(Ultra-)short-acting insulin analogues for adults with type 1 diabetes mellitus on multiple daily injections: a network meta-analysis.

rationaleUltra-short-acting insulin analogues have the potential to achieve clinical benefits for people with type 1 diabetes, but their relative effects in relation to regular human insulin remain unclear.

objectivesPrimary objective: To compare the effects of long-term treatment with (ultra-)short-acting insulin analogues to regular human insulin or another (ultra-)short-acting insulin analogue in people with type 1 diabetes on multiple daily injections through a network meta-analysis. SECONDARY

objectiveTo obtain an estimate of the relative ranking of these (ultra-)short-acting insulin analogues compared to regular human insulin. SEARCH

methodsWe searched CENTRAL, MEDLINE, Web of Science, the WHO International Clinical Trials Registry Platform and ClinicalTrials.gov from inception to 14 May 2025, without language restrictions. ELIGIBILITY CRITERIA: We included all randomised controlled trials (RCTs) with an intervention duration of at least 24 weeks that compared (ultra-)short-acting insulin analogues with regular human insulins in non-pregnant adults with type 1 diabetes on multiple daily injections. OUTCOMES: Critical outcomes included glycaemic control, hypoglycaemia, diabetic ketoacidosis, quality of life and adherence to treatment. RISK OF BIAS: We evaluated studies with the Cochrane risk of bias tool (RoB 2). SYNTHESIS

methodsWe performed statistical analyses using a frequentist network meta-analysis to compare estimates for the critical outcomes, with regular human insulin as the reference treatment, and a random-effects model for other outcomes and pairwise comparisons. We rated the certainty of the evidence using the CINeMA framework. INCLUDED STUDIES: We included 15 RCTs with 6335 participants evaluating insulin aspart, lispro, glulisine, ultra-rapid lispro (URLi) and fast-acting insulin aspart (FIAsp). The duration of interventions ranged from 24 to 52 weeks. Participants had a mean age of 36.4 years and a mean HbA1c of 8.1%. Participants had different basal insulin regimens. Thirteen studies were funded by the pharmaceutical industry. SYNTHESIS OF

resultsNetwork meta-analysis: (Ultra-)short-acting insulin analogues versus regular human insulin (RHI) Glycaemic control: haemoglobin A1c (HbA1c) Based on low-certainty evidence from 14 studies with 6039 participants, all five (ultra-)short-acting insulin analogues may result in little to no difference in HbA1c compared to RHI (mean difference (MD): URLi -0.25%, 95% confidence interval (CI) -0.51% to 0.00%; lispro -0.22%, 95% CI -0.46% to 0.02%; glulisine -0.21%, 95% CI -0.48% to 0.06%; FIAsp -0.17%, 95% CI -0.28% to -0.05%; aspart -0.14%, 95% CI -0.21% to -0.06%). Glycaemic control: fasting blood glucose (FBG) Based on extremely low-certainty evidence from four studies with 2737 participants, we are uncertain about the effects of aspart and FIAsp compared to RHI (MD: FIAsp 0.69 mmol/L, 95% CI -0.31 mmol/L to 1.69 mmol/L; aspart 0.80 mmol/L, 95% CI -0.01 mmol/L to 1.61 mmol/L). Lispro, glulisine and URLi could not be assessed in relation to RHI. Mild/moderate (non-severe) hypoglycaemia Based on very low-certainty evidence from six studies with 3967 participants, we are uncertain about the effects of aspart and FIAsp compared to RHI (risk ratio (RR): aspart 1.05, 95% CI 0.97 to 1.13; FIAsp 1.04, 95% CI 0.96 to 1.12). Lispro, glulisine and URLi could not be assessed in relation to RHI. Severe/serious hypoglycaemia Based on low- to very low-certainty evidence from 11 studies with 5390 participants, URLi may result in a reduction in severe/serious hypoglycaemia compared to RHI (RR: URLi 0.80, 95% CI 0.36 to 1.77), but the evidence is very uncertain for aspart and FIAsp (RR: aspart 0.89, 95% CI 0.73 to 1.09; FIAsp 0.79, 95% CI 0.55 to 1.15). Glulisine may result in an increase in severe/serious hypoglycaemia compared to RHI (RR glulisine 1.54, 95% CI 0.46 to 5.19). In contrast, lispro may result in little to no difference compared to RHI (RR: lispro 1.01, 95% CI 0.51 to 1.98). Mild/moderate nocturnal hypoglycaemia This outcome could not be evaluated with network meta-analysis. Severe/serious nocturnal hypoglycaemia Based on very low-certainty evidence from four studies with 3229 participants, we are uncertain about the effects of aspart and FIAsp compared to RHI (RR: aspart 0.63, 95% CI 0.46 to 0.88; FIAsp 0.52, 95% CI 0.26 to 1.04). Lispro, glulisine and URLi could not be assessed in relation to RHI. Diabetic ketoacidosis (DKA) Based on low- to very low-certainty evidence from nine studies with 4599 participants, lispro and URLi may result in a reduction in DKA compared to RHI (odds ratio (OR): lispro 0.74, 95% CI 0.10 to 5.39; URLi 0.74, 95% CI 0.03 to 16.64), whereas FIAsp may result in an increase compared to RHI (OR: FIAsp 1.18, 95% CI 0.12 to 11.24), but the evidence is very uncertain for aspart (OR: aspart 0.80, 95% CI 0.20 to 3.11). Glulisine could not be assessed in relation to RHI. Quality of life This outcome could not be evaluated with network meta-analysis. Adherence to treatment Based on very low-certainty evidence from five studies with 3579 participants, we are uncertain about the effects of aspart and FIAsp compared to RHI (RR: aspart 1.01, 95% CI 0.95 to 1.07; FIAsp 1.01, 95% CI 0.94 to 1.08). Lispro, glulisine and URLi could not be assessed in relation to RHI. AUTHORS'

conclusions(Ultra-)short-acting insulin analogues may result in little to no difference in glycaemic control, non-severe hypoglycaemia and treatment adherence at short-term follow-up, but some agents may reduce severe/serious hypoglycaemia and DKA. We are uncertain about the effects of these insulins on mild to moderate nocturnal hypoglycaemia and quality of life. There were no long-term follow-up data.

fundingWorld Health Organization (WHO). REGISTRATION: Protocol for this update: PROSPERO CRD42024599817 (https://www.crd.york.ac.uk/PROSPERO/view/CRD42024599817); DOI of the previous review: 10.1002/14651858.CD012161.