A wearable multisensor patch detected signals of patients' heart failure (HF) exacerbations days before hospital readmission, according to the LINK-HF study. Continuous 24-hour monitoring analyzed by a machine learning algorithm detected precursors of hospitalization for HF exacerbation with 76% to 88% sensitivity at 85% specificity, reported researchers led by Josef Stehlik, MD, MPH, of the Veterans Affairs (VA) Salt Lake City Health Care System. Multivariate physiological telemetry from the Vital Connect patch had "predictive accuracy comparable to implanted devices," they concluded in Circulation: Heart Failure. The patch uploaded vital data via smartphone to the PhysIQ cloud analytics platform, which created a personalized baseline model of expected physiological values used to trigger clinical alerts. Median time between initial alert and readmission was 6.5 to 8.5 days (depending on the type of hospitalization and positive window method used), which would be enough time for clinicians to introduce an intervention aimed at reversing the worsening HF, according to Stehlik and colleagues. "A time-to-HF hospitalization analysis also demonstrated a significant divergence between the group of subjects with and without a clinical alert," they added. The first 90 days after hospital discharge carry a marked risk of HF rehospitalization, making it "an opportune period for noninvasive monitoring aimed at identifying patients with incipient HF decompensation," the group wrote. Previous research has focused on using data from implantable devices to predict worsening HF. LINK-HF involved 100 adults discharged from an index HF hospitalization who were monitored 24 hours a day for up to 3 months with the Vital Connect, a disposable multisensor patch placed on the chest. During that time, there were 49 hospitalizations (27 for HF) in 38 patients. Patients were enrolled in the current study at four VA medical centers. All had New York Heart Association functional class II-IV symptoms. Average age was 68.4; 98% were men. About three-quarters had HF with reduced ejection fraction. Participants were instructed to replace the sensor patch once the battery was depleted or the adhesive started to wear off. Monitoring was complete to 30 days for 87% and to the full 90 days for 74%. "The sensor collects continuous ECG waveform, continuous 3-axis accelerometry, skin impedance, skin temperature, and information on activity and posture. Data derived from the primary information include heart rate, heart rate variability, arrhythmia burden, respiratory rate, gross activity, walking, sleep, body tilt, and body posture," the investigators noted. They cautioned that they had excluded from the analysis five events that had not been preceded by sufficient data transfer from the study subjects. "If these data were not missing at random, this could have introduced a bias. If these five events were to be included and considered failed detections due to the lack of a preceding alert, the resulting predictive platform sensitivity would be 72.4% for HF hospitalization and 67.5% for unplanned nontrauma hospitalization, at the selected specificity of 85%," Stehlik's team acknowledged. Other limitations to the study include the lack of formal testing and validation sets and the predominantly male cohort. Next steps include a randomized trial of remote monitoring with versus without alerts communicated to the clinical team, according to the investigators. Source
