This study attempted to identify factors that contributed to prolonged TIR in children and adolescents using CSII. We found that the main problem of patients not achieving the >70% target range included repeated hyperglycemia and high glycemic variability defined by coefficient of variation (CV). Participants achieving TIR >70% predominantly used predictive low glucose suspend systems, maintained adequate interval between insulin delivery and meal consumption, used bolus calculator, and more frequently created electronic reports. Moreover, in this group, we observed lower daily and basal insulin requirements.

Furthermore, disease duration >3 years in our study participants lowered the probability of exceeding 70% TIR, and one additional year of diabetes duration increased the risk by 20%. The diabetes duration in the group was not long (median, 3 and 4 years for study and control groups, respectively) and the whole cohort had good metabolic control as assessed by HbA1c levels (median 7.08%; 54 mmol/mol). The possible reason for this trend is the effect of the patients’ age (median age, 12 years); when they become adolescents, the glycemic control worsens compared to childhood. Adolescents have the highest glycemic variability and poorest metabolic control (especially those aged 13–18 years) [16, 17].

We found no differences between the insulin analogs used by the study participants. There is a huge interest now on faster acting insulin analogs. As demonstrated, faster acting insulin aspart (faster aspart) used in children prolonged TIR (38% vs. 50%) and is more effective than insulin aspart in reducing postprandial hyperglycemia during the first and second hour after consuming a meal [18, 19]. We did not note in our study group that more participants used faster aspart. We also found no influence of the type of insulin pump on the TIR.

Previous studies have not reported differences between steel and teflon infusion sets in their function over 7 days [20]. In this study, we found that teflon cannulas correlated with better TIR. The possible cause of the observed trend is the predominant use of teflon cannulas in our diabetology center, thus making them the first choice for most patients.

The general rule concerning length of use is 2 and 3 days for steel and teflon infusion sets, respectively [21, 22]. Our study participants declared changing their infusion sets regularly (78.2% declared every 3-day exchange), without differences between the two groups.

We recruited participants who used both types of CGM: real-time CGM (RT-CGM) and is-CGM. Some RT-CGMs work with insulin pump and have the following additional features: predictive low glucose suspend (PLGS) or low glucose suspend (LGS), that influence metabolic control. Among our study group, patients with TIR >70% predominantly used the PLGS system (GL3). We noted that not all participants who had the opportunity of using PLGS system took advantage of it. Thirty-four participants (in both groups) used Medtronic G640 insulin pump, but only 28 used compatible CGM system (GL3) with PLGS function. The possible cause of that is dissatisfaction with the CGM system due to inaccurate blood glucose measurements, need for calibration, and/or lack of mobile phone application. We observe that parents having younger children prefer the CGM system with a mobile phone application to manage diabetes remotely as it increases their sense of security. There is high quality evidence that PLGS leads to decreased time spent in hypoglycemia and nocturnal hypoglycemia, with no increase in the mean blood glucose concentration and hyperglycemia episodes [23,24,25].

Participants in the study group were more likely to calculate the insulin dose using “bolus calculator,” an available feature of automated bolus calculation in most insulin pumps. Adult user data indicates that the use of a “bolus calculator” improved HbA1c levels, mean blood glucose levels, and glucose variability [26, 27]. On the other hand, a randomized controlled trial in a pediatric group did not reveal any additional effect of “bolus calculator” use with regard to HbA1c levels, postprandial blood glucose values, or other study outcomes [26].

The International Society for Pediatric and Adolescent Diabetes (ISPAD) recommends carbohydrate counting from the onset of diabetes, because it is correlated with improved glycemic control and quality of life among both adults and adolescents [28]. There are few methods of calculating carbohydrate, but research found no evidence to suggest that one particular method is superior to another [28]. In our diabetology center, during the first hospitalization, patients use 10 g carbohydrate portions and are introduced to carbohydrate counting and insulin dose calculations by using an individualized insulin-to-carbohydrate ratio. It is worth emphasizing that almost all study participants declared that they were counting carbohydrates (95.5%). Some patients in our center count also FPU because those macronutrients (fat and protein) lead to delayed hyperglycemia (up to 3–6 h after the meal) [29]. Usually, patients count that 1 FPU equals 100 kcal of fat or protein and requires the same amount of insulin (as an extended bolus) as 10 g of carbohydrates [29]. Over half (56.4%) of the participants declared that they counted FPU, without any differences between the two groups.

The timing of insulin bolus plays a crucial role in achieving stable glycemic values and long TIR. The recommended insulin timing is 15–20 min before meal consumption [28, 30]. Previous studies revealed that rapid-acting insulin analogs before meals as opposed to after meals reduce postprandial glycemia by almost 30% [30]. We found also that participants in the study group were more disciplined with regard to maintaining a latency period between insulin delivery and meal consumption (72.0%) than participants in the control group.

We observed significantly higher insulin doses for both total daily dose and basal insulin dose among participants with poor metabolic control (TIR ≤70% group). Previous studies clearly indicate that uncontrolled glycemia (chronic hyperglycemia) is a risk factor of insulin resistance [29]. Interestingly, the study participants’ insulin requirements are still being recommended by ISPAD at ranges of 0.7 to 1.0 IU/kg/day [29]. During puberty, the requirements may increase even up to 2 U/kg/day [29]. The study and control group patients required about 0.71 and 0.83 IU/kg/day, respectively.

Appropriate disease self-management is a crucial factor affecting good metabolic control in diabetes. Hence, it was not surprising that creating electronic reports for glycemic trends and insulin requirements using a dedicated platform in a domestic environment was related with longer TIR. Considering recent advancements in diabetes due to the use of technology such as smartphone applications and telemedicine, there is significant opportunity to achieve better patients’ involvement in diabetes self-management and subsequently improve metabolic control and possibly ease the disease burden.