The Cardiac Safety Research Consortium enters its second decade: An invitation to participate.

The Cardiac Safety Research Consortium (CSRC), a transparent, public-private partnership established in 2005 as a Critical Path Program and formalized in 2006 under a Memorandum of Understanding between the United States Food and Drug Administration and Duke University, is entering its second decade. Our continuing goal is to advance paradigms for more efficient regulatory science related to the cardiovascular safety of new therapeutics, both in the United States and globally, particularly where such safety questions add burden to innovative research and development. Operationally, CSRC brings together a broad base of stakeholders from academia, industry, and government agencies in a collaborative forum focused on identifying barriers and then creating novel solutions through shared data, expertise, and collaborative research. This white paper provides a brief overview of the Consortium's activities in its first decade and a context for some of our current activities and future directions. The growth and success of the CSRC have been primarily driven by members' active participation and the development of goodwill and trust throughout our membership, which have facilitated novel collaborations across traditionally competitive or contentious stakeholder boundaries. The continued expansion of our base of participating academicians, industry experts, and regulators will define the Consortium's success in our second decade. It is our hope that sharing our endeavors to date will stimulate additional participation in the CSRC and also provide a model for other groups starting to develop similar collaborative forums.

The Cardiac Safety Research Consortium electrocardiogram warehouse: thorough QT database specifications and principles of use for algorithm development and testing.

This document examines the formation, structure, and principles guiding the use of electrocardiogram (ECG) data sets obtained during thorough QT studies that have been derived from the ECG Warehouse of the Cardiac Safety Research Consortium (CSRC). These principles are designed to preserve the fairness and public interest of access to these data, commensurate with the mission of the CSRC. The data sets comprise anonymized XML formatted digitized ECGs and descriptive variables from placebo and positive control arms of individual studies previously submitted on a proprietary basis to the US Food and Drug Administration by pharmaceutical sponsors. Sponsors permit the release of these studies into the public domain through the CSRC on behalf of the Food and Drug Administration's Critical Path Initiative and public health interest. For algorithm research protocols submitted to and approved by CSRC, unblinded "training" ECG data sets are provided for algorithm development and for initial evaluation, whereas separate blinded "testing" data sets are used for formal algorithm evaluation in cooperation with the CSRC according to methods detailed in this document.

Quantitative performance of E-Scribe warehouse in detecting quality issues with digital annotated ECG data from healthy subjects.

The US Food and Drug Administration recommends submission of digital electrocardiograms in the standard HL7 XML format into the electrocardiogram warehouse to support preapproval review of new drug applications. The Food and Drug Administration scrutinizes electrocardiogram quality by viewing the annotated waveforms and scoring electrocardiogram quality by the warehouse algorithms. Part of the Food and Drug Administration warehouse is commercially available to sponsors as the E-Scribe Warehouse. The authors tested the performance of E-Scribe Warehouse algorithms by quantifying electrocardiogram acquisition quality, adherence to QT annotation protocol, and T-wave signal strength in 2 data sets: "reference" (104 digital electrocardiograms from a phase I study with sotalol in 26 healthy subjects with QT annotations by computer-assisted manual adjustment) and "test" (the same electrocardiograms with an intentionally introduced predefined number of quality issues). The E-Scribe Warehouse correctly detected differences between the 2 sets expected from the number and pattern of errors in the "test" set (except for 1 subject with QT misannotated in different leads of serial electrocardiograms) and confirmed the absence of differences where none was expected. E-Scribe Warehouse scores below the threshold value identified individual electrocardiograms with questionable T-wave signal strength. The E-Scribe Warehouse showed satisfactory performance in detecting electrocardiogram quality issues that may impair reliability of QTc assessment in clinical trials in healthy subjects.