Consistency Between Network Meta-Analysis and Real-Word Data Analysis: Comparison of Changes in Hemoglobin A1c Levels After Treatment with Metformin Vs the Other Oral Antidiabetics in Japanese Patients with Type 2 Diabetes

OBJECTIVES: Network meta-analysis (NMA) with systematic review of randomized controlled trials is often used to compare multiple interventions. Real-world data (RWD) analysis is also an option to compare them. As differences in methods and data sources may lead to different results, we examined the consistency of results from a published NMA study comparing the outcome, change in hemoglobin (Hb) A1c after ≥12-week treatment, between metformin 1500 mg/day and other oral antidiabetics with results from a RWD analysis.

METHODS: We analyzed a clinical database provided by Medical Data Vision Co., Ltd. Eligibility criteria for the analysis were similar to the NMA: adult patients with type 2 diabetes who remained first-line treatment (combinations of drug types and dosages) for ≥90 days and had HbA1c data to assess the outcome. Treatments used for >100 patients were compared the outcome with metformin 1500 mg/day.

RESULTS: Of 105,096 patients who used any treatment assessed in the NMA and had ≥1 HbA1c data, 19,850 remained on the same treatment for ≥90 days and then 11,096 had HbA1c data to assess the outcome. Of the 36 treatment types, other than metformin 1500 mg/day, in the NMA, 13 types had >100 patients. Metformin 1500 mg/day decreased HbA1c greater than all the 13 types in both NMA and RWD analyses. The difference was significant for both analyses for 6 types, whereas it was significant only for the NMA for 2 types and only for the RWD analysis for 2 types. The second largest decrease of HbA1c was shown in metformin 750 mg/day followed by vildagliptin 100 mg/day in the RWD analysis, whereas it was canagliflozin 100 mg/day followed by vildagliptin 100 mg/day and tofogliflozin 20 mg/day in the NMA.

CONCLUSIONS: The largest effect was observed for metformin 1500 mg/day in both analyses; reasons for the differences will be investigated.