(PO-4409) Assessing Fit-for-purpose: Surrogate Endpoints And Clinically Important Outcomes In Japanese Real-world Data Source In Type-2 Diabetes
Monday, September 14, 2020
Jocelyn Ruoyi Wang, Amanda Patrick, Pattra Mattox and Ashley Jaksa
Background: Regulatory and payer decision-makers are increasingly relying on real-world evidence (RWE) to compliment traditionally used randomized-controlled trials (RCT). One challenge of using real-world data to generate RWE is ensuring the data is fit for use and has all the requested data to answer the research question and define the populations, exposure measurements, outcomes, confounders. In type-2 diabetes (T2D), phase 3 trials often use glycemic control as a surrogate endpoint for preventing microvascular complications, but the availability of these key study parameters within real-world data sources remains unexamined.
Objectives: To assess the availability of surrogate endpoints and clinical outcome measurements for type 2 diabetes in a Japanese administrative database.
Methods: Newly-diagnosed T2D patients were identified using ICD-10 diagnostic codes in the Japanese Medical Data Center (JMDC) claims database during 2011-2018. Patients <18 or without continuous enrollment during one-year baseline and one-year follow-up were excluded. Surrogate endpoints (glycemic control, glycated hemoglobin [HbA1c], fasting plasma glucose [FPG]) were assessed at baseline and at one-year follow-up. Patients were followed for direct clinical endpoints (e.g., microvascular complications identified using ICD-10 codes) until death, outcome occurrence, disenrollment, or the end-of-data.
Results: Among 51,559 patients, 62.1% and 60.7% of patients had at least one baseline record of HbA1c and FPG level, respectively. The median change in HbA1c was -0.10 %; FPG median change was -4.00 mg/dL. Baseline and follow-up HbA1c values were missing for 37.9% and 41.7% of T2D patients, respectively. Similar missingness was observed for baseline and follow-up FPG. The commonly reported microvascular complications were identifiable within the dataset. The risk (per 1000 patients) of microvascular complications ranged from 128.0 (neuropathy) to 322.6 (nephropathy).
Conclusions: Several potential longitudinal databases for RWE exist in Japan including JMDC. Our analysis showed the availability of key variables for diagnosis, procedures, laboratory values, and clinical outcomes required to evaluate real-world studies in a T2D population. Despite some incompleteness, the laboratory values recorded in this database were considered high-quality to allow effectiveness evaluations of antiglycemic agents. More broadly, JMDC may also allow the exploration of potential surrogate endpoints for other chronic conditions.