#implantable-cardioverter-defibrillator #sudden-cardiac-death #heart-failure #arrhythmia #primary-prevention #secondary-prevention

Implantable Cardioverter-Defibrillator (ICD)

Details

An ICD is an implanted device that monitors the cardiac rhythm and delivers high-energy shocks or antitachycardia pacing (ATP) to terminate ventricular fibrillation and sustained ventricular tachycardia. Modern devices also provide bradycardia pacing and, as CRT-D, biventricular pacing. Indications divide into secondary prevention — after a survived cardiac arrest or sustained VT — and primary prevention — before a first life-threatening event, on structural, functional or genetic risk criteria.

Device types: transvenous ICD (standard); subcutaneous ICD (S-ICD, no intracardiac leads); CRT-D (biventricular pacing plus defibrillation); wearable cardioverter-defibrillator (temporary bridge only).

Frameworks used below: the ESC 2026 heart failure guideline (Class/Level), the 2017 AHA/ACC/HRS ventricular arrhythmia and sudden death guideline, and the 2025 appropriate-use criteria for ICD and CRT (scored A/M/R from 1 to 9). The tables give these side by side because where they diverge, the divergence is usually the clinically interesting part.

Key Facts

Reading the Indications

Three timing gates decide most contested cases: ≥40 days after myocardial infarction, ≥90 days after revascularisation, and ≥3 months of optimal medical therapy before an LVEF-based decision. (sources/icd-crt-auc-2025, rating: high)
Ischaemic and non-ischaemic disease no longer carry the same class in Europe. ESC 2026 keeps ischaemic primary prevention at Class I but downgrades non-ischaemic to IIa, reflecting DANISH's null all-cause mortality result and the competing risk of non-sudden death. Benefit therefore depends on how much non-sudden death competes, and older comorbid patients derive substantially less. (sources/HF-ESC-2026, rating: very high)
Deferring drug uptitration is not free: every three months spent deferring carries a 2–4% arrhythmic death risk at six months and about 7% at one year, which is what justifies waiting three months rather than longer.
AUC scores: A(7–9) appropriate, M(4–6) may be appropriate, R(1–3) rarely appropriate.

Secondary Prevention

Scenario	ESC 2026	AHA 2017	AUC 2025
VF or sustained VT with haemodynamic instability, no reversible cause, survival >1 year, not within 48 h of MI	I A	I B-R	A(8–9)
Stable sustained VT in ischaemic heart disease	—	I B-NR	A(8–9)
Unexplained syncope + ischaemic HD + inducible sustained monomorphic VT at EPS	—	I B-NR	—
Channelopathy (LQTS, SQTS, CPVT, Brugada) after sustained VT/VF	—	I B-NR	A(9)
VF/VT within 48 h of acute MI	—	—	R(2–3)
Reversible cause (electrolytes, drug toxicity, ischaemia that is revascularised)	—	—	R(1–3)

(sources/HF-ESC-2026, rating: very high; sources/va-scd-aha-2017, rating: very high; sources/icd-crt-auc-2025, rating: high)

AVID is the anchor trial: a 27% relative and 7% absolute reduction in 2-year mortality with ICD versus antiarrhythmic drugs. Secondary prevention ICD represents intermediate value ($64,000–100,000 per QALY), and value is best when arrhythmic risk is high and non-arrhythmic mortality low. (sources/va-scd-aha-2017)

Primary Prevention — Ischaemic Cardiomyopathy

Scenario	ESC 2026	AHA 2017	AUC 2025
LVEF ≤35%, NYHA II–III, >40 days post-MI, ≥90 days post-revascularisation, on optimal therapy	I B1	I A	A(8–9)
LVEF ≤30%, NYHA I	—	I A	A(8–9)
LVEF ≤40% + NSVT + inducible sustained VT/VF at EPS	—	I B-R (MUSTT)	A(7–8) at LVEF 31–35%
LVEF 31–35%, no NSVT	—	—	M(4–5)
Within 40 days of MI	III B1	—	R(2–3)

(sources/HF-ESC-2026; sources/va-scd-aha-2017; sources/icd-crt-auc-2025)

Primary prevention in ischaemic disease is high value (<$50,000 per QALY), consistently across MADIT-I, MADIT-II and SCD-HeFT. (sources/va-scd-aha-2017)

Primary Prevention — Non-Ischaemic Cardiomyopathy

Scenario	ESC 2026	AHA 2017	AUC 2025
LVEF ≤35%, NYHA II–III, ≥3 months optimal therapy	IIa B1	I A	A(7–9) under 75; M(5–6) at ≥85
NYHA I, LVEF ≤35%	—	IIb B-R	—
<3 months of GDMT	—	—	R(1–3)
NYHA IV, not a candidate for CRT, LVAD or transplant	III C	III C-EO	R(2)
NYHA IV on the transplant list or an LVAD candidate	—	—	A(7)
LMNA mutation + ≥2 of NSVT, LVEF <45%, non-missense mutation, male sex	—	IIa B-NR	A(7) at LVEF 35–45%; A(8) at ≤35%
LGE-positive with LVEF 36–49%	—	—	M(5–6)

(sources/HF-ESC-2026; sources/va-scd-aha-2017; sources/icd-crt-auc-2025)

Primary Prevention — Disease-Specific and Genetic Conditions

These ratings apply regardless of GDMT duration, because the arrhythmic substrate is intrinsic to the disease.

Condition	Criterion	AUC 2025 (and guideline class where one exists)
Giant-cell myocarditis	LVEF ≤35%	A(7–8)
Cardiac sarcoidosis	LVEF ≤35%; LVEF 36–49% with LGE or prior AV block	A(7); M(6)
Myotonic dystrophy type 1	PR >240 ms or QRS >120 ms; no conduction disease	A(8); M(5)
Chagas cardiomyopathy	LVEF ≤35%; LVEF 36–49% + NSVT	A(8); M(6)
ATTR amyloidosis	LVEF ≤35% + NYHA II–III; NYHA IV	M(6); M(5)
Acute lymphocytic myocarditis	Recovering; once LVEF normalised	M(4); R(3)
HCM	≥1 major SCD risk factor	A(8); AHA IIa for ≥1 risk factor, I B-NR after SCA or sustained VT; EPS for risk stratification III: No Benefit
ARVC	≥1 risk factor without prior VA	A(7); AHA I B-NR after SCA, sustained VT or ventricular dysfunction
LMNA	See table above	A(7–8)
CPVT	NSVT despite beta-blocker + flecainide	M(5–6)
LQTS	Asymptomatic, QTc >500 ms, beta-blocker intolerant	M(6)
Any gene carrier	Normal echo and ECG, no phenotype	M(4)

(sources/icd-crt-auc-2025, rating: high; sources/va-scd-aha-2017)

Comorbidity, Age and Generator Replacement

Competing mortality lowers appropriateness: age ≥85 with NYHA II and LVEF ≤35% rates M(5–6), dialysis-dependent ESRD M(4–5), and severe frailty M(4–5). Active intravenous substance use rates R(2) because of device infection risk. (sources/icd-crt-auc-2025, rating: high)
Generator replacement is a new decision, not a renewal: valid original indication with LVEF still ≤35% is A(8), LVEF 36–49% M(6), LVEF normalised ≥50% with no prior VA M(4), and prior sustained VA A(8–9) regardless of current LVEF. ESC 2026 makes shared decision-making before replacement IIa C. (sources/icd-crt-auc-2025; sources/HF-ESC-2026)
Wearable cardioverter-defibrillator: a bridge to decision or to transplant, ESC IIb C; it is not covered by the AUC. (sources/HF-ESC-2026)

Evidence Base

MADIT-II (n=1,232; prior MI; LVEF ≤30%; no EPS required; mean LVEF 23%) was stopped at a mean 20 months when the efficacy boundary was reached: all-cause mortality 14.2% vs 19.8%, HR 0.69 (0.51–0.93; P=0.016), with benefit emerging at about 9 months. Heart failure hospitalisation trended higher in the ICD arm (19.9% vs 14.9%; P=0.09). (sources/icd-icm-maditii-nejm-2002, rating: very high)
SCD-HeFT (n=2,521; 52% ischaemic, 48% non-ischaemic; LVEF ≤35%; NYHA II 70%, III 30%; median 45.5 months) is the trial behind the ≤35% threshold that MADIT-II did not test: HR 0.77 (97.5% CI 0.62–0.96; P=0.007), with no aetiology interaction (P=0.68) — primary prevention benefit applied equally to non-ischaemic disease. (sources/icd-hfref-scdheft-nejm-2005, rating: very high)
DEFINITE (n=458; non-ischaemic, LVEF <36%, NSVT or ≥10 PVCs/hour; 85% on ACE inhibitor and beta-blocker; shock-only single-chamber device; mean 29 months) reduced sudden arrhythmic death (HR 0.20, 0.06–0.71; P=0.006) but missed its all-cause mortality primary end point (HR 0.65, 0.40–1.06; P=0.08). Only about one third of deaths were arrhythmic rather than the assumed half, so the trial was effectively underpowered. (sources/icd-nicm-definite-nejm-2004, rating: high)
COMPANION (n=1,520; NYHA III–IV, LVEF ≤35%, QRS ≥120 ms) showed CRT-D reduced all-cause mortality against optimal pharmacological therapy, HR 0.64 (0.48–0.86; P=0.003), including in the non-ischaemic subgroup (HR 0.50); CRT-P alone was borderline (HR 0.76; P=0.059). (sources/crt-companion-nejm-2004, rating: very high)
DANISH (n=1,116; LVEF ≤35%, NYHA II–IV, NT-proBNP >200; 58% CRT; median 67.6 months) is the trial that changed the European class: all-cause mortality HR 0.87 (0.68–1.12; P=0.28), with sudden cardiac death halved (HR 0.50; P=0.005) and cardiovascular death neutral (HR 0.77). There were 31% non-cardiovascular deaths. Benefit was confined to patients under 68 (HR 0.64; P=0.01; ≥68: HR 1.05). (sources/icd-nicm-danish-nejm-2016, rating: high)

ICD Programming

Full detail: concepts/ICD-Programming.

Detection: prolonged detection (30/40 intervals) as default, VF zone ≥200 bpm for primary prevention, SVT discriminators with an upper limit around 230 bpm. (sources/icd-programming-hrs-2015, rating: high)
Therapy: ATP first for all VT including fast VT, shocks as backup, ATP during charging in the VF zone.
Bradycardia: VVI 40 in single-chamber devices; DDD with a pacing-minimisation algorithm in dual-chamber; biventricular pacing ≥98% in CRT-D.
Defibrillation testing: safely omitted for standard left-sided transvenous implants with confirmed lead position; still reasonable for the S-ICD and uncertain systems.
Manufacturer-specific translations of these principles were revised in 2019 after some translated settings produced adverse outcomes; the templates are VVI 40, DDD with pacing minimisation, and DDD ± rate response for CRT. (sources/icd-programming-hrs-2019, rating: high)

Adjunct — Potassium Optimisation

POTCAST (n=1,200 ICD patients, baseline potassium ≤4.3 mmol/L) targeted high-normal potassium (4.5–5.0 mmol/L) with supplements, an MRA, dietary advice and diuretic reduction: the composite of sustained VT, appropriate therapy, arrhythmia or HF hospitalisation and death fell from 29.2% to 22.7% (HR 0.76, 0.61–0.95; NNT 12.3). (sources/potcast-nejm-2025, rating: very high)
The mean rise was only about 0.3 mmol/L and fewer than half reached target, yet benefit was present in those who did not — the mechanism appears to be avoidance of low-normal potassium rather than attainment of a high-normal level.
Full trial and the surrounding potassium literature: concepts/Potassium-and-Ventricular-Arrhythmias.

Complications

Device infection: in DANISH, non-CRT ICD patients had a hazard ratio of 6.35 (P=0.006) for device infection — the reason active intravenous substance use rates R(2). (sources/icd-nicm-danish-nejm-2016, rating: high)
Inappropriate therapy: 5.9% in DANISH; associated with a 2-fold mortality increase in SCD-HeFT and reducible by programming (see concepts/ICD-Programming). (sources/icd-nicm-danish-nejm-2016; sources/icd-programming-hrs-2015)
Appropriate shocks were associated with a 5-fold mortality increase in SCD-HeFT — a marker of substrate severity as well as a harm in itself. (sources/icd-programming-hrs-2015)
Lead-related complications: with a dedicated 4.7-F lumenless left bundle branch area defibrillation lead (LEADR, n=302 successful implants), freedom from lead-related major complications was 97.7% at 6 months and 96.8% at 12 months, with appropriate therapy in 7.9% and inappropriate shocks in 2.1% at 1 year. (sources/leadr-lbbap-hrs-2025, rating: high)

Contradictions / Open Questions

Non-ischaemic indications diverge between guidelines for the same patient. ESC 2026 rates LVEF ≤35% NYHA II–III non-ischaemic cardiomyopathy IIa B1; the 2017 AHA/ACC/HRS guideline still rates it I A; the 2025 AUC rates it A(7–9) under 75 but M(5–6) at ≥85. All three rest on the same trials and differ in how heavily they weight DANISH and competing mortality, so the class assigned reflects the guideline's philosophy as much as the evidence. (sources/HF-ESC-2026, rating: very high; sources/va-scd-aha-2017)
DEFINITE: a significant arrhythmic benefit without a mortality benefit. Sudden arrhythmic death fell 80% (HR 0.20) while all-cause mortality did not reach significance (HR 0.65; P=0.08). Only about a third of deaths were arrhythmic against an assumed half, so the trial was underpowered for its own primary end point — and the same competing-mortality pattern later appeared in DANISH. (sources/icd-nicm-definite-nejm-2004, rating: high)
MADIT-II was stopped early and predates modern therapy. Termination at the efficacy boundary after a mean 20 months usually overestimates effect size; enrolment (1997–2001) predates SGLT2 inhibitors, ARNI and CRT-D; and its ≤30% entry threshold is not the ≤35% used in guidelines, which comes from SCD-HeFT. (sources/icd-icm-maditii-nejm-2002, rating: very high)
Modern quadruple therapy shrinks the absolute benefit. DANISH enrolled before SGLT2 inhibitors and sacubitril–valsartan. If contemporary medical therapy further reduces background arrhythmic death and improves LVEF, the NNT rises, and no trial has re-measured ICD benefit on a current GDMT backbone. (sources/icd-nicm-danish-nejm-2016, rating: high)
LGE-guided implantation in LVEF 36–49% is unproven. Observational and registry data (PARTITA, ESTIMATE) suggest scar predicts arrhythmic risk beyond LVEF, but no RCT has tested LGE-guided implantation as a primary end point, which is why the AUC rates it M(5–6) rather than A. (sources/icd-crt-auc-2025, rating: high)
Age ≥85: appropriateness scores outrun evidence. The AUC rates these patients M(5–6), but NNT analysis and preference data are thin, and deactivation wishes are not well captured by scenario-based ratings. (sources/icd-crt-auc-2025)
The wearable defibrillator failed its randomised test. VEST showed no significant reduction in arrhythmic death, so the device remains a bridge on physiological reasoning rather than on proven benefit — consistent with its ESC IIb C rating. (sources/HF-ESC-2026, rating: very high)
Diastolic function as an alternative selection criterion. A single-centre pilot (n=210, 7-year follow-up) found grade III diastolic dysfunction independently predicted arrhythmic death or resuscitated arrest (HR 3.52, 2.00–6.22), with 41% cumulative risk at 8 years among patients with LVEF >35% — a group currently ineligible. Pilot-level only, pre-2016 ASE criteria, not guideline-incorporated. (sources/arrhythmia-diastolic-circep-2020, rating: medium)

Connections

Device and therapy

concepts/ICD-Programming — how detection, ATP, bradycardia pacing and defibrillation testing are configured
entities/S-ICD — subcutaneous alternative when transvenous access is contraindicated
entities/CRT — CRT-D for HFrEF with broad QRS

Clinical context

entities/Heart-Failure — primary prevention in HFrEF; GDMT duration sets eligibility
concepts/Sudden-Cardiac-Death — the outcome the device prevents
concepts/Potassium-and-Ventricular-Arrhythmias — POTCAST and the arrhythmia–potassium relationship

Disease-specific indications

concepts/HCM-Risk-SCD — the risk-factor model behind A(8) in HCM
entities/HCM
concepts/Arrhythmogenic-Cardiomyopathy — ARVC ratings
entities/ARVC
entities/LMNA — A(7–8) at LVEF 35–45%
concepts/VA-Risk-Stratification-DCM — LGE plus intermediate LVEF in non-ischaemic disease
concepts/LV-Diastolic-Function — grade III dysfunction as a candidate criterion

Sources

Guidelines and appropriateness criteria

Trials

Programming and leads

Adjuncts and open questions