Comparative Benefits and Harms of Basal Insulin Analogues for Type 2 Diabetes: A Systematic Review and Network Meta-analysis.

Background: Basal insulin analogues aim for protracted glycemic control with minimal adverse effects. Purpose: To assess the comparative efficacy and safety of basal insulin analogues for adults with type 2 diabetes mellitus (T2DM). Data Sources: Several databases from inception to April 2018 without language restrictions, ClinicalTrials.gov to April 2018, references of reviews, and meeting abstract books. Study Selection: Randomized trials lasting at least 12 weeks that compared efficacy (change in hemoglobin A1c [HbA1c] level from baseline [primary outcome]; percentage of patients with HbA1c level <7% at end of study and change in body weight [secondary outcomes]) and safety (hypoglycemia) of basal insulin analogues. Data Extraction: Two authors independently extracted data and assessed risk of bias for each outcome. All authors evaluated overall confidence in the evidence. Data Synthesis: Thirty-nine trials (26 195 patients) assessed 10 basal insulin analogues. Low- to very-low-quality evidence indicated that thrice-weekly degludec (Deg-3TW) was inferior to most other regimens for reducing HbA1c level, with mean differences ranging from 0.21% (vs. degludec, 100 U/mL [Deg-100]) to 0.32% (vs. glargine, 300 U/mL [Glar-300]). High- to moderate-quality evidence suggested that detemir had a favorable weight profile versus all comparators, and Glar-300 was associated with less weight gain than glargine, 100 U/mL (Glar-100); Deg-100; degludec, 200 U/mL (Deg-200); Deg-3TW; and LY2963016. Low- and very-low-quality evidence suggested that Deg-100, Deg-200, and Glar-300 were associated with lower incidence of nocturnal hypoglycemia than detemir, Glar-100, LY2963016, and neutral protamine lispro (NPL). Incidence of severe hypoglycemia did not differ among regimens, except NPL, which was associated with increased risk versus Deg-100, detemir, Glar-100, and Glar-300. Limitations: Results are based mostly on indirect comparisons. Confidence in summary estimates is low or very low due to individual-study limitations, imprecision, or inconsistency. Conclusion: Low-quality evidence suggests that basal insulin analogues for T2DM do not substantially differ in their glucose-lowering effect. Low- and very-low-quality evidence suggests some regimens may be associated with lower risk for nocturnal hypoglycemia (Deg-100, Deg-200, and Glar-300) or less weight gain (detemir and Glar-300). Primary Funding Source: None. (PROSPERO: CRD42016037055).