Abstract
Background Head-to-head comparison of clinical effectiveness between dulaglutide and liraglutide in Asia is limited. This study was aimed to assess the real-world comparative effectiveness of dulaglutide versus liraglutide. Methods We conducted a retrospective cohort study by utilizing multi-institutional electronic medical records to identify real-world type 2 diabetes patients treated with dulaglutide or liraglutide during 2016-2018 in Taiwan and followed up until 2019. Effectiveness outcomes were assessed at every three months in the one-year follow-up. Propensity score techniques were applied to enhance between-group comparability. Significant differences in changes of effectiveness outcomes between treatment groups during the follow-up were examined and further analyzed using mixed-model repeated-measures approaches. Results A total of 1,512 subjects receiving dulaglutide and 1,513 subjects receiving liraglutide were identified. At 12 months, significant HbA1c changes from baseline were found in both treatments (dulaglutide: -1.06%, p<0.001; liraglutide: -0.83%, p<0.001), with a significant between-group difference (-0.23%, 95% confidence interval: -0.38 to -0.08%, p<0.01). Both treatments yielded significant declines in weight, alanine aminotransferase level, and estimated glomerular filtration rate from baseline (dulaglutide: -1.14 kg, -3.08 U/L and -2.08 ml/min/1.73 m2, p<0.01; liraglutide: -1.64 kg, -3.65 U/L and -2.33 ml/min/1.73 m2, p<0.001), whereas only dulaglutide yielded a significant systolic blood pressure reduction (-2.47 mmHg, p<0.001). Between-group differences in changes of weight, blood pressure, and liver and renal functions at 12 months were not statistically significant. Conclusions In real-world T2D patients, dulaglutide versus liraglutide was associated with better glycemic control and comparable effects on changes of weight, blood pressure, and liver and renal functions.
Why Does Primary Angioplasty Not Work in Registries? Quantifying the Susceptibility of Real-World Comparative Effectiveness Data to Allocation Bias
