EIJC17 & Dataharvest

The information architecture of evidence-based medicine is broken - whole clinical trials are routinely left unpublished, exposing the medical literature to avoidable bias and exaggeration. When trials are published, the results are often misreported in academic journal articles. The same trials can be found reporting different results in different places, and the information in journal articles has been shown to be less complete than – and discrepant with – other sources like clinical study reports. Furthermore, key documents such as Case Report Forms and Ethics Committee paperwork are often left inaccessible, even though we know they can contain important information. We seek to help address this by building OpenTrials (explorer.opentrials.net) - an open, freely accessible database that can store, or link to, all publicly accessible data and documents on all trials in one place, building user-friendly windows onto this information. We have also built an open Application Programming Interface (API) allowing data scientists, programmers, and other power users to query the data at scale. We hope that placing all the information in one place will support informed decision-making by doctors, researchers, and patients. Additionally we hope that it will change norms surrounding sharing trial information, as it becomes easier to identify missing sources of information. From a investigative and data journalism perspective, there exist many potential stories in the data, including most/least transparent trial sponsors and focusing on particular drug/disease areas.