Tools for risk stratification of sudden cardiac death: A review of the literature in different patient populations

Abstract

While various modalities to determine risk of sudden cardiac death (SCD) have been reported in clinical studies, currently reduced left ventricular ejection fraction remains the cornerstone of SCD risk stratification. However, the absolute burden of SCD is greatest amongst populations without known cardiac disease. In this review, we summarize the evidence behind current guidelines for implantable cardioverter defibrillator (ICD) use for the prevention of SCD in patients with ischemic heart disease (IHD). We also evaluate the evidence for risk stratification tools beyond clinical guidelines in the general population, patients with IHD, and patients with other known or suspected medical conditions.

1. Introduction

In spite of the advances in modern technology, accurate identification of the patient who will experience sudden cardiac death (SCD) remains one of the holy grails of cardiology. The closest the clinician can come to prediction is an estimation of risk for this event which is likely to be terminal, and to determine an approximate categorization of patients into high and low risk groups. Appropriate high risk patients can be offered an implantable cardioverter defibrillator (ICD), the only currently available option for SCD prevention. However, the ICD is incompletely effective in preventing SCD since it treats only ventricular tachyarrhythmias but not electromechanical dissociation/pulseless electrical activity in the failing heart.

The overall annual incidence of SCD, based on extrapolation of data from the United States, is approximately 1 in 1000 adults over the age of 35 years.¹ While SCD occurs in a higher proportion of adults with traditional cardiac risk factors and a history of heart disease, the absolute number of SCDs which occur in the general population by far outnumber the absolute number of SCDs in the high risk groups. Thus the majority of SCD accrues from the general population, in whom there are no currently available screening tools.

Prevention of SCD can be categorized into primary prevention (i.e., in patients with no prior history of SCD), and secondary prevention (i.e., in patients with a history of resuscitated cardiac arrest, unstable ventricular tachycardia (VT), ventricular fibrillation (VF), or syncope with high risk features). The focus of this review will be primary prevention of SCD in adults; ICD use for secondary prevention of SCD will be discussed briefly for completeness.

We present here an enumeration and summary of prospective clinical studies evaluating SCD risk in three categories of patients: the population of patients with ischemic heart disease (IHD), populations of patients with other high-risk conditions, both cardiac and non-cardiac, and the general population. Only studies with a sample size of at least 200 patients were included in this review. The tools available for risk stratification of SCD can be broadly categorized as follows: historical factors, autonomic parameters, biomarkers, characteristics of the surface ECG, invasive electrophysiological study (EPS), left ventricular ejection fraction (LVEF), and assessment of myocardial scar burden. Populations with congenital disorders known to carry a high risk of SCD, namely long QT syndrome, short QT syndrome, Brugada syndrome, hypertrophic cardiomyopathy, arrhythmogenic right ventricular cardiomyopathy, tetratogy of Fallot, Wolff–Parkinson–White syndrome, and idiopathic VT are excluded from this review, and addressed elsewhere in this supplement.

2. Primary prevention of SCD in patients with ischemic heart disease (IHD)

2.1. Left ventricular ejection fraction (LVEF)

The mainstay of current clinical guidelines in the determination of patients at high risk for SCD is the LVEF. LVEF has been recognized as a predictor of overall cardiac mortality in IHD patients since the 1980's.² For this reason, clinical trials evaluating the efficacy of the ICDs in primary prevention of SCD have consistently used LVEF cut-offs in the selection of patients. Large clinical trials on SCD risk stratification over the last 20 years have all proven a reduction in SCD with ICD use in patients with reduced LVEF.

In 1999, the Multicenter Unsustained Tachycardia Trial (MUSTT), showed that amongst 704 coronary artery disease patients with LVEF ≤40% asymptomatic non-sustained ventricular tachycardia (NSVT), and inducible sustained ventricular tachyarrhythmias on EPS, ICD therapy decreased the risk of SCD by 27% over a 2 year follow up period. In comparison, anti-arrhythmic drug therapy was not found to be beneficial in reducing the risk of SCD. Patients who were inducible to sustained VT (whether treated with anti-arrhythmic drugs or not) fared worse than non-inducible patients, highlighting the ability of EPS to stratify risk.³ In 2002, the Multicenter Automatic Defibrillator Implantation Trial (MADIT II) showed that amongst 1232 patients following myocardial infarction (MI) with LVEF ≤30%, prophylactic ICD implantation decreased the rate of SCD by over 30% over a follow up period of 20 months.⁴ In 2005, the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT), showed that amongst 2521 New York Heart Association (NYHA) class II or III heart failure patients (due to both ischemic and non-ischemic causes) with LVEF ≤35%, ICD implantation reduced overall mortality by 23% over a median follow up period of 45.5 months.⁵

Clinical trials directly evaluating the risk of SCD among various LVEF strata are comparatively fewer. In 2008, the Improved Stratification of Autonomic Regulation (ISAR-risk) study showed that amongst 2343 survivors of MI in sinus rhythm, LVEF ≤30% predicted increased all cause mortality and SCD compared with LVEF >30%.⁶ The Risk Estimation Following Infarction, Noninvasive Evaluation (REFINE) trial in 2007 showed that amongst 322 post-MI patients, LVEF ≤30% as compared with LVEF >30% had an increased risk of SCD or resuscitated cardiac arrest (HR 3.30, p = 0.005).⁷ This paucity of trials directly comparing SCD risk in different LVEF strata contributes to the discordance of LVEF cut-offs across various published clinical guidelines for primary prevention ICD implantation.⁸ The most recent 2013 consensus guidelines on appropriate use of ICD for the primary prevention of SCD in IHD are summarized in Fig. 1.⁹ In these latest guidelines, LVEF remains the first step in determining SCD risk, and the timing of a recent MI is the second step.

Fig. 1.

Current recommendations for appropriate use of implantable cardioverter defibrillators in ischemic heart disease patients. Numbers indicate new evidence shown in Table 1 for additional risk stratification tools in the given subsets of patients. LVEF – left ventricular ejection fraction, MI – myocardial infarction, NSVT – non-sustained ventricular tachycardia, VT – ventricular tachycardia, EPS – electrophysiology study, PCI – percutaneous coronary intervention, CABG – coronary artery bypass graft, GDMT – goal-directed medical therapy, NYHA – New York Heart Association.

2.1.1. Timing of ICD implantation following MI

The only current recommendations for ICD use in patients less than 40 days following MI comes from the previously described 1999 MUSTT trial, which included patients 4 or more days after MI. For this reason, current recommendations state that ICD use is appropriate in patients 4–40 days after MI with inducible VT on EPS.⁹ However, it should be noted that only 34 of 704 patients in the MUSTT trial were within 1 month of myocardial infarction.

Apart from the above case, ICD use is not recommended in patients less than 40 days following MI. The evidence for this comes from the Defibrillator in Acute Myocardial Infarction Trial (DINAMIT) and the Immediate Risk Stratification Improves Survival (IRIS) trial. In 2004 the DINAMIT trial showed that during 6–40 days following an MI in patients with LVEF ≤35% and impaired heart rate variability, ICD implantation caused a decrease in the rate of death due to arrhythmia, but that benefit was offset by an increase in the rate of non-arrhythmic deaths.¹⁰ In 2009, the IRIS trial enrolled patients 5–31 days following MI with LVEF <=40%, and either heart rate of 90 or more on first available ECG or NSVT on Holter monitoring. Patients were randomized to receive ICD or medical therapy alone. ICD use was found not to reduce overall mortality in these patients.¹¹

2.2. Non-LVEF parameters for assessment of SCD risk

In addition to the studies for SCD risk stratification which have been incorporated into the official guidelines, there is an accumulating body of literature for various other tools in SCD risk stratification amongst IHD patients (Table 1), but these strategies have yet to be universally recommended or incorporated into guidelines. Data from the Duke Databank for Cardiovascular disease, which included patients undergoing cardiac catheterization at Duke University Medical Center and found to have at least one native coronary artery stenosis of ≥75%, have identified various factors from a patient's history which increase the risk of SCD. While each of these factors was independently predictive of SCD at a level which reached statistical significance, hazard ratios obtained were relatively low. These historical factors, LVEF, and the number of diseased coronary arteries were combined by the authors to develop the Duke SCD risk score which provides a numerical estimation of risk over a 1–10 year follow up period. In this model, LVEF still carried the greatest statistical importance compared to all other variables. This model was internally validated using a bootstrapping technique, and externally validated in the SCD-HeFT database. The major limitation to use of this clinical risk score is that even in the highest quartile of this risk score, deaths due to SCD represent only 1 in 8 deaths that occur in this population, and the number needed to treat with an ICD to prevent one SCD was 18 in the highest risk quartile.¹² Nevertheless, the Duke SCD risk score remains a promising tool for SCD risk stratification beyond current primarily LVEF-based guidelines.

Table 1.

Sudden cardiac death risk stratification tools in ischemic heart disease patients beyond current guidelines which have been evaluated in prospective clinical studies, with the endpoint being either sudden cardiac death alone, or a composite of sudden cardiac death and ventricular tachycardia/fibrillation. See Fig. 1 for correlation of where these tools may be incorporated into current guidelines.

Ref	#	Study name (location)	Population (sample size)	Risk stratification tool	Hazard ratio or relative risk for endpoint (95% CI)	p Value
14	1	(China)	ICM LVEF<50%, GDMT, NYHA II–IV (n = 1054)	β1 receptor auto-antibodies	3.749 (2.389–5.884)	<0.001
				fQRS in inferior leads	2.714 (1.809–4.072)	<0.001
12	2	Duke Databank for CVD (U.S.A.)	Patients with ≥1 native coronary artery with ≥75% stenosis (n = 37258)	History of diabetes	1.40 (1.25–1.56)	<0.0001
				History of hypertension	1.37 (1.24–1.53)	<0.0001
				History of HF	1.39 (1.23–1.58)	<0.0001
				History of cerebrovascular disease	1.49 (1.28–1.73)	<0.0001
				History of tobacco use	1.20 (1.08–1.34)	0.0008
				LVEF (per 1% decrease)	1.05 (1.04–1.05)	<0.0001
				Diseased coronary arteries (per # increase)	1.37 (1.29–1.45)	<0.0001
22	3	VALIANT (24 countries)	Acute MI with HF or LVEF ≤40% (n = 11256)	HR per 10 beats/min increase: at initial hospitalization, 6 months to 3 years later	1.20 (1.06–1.37), 1.10 (1.01–1.19)	NA, NA
				CrCl per 10 cc/min: initial hospitalization, discharge to 30 days	0.82 (0.74–0.91), 0.93 (0.87–0.99)	NA, NA
				LVEF per 10% increase below 40%: initial hospitalization, 6 months to 3 years later	0.74 (0.56–0.98), 0.67 (0.58–0.78)	NA, NA
				AF post MI: initial hospitalization, 6 months to 3 years later	2.03 (1.30–3.16), 1.65 (1.23–2.19)	NA, NA
				Re-hospitalization: discharge to 30d, 6 months to 3 years later	2.48 (1.52–4.06), 1.47 (1.17–1.86)	NA, NA
				Interval HF: discharge to 30d, 6 months to 3 years later	2.19 (1.34–3.59), 1.45 (1.05–1.99)	NA, NA
15	3	ABCD (U.S.A, Germany, Israel)	Age ≥18, ICM, NSVT, LVEF ≤40% (n = 566)	MTWA in LVEF ≤40%	2.7% vs. 0%a	0.04
				MTWA in LVEF ≤30%	8.8% vs. 2.9%a	<0.05
42	4	(USA)	ICM EF ≤35%, no prior VA (n = 768)	MTWA	2.29 (1.00–5.24)	0.049
20	5	(Denmark)	Acute MI (n = 988)	Global longitudinal strain on echo	1.24 (1.10–1.40)	0.0004
				Mechanical dispersion by echo (per 10 ms increase)	1.15 (1.01–1.31)	0.032
19	6	MERLIN-TIMI 36 trial (U.S.A., Czech republic, Netherlands)	NSTEMI ≤48 h prior (n = 6345)	4–7 beats VT on 7 days of telemetry	2.3 (1.5–3.7)	<0.001
19				≥8 beats of VT on 7 days of telemetry	2.8 (1.5–5.1)	0.001
17				VT ≥4 beats, ≥1 mm ST depression for ≥1 min	6.5	<0.001
18				QTc interval (≥450 ms in men, ≥470 ms in women)	2.3	0.005
21	7	(Norway, Belgium)	>40d post MI (n = 569)	Mechanical dispersion by echo (per 10 ms increase)	1.7 (1.2–2.5)	<0.01
43	8	MADIT II (U.S.A)	Age ≥21, Prior MI (>1 month ago), revascularization >3 months ago, LVEF ≤30% (n = 1232)	BMI (per 5 unit decrease)	1.41 (1.09–1.83)	0.01
44				SBP (per 10 mmHg increase)	0.84 (0.71–0.99)	0.04
45				Renal insufficiency	2.00 (1.01–4.02)	0.04
13				Inferior QRS fragmentation	2.05	0.007
13				Inferior QRS fragmentation in LBBB	4.24	0.002
46			Medically treated arm	QRS duration >140 ms	2.12 (1.20–3.76)	0.01

CI – confidence interval, ICM – ischemic cardiomyopathy, LVEF – left ventricular ejection fraction, GDMT – guideline directed medical therapy, NYHA – New York Heart Association, CVD – cardiovascular disease, MI – myocardial infarction, HF – heart failure, HR – heart rate, min – minute, CrCl – creatinine clearance, AF – atrial fibrillation, NSVT – non-sustained ventricular tachycardia, MTWA – microvolt T wave alternans, VA – ventricular arrhythmia, VT – ventricular tachycardia, BMI – body mass index, SBP – systolic blood pressure, LBBB – left bundle branch block.

^aHazard ratio/relative risk ratio not reported.

Various parameters from the surface ECG have increasingly been proven to reach levels of statistical significance in predicting SCD. A Chinese study and an analysis of the MADIT II study data in 2012 showed an increased risk of SCD when QRS fragmentation was present in the inferior leads.^13,14 The Alternans Before Cardioverter Defibrillator (ABCD) trial showed in 2010 that amongst patients with LVEF ≤40% and NSVT who were more than 28 days post MI, microvolt T wave alternans (MTWA) and EPS were independent predictors of SCD; MTWA predicted events better in patients with LVEF <30%, and EPS predicted events better in patients with LVEF >30%. The use of EPS in patients with LVEF 36–40% is already incorporated into clinical guidelines; however the ABCD study suggests that the use of MTWA in combination with EPS may better predict SCD risk, as each may identify different arrhythmogenic substrates.¹⁵ It should be noted that MTWA may have limited prognostic ability in guiding patients who will benefit most from ICD implantation; in CAD and reduced LVEF, MTWA results did not predict higher risk of ventricular tachyarrhythmias and ICD therapies.¹⁶

In 2010, the Metabolic Efficiency with Ranolazine for Less Ischemia in Non-ST-elevation 36 (MERLIN-TIMI 36) trial evaluated 6345 patients within 48 h after hospitalization for non-ST-elevation MI (NSTEMI) with 7 day outpatient ECG monitoring, with median follow up of 348 days. In this study, increasing number of beats of spontaneous ventricular tachycardia, and the presence of ischemia on continuous ECG monitoring was associated with an increased risk of SCD. An increased QTc interval was also associated with increased SCD risk.^17–19 This study supports the use of continuous ECG monitoring for 7 days following NSTEMI to evaluate patients at increased risk for SCD, a tool which is not currently in widespread use.

Recent studies have evaluated novel echocardiographic parameters of mechanical dispersion and strain echocardiography in the use of risk stratification in patients who are post MI, as they may reflect myocardial deformation heterogeneity. Both global longitudinal strain and increased mechanical dispersion have been shown to have a statistically significant predictor of SCD risk.^20,21

While most studies of SCD risk assessment involve a baseline assessment of a certain clinical tool, and follow up at one time point, the Valsartan in Acute Myocardial Infarction trial (VALIANT), published in 2010, evaluated the time dependence of risk factors. They found that in the immediate hospitalization after MI, higher baseline heart rate and impaired creatinine clearance were the strongest predictors of SCD among the variables that they evaluated. During long term follow up of up to 3 years, the strongest predictors of SCD were found to be prior MI, initial LVEF <40%, and recurrent cardiovascular events; however the predictive power diminished with time. The authors argue that this implies that risk stratification for SCD ought to be a dynamic and on-going process.²² This study, however, did not reassess clinical parameters at each time point of follow up and rather based their analysis on baseline values. The clinical implications of this study may be that in those for whom the short term risk of SCD based on risk stratification from baseline parameters is low, ICD therapy may be delayed to avoid ICD related complications for the time period for which they are considered low risk.

There has additionally been interest in the effect of autonomic parameters on risk of sudden cardiac death, however the majority of studies on this topic have small sample sizes. One exception is the ATRAMI study, which evaluated the effect of heart rate variability and baroreflex sensitivity on the risk of cardiac arrest.²³ Of note, the endpoint of this study included both arrhythmic and non-arrhythmic causes of cardiac arrest. The composite of autonomic parameters had a relative risk of 16.79 (95% CI 6.01 to 46.89, p < 0.0001) of cardiac arrest.²⁴ Autonomic parameters have so far had limited use in the clinical setting and need to be studied further to determine their prognostic value.

3. Primary prevention of SCD in patients with other cardiac and non-cardiac disease

3.1. Patients with non-ischemic cardiomyopathy (NICM)

Current guidelines for ICD use in patients with NICM recommend implantation in patients with LVEF ≤35%, NYHA class I–III heart failure, following goal-directed medical therapy (GDMT) for a minimum of 3 months. ICD use is also recommended in patients with specific cardiomyopathy types: those patients with LVEF ≤35% and sarcoid heart disease, myotonic dystrophy, Chagas disease, or persistent peripartum cardiomyopathy. In addition, ICD use is recommended in patients with Giant cell myocarditis, regardless of LVEF.⁹ New technologies available for quantitative characterization of T-wave alternans (TWA), and examination for structural heart disease using cardiac MRI have now been shown to have statistically significant predictive value in determining SCD risk (see Table 2).

Table 2.

Risk stratification tools for prediction of sudden cardiac death in other disease populations.

Ref	Study name (location)	Population	Risk stratification tool	Hazard ratio or relative risk for endpoint (95% CI)	p Value
Non-ischemic cardiomyopathy (NICM)
47	(Slovenia)	HF NYHA class III or IV, LVEF <40% (n = 398)	QTc increase of ≥10% after 1 year f/u	5.98 (1.05–24.74)	0.006
26	(U.K)	NICM (n = 472)	Fibrosis on LGE-CMR	4.61 (2.75–7.74)	<0.001
			Fibrosis extent on LGE-CMR (per 1% increment)	1.10 (1.05–1.16)	<0.001
27	Meta-analysis	CAD or NICM (n = 1105)	LV scar on LGE-CMR	4.33 (2.98–6.29)	NA
25	MUSIC study (Spain)	HF NYHA class II or III (n = 650)	TWA, index of average alternans>3.7uV	2.29 (1.31–4.00)	0.004
			TWA, index of average alternans at HR 90	1.07 (1.00–1.15)	0.046
14	(China)	DCM LVEF <45% despite medical therapy, NYHA II–IV	J wave in the inferior leads (n = 572)	4.095 (2.132–7.863)	<0.001
			β1 receptor auto-antibodies (n = 704)	4.514 (2.405–8.471)	<0.001
48	ALPHA study (Italy)	NICM, LVEF ≤40%, NYHA class II–III (n = 446)	TWA	5.53 (1.29–23.65)	0.004
49	(Austria)	LVEF<35% (n = 452)	log (BNP)	Chi square: 11.8125	0.0006
Suspected heart disease
50	(Argentina)	Chest pain patients with suspected ACS (n = 982)	Vitamin D (Highest quartile vs. lowest quartile)	0.32 (0.11–0.94)	0.038
30	Finnish Cardiovascular Study (Finland)	Patients getting stress test (n = 1297)	TCRT/HR loop area (<2.597 vs. >2.597)	10.7 (1.4–83.7)	0.024
			Baseline TCRT (<0 vs. >0)	5.0 (1.1–22.7)	0.038
29	LURIC study (Germany)	Patients referred for coronary angiography (n = 3303)	Plasma renin concentration (per SD log increase)	1.23 (1.10–1.38)	<0.001
28	(U.S.A)	Patients in sinus rhythm having EP study (n = 313)	TWA	12.2	<0.0001
			PVS	3.0	<0.0001
End stage renal disease (ESRD)
35	(Netherlands)	Dialysis patients (n = 277)	QRS-T angle ≥130° in men and ≥116° in women	2.99 (1.04–8.60)	NA
34	(Hong Kong)	ESRD on PD for≥3months (n = 230)	LVEF (per 1% increase)	0.94 (0.89–0.98)	0.004
			SBP (per 1 mmHg increase)	1.05 (1.02–1.08)	0.0031
			DBP (per 1 mmHg increase)	0.92 (0.87–0.97)	0.0033
			Cardiac troponin T (per 1ug/L increase)	1.14 (1.01–1.31)	0.031
Obstructive sleep apnea (OSA)
32	(U.S.A)	Adult having first diagnostic polysomnogram (n = 10701)	Apnea-hypoapnea index>20	1.60 (1.14–2.24)	0.007
			Mean nocturnal SaO2<93%	2.93 (1.98–4.33)	<0.0001
			Lowest nocturnal SaO2<78%	2.60 (1.85–3.65)	<0.0001
			Age >60 years	5.53 (3.84–7.94)	<0.0001
Other
51	SEAS study (USA, Europe)	Mild-moderate AS, age 45–85 (n = 1542)	QRS duration (<85 ms vs.>100 ms without BBB)	5.0 (1.8–13.7)	0.002
52	LIFE study (USA, Scandinavia)	HTN, LVH on ECG, sinus rhythm (n = 8831)	New onset AF	3.13 (1.87–5.24)	<0.001
53	(UK)	SVT or NSVT (n = 373)	Fibrosis on LGE-CMR	3.3 (1.8–5.8)	<0.001

NYHA – New York Heart Association, LGE-CMR – late gadolinium enhancement cardiac magnetic resonance imaging, CAD – coronary artery disease, LV – left ventricle, TWA – T wave alternans, DCM – dilated cardiomyopathy, BNP – brain natriuretic peptide, TCRT – Total cosine R-to-T, ACS – acute coronary syndrome, PVS – programmed ventricular stimulation, ESRD – end-stage renal disease, PD – peritoneal dialysis, LVEF – left ventricular ejection fraction, S/DBP – systolic/diastolic blood pressure, LVH – left ventricular hypertrophy, AF – atrial fibrillation, (N)SVT – (non-)sustained ventricular tachycardia.

The Muerte Súbita en Insuficiencia Cardiaca (MUSIC) study, published in 2011 assessed the prognostic value of TWA amongst NYHA class II and III heart failure patients, without specifying the etiology of the heart failure. Using a fully automated system for the measurement of TWA, with quantitative measures of the degree of alternans, this study showed an increased risk of SCD (HR = 2.29) in patients with a higher degree of TWA after a median follow up period of 48 months.²⁵

Quantification of myocardial scar burden using late gadolinium enhancement cardiac magnetic resonance imaging (LGE-CMR) has also been shown to be associated with SCD risk. A 2013 study by Gulati et al found that both myocardial fibrosis, and increased fibrosis extent were independently associated with SCD in patients with NICM.²⁶ A 2013 meta-analysis by Scott et al comprising a total of 1105 patients also found an increased risk of SCD with greater extent of left ventricular scar burden.²⁷

3.2. Patients with suspected cardiovascular disease

There have been several studies evaluating SCD risk in patients with suspected cardiovascular disease, with cohorts of patients being recruited from referrals for various cardiovascular testing modalities (see Table 2).

In a study published by Gold et al in 2000, 313 patients referred for electrophysiologic study were evaluated by TWA and signal averaged ECG, then followed up for 400 days. On multivariate analysis, they found that only TWA and EPS were independent predictors of SCD. Of note, the cohort of patients with TWA had an LVEF which was lower than those patients without TWA (35 ± 17% vs. 54 ± 13%, p = 0.001).²⁸

In Germany, the Ludwigshafen Risk and Cardiovascular Health (LURIC) Study enrolled a cohort of 3303 patients referred for coronary angiography. Here, an increased plasma renin concentration, a marker of activation of the renin–angiotensin–aldosterone axis, weakly correlated with increased risk of SCD at a median follow up of 9.9 years (HR 1.23, p < 0.001).²⁹

Subsequently, the Finnish Cardiovascular study evaluated ECG parameters in patients referred for bicycle stress test. They found that attenuated hysteresis of the depolarization and repolarization wave fronts (total cosine R-to-T, TCRT) during the recovery period after exercise were predictive of SCD after a mean follow up period of 45 months (HR 10.7, p = 0.024).³⁰

3.3. Patients with obstructive sleep apnea

There is a known association between obstructive sleep apnea and sudden cardiac death. Putative mechanisms proposed to explain these associations include intermittent hypoxia and recurrent arousals, leading to sympathetic activation and electrical and structural remodeling of the heart, predisposing to cardiac arrhythmias.³¹ A 2013 study by Gami et al evaluated the SCD risk at an average of 5.3 years of follow up in adults undergoing their first diagnostic polysomnogram. SCD was found to be best predicted by age >60 years, apnea-hypoapnea index >20, mean nocturnal oxygen saturation <93%, and lowest nocturnal oxygen saturation <78%.³² In OSA patients, adherence to use of continuous positive airway pressure ventilation during sleep has been shown to decrease adverse cardiac events.³¹

3.4. Patients with end-stage renal disease (ESRD)

The risk of cardiovascular mortality in ESRD patients is known to be 10–100 fold higher than age, sex, and race matched controls in the general population. In particular, SCD accounts for 26.1% of deaths in the ESRD population, and these deaths do not appear to be caused by coronary artery disease.³³ A 2010 study by Wang et al found that amongst ESRD patients on peritoneal dialysis for over 3 months, after a 5 year follow up, higher LVEF and diastolic blood pressure were associated with a decreased risk of SCD whereas increased systolic blood pressure was associated with an increased risk of SCD.³⁴ A recent study from the Netherlands on chronic dialysis patients showed that abnormal QRS-T spatial angle on surface ECG was predictive of SCD and all cause mortality.³⁵ Currently, there is no consensus on the true survival benefit derived from ICD implantation in patients with ESRD.

4. Primary prevention of SCD in the general population

As the absolute number of SCD events which occur in the general population outnumber those which occur in high risk groups,¹ a review of risk stratification for SCD would be incomplete without addressing the general population. There are no official recommendations for prophylactic ICD use for primary prevention of SCD in this group with good reason; the incidence rate is so low that any attempt at mitigating risk with ICD use would by far be overshadowed by the inherent risks of ICD use in large number of those who will never experience a fatal arrhythmic event.

Nevertheless, in recent years, there have been several long term prospective studies in large cohorts of the general population, evaluating the predictive power of various clinical tools for SCD, summarized in Table 3. These studies all evaluate various baseline characteristics, and followed patient outcomes over an average time span of between 7.7 and 30 years. Interestingly, almost all of these studies originated from patient populations in the United States and Finland. The applicability of this data to other parts of the world remains unclear.

Table 3.

Risk stratification tools for prediction of sudden cardiac death in the general population, listed in order of increasing hazard ratios.

Ref	Study name	Population	Risk stratification tool	HR or RR for SCD (95%CI)	p value
54	Kuopio (Finland)	Men 42–60 (n = 2368)	Cardiorespiratory fitness (per 1 MET increase)	0.78 (0.71–0.84)	<0.001
55	CHS (U.S.A)	Adults ≥65 (n = 3089)	Highly sensitive troponin T (1 pg/ml per year increase from baseline)	1.03 (1.01–1.06)	0.03
56	WHI (U.S.A)	Women (n = 161,808)	Older age	1.09 (1.07–1.11)	NA
57	CHS (U.S.A)	Adults ≥65 (n = 5806)	CRP (+1 log increase)	1.13 (1.00–1.28)	0.049
58	ARIC study (U.S.A)	Adults 45–64 (n = 14,574)	APCs on 2 min telemetry strip	1.15 (0.56–2.39)	NA
59	Kuopio (Finland)	Men 42–61 (n = 2666)	SBP at rest (per 10 mmHg increment)	1.15 (1.07–1.25)	<0.001
60	Health 2000 (Finland)	Adults ≥30 (n = 5618)	T-wave residuum (per SD increment)	1.2 (1.0–1.5)	0.05
55	CHS (U.S.A)	Adults ≥65 (n = 4431)	Highly sensitive troponin T (+1 log increase)	1.26 (1.01–1.57)	0.04
57	CHS (U.S.A)	Adults ≥65 (n = 5382)	IL-6 (+1 log increase)	1.26 (1.02–1.56)	0.04
61	Kuopio (Finland)	Men 42–60 (n = 2049)	QRS duration (per 10 ms increase)	1.27 (1.14–1.40)	<0.001
60	Health 2000 (Finland)	Adults ≥30 (n = 5618)	Total cosine R-to-T (per SD decrement)	1.3 (1.1–1.6)	0.013
60	Health 2000 (Finland)	Adults ≥30 (n = 5618)	T-wave morphology dispersion (per SD increment)	1.4 (1.1–1.7)	0.001
56	WHI (U.S.A)	Women (n = 161,808)	History of hypertension	1.46 (1.11–1.93)	NA
62	Kuopio (Finland)	Men 42–60 (n = 2641)	Impaired fasting glucose	1.51 (1.07–2.14)	0.02
56	WHI (U.S.A)	Women (n = 161,808)	Race (African–American vs. White)	1.61 (1.18–2.19)	NA
	WHI (U.S.A)	Women (n = 161,808)	History of carotid artery disease	1.72 (1.03–2.87)	NA
	WHI (U.S.A)	Women (n = 161,808)	Waist:hip (Highest vs. lowest quartile)	1.73 (1.20–2.48)	NA
	WHI (U.S.A)	Women (n = 161,808)	History of DM	2.00 (1.55–2.58)	NA
58	ARIC study (U.S.A)	Adults 45–64 (n = 14,574)	VPCs on 2 min telemetry strip	2.09 (1.22–3.56)	NA
63	CHD study (Finland)	Adults 30–59 (n = 10,957)	T wave axis (≤10 or≥100 vs. 0–90)	2.13 (1.63–2.79)	<0.001
64	CHS (U.S.A)	Adults ≥65 (n = 2312)	Vitamin D and PTH (<20, ≥65 vs. ≥20, <65)	2.16 (1.15–4.05)	NA
63	CHD study (Finland)	Adults 30–59 (n = 10,957)	QRST angle (>100 vs. < 100)	2.26 (1.59–3.21)	<0.001
56	WHI (U.S.A)	Women (n = 161,808)	Current smoker	2.26 (1.66–3.09)	NA
65	Nurses Health Study (USA)	Female nurses 30–55 (n = 121,701)	Current smoker	2.44 (1.80–3.31)	NA
66	Meta-analysis	Meta-analysis (n = 106,195)	VPCs >1 on ECG or >30 on 1 h telemetry	2.64 (1.93–3.63)	NA
67	CHS (U.S.A)	Adults ≥65 (n = 4465)	Cystatin C (≥1.10 mg/L vs. ≤0.91 mg/L)	2.67 (1.33–5.35)	NA
	CHS (U.S.A)	Adults ≥65 (n = 4465)	Cystatin C (0.92 mg/L to 1.09 mg/L vs. ≤0.91 mg/L)	2.72 (1.44–5.16)	NA
68	Physician's Health Study (U.S.A)	Male physicians 40–84 (n = 22,071)	CRP (Highest quartile vs. lowest quartile)	2.78 (1.35–5.72)	<0.001
62	Kuopio (Finland)	Men 42–60 (n = 2641)	History of DM	2.86 (1.87–4.38)	<0.001
58	ARIC study (U.S.A)	Adults 45–64 (n = 14,574)	APCs and VPCs on 2 min telemetry	6.39 (2.58–15.84)	NA

CRP – C-reactive protein, A/VPC – atrial/ventricular premature complex, SBP – systolic blood pressure, SD – standard deviation, DM – diabetes mellitus, PTH – parathyroid hormone, NA – not available.

5. A brief word on secondary prevention ICD use

Secondary prevention ICD use has been found to have a mortality benefit over anti-arrhythmic therapy in multiple clinical trials, and requires fewer numbers of patients to be treated per life saved compared to primary prevention populations.^36–38 Current guidelines recommend ICD use in patients with resuscitated cardiac arrest due to VT/VF which does not have a reversible cause. It is also indicated in patients following cardiac arrest due to MI who are not candidates for complete revascularization. Patients with syncope of undetermined origin, who have inducible VT/VF on EPS are also candidates for ICD therapy.³⁹

6. Conclusions

The oldest and most reproducible evidence for predictors of SCD comprise the current guidelines for ICD use. We have reviewed here the evidence that has been accrued beyond what is reflected in the guidelines, and which will likely increase the predictive power for SCD events. All studies involve a one-time evaluation of a certain parameter, so the risk estimate applies to the length of the follow up period for each study. In studies of patients with IHD, the follow up duration has been as short as one year, and in studies of the general population, up to 30 years. In the future, an important area of study will be to identify those patients at imminent risk of death in the days or weeks following their presentation to medical attention who may require sooner intervention with ICD or with bridging strategies to ICD, such as wearable cardioverter defibrillators.⁴⁰

While ICDs have clearly been shown to reduce the risk of SCD, they do not eliminate this risk. The SCD-HeFT trial showed that the deaths due to SCD in the group treated with ICD vs. placebo was 20% vs. 39%.⁴¹ This observation is specific to the trial population of patients with LVEF ≤35% and NYHA class II–III heart failure, so other populations of patients may derive different levels of benefit from ICD therapy. Furthermore, ICD's carry the risks of implant-related complications including infection, lead-related malfunction, and inappropriate shocks; some of these complications may be mitigated by recent technologic advancements such as the subcutaneous ICD. For that reason, the studies reviewed here may also help the astute clinician to identify a low risk cohort amongst those patients with ICD indications, in whom, a combination of patient preference and individual risk due to comorbidities may negate the potential benefit of ICD.

The tools reviewed here carry a range of cost, equipment, time, and personnel requirements. The least expensive tools are factors from a patient's medical history (personal and family), followed by blood pressure and heart rate measurements, then elements of the surface ECG which can be read by a practitioner without analysis by specialized software. Biomarkers have variable costs depending on complexity of the assay. Imaging and EPS require the equipment, technicians, and qualified interpreters of the results. For this reason, developing a risk stratification scheme will depend heavily upon the resources available to the medical system, the clinician and the patient, and ultimately must be a case by case interpretation of both published guidelines and the additional data presented here.

Conflicts of interest

All authors have none to declare.