Alert-Based ICD Follow-Up: A Model of Digitally Driven Remote Patient Monitoring

Abstract

Objectives: The goal of this study was to test whether continuous automatic remote patient monitoring (RPM) linked to centralized analytics reduces nonactionable in-person patient evaluation (IPE) but maintains detection of at-risk patients and provides actionable notifications.

Background: Conventional ambulatory care requires frequent IPEs. Many encounters are nonactionable, and additional unscheduled IPEs occur.

Methods: Patients receiving implantable cardioverter-defibrillators for Class I/IIa indications were randomized (2:1) to RPM or conventional follow-up, and they were followed up for 15 months. IPEs were conducted every 3 months in the conventional care group but at 3 and 15 months with RPM. Groups were compared for patient retention, nonactionable IPEs, and discovery of at-risk patients during 1 year of exclusive RPM. Frequency and value of RPM alerts were assessed.

Results: Patients enrolled (mean age 63.5 ± 12.8 years; male 71.9%; left ventricular ejection fraction 29.0 ± 10.7%; primary prevention 72.3%; n = 1450) were similar between groups (977 RPM vs. 473 conventional care). Mean follow-up durations were 407 ± 103 days for the RPM group versus 399 ± 111 days for the conventional care group (p = 0.165). Patient attrition to follow-up was 42% greater with conventional care (20.1% [87 of 431]) versus RPM (14.2% [129 of 908]; p = 0.007). Nonactionable IPEs were reduced 81% by RPM (0.7 per patient year) compared with conventional care (3.6 per patient year; p < 0.001) but event discoveries remained similar (2.9 per patient year). In RPM, alert rate was median 1 per patient (interquartile range: 0 to 3) with >50% actionability, indicating low volume but high clinical value. Unscheduled IPE was the basis for discovery of 100% of intercurrent problems in RPM and also 75% in conventional care, indicating limited value of appointment-based follow-up for problem discovery. The number of IPEs needed to discover an actionable event was 8.2 in Conventional, 4.9 in RPM, and 2.1 when alert driven (p < 0.001).

Conclusions: RPM transformed ambulatory care to IPE directed to those patients with clinically actionable events when required. Filtering patient information by digitally driven remote monitoring expends fewer clinic resources while providing a greater yield of actionable interventions. (Lumos-T Safely Reduces Routine Office Device Follow-up [TRUST]; NCT00336284).

Keywords: ambulatory care; digital medicine; follow-up; implantable cardioverter-defibrillator; patient monitoring; remote monitoring.