Author + information
- Suveen Angraal, MBBS,
- Joseph S. Ross, MD, MHS,
- Sanket S. Dhruva, MD,
- Nihar R. Desai, MD, MPH,
- John W. Welsh, BA and
- Harlan M. Krumholz, MD, SM∗ ()
- ↵∗Yale Center for Outcomes Research and Evaluation, 1 Church Street, Suite 200, New Haven, Connecticut 06510
Data sharing is important to maximize what can be learned from clinical trials (1). The DIG (Digitalis Investigation Group) trial is ideal to assess the effects of data sharing. A federally funded trial conducted from 1991 to 1995 and published in 1997, the DIG trial randomized 6,800 people with stable heart failure to digoxin or placebo (2). Individual patient-level data were made available from 2000 to 2002 through the National Heart, Lung, and Blood Institute’s Limited Access Datasets program. In 2009, data were made widely available through the National Institutes of Health’s Biologic Specimen and Data Repository Information Coordinating Center (BioLINCC). We sought to determine the contributions of outside investigators to the trial’s publication yield as a metric for the benefits of data sharing, and compare publications by trial investigators with those of outside investigators. We also assessed the use of various variables collected in the trial to measure the extent to which the data have been harvested.
We searched PubMed and used BioLINCC’s data repository to identify all published studies that used the DIG trial data. We compared the number of studies involving ≥1 trial investigator with those involving only outside investigator from 2002 to 2009 (pre-BioLINCC availability) and 2010 to 2016 (post-BioLINCC availability). To represent the timeline and impact of the studies, each article’s publication date was plotted against the most recently available impact factor of the publishing journal. We calculated citations per article-year by dividing the number of citations of each study by the number of years since its publication. Median citations per article-year (interquartile range [IQR]) were calculated for studies by trial investigators and outside investigators to assess the impact of publications by the 2 groups.
To assess the data utilization from the DIG trial, we identified the variables collected from the data dictionary and included those related to patient demographics, assessment, follow-up, events, and treatment. We then determined which variables were reported in the trial’s original publication (2), and which were reported in subsequent publications. Identifications were made by one author and confirmed by a second author.
After the trial’s main publication in 1997, there were no subsequent publications through 2002. Since 2002 until December 31, 2016, 75 studies have been published, of which 41 (55%) were conducted by outside investigators. Outside investigators published 17 studies from 2002 to 2009, and 24 from 2010 to 2016. Trial investigators published 28 studies from 2002 to 2009 and 6 from 2010 to 2016, but none since 2014 (Figure 1).
Of the 41 studies by outside investigators, 5 were published in journals with an impact factor ≥10. The median citations per article-year were 4.8 (IQR: 2.2 to 7.7). The top 2 cited studies had 652 and 637 citations. Of 34 studies by trial investigators, 7 were published in journals with an impact factor ≥10. The median citations per article-year were 6.9 (IQR: 3.8 to 11.2). The top 2 cited studies had 424 and 338 citations.
We identified 230 distinct variables collected in the DIG trial, of which 58 (25%) were reported in the initial publication. As of December 31, 2016, 149 (65%) variables have been reported in various studies published using the trial data.
In this study, we found that outside investigators have authored a significant proportion of publications from the DIG trial and have been instrumental in harvesting the data. Based on the timeline of publications, making trial data publicly available seems to spur publications by both outside and trial investigator groups. Studies by outside investigators have produced key practice-changing insights; for example, digoxin was found to increase mortality in women but not men (3), and was associated with increasing mortality with higher serum concentration (4).
The impact factors of the journals and the citations among trial investigators and outside investigators were similar, speaking for both the quality and the rigor of both groups of investigators. However, one-third of the variables collected in the trial remain unreported, indicating that there may still be information remaining to be harvested from the DIG trial.
Although our study relied on metrics of publication and citation that may not perfectly measure the importance of publications, it reveals the potential for data sharing to facilitate knowledge generation by enabling investigators who were not part of a trial to make substantive contributions to the ultimate yield of the effort.
Please note: Dr. Desai is funded by grant K12 HS023000-03 from the Agency for Healthcare Research and Quality. Dr. Ross is supported by the Laura and John Arnold Foundation for the Collaboration on Research Integrity and Transparency at Yale and by the Food and Drug Administration for the Center for Excellence in Regulatory Science and Innovation program. The sponsors did not play a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Dr. Ross has received research support from the Blue Cross Blue Shield Association. Drs. Ross, Desai, and Krumholz have received research grant support from Medtronic and Johnson & Johnson (Janssen), through Yale, to develop methods of clinical trial data sharing. Drs. Ross and Krumholz have received research grant support from Medtronic and the Food and Drug Administration, through Yale, to develop methods for post-market surveillance of medical devices. Dr. Krumholz has contracts with the Centers for Medicare & Medicaid Services; has served as the chair of a cardiac scientific advisory board for UnitedHealth; has served as a participant/participant representative of the IBM Watson Health Life Sciences Board; has served on the advisory board for Element Science; has served on the physician advisory board for Aetna; has served on the open trials advisory board for the Laura and John Arnold Foundation through the Center for Open Science; and is the founder of Hugo, a personal health information platform. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- 2017 American College of Cardiology Foundation