DCRI Researchers Weigh In On Patient Data-Sharing

July 10, 2019 – The paper outlines potential benefits and challenges for adopting initiatives that would enable data sharing at an individual patient level.

Sharing data across clinical trials and cohort studies is key to continuous knowledge generation and healthcare innovation, according to a recent discussion paper from the National Academy of Medicine (NAM) that features contributions from the DCRI’s Eric Peterson, MD, MPH, FAHA, FACC, and Robert Califf, MD, MACC.

Data sharing is becoming increasingly prevalent in research; for example, all journals that are members of the International Committee of Medical Journal Editors (ICMJE) now required that articles reporting clinical trial results include a data sharing statement that details the study teams’ plans to share their data. The NAM writing group laid out a strategy for how sharing data from clinical trials can be made practical. They envision a learning health system in which knowledge gleaned from patient data leads to real-time improvements.

The paper outlined several examples of organizations that have successfully implemented strategies for data sharing. The DCRI’s Supporting Open Access for Researchers (SOAR) initiative was prominently featured as a pioneer in data sharing. SOAR has facilitated the sharing of clinical trials data from Bristol-Myers Squibb as well as clinical data from Duke University. Since SOAR’s inception in 2013, it has received 57 data requests, and 22 of these to date have been approved and are moving forward.

The paper’s authors see patient-level data sharing as not only a boon for patient care, but also as a positive for both the research and academic community. A data sharing strategy could lead to enhanced transparency in research and increased opportunity for collaboration. The paper also addresses potential barriers that may need to be addressed for successful data sharing strategies to take hold, including the need to protect patient privacy, as well as inconsistencies in data definitions across data sources.   The authors concluded that they feel that the future of data sharing is quite bright.

“With initiatives like SOAR, the DCRI is positioning itself as a thought leader in the data-sharing space,” Peterson said. “This effort is also allowing Duke to be one of the first institutions to freely share our clinical data to support researchers around the world.”

Warfarin Control Measures May Not Accurately Predict Risk in Atrial Fibrillation Patients

July 8, 2019 – The study results suggest that clinical factors may be more useful than international normalized ratio metrics in predicting future risk of bleeding or thrombotic events.

A metric typically used to determine doses of warfarin in patients with atrial fibrillation, international normalized ratio (INR), was once thought to be a potential indicator of future risk levels for bleeding or thrombotic events. However, research published recently in JAMA Cardiology by the DCRI’s Sean Pokorney, MD, MBA, suggests that historical INR values may not be the most accurate predictor of future events.

The study examined patients from the ORBIT-AF registry—5,545 patients for the bleeding analysis and 5,635 patients for the thrombotic event analysis. Data analysis was performed from August 2016 to February 2019.

Traditionally, patients’ risk of bleeding and thrombotic events has been estimated based on clinical factors. However, this study sought to determine whether past measures of INR control factored into this risk independently of clinical factors. It had been previously described that in patients with atrial fibrillation taking vitamin K antagonists, low international normalized ratio (INR) values mean higher risk of thrombotic events, while high INR values may signal higher risk for bleeding events.

“These results suggest that using metrics for warfarin control may not be helpful in predicting patients’ potential for future risk beyond the predictive capacity of risk scores using clinical factors,” Pokorney said. “It is just challenging to predict future bleeding or stroke events for patients on warfarin, since historical INR values are not predictive of future INR values.”

Other DCRI contributors to this study include Laine Thomas, PhD; Jonathan Piccini, MD, MHS; and Eric Peterson, MD, MPH.