The QT interval prolongation beyond a certain threshold, also called Long QT,
reflects a ventricular repolarization disorder. This condition could be caused by
congenital syndromes or acquired factors like electrolytes impairment and
drugs. A major complication is Torsade-de-Pointes (TdP), and the risk of this
occurrence becomes significantly higher when the heart rate corrected QT
interval (QTc), by dedicated formulae (such as the Bazett one), exceeds 500
msec. TdP can spontaneously terminate or degenerate into ventricular
fibrillation and possibly cause sudden cardiac death. The electrocardiogram
(ECG) is a simple, inexpensive, sensitive and accessible tool that could detect
and assess the risk of TdP. In this context, it is worth considering the
development of a clinical decision support with the ability to integrate the QT
value of the last ECG done, drug prescription and the various aggravating
factors, such as electrolyte disorders, to detect a possible risk of QT
prolongation. The principal goal of this research was to identify factors that
could improve the relevance of the alert emanating from the Electronic Health
Record (EHR) after the algorithm detects a prolongation of the QT interval, and
so, a possible risk of TdP.
Our objective is to standardize and systematize the recognition of the QT
interval prolongation in hospitalized patients, and to participate in the
development, implementation, and evaluation of a new version of this clinical
decision support for all the Departments of the Centre Hospitalier Universitaire
Vaudois (CHUV). This Master’s Thesis is based on real-world clinical data,
where the alerts have been activated, to identify the variables of interest and
build a database which became the starting point for statistical analyses, such
as linear/logistic regression and chi-square. The results will lead to the
development of an updated version of this decision aid, that could be used to
enhance the quality and safety of Long QT patients’ clinical detection and
management. The clinical decision support will be implemented in all the
Departments at the CHUV. Our expected results are the identification of
patients with a Long QT (acquired and congenital one), and those who could
be associated more generally with repolarization disorders in the electronic
clinical documentation in at-risk situations, such as renal and hepatic failure or
polypharmacy. This approach is expected to be able to systematize the
development and implementation of future clinical decision support systems
based on rule engines, or artificial intelligence algorithms.