< Back to previous page

Project

Development of decision support systems to manage QT-prolongation in clinical practice.

The risk of QTc-prolongation has become an important issue in drug safety. In rare cases it can lead to serious adverse events like Torsade de Pointes and sudden cardiac death. In the last decades, several drugs have been removed from the market (e.g. cisapride) or restricted in use (e.g. domperidone, (es)citalopram) because of this risk. At the moment, more than 170 drugs are linked with this risk of QTc-prolongation, as defined in the QT-drug lists of CredibleMeds. Furthermore, at lot of other risk factors (e.g. age, female gender, electrolyte disturbances, cardiovascular and other comorbidities, congenital long QT-syndrome) can increase the risk. Due to possible combinations of these factors with additive effect on the QTc-interval, a complex risk estimation is warranted for each individual patient who receives a QTc-prolonging drug.

The overall aim of this PhD-project was to study the risk of QTc-prolongation and Torsade de Pointes in clinical practice and to develop tools to help healthcare professionals deal with this risk. To attain this overall aim, specific research questions were defined and investigated in different types of studies.

In general, the project can be divided in four major parts. In the first part, epidemiological studies in different settings (psychiatry, University Hospitals of Leuven, community pharmacies) were performed to study the use of QTc-prolonging drugs, concurrent risk factors, and the current risk management of QTc-prolongation in these settings. The second part starts with a systematic review which summarized and assessed the evidence for different risk factors for QTc-prolongation. Further, two clinical studies (in University Hospitals of Leuven and in psychiatric hospitals) were set up to investigate the prevalence of QTc-prolongation in patients starting with a(n) (additional) QTc-prolonging drug, and a preliminary risk score was developed to distinguish between low- and high risk patients. In the third part, Belgian pharmacovigilance data and a diagnostic registry of the University Hospitals of Leuven were screened to select cases of Torsade de Pointes and to analyze the characteristics of these cases. Finally, in the fourth part, tools were developed for clinical practice. On the one hand, the risk score was further optimized and validated, and a first version of an algorithm was proposed that can be translated in an advanced clinical decision support system. On the other hand, an e-learning program was developed for community pharmacists, and the impact on knowledge and clinical practice was investigated.

This PhD-thesis starts with a general introduction on QTc-prolongation and Torsade de Pointes and the current guidelines to manage this risk (Chapter 1). The objectives and methods of the project are explained in Chapter 2.

The results of the epidemiological studies are presented in Part 2 of the thesis (Chapter 3: psychiatric hospitals; Chapter 4: university hospital; Chapter 5: community pharmacies). All three studies indicated that QTc-prolonging drugs are often prescribed and combined, even in patients who have other risk factors for QTc-prolongation. On top of that, the current risk management and awareness of QTc-prolongation is very limited in all settings.

Part 3 consists of a systematic review, two clinical studies and the development of a risk score for QTc-prolongation (the RISQ-PATH score). The systematic review (Chapter 6) gives a clear overview of the available evidence for a broad range of risk factors for QTc-prolongation, including demographic factors, cardiovascular factors, other comorbidities, electrolyte disturbances and QTc-prolonging drugs. Very strong evidence was found for hypokalemia, use of diuretics, antiarrhythmic drugs and QTc-prolonging drugs of list 1 of CredibleMeds. Based on these results, the preliminary RISQ-PATH score was developed.

In both clinical studies (Chapter 7: RISQ-PATH study in University Hospitals of Leuven; Chapter 8: QT-study in psychiatry), the mean QTc-intervals did not highly differ between the baseline and follow-up electrocardiograms after the start of a QTc-prolonging drug, but an important number of risk patients did develop QTc-prolongation in a follow-up ECG. The preliminary RISQ-PATH score, with a cut-off of 10 points, was able to distinguish between low- and high-risk patients with a sensitivity of 96% and a negative predictive value of 98%. Consequently, this risk score is promising to identify patients who do not need further ECG follow-up when starting QTc-prolonging drugs.

In Part 4, we focused on cases of Torsade de Pointes. In the University Hospitals of Leuven (Chapter 9), a retrospective analysis of a diagnostic registry was performed using the ICD-9 code ‘427.1 paroxysmal ventricular tachycardia’. From 2011 until 2013, 19 cases of Torsade de Pointes secondary to the acquired long QT-syndrome were detected. This number corresponds with an incidence of 0.16 cases of Torsade de Pointes per thousand patients per year, or extrapolated to all Belgian hospitals, approximately 173 possibly lethal cases of Torsade de Pointes in Belgian hospitals each year. The preliminary RISQ-PATH score was able to predict the detected cases. Secondary, with the help of MedDRA preferred terms, the EudraVigilance database was searched for Belgian cases of Torsade de Pointes (Chapter 10). Over 15 years, only 31 cases specifically coded as Torsade de Pointes were identified. These results underline the clear underreporting of QTc-prolongation and Torsade de Pointes to the national authorities in Belgium.

Finally, in Part 5, tools for clinical practice were developed. In Chapter 11, the preliminary RISQ-PATH score was further optimized, resulting in the RISQ-PATH model, and validated in a large hospital population (training group: >60,000 patients with an electrocardiogram in 2015 in one of the Nexus hospitals; validation group: >28,000 patients with an electrocardiogram in January-April 2016). With multiple logistic regression analysis, the most optimal prediction model could be designed to predict a QTc-interval ≥450(♂)/470(♀)ms. The RISQ-PATH model, with a cut-off probability of 0.035, could predict a prolonged QTc-interval with an area under the ROC-curve of 0.772, a sensitivity of 87% and a specificity of 45%. These results were confirmed in the validation dataset. A first version of an algorithm was proposed that can be incorporated in a clinical decision support system.

As second tool, an e-learning for community pharmacists was developed and implemented in a Flemish pharmacy network (Chapter 12). The e-learning consists of 9 modules, including basic concepts of QTc-prolongation, pharmacovigilance, QTc-prolonging drugs, other risk factors summarized in a risk score, handling drug-drug interactions with an algorithm, communication with physicians and patients, guidelines to plan a local concertation and a library section. The impact and satisfaction of the e-learning was measured with a pre-post knowledge test and a satisfaction questionnaire. The participating pharmacists (response rate of 88.8%) were very satisfied with this training, resulting in a median satisfaction score of 9/10. Furthermore, we found a significant increase in knowledge about this topic immediately after following the e-learning. The pharmacists also indicated that they had more confidence in their own knowledge. In a follow-up study (Chapter 13), we measured the further impact of the e-learning on the activation of drug-drug interaction alerts for QTc-prolongation in the pharmacy software, and the current risk management of QTc-prolongation (by repeating the epidemiological study, see Chapter 5). There was a significant increase in the number of pharmacies in which the drug-drug interaction alerts for QTc-prolonging antibiotics were activated before and after the implementation of the e-learning. Moreover, in comparison with 2014, more actions were taken to deal with this risk (gathering more information of the patient, reviewing the risk factors of the patient, contacting the physician, warning the patient for the risk of QTc-prolongation, replacing one of the concerning QTc-prolonging drugs by an alternative). However, the community pharmacists still have to be encouraged to further improve this risk management.

In the closing part of this PhD-thesis (Part 6), a general discussion of the findings is provided, including methodological considerations and a SWOT-analysis of the RISQ-PATH model. In addition, recommendation for the implementation of the proposed tools (clinical decision support and training) and further research perspectives are discussed.

Date:1 Jan 2013 →  31 Dec 2016
Keywords:QT interval, QT-prolongation, Torsade de Pointes, Sudden cardiac death, Drug safety, Drug interactions, Decision support systems
Disciplines:Biomarker discovery and evaluation, Drug discovery and development, Medicinal products, Pharmaceutics, Pharmacognosy and phytochemistry, Pharmacology, Pharmacotherapy, Toxicology and toxinology, Other pharmaceutical sciences, Cardiac and vascular medicine, Psychiatry and psychotherapy, Nursing, Other paramedical sciences, Clinical and counselling psychology, Other psychology and cognitive sciences
Project type:PhD project