Evidence review

Clinical and technical evidence

Regulatory bodies

A search of the Medicines and Healthcare Products Regulatory Agency website revealed no manufacturer Field Safety Notices or Medical Device Alerts for this device.

A search of the US Food and Drug Administration (FDA) database: Manufacturer and User Device Facility Experience (MAUDE) between January 2006 and January 2016 found 461 medical device reports relating to 'Medtronic CareLink', most of which appear to be related to a similarly named device that is not part of the CareLink network service. No patients were harmed by the device problems.

It should be noted that the MAUDE database is a passive surveillance system and potentially includes incomplete, inaccurate, untimely, unverified or biased data. The incidence of an event cannot be determined from this reporting system alone due to potential under-reporting of events and lack of information about frequency of device use (FDA, 2015).

Clinical evidence

Table 1 Summary of primary results from selected studies

Study

Study details

Results

Summary of findings

Al-Khatib et al. (2010)

Prospective, randomised study

Single centre (USA)

Remote monitoring (CareLink; n=76) at 3‑month intervals for 12 months with a telephone call at 6 months and a face-to-face follow‑up visit at 12 months.

Control (n=75) patients attended face-to-face follow‑ups at 3‑month intervals for 12 months, unless there was a device-related issue.

Median age (years): 63 CareLink; 63 control.

Rate of composite hospitalisations, emergency room visits and unscheduled visits to the EP clinic: CareLink 32%; control 34% (p=0.77).

Number of hospitalisations: CareLink 23%; control 24% (p=0.88).

Number of emergency room visits for cardiac cause: CareLink 7%; control 5% (p=0.74).

Number of unscheduled visits to the EP clinic : CareLink 7%; control 7% (p=0.98).

There was no significant difference in the composite of cardiovascular hospitalisations, emergency room visits for a cardiac cause, and unscheduled visits to the EP clinic for device-related issues at 1 year. QoL and patient satisfaction were significantly better in the control arm than in the remote arm at 6 months but not at 12 months.

Boriani et al. (2013).

Prospective, RCT

Multicentre, 32 centres from 6 countries (France, Hungary, Israel, Italy, Spain and Switzerland)

Remote monitoring (CareLink; n=76) at 4 and 12 months with activation of automatic alerts. Face-to-face follow‑ups were scheduled at baseline and at 8 months.

Control (n=72) patients attended face-to-face follow‑ups at baseline and every 4 months for 12 months.

Audible alerts for device integrity issues and VF detection were enabled for both groups.

Mean age (years): 68 (CareLink); 67 (control).

Median delay between alert triggering to event review (days; 25th–75th percentile): CareLink 3 (1–10); control 37 (14–71; p<0.001)

The median delay from device-detected events to clinical decisions was considerably and significantly shorter in the CareLink group compared with the control group. In-hospital visits were significantly reduced in the CareLink group. The annual rate of all-cause hospitalisations per patient did not differ between the 2 groups.

Crossley et al. (2009)

Prospective, randomised, parallel, unblinded, open-label clinical trial

Multicentre, 50 centres (USA)

Remote monitoring (CareLink; n=602) at 3, 6 and 9 months and 1 face-to-face follow‑up visit at 12 months.

Control (n=295) patients had transtelephonic monitoring at 2, 4, 8 and 10 months and 1 face-to-face follow‑up visit at 12 months. Patients with dual-chamber pacemakers had a face-to-face follow‑up visit at 6 months.

Mean age (years): 68 (CareLink); 69 (control).

Median time to first CAE (months): CareLink 4.9; control 6.3 (p<0.0001).

Remote pacemaker interrogation follow‑up using the CareLink network service detected CAEs that were potentially important more quickly and more frequently than transtelephonic rhythm strip recordings.

Crossley et al. (2011)

Prospective and randomised

Multicentre, 136 centres (USA)

Remote monitoring (CareLink; n=1014) at 3, 6, 9 and 12 months with face-to-face follow‑ups at 1 and 15 months post-implantation.

Control (n=983) patients had face-to-face follow‑ups at 3, 6, 9 and 12 months.

All audible patient alerts were disabled, except for those associated with lead and device integrity.

Mean age (years): 65.2 (CareLink); 64.9 (control).

Median time from an event to clinical decision (days): CareLink 4.6; control 22 (p<0.001).

Wireless remote monitoring with automatic clinician alerts compared with standard face-to-face follow‑up significantly reduced the time to a clinical decision in response to clinical events. Wireless monitoring was associated with a significant reduction in mean length of CV-related hospital stay.

de Ruvo et al. (2015)

Prospective, non-randomised study

Single centre (Italy)

Remote monitoring with 4 different devices (n=211) with office face-to-face follow‑ups at 1 and 12 months after implantation:

CareLink (n=65); BIOTRONIK Home Monitoring (BHM; n=61); Boston Latitude (LAT; 49); St. Jude Merlin (SJM; 36).

Mean age (years): 70 (CareLink); 70 (BHM); 66 (LAT); 67 (SJM).

Event notification through remote monitoring: CareLink 46/46; BHM 62/69; LAT 33/34; SJM 38/40.

False positive remote monitoring detected events: CareLink 11; BHM 1; LAT 4; SJM 0.

False negative remote monitoring-detected events: CareLink 0; BHM 7; LAT 1; SJM 2 (p≤0.008 after Bonferroni correction).

Actionable events detected by remote monitoring: CareLink 12/14; BHM 31/34; LAT 20/24; SJM 6/8.

Although all remote monitoring systems effectively detected major events, daily transmission (using BHM) was independently associated with an increased probability of event detection compared with periodic transmission systems. The CareLink network service had fewer false negatives when compared with other home monitoring devices and had 100% event notification.

Landolina et al. (2012)

Prospective, randomised, open, study

Multicentre, 6 centres (Italy)

Remote monitoring (CareLink; n=99) with audible alerts disabled, at 4 and 12 months with face-to-face follow‑ups at 8 and 16 months.

Control (n=101) patients had standard management consisting of scheduled visits at 4, 8, 12 and 16 months and patient response to audible alerts.

Mean age (years): 66 (CareLink), 69 (control).

All emergency department and urgent face-to-face follow‑ups: CareLink 75 (0.59 events per year); control 117 (0.93 events per year; IRR 0.65 95% CI 0.49–0.88; p=0.005).

Remote monitoring reduced emergency department or urgent face-to-face follow‑up and, in general, total healthcare use by patients with HF with modern ICD/CRT‑Ds. Compared with standard follow‑up through face-to-face follow‑up and audible ICD alerts, remote monitoring resulted in increased efficiency for healthcare providers and improved quality of care for patients.

Luthje et al. (2015)

Prospective, pilot randomised study

Single centre (Germany)

Remote monitoring (CareLink; n=87) with OptiVol audible alerts turned off.

Control (n=89) patients had face-to-face follow‑up every 3 months with OptiVol audible alerts turned off.

Mean age (years): CareLink 66; control 65.9.

Patients hospitalised for worsened HF during follow‑up: CareLink 20; control 22.

Mean number of emergency department visits: CareLink 0.10±0.25; control 0.10±0.23.

Mean number of urgent care visits: CareLink 0.30±0.50; control 0.10±0.30 (p=0.0332)

Total number of patients having ICD shocks: CareLink 15%; control 11%.

CareLink with fluid monitoring had no significant effect on HF-related hospitalisations, ICD shocks or mortality.

Abbreviations: CAE, clinically actionable event; CV, cardiovascular; CI, confidence interval; EP, electrophysiology; HF, heart failure; ICD, implantable cardioverter defibrillator; CRT‑D, cardiac resynchronisation therapy with a defibrillator; IRR, incident rate ratio; RCT, randomised controlled trial; QoL, quality of life; VF, ventricular fibrillation.

Recent and ongoing studies

REM-HF is a multicentre randomised controlled trial (RCT) comparing routine downloads with weekly downloads from 3 remote monitoring systems: CareLink (Medtronic), Merlin@home (St Jude's Medical) and Latitude (Boston Scientific). The study protocol has been published (Morgan et al. 2014) and recruitment has been completed; results are expected in the near future.

Costs and resource consequences

A cost-utility analysis based on the EVOLVO study (Landolina et al. 2012) has been carried out (Zanaboni et al. 2013). The EVOLVO study was a multicentre RCT in Italy, which compared a remote monitoring arm (CareLink network service) with a standard arm (face-to-face follow‑up). The mean annual costs were €1,962.78 (£1,485.82) for patients on the home-monitoring arm and €2,130.01 (£1,612.42) for those on the standard arm. Although the home-monitoring arm showed a slight cost-saving, this was not significant (p=0.8). The authors did report a significant difference in the annual cost to the patients and family. The mean annual costs to patients were €291.36 (£220.56) for the home-monitoring arm and €381.34 (£288.67) for the standard arm (p=0.01). A cost-utility analysis on 180 of the patients showed a significant 0.065 quality-adjusted life-year increase (p=0.03) and a non-significant cost-saving of €888.10 (£672.29) per patient (p=0.33) in the home-monitoring arm over 16 months. Although this study was carried out in Italy, the results may be applicable to UK. Euro values have been estimated in pounds using an exchange rate of 1 EUR=0.757 GBP.

The following 2014/15 National Tariff Payment System costs for NHS (2013) cardiology outpatient attendances have been listed for information:

  • first attendance, single professional: £164

  • first attendance, multi-professional: £189

  • follow‑up attendance, single professional: £92

  • follow‑up attendance, multi-professional: £131.

Current reimbursement across the NHS for remote follow‑up varies according to local arrangements.

Strengths and limitations of the evidence

The 7 included studies varied in quality but generally were of a reasonable standard. All included studies were prospective; 6 were randomised and compared CareLink monitoring with a control. In general, patients in the control groups did not have remote monitoring and instead attended face-to-face follow‑up. One study (de Ruvo et al. 2015) compared CareLink with other home monitoring devices. Four studies (Al-Khatib et al. 2010; Boriani et al. 2013; de Ruvo et al. 2015; Luthje et al. 2015) had low patient numbers. The remaining studies included a reasonable number of patients. Patient follow‑ups were relatively short and in most studies were around 12 months. Two studies (Crossley et al. 2011; Luthje et al. 2015) had follow‑ups at 15 months and 1 study (Landolina et al. 2012) had a 16‑month follow‑up. Although these 3 studies had longer follow‑up periods, the follow‑ups were short compared with functioning times for implantable cardioverter defibrillators or cardiac resynchronisation therapy devices, which can be between 6–10 years (NHS Choices).

Three studies (Al-Khatib et al. 2010; de Ruvo et al. 2015; Luthje et al. 2015) were carried out in a single centre, which could decrease selection bias because all patients were randomised to groups unlike some multicentre studies. For example, the study by Boriani et al. (2013) included patients from 32 centres across 6 countries. The large number of study sites coupled with the relatively low patient numbers could lead to participant selection bias, because low numbers of patients were recruited from each study site. However the results from multicentre studies (Boriani et al. 2013; Crossley et al. 2009; Crossley et al. 2011; Landolina et al. 2012) may be more generalisable to clinical practice than those from single-centre studies because data were collected at multiple sites and settings.

None of the included studies were carried out in the UK. Three studies (Al-Khatib et al. 2010; Crossley et al. 2009; Crossley et al. 2011) were done in the USA, so the results may not apply to UK practice. The remaining studies were carried out in Europe, except for Boriani et al. (2013) which included 1 centre in Israel. The results of these European-based studies may be more applicable to the UK.

Funding for 4 studies came from Medtronic (Boriani et al. 2013; Crossley et al. 2009; Crossley et al. 2011; Luthje et al. 2015). One or more authors in 2 studies (Al-Khatib et al. 2010; Luthje et al. 2015) were also previously funded by Medtronic and 1 or more authors in 3 studies (Boriani et al. 2013; Crossley et al. 2009; Landolina et al. 2012) were employed by Medtronic. The study by de Ruvo et al. (2015) had no funding from or affiliations with manufacturers.

Two papers (Crossley et al. 2009; Crossley et al. 2011) did not collect data on key outcomes, such as stroke or heart failure. Data on mortality were presented by Crossley et al. (2011). The paper by Crossley et al. (2011) did not include patients with atrial fibrillation or patients having warfarin (anticoagulation therapy). The results observed in this study may not be seen in the 'real-world' where atrial fibrillation and anticoagulation therapy are both widespread.

The study by Landolina et al. (2012) did not describe what interventions, if any, were made by the clinicians to reduce emergency visits and did not show reduced hospitalisations as a result of using the device. Also, the reduction of emergency department or urgent office visits was balanced by increased additional visits as a result of alerts.