Strict LBBB

The presence of left bundle branch block (LBBB) is an important predictor of benefit from cardiac resynchronization therapy (CRT), a heart failure device therapy applied in ~75,000 patients per year in the United States. However, approximately one-third of patients diagnosed with LBBB may not benefit from CRT because they do not have activation consistent with a true complete LBBB. To address the need for improved LBBB detection, new “strict” ECG criteria for LBBB, that were indeed shown to better predict benefit from CRT, were recently proposed.

The newly proposed “strict” LBBB criteria include: QRS duration ≥140 milliseconds (men) or ≥130 milliseconds (women), QS- or rS-configurations of the QRS complex in leads V1 and V2, and mid-QRS notching or slurring in ≥2 of leads V1, V2, V5, V6, I and aVL(Strauss DG, et al. Am J Cardiol 2011; https://doi.org/10.1016/j.amjcard.2010.11.010) These new “strict” LBBB criteria are not regularly used and most hospital systems and physicians still use more conventional criteria to diagnose LBBB which could lead to a higher false-positive rate of LBBBs.

This year’s International Society for Computerized Electrocardiology (ISCE) meeting will contain a session on Device Therapies for Heart Failure and Sudden Cardiac Death. One part of the session will focus on diagnosis of LBBB, due to its importance in identifying which patients will benefit from CRT. ISCE and the Telemetric and Holter Warehouse Initiative (THEW-project.org) invite any interested parties (academic, industry, government) to study the “strict” LBBB diagnosis using their existing or novel methods. The data used in this initiative are from the MADIT-CRT trial conducted at the University of Rochester (Rochester, NY). The data will be shared in ISHNE format and available for download in December 2017 with agreement from Boston-Scientific Inc.
This initiative is now closed.

Goals:

The objective of the initiative is to give investigators interested in this field of computerized electrocardiography an opportunity to present the outcome of their automatic algorithm for “strict” LBBB diagnosis as well as their potential algorithm improvements for LBBB detection in heart failure patients included in the MADIT-CRT clinical trial. The 12-lead high-resolution ECGs were recorded before CRT implantation using 24-hours Holter recorders (H12+, Mortara Instruments, Milwaukee, WI, USA) with Mason-Likar lead configuration (the Mortara system provides 10 electrodes and records 8-lead signals [I, II, V1-V6] the other 4 leads are calculated). The first 20-minute ECG signals were recorded while the patients were in a supine position. The sampling frequency was 1 kHz and amplitude resolution 3.75 microVolts.

The investigators can download two datasets of 10-second recordings in ISHNE format and will be expected to electronically submit a set of results in a pre-defined comma separated value (CSV) format. These results will then be processed using a standard reporting system developed at the FDA, and the results will be presented during the ISCE session in April 2018. To facilitate participant’s algorithm development and testing, the initiative provided a training dataset (including ECGs and LBBB labels, N= 300 and a validation dataset (ECGs without LBBB labels, N= 302). The LBBB labels of the training and validation datasets will be released on the THEW website after ISCE 2018 for transparency and to facilitate further research. All ECGs provided in the learning and validation sets have been manually adjudicated for “strict” LBBB presence by 2 independent physicians and differences were resolved by consensus in a tie-break with a third reviewer. In detail, the adjudication process was as follows:

1. A median beat was derived from the 10-second ECG. These were constructed using a 10-second window of sinus rhythm beats.

2. The global QRS duration was determined on the median beat across all 12-leads (i.e. butterfly plot). QRS-onset and QRS-offset annotations were manually adjusted where necessary.

3. Visual determination of QS- or rS-configurations in leads V1 and V2 on the median beat in these leads.

4. Visual determination of mid-QRS notching or slurring in ≥2 of leads V1, V2, V5, V6, I and aVL and their starting position relative to the beginning of the global QRS complex (in ms). A slur or notch was required to begin after the first 40 ms and before 50% of the QRS duration.

Of note: For the manual adjudication, if steps 2 and 3 above were found not to meet strict LBBB criteria the ECG was not adjudicated further (i.e. no notches/slurs were annotated even though they may be present).

Of note, the datasets include other forms of conduction (normal and left or right ventricular conduction delay that are not “strict” LBBB). The initiative will provide labels for whether “strict” LBBB is present or not.

The provided data only includes the ECG signals (10-second strips and medians) and whether the specific subject is male or female. No other subject information form the trial is or will be made available.

Overall the set of data includes: 602 files recorded in the MADIT-CRT clinical trial (~72% men, ~28% women). The dataset includes 10-second 10-electrode 12-lead Mason-Likar ECGs extracted from 20-minutes continuous ECG recorded in supine position.

ISCE LBBB unblided final with comments

**OF NOTE: The specific location of notches/slurs in the individual leads is not required (see table below) but is very much encouraged to maximize the goals and impact of the initiative.