The Clinical Evaluation of Remote Notification to Reduce Time to Clinical Decision (CONNECT) Trial The Value of Remote Monitoring

The Clinical Evaluation of Remote Notification to Reduce Time to Clinical Decision (CONNECT) Trial The Value of Remote Monitoring George H. Crossley, MD President, Mid State Cardiology, a unit of St. Thomas Heart Clinical Professor of Medicine, University of Tennessee College of Medicine ACC Governor, Tennessee

Disclosures CONSULTING FEES/HONORARIA Medtronic Inc Boeringer SPEAKER'S BUREAU Medtronic Sanofi RESEARCH/RESEARCH GRANTS Medtronic St Jude Sanofi

Study Purpose To demonstrate that remote monitoring with automatic clinician notifications reduces the time from a clinical event to a clinical decision in response to arrhythmias, cardiovascular disease progression, and device issues as compared to standard in-office care. Rates of cardiovascular health care utilization (HCU) between treatment groups

Study Design Randomized, multi-center prospective study N = 1,997 newly implanted CRT-D and DR-ICD patients 136 US centers Remote management system vs. standard In-office care Patients followed remotely for 12 months (Remote Arm) Patient signs Informed Consent/HIPAA Authorization implanted with a study device and randomized Remote Arm 1014 pts Enrollment 1 Month Office F/U In office Arm 983 pts Enrollment 1 Month Office F/U 3 Month Remote F/U 3 Month Office F/U 6 Month Remote F/U 6 Month Office F/U 9 Month Remote F/U 9 Month Office F/U 12 Month Remote F/U 12 Month Office F/U 15 Month Office F/U 15 Month Office F/U Crossley G, Boyle A, Vitense H, Sherfesee L, Mead RH. Trial design of the clinical evaluation of remote notification to reduce time to clinical decision: the Clinical evaluation Of remote NotificatioN to reduce Time to clinical decision (CONNECT) study. Am Heart J. 2008 Nov;156(5):840-6. Epub 2008 Sep 11.

Required Study Programming Remote In office Medtronic CareLink Home Monitor Provided Not Provided Clinical Management Alerts AT/AF Burden Automatic Clinician Alert, 12 hrs/day Off Fast V. Rate during AT/AF Automatic Clinician Alert, 120 bpm x 6 hrs AT/AF /day Off Number of Shocks Delivered Automatic Clinician Alert, 2 Shocks Delivered Off All Therapies Exhausted in a Zone Automatic Clinician Alert Off Lead / Device Integrity Alerts Lead Impedance Out of Range Automatic Clinician and Audible Patient Alert Audible Patient Alert VF Detection / Therapy Off Automatic Clinician and Audible Patient Alert Audible Patient Alert Low Battery Voltage Automatic Clinician and Audible Patient Alert Audible Patient Alert Excessive Charge Time Automatic Clinician and Audible Patient Alert Audible Patient Alert

Required Study Programming Atrial fibrillation Midnight Day 1 Midnight Day 2 Midnight Day 3 Atrial fibrillation Midnight Day 1 Midnight Day 2 Midnight Day 3

Study Methods All events that did or would have triggered alerts if device programmed accordingly included Events that triggered alerts: the center logged date of clinical decision Events that did not trigger alerts: date of decision was date of first device interrogation following event Time to decision determined for each event, and for each subject with an event, these times were averaged Due to skewness of data, nonparametric test used to compare time to decision per patient between arms For health care utilization, multiple events proportional hazards models used to compare rates of each of the following between arms: Cardiovascular hospitalizations ED visits Unscheduled clinic visits, including urgent care visits

Study Demographics Patient Characteristics Remote (n=1014) In office (n=983) Male 70.5% 71.7% Age (years) 65.2 ± 12.4 64.9 ± 11.9 CRT D 36.4% 35.3% LVEF (%) 28.6 ± 10.0 29.2 ± 10.3 NYHA No HF Class I Class II Class III Class IV 5.3% 3.9% 40.9% 48.5% 1.5% 6.7% 4.7% 39.5% 47.5% 1.5%

Primary Endpoint Time from event to clinical decision in the Remote Arm was significantly shorter than in the In office Arm (p<0.001) Event to Clinical Decision (median time) (per patient with at least one event) Median time in the Remote arm was 4.6 days vs. 22 days in the In office arm Note: Data includes events for patients who crossed over, were noncompliant or had alerts occur prior to home monitor setup

Time from Event to Decision by Alert Type (median days) Device Event No. of Events (No. of Patients) No. of Days from Event Onset To Clinical Decision Median (Interquartile Range) Remote In office Remote In office AT/AF burden at least 12 hrs 437 (107) 280 (105) 3 (1, 15) 24 (7, 57) Fast V rate at least 120 bpm during at least 6 hrs AT/AF 41 (26) 47 (37) 4 (2, 13) 23 (5, 40) At least 2 shocks delivered in an episode 44 (35) 32 (23) 0 (0, 1.5) 0 (0, 2) Lead impedances out of range 26 (18) 12 (6) 0 (0, 9) 17 (5.5, 45) All therapies in a zone exhausted for an episode 16 (12) 11 (6) 0 (0, 1) 9 (0, 36) VF detection/therapy off 10 (10) 8 (8) 0 (0, 0) 0 (0, 84) Low battery 1 (1) 1 (1) 30 0 Overall 575 (172) 391 (145) 3 (0, 13) 20 (4, 52)

Results of Clinician Alert Transmissions Clincian alert unable to transmit due to other reasons (patient not home, monitor not plugged in, etc.) 9% Clinician alert unable to transmit due to home monitor not set up 25% (Remote Arm) Successful clinician alert transmission & device data viewed prior to clinicial decision 41% Patient seen inoffice prior to clincian alert transmission 13% Successful clinician alert transmission but decision made prior to viewing data 12%

Clinician Alert Transmissions

Clinic Visits (Scheduled and Unscheduled) By replacing routine clinic visits with remote monitoring, the observed rate of total clinic visits per patient year was Remote (3.92) vs. In office (6.27)

Health Care Utilization Visits by Treatment Arm * Includes Urgent Care Visits

Impact of Remote Management This study showed the Remote Arm had significantly shorter hospitalization length of stays than In office Arm (p=0.002) Remote Arm = 3.3 days per hospitalization In office Arm = 4 days per hospitalization Mean reduction 18% Estimated savings per hospitalization $1,659* (p = 0.002) * Estimated using the Medicare Limited Data Set Standard Analytic Files from 2002 2007

Conclusions In this study monitoring patients remotely with automatic clinician alerts showed: A significant reduction in time from onset of events to clinical decisions in response to arrhythmias, and device issues Replacement of routine in clinic visits with remote transmissions did not significantly increase other health care utilizations (cardiovascular hospitalizations, emergency department, and unscheduled clinic visits) A significant reduction in mean length of stay per cardiovascular hospitalization