control group (SmartGuard off), the glucose level in 9 of 10 patients fell below the 2.8 mmol/l threshold. Only 2 of 10 patients reached the 2.8 mmol/l threshold when SmartGuard was on. In the other 8 patients in the intervention arm, the mean ± SD lowest glucose sensor level was 3.5 mmol/l (±0.75 mmol/l). For those patients whose glucose did not drop to 2.8 mmol/l, the mean rate of glucose fall before pump suspension was 1.0 mmol/l/hr (±0.52 mmol/l/hr). This was compared with a rate of 1.6 mmol/l/hr (±0.49 mmol/l/hr) in people whose blood glucose level dropped to 2.8 mmol/l. The mean glucose level when the pumps automatically restarted after suspension was 4.5 mmol/l (±1.2 mmol/l).

Conclusions

The authors concluded that SmartGuard has the potential to reduce overnight hypoglycaemia in patients on insulin pump therapy.

Abbreviations: n, number of patients; mmol/l, millimoles per litre; mmol/l/hr, millimoles per litre per hour; SD, standard deviation; SG, sensor glucose.