Retain
- Heartinsight algorithm was evaluated in a large heterogeneous cohort of 1,352 patients, integrating more recent generations of medical devices and including patients treated according to the most recent therapeutic recommendations.
- On a median monitoring of 599 days, the algorithm demonstrated a sensitivity of 51.5 % to predict hospitalizations for heart failure, with a rate of false alerts of 0.85 per patient-year and a specificity of 81.4 %.
Methodology
- The Heartinsight algorithm was retrospectively applied to the data of 4 clinical trials, where hospitalization events for worsening heart failure have been evaluated by independent external committees.
- Analysis included 1,352 patients with a class II/III of the New York Heart Associationwithout long -lasting atrial fibrillation and equipped with remote active monitoring devices.
- Patients had to have an implantable cardiac defibrillator double bedroom (or a ” DX ICD ») With atrial dipole floating on the probe of the defibrillator, an activated thoracic impedance function and active remote monitoring.
- The algorithm evaluates daily trends of 7 physiological parameters collected during the previous 90 days, including 24 -hour heart rate, nocturnal heart rate, heart rate variability, atrial tachyarhythmia, ventricular extrasystoles, physical activity and thoracic impedance.
Main results
- During a median follow -up of 599 days, 110 patients (median age 68 years [intervalle interquartile – IIQ – 61-75]75.7 % of men) had a total of 165 hospitalizations for worsening heart failure.
- The estimated sensitivity for the prediction of hospitalizations, determined by generalized estimation equations, was 51.5 % [43,0-59,9 %]with a rate of false alerts of 0.85 per patient-year.
- The median alert time was 34 days (IIQ, 16-78) and the specificity was 81.4 % [80,4-82,4 %].
- The results were checked in multivariate analysis with 2 adjustment covariables and univariate analysis in prespeccified subgroups and no significant difference has been highlighted.
In practice
Despite improvements in treatments, heart failure remains associated with poor prognosis and high levels of expensive unplanned hospitalizations. Early clinical intervention for heart failure events is an essential strategy for delaying the progression of the disease, improving patient results and reducing management costs. Heartinsight is based on alerts designed to predict the high risk of heart failure events by assessing temporal trends in physiological parameters. The data is obtained through remote automatic daily monitoring.
Main limitations
The researchers noted several important limitations. The exclusion of patients due to inclusion/exclusion criteria could be a potential bias source. Another limitation concerns differences in basic drugs over time, as only a minority of patients has received medical treatment meeting the most recent recommendations. There was also a certain inconsistency in the declaration of basic data, such as comorbidities and drugs, between the trials included. In addition, highly selected heart failure events do not fully reflect the real situation, as secondary heart failure events and outpatient events have not been taken into account.
Funding and interest links
The study was supported by Biotronik SE & CO KG. Alan Bulava has received fees from consultant and/or subsidies from Abbott, Biotronik and Boston Scientific. Jodo de Sousa received subsidies and/or fees from Microport and Medtronic consultant. Morio Shoda received subsidies from Abbott, Biotronik, Boston Scientific and Medtronic. Laurence Guédon-Moreau has received fees for conferences and/or consultations from Novartis, Microport, Boston Scientific and Medtronic. Tobias Timmel and Sally Thompson Hilpert are employed by Biotronik SE & CO KG.
This article was created using several editorial tools, including AI, as part of the process. The editorial team saw this content before its publication.