The New England Journal of Medicine AUGUST 28, 2016

Defibrillator Implantation in Patients with Nonischemic Systolic Heart Failure

Lars Køber, Jens J. Thune, Jens C. Nielsen, et al.

Bottom Line

In this randomized, open-label trial, prophylactic implantable cardioverter-defibrillator (ICD) implantation in patients with symptomatic nonischemic systolic heart failure did not significantly reduce the rate of all-cause mortality compared with usual clinical care.

Key Findings

1. The primary endpoint of all-cause mortality occurred in 21.6% of patients in the ICD group compared to 23.4% in the control group (hazard ratio, 0.87; 95% confidence interval, 0.68 to 1.12; P = 0.28).
2. ICD implantation resulted in a significant reduction in sudden cardiac death (4.3% in the ICD group vs. 8.2% in the control group; hazard ratio, 0.50; 95% confidence interval, 0.31 to 0.82; P = 0.005).
3. Prespecified subgroup analysis suggested a potential survival benefit of ICD implantation in younger patients (age ≤70 years), though the trial was not powered to confirm this as a definitive finding for this subset.

Study Design

Design
RCT
Open-Label
Sample
1,116
Patients
Duration
5.6 yr
Median
Setting
Multicenter, Denmark
Population Patients with symptomatic nonischemic systolic heart failure (left ventricular ejection fraction ≤35%) receiving guideline-directed medical therapy.
Intervention Implantation of an ICD (with or without CRT).
Comparator Usual clinical care (without ICD, with or without CRT).
Outcome All-cause mortality.

Study Limitations

The trial was open-label, which could introduce potential bias, although the primary outcome of all-cause mortality is objective and typically not subject to bias.
A substantial proportion of patients (58%) received cardiac resynchronization therapy (CRT), which may have reduced the absolute risk of arrhythmic death and attenuated the potential mortality benefit of the ICD alone.
The findings may not be generalizable to all patient populations, as the study cohort primarily consisted of patients receiving modern, guideline-directed medical therapy for heart failure.

Clinical Significance

The DANISH trial challenged the routine recommendation for primary prevention ICDs in all patients with nonischemic cardiomyopathy, highlighting that modern medical therapy and CRT have likely altered the risk-benefit profile, necessitating a more personalized approach to patient selection.

Historical Context

Prior to the DANISH trial, ICD indications for nonischemic cardiomyopathy were largely extrapolated from trials involving ischemic heart disease or mixed populations (such as SCD-HeFT). DANISH was the first large-scale, contemporary, randomized trial to specifically investigate the survival benefit of primary prevention ICDs in a dedicated cohort of patients with symptomatic nonischemic heart failure.

Guided Discussion

High-yield insights from every perspective

Med Student
Medical Student

The DANISH trial observed that while sudden cardiac death was significantly reduced by the ICD, all-cause mortality was not. What is the pathophysiological explanation for why preventing a specific cause of death (arrhythmia) might not translate into a survival benefit in patients with nonischemic cardiomyopathy compared to those with ischemic cardiomyopathy?

Key Response

In nonischemic cardiomyopathy, patients are more likely to die from progressive pump failure (heart failure) or non-cardiac comorbidities rather than the ventricular arrhythmias common in ischemic heart disease (scar-mediated re-entry). The 'competing risk' of death from non-arrhythmic causes is higher in this population, meaning that even if the ICD successfully terminates a lethal rhythm, the underlying disease process or other conditions often claim the patient shortly thereafter.

Resident
Resident

A 72-year-old male with nonischemic cardiomyopathy and an LVEF of 30% asks if he should receive an ICD. Based on the subgroup analysis of the DANISH trial, how does patient age modify the expected benefit of prophylactic ICD implantation in this population?

Key Response

The DANISH trial found a significant interaction between age and the benefit of an ICD. Patients younger than 68 years experienced a significant reduction in all-cause mortality (HR 0.64), whereas older patients (≥68 years) saw no survival benefit. For a resident, this highlights the importance of clinical reasoning: in older patients, the competing risks of death from pump failure or age-related comorbidities outweigh the protection offered by an ICD against sudden cardiac death.

Fellow
Fellow

In the DANISH trial, approximately 58% of patients in both arms were treated with Cardiac Resynchronization Therapy (CRT). How does the inclusion of CRT-Pacemaker (CRT-P) and CRT-Defibrillator (CRT-D) patients potentially dilute the observed treatment effect of the ICD in a trial of nonischemic heart failure?

Key Response

CRT itself promotes reverse remodeling, improves LVEF, and reduces both heart failure and sudden cardiac death. By including a large population receiving CRT, the 'baseline' mortality in the control group is lowered, and the substrate for arrhythmia may be reduced. This makes it significantly harder to demonstrate an incremental survival benefit of an ICD (CRT-D vs. CRT-P) on top of modern guideline-directed medical therapy and resynchronization.

Attending
Attending

Given the DANISH results, should we move toward a 'selective' rather than 'blanket' application of primary prevention ICDs in nonischemic patients, and what parameters beyond LVEF should now be prioritized in the shared decision-making process?

Key Response

The trial forces a shift from LVEF-only criteria to a more nuanced risk stratification. Attendings should teach that age, the presence/extent of myocardial fibrosis on Cardiac MRI (Late Gadolinium Enhancement), and the patient's overall frailty are more predictive of benefit than EF alone. In NICM, an ICD is less a 'default' and more a discussion about the 'mode of death'—preventing a quick death in a patient who might otherwise have many productive years ahead.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD
PhD

The DANISH trial utilized a traditional Cox proportional-hazards model for its primary analysis. Critique the use of this model in the presence of 'competing risks' (e.g., death from progressive heart failure vs. sudden cardiac death) and how a Fine-Gray subdistribution hazard model might have altered the interpretation of the device's efficacy.

Key Response

In device trials, competing risks are critical. If a patient dies of pump failure, they can no longer experience sudden cardiac death. The Cox model can overestimate the absolute risk of the event of interest in these scenarios. A Fine-Gray model would more accurately account for these competing events, likely further emphasizing that the absolute benefit of an ICD in the nonischemic population is smaller than previously estimated due to the high incidence of non-arrhythmic mortality.

Journal Editor
Journal Editor

The DANISH trial was an open-label study where clinicians were aware of which patients had an ICD. As a reviewer, what are the primary threats to internal validity regarding 'performance bias' and 'ascertainment bias' in this trial, and how might they have influenced the null result?

Key Response

A reviewer would flag that clinicians might have managed patients in the control group more aggressively (closer follow-up, more frequent medication titration) because they lacked the 'safety net' of an ICD (performance bias). Additionally, the determination of 'sudden cardiac death' is notoriously subjective; knowledge of the presence of a device could influence how a physician adjudicates the cause of death (ascertainment bias), though all-cause mortality remains a robust, objective endpoint.

Guideline Committee
Guideline Committee

How do the results of the DANISH trial reconcile the discrepancy between the 2021 ESC Guidelines and the 2022 AHA/ACC/HFSA Guidelines regarding the strength of recommendation for primary prevention ICDs in nonischemic cardiomyopathy?

Key Response

The DANISH trial prompted the ESC to downgrade the recommendation for ICDs in nonischemic cardiomyopathy from Class I to Class IIa. Conversely, the 2022 AHA/ACC/HFSA guidelines maintained a Class I recommendation (Level of Evidence: A), citing meta-analyses that include older trials (like DEFINITE) which suggest a benefit. The committee must decide if one high-quality, modern trial (DANISH) should override older evidence that was gathered before the widespread use of ARNI, SGLT2i, and CRT.

Clinical Landscape

Noteworthy Related Trials

2002

MADIT-II Trial

n = 1,232 · NEJM

Tested

Prophylactic ICD implantation

Population

Patients with a prior myocardial infarction and an LVEF of 30% or less

Comparator

Conventional medical therapy

Endpoint

All-cause mortality

Key result: Prophylactic implantation of a defibrillator resulted in a 31% reduction in mortality among patients with prior myocardial infarction and reduced LVEF.
2004

COMPANION Trial

n = 1,520 · NEJM

Tested

Cardiac resynchronization therapy (CRT) with or without ICD

Population

Patients with NYHA class III or IV heart failure and a prolonged QRS interval

Comparator

Optimal pharmacological therapy alone

Endpoint

All-cause mortality or hospitalization for heart failure

Key result: CRT with an ICD significantly reduced the risk of death or hospitalization compared to pharmacological therapy alone in advanced heart failure patients.
2005

SCD-HeFT Trial

n = 2,521 · NEJM

Tested

Implantable cardioverter-defibrillator (ICD)

Population

Patients with New York Heart Association class II or III heart failure and an LVEF of 35% or less

Comparator

Placebo or Amiodarone

Endpoint

All-cause mortality

Key result: ICD therapy significantly reduced mortality by 23% compared to placebo in patients with heart failure of both ischemic and nonischemic etiology.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis