Association between socioeconomic factors and ICD implantation in a publicly financed health care system: a Danish nationwide study

Abstract

Aims

For patients surviving out-of-hospital cardiac arrest (OHCA) with a shockable rhythm, implantable cardioverter defibrillator (ICD) is recommended for non-reversible causes of arrest. We aimed to determine factors associated with implantation of ICD and survival in patients surviving non-AMI OHCA in a nationwide register covering all OHCAs in Denmark.

Methods and results

We identified 36 950 OHCAs between 2001 and 2012, 1700 of whom were ICD naïve, ≥18 years, of non-AMI cardiac aetiology and surviving until discharge. Six hundred fifty eight patients had ICD implanted during index admission. Association to ICD implantation during index admission was analysed in logistic regression, survival was assessed using Cox regression. Implantable cardioverter defibrillator implantation increased during the study period [odds ratio (OR) 1-year increase: 1.04, 95% confidence intervals (95% CI): 1.00–1.08, P = 0.03]. Non-shockable rhythm and age ≥70 years were associated with lower odds of ICD implantation (OR_{non-shockable}: 0.27, 95% CI: 0.19–0.37, P < 0.001, OR_{70–79 years}: 0.71, 95% CI: 0.52–0.98, P = 0.04, OR_{≥80 years}: 0.13, 95% CI: 0.07–0.22, P < 0.001). Non-AMI ischaemic heart disease, highest income tertile and chronic heart failure were associated with higher odds (OR_IHD: 2.51, 95% CI: 1.77–3.60, P < 0.001, OR_{highest income tertile}: 1.58, 95% CI: 1.06–2.23, P = 0.02, OR_HF: 1.77, 95% CI: 1.35–2.32, P < 0.001). Implantable cardioverter defibrillator implantation was associated with a lower risk of mortality (HR: 0.70, 95% CI: 0.53–0.92, P = 0.01).

Conclusion

Implantable cardioverter defibrillator implantation rates increased over the study period. CHF, previous IHD and high income were associated with ICD implantation, while older age and non-shockable rhythm was associated with lower odds of ICD implantation. Implantable cardioverter defibrillator implantation was associated with higher survival rates.

Introduction

Since three randomized clinical trials showed that implantable cardioverter defibrillators (ICD) reduce arrhythmic death and all-cause mortality in patients surviving out-of-hospital cardiac arrest (OHCA),^1–3 ICD implantation has been recommended for this patient group, in the absence of reversible causes of arrest.⁴ Acute myocardial infarction (AMI) with revascularization is generally considered a reversible cause of arrest, and ICD implantation is not usually indicated in these patients unless left ventricular ejection fraction remains below 35% at assessment 40 days post OHCA.⁴ However, in patients without AMI as cause of arrest, ICD should be considered given the patient has an expected survival >1 year and a good neurological function.⁴

However, several clinical and demographic characteristics may affect whether ICD is implanted or not. Previous register studies have shown that patients with ICD implantation were younger, had better neurological function and more often had bystander CPR performed compared to patients not considered for ICD.⁵ Other studies have found that women and patients with more severe comorbidities less often have ICD implanted,^6,7 but recent register data suggest that this difference has become less apparent in recent years.⁷ In the USA uninsured patients less often have ICD implanted,⁸ but in the Danish publicly financed health care system income disparities in utilization should be less likely. At present, pre-arrest factors such as witnessed arrest, bystander CPR and place of arrest and demographic factors such as comorbidity, income and educational level have to our knowledge not been subject to ICD-related studies in a non-AMI OHCA population.

Therefore, in this nationwide study we aimed to (i) assess implantation rates from 2001 to 2012 in patients surviving OHCA not caused by AMI, (ii) investigate which socioeconomic, demographic and clinical factors that were associated with ICD implantation in this patient group, and (iii) describe the subsequent survival in patients with and without ICD implantation post-OHCA.

Methods

Register and study population

This is a retrospective register-based follow-up study, using the Danish Cardiac Arrest Register to assess all OHCA cases in Denmark between 1 June 2001 and 31 December 2012. All OHCA patients with attempted resuscitation by Emergency Medical Services (EMS) or bystanders are recorded in the Danish Cardiac Arrest Register, excluding patients with obvious signs of death, i.e. decapitation. The EMS response consists of an ambulance operated by emergency medical technicians or paramedics, equipped with basic life support equipment. In many cases, a separate rapid response unit consisting of paramedics or emergency trained nurses or a mobile emergency care unit, operated by a paramedic and an anaesthesiologist is also dispatched. The registration is close to complete, as the EMS is dispatched to all OHCA cases in Denmark and EMS personnel are required through contractual agreements to fill out a case report to the Danish Cardiac Arrest Register for every OHCA case, in whom resuscitation was initiated. In terms of health care administration, Denmark is divided in five administrative regions, managing hospitals and the EMS services.

Link to other registries

Through the unique and personal ID-number provided to each Danish citizen, we were able to link information from other registries, as the ID-number is used in all contacts with medical service and government bodies such as tax authorities. Thus, we obtained civil status, age and sex via the Civil Personal Registries, and income and education from Statistics Denmark. Income was divided into three groups based on the average income for the past 5 years prior to the OHCA. The highest education level completed was divided into three groups corresponding to the International Standard Classification of Education (ISCED). Group 1 consists of pre-primary, primary and lower secondary education, corresponding to ISCED 0–2. Group 2 consists of high school or vocational education as well as short-cycle tertiary education, corresponding to ISCED 3 + 5 (There is no education corresponding to ISCED 4 in Denmark). Finally, Group 3 consists of median-length tertiary education, bachelor, master and PhD-degrees, corresponding to ISCED 6–8.

From the Danish National Patient Register, diagnoses necessary to create a Charlson comorbidity index based on diagnoses 10 years prior to OHCA were identified. The Charlson comorbidity index is a weighted, validated index used to predict short-term mortality.⁹ Heart failure diagnoses post-OHCA until ICD implantation or 90 days post-OHCA were also identified and added to the sample (I50).

Patients were followed until death, emigration or 31 December 2012, ensuring between 1 and up to 10 years of follow-up.

Defining an implantable cardioverter defibrillator-eligible population

Implantable cardioverter defibrillator implantation was defined as presence of one of the following Danish procedure codes in the Danish National Patient Register: BFCB0, BFCB00, BFCB01, BFCB02, BFCB03, during index admission post-OHCA. The ICD population was defined as ICD-naïve patients surviving to discharge from index admission post-OHCA, 18 years or older at time of OHCA and with non-AMI cardiac aetiology of arrest (Figure 1). Patients with unknown or missing aetiology of arrest were also included (n = 70). Patients with acute myocardial infarctions (I21–24) as presumed cause of arrest were excluded from this study, as use of ICD and subsequent mortality in this patient group have been described previously.¹⁰ Diagnoses on non-AMI ischaemic heart disease in relation to OHCA was extracted (I25).

Statistics

The main outcomes of this study were factors associated with ICD implantation and 10-year survival. We examined factors associated with ICD implantation during the index hospitalization post-OHCA, temporal trends in implantation and mortality after OHCA. We tested baseline differences between patients with and without ICD implantation during the index hospitalization using chi square test for categorical variables and Student’s t-tests for continuous variables. Increase in comorbidities during the study period in patients with and without ICD implantation was tested using Cochran–Armitage trend test in each group.

In order to assess factors associated with ICD implantation post-OHCA and temporal trends in implantation we performed multiple logistic regression, adjusting for patient age, presence of non-AMI ischaemic heart disease, sex, year of arrest, healthcare administrative region, civil status, Charlson comorbidity index (CCI), income, arrest in public, witnessed arrest, bystander CPR, bystander defibrillation, primary rhythm, and chronic heart failure. We present odds ratios (OR) with 95% confidence intervals (95% CI) for each factor. In addition, we performed logistic regression with educational level and income, educational level and CCI, and educational level, CCI, and income. In order to assess competing risks, we performed analysis assessing factors associated with ICD implantation within 1 year, with death as a competing risk.

As patients receiving an ICD are likely to be less frail than patients not receiving ICD, there is a risk of bias in the survival analysis. Therefore, we only assessed survival in patients who survived or were followed at least 1 year post-OHCA in compliance with guidelines recommending at least 1 year of expected lifetime when considering an ICD implantation.⁴ Using Cox regression, we present hazard ratios (HR) and 95% CI for ICD implantation, age, non-AMI ischaemic heart disease, sex, health care administrative region, civil status, Charlson comorbidity index, income, arrest in public, witnessed arrest, bystander CPR, bystander defibrillation, primary rhythm, and presence of chronic heart failure and year of arrest. We also assessed competing risks between cardiovascular death and death from other causes, adjusting for confounders.

In order to evaluate the association between ICD implantation and mortality during different time periods we divided the study period into periods 2001–2004, 2005–2008, and 2009–2012 and analysed association between ICD implantation and mortality in each period, using Cox regression adjusting for ICD implantation, age, health care administrative region, income, sex, CCI, educational level, chronic heart failure, and primary rhythm.

In all models, age was entered as the following categories: 18–59, 60–69, 70–79, and ≥80 years, as age was not linearly associated with outcomes.

We performed multiple imputation by chained equations to create 50 imputed datasets to account for cumulative missingness in the datasets and analysis was performed on the imputed datasets and pooled.

SAS 9.4 (SAS Institute Inc.) and R 3.3 were used for data management and analysis. For survival analysis we used the R package ‘survival’ and the R package ‘mice’ for imputation.

Ethics

Register studies do not require ethical approval in Denmark. The Danish Data Protection Agency approved the study (2007-58-0015, internal reference GEH-2014-017, I-suite: 02735).

Results

Patients and characteristics

In this study we identified 1700 adult, ICD-naïve patients surviving to hospital discharge after OHCA from presumed non-AMI cardiac aetiology (Figure 1). A total of 658 patients had an ICD implanted during index admission, while 1042 patients did not (Table 1). Age was lower in patients with ICD implantation vs. patients without (59 vs. 65 years, P < 0.001, Table 1). Patients with ICD implantation more often had chronic heart failure, a shockable rhythm, and a favourable pre-hospital profile by having witnessed arrest, bystander CPR, arrest in public, and bystander defibrillation (31% vs. 24%, P = 0.003, 11% vs. 34%, P < 0.001, 81% vs. 57%, P < 0.001, 68% vs. 44%, P < 0.001, 52% vs. 46%, P = 0.01, 9% vs. 4%, P < 0.001).

Patients with ICD implantation were more often diagnosed with non-AMI ischaemic heart disease during the index hospitalization (18% vs. 8%, P < 0.001), but more often had a CCI score of 0 (67% vs. 60%, P < 0.001). Patients with ICD implantation were more often male, not living alone, with higher educational level and income level above the lowest tertile (80% vs. 71%, P < 0.001, 25% vs. 40%, P < 0.001, 64% vs. 51%, P = 0.002, 87% vs. 72%, P < 0.001).

Outcomes

Only 2% of patients with ICD implantation died within the first year post OHCA after being discharged alive, compared to 19% of patients in whom an ICD was not implanted (P < 0.001).

Over time, there was no trend towards increasing comorbidity in patients with ICD, P = 0.27, but this was found in patients without ICD implantation (P = 0.01).

In multiple logistic regression, a non-shockable rhythm and age groups 70–79 and ≥80 years were associated with lower odds of ICD implantation (Table 2) while non-AMI ischaemic heart disease, highest income tertile, witnessed arrest, bystander CPR, bystander defibrillation, health care administrative region 4, and chronic heart failure were associated with higher odds of ICD implantation. Implantation rates increased during the study period (Table 2 and Figure 2). There were no interaction between year of arrest and sex, P = 0.54. In univariate analysis higher educational level was associated with higher odds of ICD implantation; this did not persist when adjusting for income (OR_{ISCED3 + 5}: 1.15, 95% CI: 0.91–1.43, P = 0.23, OR_ISCED6–8: 1.32, 0.84–2.08, P = 0.23). When assessing ICD implantation within the first year in competing risk analysis, age ≥ 80, CCI ≥ 3, and not cohabiting with a partner were associated with lower odds of ICD implantation (OR_age80: 0.17, 95% CI: 0.09–0.31, P < 0.001, OR_CCI3: 0.52, 95% CI: 0.33–0.84, P = 0.01, OR_{not cohabiting}: 0.64, 95% CI: 0.46–0.9, P = 0.01) while non-AMI ischaemic heart disease, bystander CPR, year of arrest (1 year increase), witnessed arrest, and chronic heart failure were associated with higher odds of ICD implantation (OR_{non-AMI IHD}: 2.69, 95% CI: 1.88–3.86, P < 0.001, OR_CPR: 1.37, 95% CI: 1.02–1.83, P = 0.03, OR_year: 1.07, 1.03–1.12, P = 0.001, OR_witnessed: 2.62, 95% CI: 1.90–3.61, P < 0.001, OR_HF: 1.8, 95% CI: 1.35–2.39, P < 0.001).

Implantable cardioverter defibrillator implantation and highest income was associated with lower mortality (Table 3), while all ages older than 60, CCI scores 2 and ≥3, chronic heart failure, non-shockable rhythm and year of arrest were associated with higher mortality. There was no interaction, neither between ICD implantation and rhythm, P = 0.73, or between sex and year of arrest, P = 0.13.

In log-rank test, ICD implantation was associated with reduced mortality in 2001–2004 as well as 2005–2008 and 2009–2012. This effect persisted in 2001–2004 upon adjusting for confounders, HR: 0.44, 95% CI: 0.26–0.73, P = 0.001, but this was not the case for 2005–2008 and 2009–2012 at HR: 1.00, 95% CI 0.64–1.54, P = 0.99 for 2005–2008 and HR: 0.72, 95% CI: 0.44–1.19, P = 0.20.

Implantable cardioverter defibrillator implantation was associated with reduced cardiovascular mortality (OR: 0.35, 95% CI: 0.22–0.55, P < 0.001) and there was a trend towards reduced mortality from other causes (OR: 0.59, 95% CI: 0.32–1.10, P = 0.10).

Discussion

Patients with highest level of income had higher odds of ICD implantation following a non-AMI related OHCA even after correction for available confounding factors. This finding was apparent despite the access to health care is free of charge to all Danish citizens and despite adjusting for confounders. We found several other factors to be associated with odds of ICD implantation: age older than 70 and non-shockable rhythm decreased odds of ICD, while witnessed arrest, bystander CPR, bystander defibrillation, non-AMI ischaemic heart disease, chronic heart failure, and health care administrative region 4 increased the odds of ICD implantation.

We also observed that ICD implantation increased significantly during the study period.

Finally, we observed a significant reduction in overall mortality in patients with ICD implanted, this was significant in death from cardiovascular causes, and again, a significantly higher survival was found in patients with higher income.

Income disparities in implantable cardioverter defibrillator implantation

Disparities have previously been observed for bystander CPR rates and outcome in OHCA in areas with lower socioeconomic status.¹¹ The reasons for our observation is likely multifactorial. It is possible that this result is in part due to a higher frailty burden in the lowest income tertile beyond what can be captured by the CCI, but at present, we cannot exclude an underutilization of ICD implantation in the lowest income tertile in patients with non-AMI ischaemic heart disease. We also found that highest income tertile was associated with lower mortality. Others have found an association between education and survival after OHCA,¹² and Rajan et al.¹³ previously described differences in 30-day survival in children based on parental income in Denmark. As the survival assessed in this study is 10-year survival, we cannot rule out that the observed higher survival in highest income group is to some extent attributed to a healthier lifestyle leading to a longer life in general.¹⁴ In our study, educational level was univariately associated with higher odds of ICD implantation, but this effect was not significant upon adjusting for income.

Factors associated with ICD implantation post-OHCA

Our observed association between higher age (older than 70 years) and lower odds of ICD implantation is not novel.¹⁵ Guidelines do not recommend using a cut-off age for deciding not to implant,⁴ but rather an expected lifetime >1 year post-OHCA. However, studies point towards less benefit of ICD implantation in patients older than 70 years.¹⁶ As such, it may not be surprising that patients older than 70 years of age have lower odds of ICD implantation.

Pre-hospital circumstances including witnessed arrest, bystander CPR and bystander defibrillation were all associated with higher odds of ICD implantation. This observation should likely be interpreted as patients with favourable pre-hospital circumstances having a good prognosis and thus being more likely candidates for ICD implantation, as clinicians may have selected these patients with greater odds of survival >1 year post-OHCA for ICD implantation.

Implantation rates of implantable cardioverter defibrillator between 2001 and 2012

In this study, we found an increase in ICD implantation rates during the last decade. A recent study from Switzerland also assessed implantation rates in secondary prevention, and also found increasing implantation rates in non-ischaemic cardiomyopathy.¹⁷ Even though some find low ICD implantation rates,¹⁸ older studies find increasing implantation rates in the USA and Canada from mid-1990s to early 2000s.⁶ Implantation rates are generally higher in the USA than in Europe, and there is great variation between countries in Europe.¹⁹ We did not find different implantation rates for men and women in our population during the study period.

Survival after implantable cardioverter defibrillator implantation in out-of-hospital cardiac arrest survivors

This study finds a survival benefit in patients with ICD implantation and a reduced mortality in death from cardiovascular causes, and a trend towards lower mortality from other causes. Furthermore, we find that ICD implantation was associated with reduced mortality in 2001–2004, but not in 2005–2008 and 2009–2012. This may indicate that the benefit of ICD implantation is reduced when more patients have ICD implanted. However, as this is a retrospective observational study, there is likely confounding by indication in relation to these results.

Additionally, we found an increasing mortality with increasing age group, this is to be expected and well documented in OHCA populations, even though it must be acknowledged that elderly can be resuscitated and live long lives post-OHCA.²⁰ Charlson comorbidity index scores of 2 and ≥3 as well as chronic heart failure were all associated with higher mortality; this was to be expected. When considering long-term mortality, comorbidity burden is likely to be influential, but we did not find CCI to increase over time in ICD-patients during the study period.

Limitations

In addition to previously noted limitations, there is a risk of residual confounding in this study as in all observational studies. Furthermore, we cannot detect causal relationships, but merely observe associations that are not necessarily causal. Information on reasons not to implant or implant an ICD was not available. As such, it is possible that some or all of the survival benefit are driven by confounding by indication in that patients who had an ICD implantation had a number of favourable characteristics and therefore carry a better prognosis that those who did not have an ICD implanted.

Conclusion

Implantable cardioverter defibrillator implantation rates increased in patients surviving OHCA not caused by AMI between 2001 and 2012. Several factors were associated with higher odds of ICD, most notably; high income was associated with higher odds of ICD implantation, even in a public health care system.

Implantable cardioverter defibrillator implantation was associated with a survival benefit overall, this was significant in the years 2001–2004, but not in 2005–2008 and 2009–2012.

Funding

This work was supported by a grant from the Research Committee at Rigshospitalet to MSc. Winther-Jensen. Dr Lippert reports unrestricted research grants from The Danish foundation TrygFonden and the Laerdal foundation. Dr Steen Hansen has received a research grant from the Danish Heart Foundation and the Laerdal Foundation. Dr Kragholm reports having received grants from The Laerdal Foundation, outside the present study. The Danish Out-of-hospital Cardiac Arrest Registry is supported by TrygFonden. Professor Christensen is supported by TrygFonden, through a non-personal grant to Aalborg University.

Conflicts of interest: Dr Hassager reports personal lecture fee from Novartis, Orion, TEVA, and ViCare, outside the submitted work. Dr Kjaergaard reports personal lecture fees from Astra Zeneca, Orion Pharma and Bayer, outside the submitted work. Dr Lassen reports personal lecture fees from Astra Zeneca, and Boehringer Ingelheim outside the present study. Dr Kragholm reports having received speaker’s honorarium from Novartis. Professor Køber reports personal lecture fee outside this work from Novartis and Sanofi-Aventis. Professor Torp-Pedersen has received grants and speakers honoraria from Bayer and a grant from Biotronic, outside of the present study. All other authors report no conflicts of interest.

References