LINK-HF2 - Remote Monitoring Analytics in Heart Failure

Heart Failure

heart failure, remote monitoring, predictive analytics, artificial intelligence

Heart failure (HF) is a type of heart disease that leads to need of admissions to the hospital during worsening of symptoms. These admissions are expensive and very inconvenient for patients. The investigators have previously shown that monitoring of patients with a using a small wearable sensor combined with a mathematical model can detect worsening of HF before the patient needs medical care. In this study the investigators will test whether the remote monitoring and prediction of HF worsening can be used to find out when patients are at risk, change their treatment and avoid a hospitalization. The study will enroll 240 Veterans with HF and randomly assign half of them to monitoring and communication of the information on HF worsening to their medical teams. The investigators hope to find our how to best use this approach in routine care of HF. The investigators also plan to determine if this approach will indeed led to less admissions to the hospital among these patients, shorter hospital stays and better quality of life.

Heart failure (HF) represents a major health burden, with 80% of the HF health care costs attributable to hospitalizations. In a pilot multicenter study funded by the VA Center for Innovation, the investigators demonstrated that multivariate physiological telemetry using a small wearable sensor has a high compliance rate and provides accurate early detection of impending readmission for HF. In this study the investigators will implement non-invasive remote monitoring within the VA system and perform a feasibility evaluation of the intervention and its programmatic effectiveness after implementation. Our hypothesis is that the implementation of this program will be feasible and acceptable to clinicians working in VA HF clinics. The investigators also hypothesize that algorithmic response to an alert generated by the predictive algorithm using a continuous stream of remote monitoring data will be feasible and provide the basis for further testing of this approach to decrease the risk of hospitalization for HF and improve other key clinical outcomes. The specific aims of our study are: Aim 1. Implement remote monitoring into the clinical workflow of HF care. Aim 1a. Design implementation strategies for non-invasive remote monitoring and algorithmic response to clinical alerts generated by the predictive analytics platform. In HF programs at five VA medical centers, eligible patients will be enrolled at the time of hospital discharge for HF exacerbation and receive a wearable monitor and a smart phone with cellular service. Data continuously uploaded to a secure server will be analyzed by the predictive analytics algorithm and a clinical alert will be generated when physiological derangements correlated with impending HF exacerbation are identified. A clinical response algorithm will provide instructions for management response to the alert, to include medication changes and/or urgent/non-urgent outpatient assessment. The intervention will include electronic health record integration. The investigators will design implementation processes for this program using the integrated Promoting Action on Research Implementation in Health Services (i-PARiHS) framework, adapted for the VA QUERI. The investigators will design 3 phases of implementation: 1) implementation intervention planning through workflow analysis, technology assessments, and recipient/stakeholder interviews; 2) formative evaluation of pilot implementation at two vanguard sites to test initial acceptability, reliability, and equipment performance; and 3) implementation fidelity monitoring by assessing consistency, safety and satisfaction. Aim 1b. Evaluate implementation outcomes, including clinician and patient perceptions and adoption of the use of ambulatory remote monitoring data. The investigators will use both quantitative and qualitative research methods to examine the eight core dimensions of implementation outcomes. Focus groups and semi-structured interviews will be done to assess clinician and patient perceptions of acceptability and feasibility. Adoption behaviors will be tracked including alert response rates and appropriateness of decisions. Fidelity of implementation will be monitored by assessing compliance with all aspects of the study protocol. Penetration and sustainability will be evaluated by assessing variation in implementation outcomes across the five study sites as well as participant perceptions from the qualitative work at the end of the study. Aim 2. Conduct a feasibility study of non-invasive remote monitoring in chronic HF. Aim 2a. Define key characteristics that will inform design of a pivotal trial of non-invasive remote monitoring aimed at reducing rehospitalization and improving quality of life in HF. The investigators will enroll 240 patients hospitalized for HF exacerbation. At enrollment, subjects will undergo 1:1 randomization to intervention or control arm. While all study subjects will use the monitoring device for 90 days after discharge, in the intervention arm, clinicians will be notified of clinical alerts and will follow the response algorithm to modify HF treatment and/or recommend urgent clinic visit/emergency room visit. In the control arm, information from the sensor will be collected, but clinical alerts will not be generated or communicated to providers. The main study outcomes will include the proportion of randomized patients who meet the algorithm's criteria for at least one alert, the proportion of time the remote monitor is in use and functioning properly, HF hospitalization rate, length of hospital stay, and health-related quality of life. Implementation factors identified in Aim 1 will help clarify the results of this aim. Aim 2b. Identify costs associated with implementation and clinical use of non-invasive remote monitoring in HF. Correct classification of costs associated with implementation of non-invasive remote monitoring will set the stage for cost-effectiveness analyses in a future pivotal trial. Recent advances in technology and in machine learning provide an opportunity for processing of new sources of real-time patient-level data to generate clinically actionable information. An important knowledge gap is how to best implement this technology-based approach into clinical practice. Our study addresses this critical question of clinical implementation, and will generate feasibility data for a design of a pivotal clinical trial of non-invasive remote monitoring with predictive analytics during the high-risk period after hospital discharge. This work has potential to result in changes to care of Veterans with HF and other chronic health conditions.

No related papers found yet.

United States