Type 2 diabetes in the real world vs. clinical trials

About 900 real-world patients with type 2 diabetes were retrospectively selected to have similar features as those from randomized clinical trials. Participants had received either GLP-1 agonists or DDP-4 inhibitors.

After 12 month of observation, authors found that both GLP-1 agonists and DDP-4 inhibitors were less effective in real-world patients than those in controlled clinical trials. In other words, A1c reduction was less then expected by clinical experiments. Analysis indicates that the main reason for such discrepancy was low adherence with real-world therapy.

When outcomes are not as expected, medication compliance may be an issue, which subsequently needs to be addressed.



Diabetes Care


August 2017


Objective: This objective of this study was to estimate and explain the gap between clinical efficacy and real-world (RW) effectiveness of type 2 diabetes medications.

Research design and methods: This mixed-methods quasi-experimental study used retrospective claims (Optum/Humedica) to compare the change in HbA1c of RW patients with type 2 diabetes 12 months after starting a glucagon-like peptide-1 receptor agonist (GLP-1RA) or dipeptidyl peptidase-4 inhibitor (DPP4i) with published findings from randomized controlled trials (RCTs) evaluating these drugs. Selected RW patients were similar to RCT patients, and regression analysis was used in the RW data to adjust for differences between poorly adherent and adherent patients to explain why RCT and RW findings may differ.


RW patients initiating a GLP-1RA (n = 221) or a DPP4i (n = 652) experienced smaller reductions in HbA1c (GLP-1RA: βˆ’0.52%, DPP4i: βˆ’0.51%) than reported in RCTs (βˆ’1.30% from seven GLP-1RA RCTs, n = 2,600; βˆ’0.68% from four DPP4i RCTs, n = 1,889).

Baseline HbA1c, additional medications, and adherence were significant explanatory factors in the RW HbA1c change. Modeled estimates of RCT efficacy (βˆ’1.04% GLP-1RA, βˆ’0.69% DPP4) were within the RCTs’ reported range (GLP-1RA: βˆ’0.84% to βˆ’1.60%; DPP4: βˆ’0.47% to βˆ’0.90%). Poor medication adherence accounted for approximately three-fourths of the gap between RW and expected RCT results (gap = 0.51% GLP1-RA; 0.18% DPP4).


Poor medication adherence is primarily why RW effectiveness is significantly less than RCT efficacy, suggesting an urgent need to effectively address adherence among patients with type 2 diabetes.