Efficacy of automated insulin delivery systems in people with type 1 diabetes: a systematic review and network meta-analysis of outpatient randomised controlled trials
Anna Stahl‐Pehe, Nafiseh Shokri Mashhadi, Marielle Wirth, Sabrina Schlesinger, Oliver Kuß, Reinhard W. Holl, Christina Baechle, Klaus-D Warz, Jutta Bürger-Büsing, Olaf Spörkel, Joachim Rosenbauer
Abstract
Background: The comparative efficacy of automated insulin delivery (AID) systems and other treatment options for type 1 diabetes, accounting for the certainty of evidence (CoE), is unknown. Methods: We searched PubMed, EMBASE, the Cochrane Central Register of Controlled Trials and ClinicalTrials.gov and included outpatient randomised controlled trials (RCTs) published until January 8, 2025, in people with type 1 diabetes with a three-week or longer intervention of AID systems (PROSPERO registration number: CRD42023395492). We performed pairwise and network meta-analyses and used the Risk of Bias tool 2 and the Grading of Recommendations Assessment, Development and Evaluation methods to determine the CoE for each outcome. Findings: A total of 46 studies involving seven insulin treatment options and 4113 participants were included, of which 29 and 17 had low and moderate risks of bias, respectively. The intervention AID systems, including the hybrid closed-loop (HCL), advanced HCL (AHCL) and full closed-loop (FCL) systems, were evaluated in 20, 25 and 1 studies, respectively. The network meta-analysis did not indicate global inconsistencies but did indicate global publication bias for all glycaemic outcomes. The CoE varied between very low and high, depending on the treatment and outcome under consideration. Compared with pump therapy, the percentage of time in the range 70-180 mg/dl was greater with AID use (HCL: 19.7% [95% confidence interval 13.2%; 26.1%], moderate CoE; AHCL: 24.1% [18.2%; 29.9%], moderate CoE; FCL: 25.5% [11.1%; 39.9%], high CoE). Compared with pump therapy, the percentage of time above 180 mg/dl and 250 mg/dl was lower with AHCL, on average, by 19.6% (14.0%; 25.1%), moderate CoE, and 14.8% (8.8%; 20.8%), moderate CoE, respectively. The CoE was very uncertain regarding the overall effect of AID systems on the percentage of time below 70 mg/dl and 54 mg/dl and the HbA1c. Interpretation: AID systems improve glycaemic outcomes to varying degrees and with varying CoE. Funding: German Federal Ministry of Education and Research (BMBF; grant 01KG2203).