Efficacy and safety of closed-loop control system for type one diabetes in adolescents a meta analysis – Scientific Reports

Study search and data extraction

An initial search gave 869 articles from the keyword combinations. The included trials were published between 2015 and 2022. After strict screening and quality assessment 11^{37,38,39,40,41,42,43,44,45,46,47} Randomized Controlled Trials (RCT) were selected for the review which were relevant to the search terms and criteria. A total of 570 adolescent patients were included in this study from the selected studies. This included 298 patients who took CLC insulin therapy and 272 patients who had SAP as their insulin therapy. Only 10^{37,38,39,40,41,42,43,44,46,47} articles were included in the meta-analysis after reviewing and accounting for heterogeneity. (Fig. 1). Table 3 provides the summary of the data extracted from the attributes of the included studies.

Characteristics and quality of trials

In relation to the masking of participants and personnel, almost all of the trials were rated at ‘‘low risk of bias’’ (9 of 11 trials, 81.81%); as for attrition bias and reporting bias, almost all the trials were rated at ‘‘low risk of bias,’’ because they reported the complete outcome data (10 out of 11 trials, 90.90%). There were no studies at ‘‘high risk of bias’’ with any issues relating to random sequence generation, allocation concealment and masking of outcome assessment (11 out of 11 trials, 100%). Figure 2 shows the risk of bias summary based on review quality appraisal judgements about each risk of bias item for each included study.

Efficacy: glycemic outcomes during day, night and during strenuous physical activities

The results from the included studies were pooled by unifying the measurement units to mg/dL. Hence all the pooled comparison results in this meta-analysis are in mg/dL.

Mean blood glucose (BG) level: day, night and during strenuous physical activities

The average BG was compared in 9 studies^{37,39,40,41,42,43,44,46}. The day monitoring comparison of the BG level showed [Mean Difference (IV, Random, 95% CI) − 4.33 [− 6.70, − 1.96]]. Pooled studies show [Heterogeneity: Tau² = 6.24; Chi² = 77.30, df = 8 (P < 0.00001); I² = 90%. Test for overall effect: Z = 3.58 (P = 0.0003)]. The night monitoring level of BG was reported by 5 studies^{37,39,42,46,47}, and it was compared. The results showed (Mean Difference (IV, Random, 95% CI) − 16.61 [− 31.68, − 1.54]). Pooled studies show [Heterogeneity: Tau² = 215.07; Chi² = 75.06, df = 4 (P < 0.00001); I² = 95%. Test for overall effect: Z = 2.16 (P = 0.03)]. The Forest plot in Fig. 3a and b are illustrative of these results. Only two studies^37,39 showed the results for glycemic outcome during extreme physical activities like winter sports. The physical activity monitoring comparison of the BG level demonstrates [Mean Difference (IV, Random, 95% CI) − 8.27 [− 19.52, 2.99]]. Pooled studies show [Heterogeneity: Chi² = 2.95, df = 1 (P = 0.09); I² = 66%. Test for overall effect: Z = 1.44 (P = 0.15)]. Figure 3c depicts this observation.

Time in range (TIR): day, night and during strenuous physical activities

Time in range (TIR) for the percentage of time spent in normoglycemia, during the day 70–180 mg/dL was compared in all the 10 studies^{37,38,39,40,41,42,43,44,46,47}. The results showed (Mean Difference (IV, Random, 95% CI) − 13.18 [− 19.18, − 7.17]. Pooled daytime study aggregate show [Heterogeneity: Tau² = 60.11; Chi² = 67.63, df = 7 (P < 0.00001); I² = 90%. Test for overall effect: Z = 4.30 (P < 0.0001)] Excluding 4 studies^40,43,44,46 to due to statistical heterogeneity only four studies^37,38,39,46 which reported the night time monitoring results of TIR were pooled. The comparison meta results show [Mean Difference (IV, Random, 95% CI) − 15.36 [− 26.81, − 3.92]]. Pooled studies for nighttime data shows [Heterogeneity: Tau² = 94.12; Chi² = 16.83, df = 3 (P = 0.0008); I² = 82%.Test for overall effect: Z = 2.63 (P = 0.009)]. The Forest plot in the Fig. 4a and b elaborates these findings. Two studies^37,39 recorded the TIR during strenuous physical activities and the results are as follows show [Mean Difference (IV, Random, 95% CI) − 7.39 [− 17.65, 2.87]]. Pooled studies for nighttime data shows [Heterogeneity: Chi² = 3.53, df = 1 (P = 0.06); I² = 72%. Test for overall effect: Z = 1.41 (P = 0.16)]. Figure 4a–c are illustrative of these results.

(a) Forest plot of comparison: Time in Range TIR-Day. (b) Forest plot of comparison: Time in Range TIR-Night. (c) Forest plot of comparison: Time in Range TIR-Physical activity.

Standard deviation (SD) of glucose variability: day and night

The SD of glucose variability for day time monitoring was compared all the 10 studies^{37,38,39,40,41,42,43,44,46,47} (Mean Difference (IV, Random, 95% CI) − 0.40 [− 0.79, − 0.00]]. Pooled studies for daytime measurements were homogeneous with results as shown [Heterogeneity: Tau² = 0.05; Chi² = 8.09, df = 7 (P = 0.32); I² = 13%. Test for overall effect: Z = 1.98 (P = 0.05)]. The night SD of glucose variability was compared in 5 studies^{38,39,43,45,46} as reported. The results of the pooled studies were (Mean Difference (IV, Random, 95% CI) -0.86 [− 2.67, 0.95] Pooled studies were homogeneous with results as shown [Heterogeneity: Tau² = 1.46; Chi² = 6.28, df = 3 (P = 0.10); I² = 52%. Test for overall effect: Z = 0.93 (P = 0.35)]. Figure 5a and b shows the Forest plot of this stated results. Figure 5a and b are explaining these results. Two studies^39,50 reported the variability as the coefficient of variation and it was converted to standard deviation and pooled. It can be noted that no studies directly reported the standard deviation of glucose variability during physical activities.

(a) Forest plot of comparison: SD of glucose variability-Day. 5(b) Forest plot of comparison: SD of glucose variability-Night.

Safety: adverse effects (AE) outcomes

Hypoglycemic events: day, night and during strenuous physical activities

AEs were compared in 10 studies^{37,38,39,40,41,42,43,44,46,47}, including a total of 570 subjects. After excluding studies to account for statistical heterogeneity, the day time reporting results for hypoglycemic events was pooled in from 9^{37,38,39,41,42,43,44,46,47} of the included studies are (Mean Difference (IV, Random, 95% CI) − 0.54 [− 1.86, 0.79]]. Pooled studies exhibits heterogeneity, which is shown as [Heterogeneity: Tau² = 2.78; Chi² = 58.72, df = 8 (P < 0.00001); I² = 86%. Test for overall effect: Test for overall effect: Z = 0.80 (P = 0.43)]. The 7 studies^{37,38,39,40,43,44,46,47} were pooled in after accounting for heterogeneity, reported the night comparison results for hypoglycemia and the results shows (Mean Difference (IV, Random, 95% CI) 0.04 [− 0.20, 0.27]). Pooled studies show homogenous results which were, [Heterogeneity: Tau² = 0.00; Chi² = 3.29, df = 6 (P = 0.77); I² = 0%. Test for overall effect: Z = 0.30 (P = 0.77)]. Forest plot in Fig. 6a and b are descriptive of these meta results. The pooled results from the two studies^37,39 which included results for hypoglycemia during physical activities are as follows, (Mean Difference (IV, Random, 95% CI) 0.00 [− 0.25, 0.25]). Pooled studies show no heterogeneity, which is shown as [Heterogeneity: Chi² = 0.00, df = 1 (P = 1.00); I² = 0%. Test for overall effect: Z = 0.00 (P = 1.00)]. Figure 6a–c are descriptive of these results.

(a) Forest plot of comparison: Hypoglycemia-Day. (b) Forest plot of comparison: Hypoglycemia-Night. (c) Forest plot of comparison: Hypoglycemia-Physical activity.

Hyperglycemic events: day, night and during strenuous physical activities

Hyperglycemic events were monitored to assess the AEs of both the insulin delivery systems under comparison. Six studies^{37,38,39,40,41,43} reported day time results for hyperglycemic events and the pooled results are [Mean Difference (IV, Random, 95% CI) − 0.48 [− 2.62, 1.65]]. Pooled studies show [Heterogeneity: Tau² = 4.16; Chi² = 31.45, df = 5 (P < 0.00001); I² = 84%. Test for overall effect: Z = 0.44 (P = 0.66)]. From the six studies^{37,38,39,40,41,43} which reported the night comparison results for hyperglycemia were pooled and the results shows [Mean Difference (IV, Random, 95% CI) − 7.11 [− 12.77, − 1.45]]. Pooled studies show [Heterogeneity: Tau² = 36.21; Chi² = 83.74, df = 5 (P < 0.00001); I² = 94%. Test for overall effect: Z = 2.46 (P = 0.01)]. Hyperglycemia during physical activities were reported by two studies^37,39 which were pooled in this meta-analysis and the results are as follows, [Mean Difference (IV, Random, 95% CI) − 0.00 [− 0.10, 0.10]] Pooled studies show [Heterogeneity: Chi² = 3.45, df = 1 (P = 0.06); I² = 71%. Test for overall effect: Z = 0.04 (P = 0.97)]. Forest plot in Fig. 7a–c are illustrative of these pooled results.

(a) Forest plot of comparison: Hyperglycemia-Day. (b) Forest plot of comparison: Hyperglycemia-Night. (c) Forest plot of comparison: Hyperglycemia-Physical activity.