Table 1 List of applied quality indicators

Hypoglycaemia^a

Represented in different ways. In most studies it is represented as the number or percentage of measurements below a certain BGL value and/or number or percentage of patients who experienced at least one measurement below a predefined BGL value

43 articles [8–13, 15–35, 42–57]

BGLs over time

Represented as mean and/or median BGL values. Each BGL measurement or each patient was regarded the unit of observation. In one study, it was calculated as area under the glucose/time curve divided by total time per admission [10]. Mean or median BGL was also calculated:

• at the end of IIT [52];

• after the target range was achieved [9, 33, 49, 54];

• after stopping IIT [45, 55];

• for patients who had admission BGL above a threshold [10];

• during IIT [27] or during the last 5 hours of IIT [57];

• after 24 hours [15]; or

• 24 hours before, within and after IIT [39] or 48 hours after IIT [11]

37 articles [8–12, 15, 17, 18, 20–22, 24, 26, 27, 29, 30, 32–36, 39, 42–47, 49–52, 54–58]

Measurements in predefined BGL ranges

Represented as the number or percentage of measurements in a predefined BGL range:

• during the study;

• after the target is achieved [27, 49, 55, 56]; or

• at the start of IIT [21]

Each BGL measurement, and in one study each patient [11], was considered the unit of observation

31 articles [9, 11, 12, 17–21, 25–28, 30–33, 35, 36, 42–45, 47, 49–51, 55–57, 59, 60]

Time to capture defined BGL target

Represented as:

• mean and median of time; or

• by Kaplan-Meier curve, as in [17, 20, 23, 25, 49]

In one study two successive BGL measurements in the target range were required before calculating time [20]

25 articles [8, 9, 11, 15–17, 20, 21, 23, 25, 27, 29, 30, 32–34, 43, 45, 46, 49, 52, 53, 55, 56, 59]

Frequency of measurements during the study

Represented as:

• mean or median per patient [11, 16, 17, 23–28];

• mean or median per patient treatment day or days [12, 15, 21, 22, 26, 27, 29–31];

• sampling interval (time) [13, 31–36];

• median frequency per patient-hour [23];

• percentage of patients with more than one measurement in predefined time interval (every 2 hours, and so on) [31];

• frequency overall per day [12, 21]; or

• percentage of time in which at least one measurement per 2 hours was taken [32].

23 articles [11–13, 15–17, 20–36]

BGL at starting IIT

Represented as mean and median BGL

14 articles [9, 11, 13, 16, 21, 27, 29, 33, 46, 52, 54–56, 59]

Protocol compliance

Compare measurement times suggested by protocol with actual times of measurements and/or pump speed during IIT and/or at the time of hypoglycaemic events [9, 21]. In one studya time-motion method was used to measure discrepancy in timing of BGL measurements between protocol and actual [28]

13 articles [9, 11, 13, 14, 16, 21, 28–31, 33, 46, 57]

Time in predefined range

Represented as:

• mean of percentage of time per patient [12, 32, 58];

• median of fraction of time per patient [13, 15, 21, 34];

• median of fraction of time per day [29];

• percentage of time for all patients [14, 17, 21, 25]; or

• percentage of time 24 hours before, within and after trial [39]

12 articles [12–15, 17, 21, 25, 29, 32, 34, 39, 58]

Hyperglycaemic events

Represented as:

• percentage of time >180^b, >250 (severe hyperglycaemia) [16], or between 151 and 200 and >200 (severe hyperglycaemia) [21];

• percentage of patients with at least one measurement per day ≥ 250 and ≥ 200) [22];

• percentage of measurements above 150 [11] or 180 [17–20]; or

• percentage of measurements and patients with at least one BGL above the 180 level for more than 2 hours [13]

9 articles [11, 13, 16–22]

Morning BGLs

Represent as:

• mean BGL around 06:00 hours [43], between 06:00 and 12:00 hours [50], or between 06:00 and 09:00 hours [19];

• mean lowest BGL between 06:00 and 09:00 [43];

• median between 06:00 and 08:00 hours [24] or between 03:00 and 06:00 hours [27]; or

• mean of BGL, but morning time was not mentioned [20]

6 articles [19, 20, 24, 27, 43, 50]

Hyperglycaemic index

Represented as median area between glucose-time curve and upper normal range divided by time per patient during the trial [15, 24], in first 24 hours [23, 25] or in first 48 hours [17]. Upper normal range was 207 [23], 117 [15], 108 [24], 120 [17], and 150 [25]. It was calculated with the same definition but without labeling as hypoglycaemic index [23, 25]

5 articles [15, 17, 23–25]

Time until starting or adjusting IIT

Represented as mean or median of time until starting and/or adjusting IIT [9, 10, 27, 28], or proportion of patients per time until starting IIT [12]. In one study a time-motion method was used [28]

5 articles [9, 10, 12, 27, 28]

Minimum and maximum recorded BGL

Represented as:

• minimum and maximum recorded BGL over all patients [35, 47]; or

• median of minimum and maximum recorded BGL per patients [24] or per patient-day [29]

4 articles [24, 29, 35, 47]

Number of patients with at least one BGL in predefined range

Represented as number and percentage per month [31] or at defined time interval after starting TGC [11] or during the study periods [60]

3 articles [11, 31, 60]

BGL change over time

Represented as:

• speed of BGL change per hour [54]; or

• BGL change in first 24 hours [15]

2 articles [15, 54]

Number of patients who achieved or did not achieve target or predefined range

Represented as number and percentage

2 articles [17, 27]

Number of positive culture

Represented as median (per patient) or rate (per year per patient).

2 article [31, 59]

Target acquisition error

Represented as absolute value and percentage of difference between the target BGL and achieved BGL

2 articles [32, 48]