OUP user menu

Mortality and appropriate and inappropriate therapy in patients with ischaemic heart disease and implanted cardioverter-defibrillators for primary prevention: data from the Danish ICD Register

Peter Weeke, Jens Brock Johansen, Ole Dan Jørgensen, Jens Cosedis Nielsen, Mogens Møller, Regitze Videbæk, Michael Vinther Højgaard, Sam Riahi, Peter Karl Jacobsen
DOI: http://dx.doi.org/10.1093/europace/eut017 1150-1157 First published online: 13 February 2013


Aims To evaluate the incidences of and risk factors predisposing to appropriate and inappropriate shocks and mortality in a ‘real-world’ population of patients with ischaemic heart disease (IHD) and implantable cardioverter defibrillators (ICD) for primary prevention of sudden cardiac death (SCD).

Methods and results In this nationwide cohort, we prospectively followed 1609 patients with IHD and left ventricular dysfunction from the Danish ICD Register who received an ICD for primary prevention of SCD (1 January 2007 to 30 November 2011). Primary study outcomes were appropriate shocks, inappropriate shocks, and all-cause mortality. Secondary study outcomes included the composite endpoints: appropriate therapy and inappropriate therapy (defined as shock or anti-tachycardia pacing). All-cause mortality was 12.1% during a mean follow-up time of 1.9 ± 1.3 years. Inappropriate shocks and therapy occurred in 2.6% and 3.7% during follow-up, respectively. Appropriate shocks and therapy was identified in 7.8% and 13.4%, respectively. Time-dependent multivariable Cox regression analyses were used to identify risk factors of inappropriate/appropriate shock, therapy and mortality. Implantation of a dual-chamber ICD was associated with increased risk of both inappropriate shocks and any inappropriate therapy compared to single-chamber ICD [hazard ratios (HR) = 2.45; confidence intervals (CI):1.16–5.14 and HR = 2.38; CI:1.28–4.42, respectively]. No excess risk of mortality was associated with any device type.

Conclusion In this nationwide study of IHD patients with an ICD for primary prevention of SCD, the incidence of appropriate shocks as well as inappropriate shocks were significantly lower than reported in randomized trials. Implantation of a dual-chamber ICD was associated with more inappropriate shocks compared with single chamber devices.

  • Ischaemic heart disease
  • ICD
  • CRT-D
  • Primary prevention

What's new?

  • The number of appropriate shocks and inappropriate shocks are lower in a ‘real-world’ cohort with recent ICD implantations compared with previous trials

  • Low rates of shocks and unchanged mortality could suggest a shift in the risk profile from arrhythmia events towards other causes of death in patients with IHD receiving an ICD for primary prevention of SCD


Patients with ischaemic heart disease (IHD) and left ventricular dysfunction have an increased risk of sudden cardiac death (SCD) despite improvements in revascularization and pharmacological management. Therapy with implantable cardioverter-defibrillators (ICDs) for both primary and secondary prevention of SCD has been shown to reduce mortality in clinical trials.15 In particular, the Multicenter Automatic Defibrillator Implantation Trial II (MADIT-II) and subgroup analyses from the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT) showed improved survival rates associated with primary preventive ICD therapy among patients with IHD and symptomatic heart failure, which is reflected in current guidelines.3,4,6 Consequently, the proportion of implanted ICDs for primary prevention have increased and accounted for approximately half of all implanted ICDs in 2008.7 However, ICD associated drawbacks such as procedure-related complications, long-term device and lead complications, and inappropriate shocks are potential limitations. Consequently, the current treatment practice when using ICD's for primary prevention of SCD and patient selection criteria have been questioned.8,9 Reports of high complication rates and increased mortality associated with dual-chamber ICDs compared with single-chamber ICD have also added to this concern.10,11 Thus, it is essential to validate whether the survival benefits and complication rates seen in the clinical trials are similar in a ‘real-life’ setting in an unselected cohort of patients treated with a primary preventive ICD.

The goal of this study was to evaluate the incidences of and risk factors predisposing to appropriate and inappropriate ICD therapy and mortality among IHD patients with left ventricular dysfunction receiving an ICD (single-chamber or dual-chamber) or cardiac resynchronization therapy-defibrillator (CRT-D) for primary prevention, identified in the nationwide Danish ICD Register (DIR).


Data sources and study design

The DIR holds information on all ICD and CRT-D device implantations and replacement procedures throughout Denmark since 1989.12 Since 1 January 2007 all patients with IHD and left ventricular dysfunction in Denmark have been offered a primary preventive ICD after recommendations from the Danish Society of Cardiology. Importantly, patients with non-IHD and left ventricular dysfunction do not routinely qualify for a primary preventive ICD although the ongoing DANish Randomized, Controlled, Multicenter Study to Assess the Efficacy of Implantable Cardioverter Defibrillator in Patients With Non-ischaemic Systolic Heart Failure on Mortality (DANISH) may change that (ClinicalTrials.gov: NCT00542945). Patient-related factors are recorded in the DIR at time of implantation by the implanting physician. Since 2007, follow-up data related to the implanted device are recorded prospectively including reasons for shock or therapy [therapy is defined as first occurrence of shock or anti-tachycardia pacing (ATP) as done previously].13 Representatives from the implanting centres meet several times each year to ensure a highly accurate, valid, and complete registration of data. Implantable cardioverter defibrillator and CRT-D treatment in Denmark is standardized through a national reference programme to which all clinics adhere. Device programming and settings are not registered in the DIR, but assigned by the implanting physician with reference to current guidelines. The ventricular fibrillation/ventricular tachycardia detection zones are traditionally set relatively high with optimization of ATP to actively avoid inappropriate shocks.14

All Danish citizens are given a unique and permanent civil registration number which enables linkage of nationwide registers on an individual level. The Civil Registration System holds information on vital status for all Danish citizens with deaths being registered within 2 weeks of occurrence.

Study cohort

From the DIR we identified a nationwide cohort of consecutive Danish patients with IHD and left ventricular dysfunction who underwent implantation of an ICD or CRT-D for primary prevention of SCD6,15 from 1 January 2007 to 30 November 2011 (n = 1731). Patients without a follow-up visit after their initial implantation were excluded from further analysis (n = 122), and our final study population comprised a total of 1609 patients. All patients were followed from the time of their first ICD or CRT-D implantation until the first occurrence of either death, explantation of the device (not related to system up/downgrade or normal device end-of-life), or end of the study period (30 November 2011). Information of causes of death was not available. An electrical storm was defined as ≥3 shocks/24 h as done previously.16

The brands of the devices implanted were: St Jude Medical (n = 629, 39.1%), Medtronic (n = 574, 35.7%), Biotronik (n = 222, 13.8%), Boston Scientific (n = 93, 5.8%), Guidant (n = 87, 5.4%), and ELA Medical (n = 4, 0.2%). Patients were grouped according to the type of implanted device: single-chamber ICD, dual-chamber ICD, and CRT-D.

Study outcomes

The primary study outcomes were all-cause mortality, appropriate shocks, and inappropriate shocks. Secondary study outcomes included the composite endpoints appropriate therapy and inappropriate therapy (shock or ATP).

Statistical analysis

Comparisons of continuous variables were performed with the Kruskal–Wallis test and categorical variables were compared with the χ2 test or Fisher's exact test where appropriate. Risk factors for mortality, appropriate and inappropriate shocks, as well as appropriate and inappropriate therapy were estimated using time-dependent multivariate Cox proportional-hazard models. Results from the Cox models are presented as hazard ratios (HR) and 95% confidence intervals (CI). All models were controlled for sex, QRS duration, left ventricular ejection fraction (LVEF), year of implantation, previous coronary artery bypass graft, previous percutaneous coronary intervention, type of implanted device (single-chamber ICD, dual-chamber ICD, or CRT-D), atrioventricular conduction disease, and device upgrade/downgrade. Type of implanted device was entered as a time-dependent covariate where patients were allowed to switch device type if they received a new type of device because of a system upgrade/downgrade or uncomplicated end-of-life. Additional analyses were performed with information on pharmacotherapy (β-blocker and amiodarone) during follow-up included in the models. Time from initial implantation was used as the underlying time scale. For each model we looked at time to first occurrence of the respective endpoints. Model assumptions of proportional hazards, no interactions, and linearity of continuous traits were found to be valid unless otherwise noted.

All statistical analyses were performed using the Stata software package (Stata version 11.1, StataCorp LP, College Station, TX, USA, 2009). For all analyses, a two-sided P value of <0.05 was considered statistically significant.


This study was approved by the Danish Data Protection Agency. No ethical approval is required for retrospective register-based studies in Denmark.


Study population

Overall, 1609 patients received their first ICD (single-chamber or dual-chamber) or CRT-D for primary prevention of SCD (46.2, 16.8, and 37.0%, respectively). Eighty-four percent of patients were men. The median age for men was 67.5 (interquartile range : 60.7–73.2) years and for women 69.0 (interquartile range: 61.0–74.6) years (P = 0.05). In brief, patients who received a CRT-D were older, with more advanced New York Heart Association (NYHA) class, and wider QRS intervals than patients who received a single-chamber or dual-chamber ICD (P < 0.001). Baseline characteristics are shown in Table 1.

View this table:
Table 1

Baseline characteristics of patients with ischaemic heart disease with a single-chamber ICD, dual-chamber ICD, or CRT-D for primary prevention


Mean follow-up was 1.9 ± 1.3 years and a total of 194 (12.1%) patients died during follow-up (mean follow-up time 1.5 ± 1.1 years). During follow-up a total of 7.8% (n = 126) of patients experienced an appropriate ICD shock (mean follow-up time 1.4 ± 1.2 years). Fifty-six patients (44%) experienced one appropriate shock, 53 patients (42%) 2–5 appropriate shocks, 11 patients (9%) 6–10 appropriate shocks, and 6 patients (5%) >10 appropriate shocks. A total of 216 patients (13.5%) experienced appropriate therapy. Inappropriate shocks were identified in 41 patients (2.6%) after a mean 1.3 ± 1.2 years. Twenty-two patients (54%) experienced 1 inappropriate shock, 13 patients (32%) 2–5 inappropriate shocks, 3 patients (7%) 6–10 inappropriate shocks, and 3 patients (7%) >10 inappropriate shocks. Inappropriate therapy was identified in 59 (3.7%) of all patients.

A total of 38 patients experienced an electrical storm and, of these, 30 individuals experienced ≥3 appropriate shocks/24 h and 8 individuals experienced ≥3 inappropriate shocks/24 h of which the latter was caused by a supraventricular tachycardia (n = 5), lead-related problems (n = 2), and T-wave oversensing (n = 1).

Treatment with a β-blocker or amiodarone during follow-up was not different between single-chamber ICD, dual-chamber chamber ICD, or CRT-D (P≥ 0.05) (Table 1).

Risk of appropriate and inappropriate shock and therapy

Figure 1A and B depicts the cumulative event plots for the probability of appropriate shock and appropriate therapy, stratified by single-chamber ICD, dual-chamber ICD, and CRT-D. Table 2 shows the results from the multivariable cox regression model on risk factors of appropriate and inappropriate shock or therapy. Male gender was identified as a risk factor for both appropriate shock and appropriate therapy (HR = 3.99; CI: 1.75–9.12 and HR = 2.99; CI: 1.73–5.17, respectively). Device type was not associated with the risk of appropriate shocks (Table 2).

View this table:
Table 2

Cox proportional hazards analysis for appropriate and inappropriate DC-shock/therapy

Figure 1

Cumulative event plots of probability of appropriate shock and therapy. Figures estimate the risk of appropriate shock or therapy (defined as shock or anti-tachycardia pacing) stratified by the type of implantable device.

Figure 2A and B depicts cumulative event plots for the risk of inappropriate shock or therapy stratified by type of device. Patients with a dual-chamber ICD had a higher risk of both inappropriate shocks and therapy compared to patients with a single-chamber ICD (HR = 2.45; CI: 1.16–5.14 and HR = 2.38; CI: 1.28–4.42, respectively). No such risks were found for CRT-D (Table 2). Patients with LVEF <25% had increased risks of both inappropriate shock and inappropriate therapy (HR = 2.65; CI: 1.39-5-05 and HR = 2.05; CI: 1.21–3.48, respectively) (Table 2). Furthermore, patients aged 65–74 years were less likely to experience an inappropriate shock compared to patients <65 years (HR = 0.43; CI: 0.21–0.89). A similar trend was made for patients aged ≥75 years although not statistically significant (HR = 0.30; CI: 0.09–1.02).

Figure 2

Cumulative event plots of probability of inappropriate shock and therapy. Figures estimate the risk of inappropriate shock or therapy (defined as shock or anti-tachycardia pacing) stratified by the type of implantable device.

Figure 3

Kaplan–Meier estimates of the probability of survival.

Risk of mortality

Device type or shock/therapy was not associated with mortality (Figure 3 and Table 3). Table 3 lists the results from the multivariable cox regression analyses for all-cause mortality. Patients with LVEF of <25% had a higher mortality compared with patients with an LVEF ≥25% (HR = 1.26; CI: 1.19–2.14) as did patients ≥65 years (Table 3).

View this table:
Table 3

Cox proportional hazards analysis for all-cause mortality

Other analyses

We also performed multivariable cox regression analysis on the risk of mortality with NYHA class instead of LVEF. Here we found an increased mortality among patients with more severe NYHA function class (data not shown). Further, we also repeated all of our analyses including information on β-blocker or amiodarone pharmacotherapy during follow-up as independent variables. No change in risk associated with any existing covariates was identified.


In this nationwide cohort, we studied 1609 patients with IHD and left ventricular dysfunction who received an ICD (single-chamber or dual-chamber) or CRT-D for primary prevention of SCD. Mortality was 12.1% and risk of appropriate shock was 7.8% during a mean follow-up of 1.9 years. Incidence of inappropriate shock and therapy rates were low. Patients with a dual-chamber ICD were more likely to receive both inappropriate shocks and inappropriate therapy compared to patients with a single-chamber ICD.


We found an annual mortality rate of 6–7% in this study. This is similar to results from previous studies of patients with IHD and reduced LVEF receiving an ICD. In MADIT I, MADIT II, and the SCD-HeFt trial, the 2 year mortality in the ICD groups were ∼14,17 ∼15,4 and ∼14%,3 respectively. Of importance, patients in this study were older (∼68 vs. 61–64 years), but received more β-blockers (∼89% vs. 70–82%) compared to these studies.3,4,17 Of note, appropriate and inappropriate shocks were not associated with mortality risk in this ‘real-life’ setting of patients receiving an primary preventive ICD which is in contrast to previous study findings.3,4 Since the mortality rate and the number of shocks (especially inappropriate) were low, lack of power may have influenced these findings and interpretations should be done with caution.

Appropriate therapy

In this study 7.8% of patients experienced an appropriate shock during the study period which corresponds to an annual shock rate of ∼4%. This incidence is significantly lower than the annual shock rates of ∼30% in MADIT, ∼13% in MADIT II and 5–6% in the SCD-HeFt trial (combined IHD and non-IHD). There are several possible reasons for this lower shock rate including differences in ICD programming where higher cut-off for detection of ventricular tachycardia (VT) / ventricular fibrillation (VF) translates into lower rates of appropriate shocks , population differences, and the fact that no patients had a CRT-D in the previous trials.18 A modern, individualized programming utilizing relatively higher rates for VT/VF detection and several attempts of ATP before shock delivery is safe and will also reduce shock incidence and improve mortality as compared with uniform ‘shock-only’ programming used in previous trials.3,14,19,20 Importantly, the latter notion was confirmed in the recent Multicenter Automatic Defibrillator Implantation Trial-Reduce Inappropriate Therapy (MADIT-RIT) which reported risk reduction for both inappropriate shocks and mortality (∼80 and ∼55%, respectively) if alternate programming of the ICD devices were used compared with conventional programming.20

A recent study using ‘real-world’ data from the Ontario ICD database (one-third primary prevention and two-third secondary prevention) using a similar ICD programming with ATP found that 7.3% of patients experienced an appropriate shock and 15–20% received appropriate ATP after one year of follow-up which is higher than this study.13 We cannot rule out that some underreporting of ICD therapy occurred in our study; however, the population in the Ontario study was significantly different, as only one-third of patients in the Ontario ICD database received a primary ICD compared to all in this study. Another explanation for the lower appropriate shock rates in this study could be due to lower arrhythmia rates. This may be a consequence of different baseline comorbidity, since less patients included in this study have had a myocardial infarction and bypass surgery, and β-blocker usage was more frequent compared to MADIT II.4 A similar comparison with the SCD-HeFT study population show how this study population was older and with a greater proportion on β-blockers.3 The latter finding on β-blocker usage is important as patients in optimal medical therapy may experience significant improvements in their LVEF and subsequently lowering their risk of arrhythmias and shocks.

We cannot differentiate how much of the lower shock rate is due to an optimized ICD programing with ATP to avoid shock and how much is related to lower incidence of arrhythmias. In this context, it is important to note that ICD shocks are only surrogate markers for malignant arrhythmia as some arrhythmias are self-terminating. However, if rates VT/VF were low in this study, but all-cause mortality study was high or unchanged compared to previous studies, our findings could suggest a change in patient risk profile from arrhythmia-related deaths towards a non-arrhythmia-related death such as heart failure, as seen in older populations. This observation underlines the importance of additional focus on selecting patients for primary prevention ICD that still have high risk of SCD, but otherwise reasonable life expectancy and lower risk of non-arrhythmic death.

Inappropriate therapy

Inappropriate shock therapy is a severe problem associated with ICD/CRT-D therapy. Importantly, the incidence of inappropriate shocks in this study was low (2.6%) compared to previous reports.21 Improvements in the detection algorithms in dual-chamber ICD technology have been associated with less inappropriate therapy and composite outcomes of clinical adverse events in some studies,22,23 although other studies have not found such improvements.24,25 We identified a two-fold increase in risk of inappropriate shocks and therapy associated with a dual-chamber ICD compared to single-chamber ICD (HR = 2.45; CI: 1.16–5.14 and HR = 2.38; CI: 1.28–4.42, respectively) (Table 2). However, these findings could be attributed to selection bias or confounding by indication. Thus, patients who received a dual-chamber ICD had more evidence of conduction disease, were more likely to be female and were older compared with patients with a single-chamber ICD (Table 1). However, since 70% of inappropriate shocks in our study were related to supraventricular tachycardia, which is in agreement with previous study findings, our findings does not support improved discrimination of VT from supraventricular tachycardia with a dual-chamber device in a ‘real-life’ setting.26

A recent study by Dewland et al.10 compared complications and cardiovascular outcomes associated with dual-chamber ICD vs. single-chamber ICD and found increased risk of both peri-procedural complications and in-hospital mortality with dual-chamber ICDs. We did not find any difference in risk of all-cause mortality associated with device type (Table 3). Nor did we see any increased risk of all-cause mortality associated with appropriate or inappropriate shocks or inappropriate therapy. These findings are important considering that previous studies found inappropriate shocks to be an independent risk factor of mortality among patients with heart failure.27

Recent clinical trials have sought to identify subgroups with the biggest morbidity and mortality yield from primary prevention with an ICD or CRT-D.35,17,28 However, in contrast to the well-defined role of ICD or CRT-D therapy in secondary prevention, identifying distinct subgroups (aside from male gender) for primary prevention has proven more difficult which in part owes to limited stratification opportunities in clinical trials.29 Fueled by the findings from MADIT-II and sub-group analyses from SCD-HeFT current guidelines favour primary preventive ICD in select patients with IHD (evidence level A) over patients with non-IHD (evidence level B).3,4,6 Consequently, primary preventive ICDs have been offered to all patients with left ventricular dysfunction and IHD since 1 January 2007 in Denmark. The Defibrillators in Non-ischaemic Cardiomyopathy Treatment Evaluation trial had a significant reduction in sudden arrhythmic deaths but no overall survival gain among non-IHD patients.21 Patients with non-IHD and left ventricular dysfunction do not routinely receive primary ICDs in Denmark. However, the ongoing DANISH study will provide needed insights into the ICD efficacy on patients with non-ischaemic heart failure.


The main limitation of this study is inherent to the observational nature of the study design although data were collected and entered prospectively. We acknowledge the possibility of selection bias and confounding by indication may have influenced our study findings, including risk of appropriate and inappropriate shocks. In particular, specific device programming settings was not available and we acknowledge that our findings may have been influenced by the lack of such information (i.e. risk of shock with a dual-chamber ICD vs. single-chamber ICD and differences in treatment algorithms between brands). Another limitation relates to the risk of underreporting of events from the reporting centers, in particular relating to episodes of ATP. An audit performed in 1999 on the accuracy and completeness of complication data found that 80% of all complications were reported. Moreover, all centres have a stable and high annual reporting rate with no systematic bias in between reporting centres.30 Of note, since 1999 all participating centres have been encouraged to report events even more carefully. Lastly, we cannot exclude the effect of unmeasured confounders that may have influenced or findings.

The main strength of this study is the ability to evaluate a highly select cohort on a national level and the completeness of data. The study population comprises individuals both in and out of the labour market, independent of sex, age, ethnicity, and participation in insurance programmes which reduces risk of selection bias. Furthermore, nonparticipation was not a concern due to the fact that register-based studies in Denmark do not require informed consent.


In this nationwide study of ‘real-world’ IHD patients with left ventricular dysfunction and a primary preventive ICD, we identified low appropriate shock rates (7.8%) compared to previous clinical trials. This means, only a minority of such patients will benefit from their ICD. Inappropriate shock and inappropriate therapy (shock or ATP) rates were also low (2.6 and 3.7%, respectively). Considering the low rate of both shocks and ATP, our findings could suggest a shift in the risk profile from arrhythmia events towards other causes of death in patients with IHD receiving an ICD for primary prevention of SCD. Such changes may disturb the balance between risk and benefit in the individual patient considered for ICD implantation in the future. The study demonstrates how ‘real-world’ data are important in order to evaluate the safety and efficacy of guideline recommended treatment.

Conflict of interest: None declared.


P.W. was funded by an unrestricted research grant from the Tryg Foundation (J.nr. 7343-09, TrygFonden, Denmark).

Author contributions

P.W. wrote the initial manuscript. All analyses were performed by P.W. and P.K.J. All authors contributed to study design, interpretation of data, and critical revision of the manuscript for important intellectual content. All authors read and approved the final version of the manuscript.


View Abstract