Assessing the Heartfelt Device's Predictive Value in Heart Failure Management: Findings from the Multicenter FOOT Trial (UK)

Abstract

Purpose

This study (NCT04072744) investigates the Heartfelt device's ability to predict heart failure events by monitoring peripheral edema, comparing its accuracy and data reliability against traditional daily weight monitoring. It aims to determine the device's effectiveness in providing early warnings to healthcare providers, patients, and caregivers about impending heart failure decompensations and assesses whether adjustments in alert sensitivity could improve patient outcomes, identifying the most efficient approach for ongoing heart failure management.

Background

The Heartfelt device, a CE-marked device covered by FDA exemption, allows heart failure monitoring at home by using 3D imaging and AI to detect peripheral edema changes autonomously, requiring no patient input to collect data. In Europe it is regulated to send alerts on potential heart failure risks to patients, families, or providers based on its findings.

Methods

The FOOT study enrolled 26 patients from five UK NHS hospitals between February 2020 and June 2022. They used the Heartfelt HF3 device at home, which operated normally but logged data and alerts without informing patients or healthcare teams to prevent bias, as well as Bluetooth scales for daily weight tracking as part of standard care. After six months, clinical records were reviewed for heart failure incidents.

Results

In the cohort using the Heartfelt device, 9 heart failure events were recorded in 8 patients, including 2 outpatient visits, 6 hospital admissions, and one fatality, with the device providing an average lead time of 13 days (6.5-20) for deterioration. For standard care (daily weighing), there was an average of 1.1 days [0.3-4.1] of data each week versus 5.9 days [4.9-6.6] of foot volume data weekly recorded by the Heartfelt device. Before 5 of the hospitalizations and one death, the Heartfelt device issued alerts within the preceding two weeks. Alerts also preceded both outpatient incidents; however, the devices had been persistently alerting approximately weekly for several weeks prior, potentially due to excessively sensitive thresholds, so the outpatient events were counted as missed. In total, 6 out of the 9 heart failure events were deemed to be successfully predicted in advance.

Most patients lacked consistent weighing scale data due to adherence issues despite patients being told to follow the advice of regular weighing during the study. Yet, the weight data available indicated that at least 2 events (1 hospitalization and 1 death) could have been predicted.

Conclusion

The Heartfelt device demonstrates superior capability in detecting peripheral edema indicative of heart failure worsening, offering more reliable data collection than traditional weight monitoring. Its predictive alerts potentially afford clinicians and caregivers an early warning system, although adjustments in sensitivity thresholds may be required to optimize patient care outcomes. The pivotal trial for the device is expected to take place in the US in 2025.