Automated Insulin Delivery vs Sensor-Augmented Pump and MDI in Type 1 Diabetes: A Network Meta-analysis of HbA1c, Time-in-Range, and Hypoglycemia

Abstract

Background
Automated insulin delivery (AID) systems promise superior glycaemic control in type 1 diabetes (T1D) versus sensor-augmented pumps (SAP) and multiple daily injections (MDI), but few trials directly compare all three. We conducted a network meta-analysis (NMA) of HbA1c, time-in-range (TIR 70–180 mg/dL), and hypoglycaemia.

Methods
Parallel or crossover randomized controlled trials in children, adolescents, or adults with T1D were eligible if they compared AID, SAP, or MDI for ≥8 weeks and reported HbA1c and/or CGM metrics. Outcomes were analysed as mean differences (MD) for HbA1c and %TIR, and risk/rate ratios for hypoglycaemia. A frequentist random-effects NMA (netmeta, R) estimated relative effects and P-scores; heterogeneity (τ²/I²), transitivity, and incoherence (design-by-treatment, node-splitting) were assessed. Prespecified sensitivity analyses excluded higher-risk studies, crossover designs, and mixed “standard care” comparators.

Results
Thirteen RCTs (N=1,743) met criteria. The network included AID–SAP (k=6), SAP–MDI (k=1; STAR-3), AID–MDI (k=1), and three AID–standard-care links handled in sensitivity analyses. Risk of bias was low-to-moderate with objective outcomes and minimal attrition. Versus SAP, AID reduced HbA1c by ~0.3–0.6 percentage points and increased TIR by ~6.7–16 percentage points (≈1.6–3.8 h/day). SAP lowered HbA1c by ~0.6% versus MDI. AID did not increase time-below-range or severe hypoglycaemia; events were rare in all arms. Findings were consistent across age groups and AID platforms, robust to all sensitivity analyses, and coherent across direct/indirect evidence. P-scores ranked treatments AID > SAP > MDI for HbA1c and TIR.

Conclusions
Across diverse populations, AID provides the greatest overall benefit—substantial TIR gains, modest HbA1c reductions beyond SAP, and no hypoglycaemia penalty—establishing a clear hierarchy (AID > SAP > MDI). Results support prioritizing AID where feasible and inform technology escalation pathways in contemporary T1D care.