Socio-economic position and sudden cardiac death: a Danish nationwide study

Abstract

Aims

The aim of this study was to examine differences in incidence rates of all-cause mortality (ACM) and sudden cardiac death (SCD) in persons of differing socio-economic position (SEP).

Methods and results

All deaths in Denmark from 1 January to 31 December 2010 (1 year) were included. Autopsy reports, death certificates, discharge summaries, and nationwide health registries were reviewed to identify cases of SCD. Socio-economic position was measured as either household income or highest achieved educational level and analysed separately. Hazard rates were calculated using univariate and multivariable Cox regression models adjusting for age, sex, and selected comorbidities. A total of 53 452 deaths were included, of which 6820 were classified as SCDs. Incidence rates of ACM and SCD increased with age and were higher in the lower SEP groups. The greatest difference in SCD incidence was found between the low and high education level groups, with an incidence rate ratio of 5.1 (95% confidence interval 3.8–6.8). The hazard ratios for ACM and SCD were significantly higher for low SEP groups, independent of comorbidities. Compared with the highest income group, the low-income group had adjusted hazard ratios of ACM and SCD that were 2.17 (2.01–2.34) and 1.72 (1.67–1.76), respectively.

Conclusion

We observed an inverse association between both income and education level and the risk of ACM and SCD in the general population, which persisted independently of baseline comorbidities. Our results indicate a need for further research into the mechanisms behind socio-economic disparities in healthcare and targeted preventative strategies.

Graphical Abstract

Graphical Abstract.

What’s new?

• This was a nationwide study of all-cause mortality (ACM) and sudden cardiac death (SCD) across socio-economic position (SEP) in the general population of Denmark in the year 2010.

• A low income and education level were both associated with an increased risk of ACM and SCD, which persisted when adjusting for age, sex, and comorbidities.

• Further research is needed to clarify the mechanisms that lead to higher SCD risk in groups of low SEP.

Introduction

Sudden cardiac death (SCD) refers to a sudden, unexpected death caused by a cardiovascular or unknown cause.¹ Sudden cardiac death continues to be a major public health problem, accounting for a substantial number of deaths globally.^2–5 Approximately 13% of all deaths in Denmark can be ascribed to SCD.⁵

Previous studies have demonstrated a consistent association between socio-economic position (SEP) and risk of cardiovascular disease (CVD), myocardial infarction, and SCD.^6–9 The effect of SEP on cardiovascular mortality is complex and is mediated by a multitude of behavioural, biological, and psychological mechanisms,¹⁰ such as differences in the risk factors of CVD, including differences in diet, body mass index (BMI), physical activity, and smoking.^10–12 Furthermore, societal factors such as unequal access to healthcare¹³ and disparate standards of care¹⁴ also impact the relation between SEP and SCD.

While earlier regional studies have described this relationship, these have been conducted in countries with varying access to healthcare and education.^8,9,15,16 Demographic variations may obscure the effect of SEP in a regional study compared with an unselected population. In Denmark, there is free and universal healthcare for all residents, thus lowering the financial barriers in healthcare access. Despite low structural barriers to healthcare access, discrepancies in health outcomes across SEP persist.¹⁷

Sudden cardiac death prediction and prevention are currently challenging in the general population, and it would be preferable if better risk prediction tools could be developed in the future to identify individuals at high risk.¹⁸ Therefore, further investigation in unselected populations into the relationship between SEP and SCD is paramount to ensure that future prevention strategies are equitable. Consequently, the aim of this study was to explore how SEP is associated with incidence rates and risk of all-cause mortality (ACM) and SCD in a nationwide and unselected setting.

Methods

Study design and population

All living persons in Denmark on 1 January 2010 were included in this cohort and followed until death or the end of 2010. All deaths were manually reviewed based on the information from death certificates, autopsy reports, if performed, and discharge summaries. All deaths in Denmark were included in the analysis of ACM. Sudden cardiac death cases were specifically identified and adjudicated through the following process.⁵

One physician screened all death certificates and excluded cases that were obviously non-sudden (e.g. non-natural deaths and deaths in patients with terminal cancer). On any indication that the case might be sudden and unexpected (SD), the case was reviewed independently by two physicians to identify SD cases. In cases of disagreement, all available information were re-evaluated to reach consensus. Sudden cardiac death was identified as SD with an underlying or suspected cardiac cause. Sudden death with an underlying cardiac cause (i.e. SCD) was identified through a comprehensive review of autopsy records, the Danish National Patient Register, and the National Causes of Death Register.

Definition of sudden cardiac death

Sudden death was defined as a sudden, natural, and unexpected death. Autopsied cases of SD of unknown or cardiac cause, as well as cases of SD presumed to be of cardiac origin, were classified as SCD and further subcategorized based on the criteria provided in Supplementary material online, Table S1.

Danish healthcare system

The Danish healthcare system is publicly funded and provided by the Danish government for all permanent residents. Residents of Denmark have a unique Civil Registration Number, which makes all interactions between patients and medical professionals traceable. Furthermore, medical records can be retrieved from the Danish National Patient Register which contain information on all in- and out-patient activities at Danish hospitals since 1994 using the International Classification of Disease, Tenth Revision (ICD-10). Information on the cause of death can be obtained from the Danish Cause of Death Register, in which immediate, contributory and underlying causes of death are recorded using ICD-10 codes.¹⁹

Death certificates and autopsy

Danish law requires that death certificates must always be issued by a medical doctor, who based on all available information, determines the most likely cause of death. Both immediate and underlying causes of death are included in the death certificate. In cases of sudden or unexpected death, the police are, by law, notified and may enquire about a medicolegal external examination in cooperation with a public health medical officer. The medical officer has access to the medical records of the deceased, including first responder calls, police reports including eyewitness accounts and statements from general practitioners, as well as the body of the deceased. This information is included in the supplementary information field of the death certificate. Information on the circumstances of death is often included in the supplementary information field, even in the absence of an external examination, which makes the Danish death certificates a useful tool for identifying SD.^20,21 A forensic autopsy may be conducted if the manner of death is unknown. If a forensic autopsy is not performed, a local pathology department may conduct a hospital autopsy upon the request of a physician and/or relatives of the deceased.

Socio-economic position

We chose equivalized disposable income as our primary indicator for SEP. This is defined as the total disposable income of a household divided by the individuals living in the household, applying a weighted scale where the first adult counts as 1, further adults as 0.5, and individuals younger than 14 years as 0.3. Any persons above the age of 24 years still living with their parents were defined as a separate household.²² Therefore, the equivalized disposable income provides a representation of the economic resources within a household, considering economies of scale and varying needs based on age. To account for yearly variations, the average equivalized income over 10 years prior to the study start was used to classify persons in income quartiles. The study population was divided into age-group-specific income quartiles in the following age groups: 0–25, 25–50, 50–75, and +75 years. The cohort was assigned to the income quartiles Q1, Q2, Q3, and Q4 corresponding to the lowest, second lowest, second highest, and highest income levels. For persons under the age of 10 years, income was averaged in years corresponding to their age. If income data were missing, or average income was below 0, the person was excluded from the study. Income data were obtained from the Danish Income Statistics Register.²³

Secondary analysis was done according to the highest attained educational level. The Danish Education Register categorizes completed education levels based on the International Standard Classification of Education into three groups: (i) lower secondary education (ISCED 0–2), (ii) upper secondary education (ISCED 3), and (iii) tertiary education (ISCED 4–8), encompassing bachelor’s, master’s, and doctoral degrees or equivalents. The secondary analysis only included patients above the age of 25 and was divided into age groups of 25–50, 50–75, and +75 years. Education data were obtained from the Danish Education Registers.²⁴

Statistical methods

Statistical analysis was performed with the use of SAS Enterprise V.7.15 as well as R 4.4.1. Population data for incidence calculations were retrieved via Statistics Denmark, using the population on 1 January 2010, as reference. Incidence rates are given per 100 000 person-years. The 95% confidence interval (CI) for incidence rates was calculated using a Poisson distribution. We used univariate and multivariable Cox analyses to calculate the hazard ratios (HRs) of SCD and ACM between income and education groups, respectively. The multivariable Cox proportional hazards model was adjusted for age, sex, and baseline comorbidities (heart failure, arrhythmia, ischaemic heart disease, cardiomyopathy, peripheral artery disease, diabetes mellitus, chronic obstructive pulmonary disease, cancer, cerebrovascular disease, epilepsy, liver disease, neurological disease, psychiatric disease, and sleep apnoea). The ICD-10 codes for the comorbidities listed are found in Supplementary material online, Table S2.

Results

Study population

The population of Denmark was ∼5.5 million as of 1 January 2010. There were a total of 54 028 deaths throughout 2010, where 7627 (14.1%) were classified as SDs and 6867 (12.7%) were classified as SCDs. After applying the exclusion criteria, there was sufficient income data for a background population of ∼5.4 million people, 53 452 deaths, and 6820 SCDs (Figure 1). The median age of the population was 41 (interquartile range 20–59) and 51% were female. The total time of follow-up was 5.38 million person-years. The population was split into age-group-specific income quartiles, with 1.35 million in each quartile with varying baseline characteristics (Table 1). People in the low-income group were less educated, more likely to be female, had a higher comorbidity burden, and accounted for a higher number of total deaths and SCDs. Characteristics of SCD victims are summarized in Supplementary material online, Table S3.

Figure 1.

Flow chart of the study population and sudden cardiac death case identification process.²⁵

Table 1.

Clinical and demographic characteristics of the background population by income quartiles

Characteristic	Overall, n = 5 406 051a	Q1, n = 1 351 515a	Q2, n = 1 351 512a	Q3, n = 1 351 511a	Q4, n = 1 351 513a
Age	41 (20, 59)	37 (22, 62)	40 (21, 58)	41 (19, 57)	44 (18, 59)
Age group
0–25	1 656 240 (31)	414 061 (31)	414 059 (31)	414 060 (31)	414 060 (31)
25–50	1 779 504 (33)	444 876 (33)	444 876 (33)	444 876 (33)	444 876 (33)
50–75	1 584 324 (29)	396 081 (29)	396 082 (29)	396 080 (29)	396 081 (29)
+75	385 983 (7.1)	96 497 (7.1)	96 495 (7.1)	96 495 (7.1)	96 496 (7.1)
Female sex	2 731 402 (51)	712 285 (53)	698 589 (52)	665 867 (49)	654 661 (48)
Education level
Low	1 600 451 (36)	567 113 (52)	447 103 (40)	345 853 (31)	240 382 (22)
Medium	1 916 496 (44)	376 542 (35)	505 383 (45)	550 645 (50)	483 926 (45)
High	870 480 (20)	137 487 (13)	164 981 (15)	208 860 (19)	359 152 (33)
Unknown	1 018 624	270 373	234 045	246 153	268 053
Death	53 452 (1.0)	19 226 (1.4)	14 537 (1.1)	10 939 (0.8)	8750 (0.6)
Sudden cardiac death	6820 (0.1)	2693 (0.2)	1876 (0.1)	1295 (<0.1)	956 (<0.1)
Comorbidities
Diabetes mellitus	139 036 (2.6)	49 669 (3.7)	37 608 (2.8)	28 892 (2.1)	22 867 (1.7)
Chronic kidney disease	39 824 (0.7)	12 976 (1.0)	10 678 (0.8)	8535 (0.6)	7635 (0.6)
Chronic obstructive pulmonary disease	86 690 (1.6)	34 526 (2.6)	23 808 (1.8)	16 664 (1.2)	11 692 (0.9)
Cancer	173 065 (3.2)	42 726 (3.2)	42 187 (3.1)	41 498 (3.1)	46 654 (3.5)
Cerebrovascular disease	104 020 (1.9)	32 071 (2.4)	27 807 (2.1)	22 980 (1.7)	21 162 (1.6)
Neurological disease	348 514 (6.4)	95 488 (7.1)	94 204 (7.0)	82 685 (6.1)	76 137 (5.6)
Epilepsy	50 382 (0.9)	17 451 (1.3)	14 243 (1.1)	10 088 (0.7)	8600 (0.6)
Psychiatric diagnosis	239 380 (4.4)	101 504 (7.5)	63 785 (4.7)	42 732 (3.2)	31 359 (2.3)
Liver disease	35 082 (0.6)	15 904 (1.2)	8307 (0.6)	5902 (0.4)	4969 (0.4)
Sleep apnoea	33 784 (0.6)	7041 (0.5)	9082 (0.7)	9057 (0.7)	8604 (0.6)
Cardiovascular disease	724 206 (13)	201 976 (15)	188 693 (14)	172 570 (13)	160 967 (12)
Heart failure	59 885 (1.1)	20 902 (1.5)	16 129 (1.2)	12 423 (0.9)	10 431 (0.8)
Arrhythmia	154 425 (2.9)	42 578 (3.2)	38 847 (2.9)	36 056 (2.7)	36 944 (2.7)
Ischaemic heart disease	193 152 (3.6)	60 740 (4.5)	51 113 (3.8)	43 107 (3.2)	38 192 (2.8)
Myocardial infarction	63 568 (1.2)	19 948 (1.5)	16 969 (1.3)	14 636 (1.1)	12 015 (0.9)
Cardiomyopathy	11 468 (0.2)	3639 (0.3)	3062 (0.2)	2475 (0.2)	2292 (0.2)

^aMedian (interquartile range); n (%).

All-cause mortality and sudden cardiac death

Income

The incidence rates across income levels are shown in Table 2 and further stratified by age group in Table 3.

Table 2.

Incidence rates per 100 000 person-years and incidence rate ratios of sudden cardiac death and all-cause mortality by income group with 95% confidence interval

Quartile groups	IR SCD	IRR	IR ACM	IRR
Q1	201 (193–208)	2.8 (2.2–3.8)	1433 (1413–1453)	2.2 (2–2.4)
Q2	140 (133–146)	2 (1.5–2.7)	1082 (1064–1099)	1.7 (1.5–1.8)
Q3	96 (91–102)	1.4 (1–1.9)	813 (798–828)	1.9 (1.3–1.1)
Q4	71 (67–76)	–	650 (636–663)	–

ACM, all-cause mortality; IR, incidence rate; IRR, incidence rate ratio; SCD, sudden cardiac death.

Table 3.

Incidence rates per 100 000 person-years of sudden cardiac death and all-cause mortality by income group stratified by age groups with 95% confidence interval

Quartile group	Age group	IR SCD	IR ACM
Q1	0–25	2.9 (1.5–5.1)	28 (23–33)
Q2		2.2 (1.0–4.1)	17 (14–22)
Q3		1.2 (0.4–2.8)	12 (9–16)
Q4		1.0 (0.3–2.5)	17 (13–22)
Q1	25–50	32 (27–37)	208 (195–222
Q2		23 (19–28)	121 (111–131)
Q3		10 (7–13)	86 (77–95)
Q4		11 (8–14)	74 (66–83
Q1	50–75	354 (335–373)	2210 (2164–2257)
Q2		171 (159–185)	1169 (1136–1204)
Q3		83 (75–93)	698 (672–725)
Q4		62 (55–71)	553 (530–576)
Q1	+75	1260 (1189–1336)	10 399 (10 191–10 610)
Q2		1186 (1117–1259)	10 162 (9957–10 371)
Q3		991 (928–1058)	8381 (8195–8569)
Q4		706 (653–762)	6604 (6440–6771)

ACM, all-cause mortality; CI, confidence interval; IR, incidence rate; SCD, sudden cardiac death.

The incidence rate per 100 000 person-years for ACM was significantly higher for Q1 than Q4 being 1433 (95% CI: 1413–1453) vs. 650 (95% CI: 636–663) with an incidence rate ratio (IRR) of 2.2 (95% CI: 2–2.4). All-cause mortality was significantly higher for Q1 than Q4 across all age groups (Figure 2).

Figure 2.

Incidence rates per 100 000 person-years of all-cause mortality (A) and sudden cardiac death (B) by age-adjusted income quartile and age group.

The incidence rate of SCD per 100 000 person-years for Q1 was significantly higher than Q4 being 201 (95% CI: 193–208) and 71 (95% CI: 67–76), respectively, resulting in an IRR of 2.8 (95% CI: 2.2–3.8). Across the age groups 25–50, 50–75, and +75 incidence rates for SCD were significantly higher in Q1 compared with Q4 (Figure 2). Both ACM and SCD showed an inverse relationship to income across all age groups, though the results of the 0–25 age group of SCD victims were non-significant.

The results of the multivariable Cox analysis are shown in Figure 3. The Q1 group had a SCD HR of 2.2 (95% CI: 2.01–2.34) and an ACM HR of 1.7 (95% CI: 1.67–1.76) when referenced to Q4. Both the Q2 and Q3 quartile groups had a significantly higher HR of SCD, and ACM compared with Q4. The results of the univariate Cox analysis are seen in Table 4.

Figure 3.

Results of the multivariable Cox analysis of all-cause mortality (A) and sudden cardiac death risk (B) across age-adjusted income quartiles.

Table 4.

Univariate and multivariable Cox analyses of sudden cardiac death and all-cause mortality risk across income quartiles

	SCD risk—Hazard ratio		ACM risk—Hazard ratio
Income group	Univariate	Multivariablea	Univariate	Multivariablea
Q1	2.82 (2.62–3.0)	2.17 (2.01–2.34)	2.20 (2.15–2.26)	1.72 (1.67–1.76)
Q2	1.96 (1.81–2.12)	1.72 (1.59–1.86)	1.66 (1.62–1.71)	1.46 (1.43–1.50)
Q3	1.36 (1.25–1.47)	1.32 (1.21–1.44)	1.25 (1.22–1.29)	1.23 (1.20–1.27)
Q4	1.0 (Ref.)	–	–	–

ACM, all-cause mortality; SCD, sudden cardiac death.

^aThe multivariable Cox analysis was adjusted for age, sex, and comorbidities (heart failure, arrhythmia, ischaemic heart disease, cardiomyopathy, peripheral artery disease, diabetes mellitus, chronic obstructive pulmonary disease, cancer, cerebrovascular disease, epilepsy, liver disease, neurological disease, psychiatric disease, and sleep apnoea).

Education

After excluding persons with missing education data and persons under the age of 25, there were 3.7 million persons in the secondary analysis. There were 43 326 deaths and 5743 SCDs with available education data. Clinical characteristics of the population are summarized in Supplementary material online, Table S4 and the SCD victims in Supplementary material online, Table S5.

The incidence of ACM and SCD was inversely associated with education level in a pattern similar to the association of income levels (see Supplementary material online, Table S6 and Figure S1). The incidence rates per 100 000 person-years of ACM for the low and high education level groups were 2245 (95% CI: 2216–2264) and 515 (95% CI: 500–530), respectively, with an IRR of 4.4 (4–4.8), while the SCD incidence rate was 302 (95% CI: 291–312) and 60 (95% CI: 55–65) with an IRR of 5.1 (95% CI: 3.9–6.8). When stratifying by age groups, the incidence rates of ACM and SCD were significantly higher for the low and medium education levels when compared with the high education level (see Supplementary material online, Table S7). The HRs for the low and medium education level groups were 1.6 (95% CI: 1.58–1.68) and 1.3 (95% CI: 1.25–1.34) for ACM risk (see Supplementary material online, Figure S2) and 1.9 (95% CI: 1.77–2.13) and 1.5 (95% CI: 1.33–1.61) for SCD risk (see Supplementary material online, Figure S2). The results of the univariate and multivariable Cox analyses of education level and ACM and SCD risk are summarized in Supplementary material online, Table S8.

Discussion

In this nationwide study of SEP and SCD, we found that SEP expressed either as income or education was inversely associated with a higher incidence of ACM and SCD. Our results are largely in line with previous studies that have found an inverse association between household income and education with SCD risk.^6,8,9 Nonetheless, our study provides novel findings regarding ACM and SCD risk and incidence rates in relation to SEP. To the best of our knowledge, this is the largest nationwide, unselected study of the socio-economic factors of SCD across all age groups. Furthermore, this study is conducted within a setting where universal access to healthcare and education lowers barriers, which may provide a clearer assessment of the influence of SEP on SCD risk.

The differences in incidence rate for SCD and ACM were most pronounced in the low education level group, having a five- and four-fold increase in incidence rate, respectively, compared with the high education level group. These differences were found across all age groups of both SEP measures except in the 0–25 age group of the income-based analysis of SCD incidence, though a non-significant trend was noticeable. The relative difference in incidence rate of SCD and ACM increased across age groups in the income analysis until the +75 age groups (see Supplementary material online, Table S9). In contrast, the relative difference decreased in the education analysis across age. Similar results have been noted by Jonsson et al.²⁶ in out-of-hospital cardiac arrest (OHCA) though the authors used an area-based SEP index consisting of both income and education data. Our results may add to the discussion of the cumulative disadvantage hypothesis (CDA) and the age-as-leveller hypothesis (AAL).²⁷ The CDA hypothesis explores how inequality develops over the life course and why the relationship between factors such as SEP and health may strengthen with age.²⁸ The AAL hypothesis finds that health inequalities may weaken as age increases.²⁹ Our results signal that both may be true in regard to the effect of income, while the effect of education points more towards AAL. Furthermore, there was a significant difference in the HRs for ACM and SCD between all levels of SEP compared with the highest SEP groups. This increased risk persisted independently of age, sex, and comorbidity load, and was most prominent in the low-income group with an over two-fold increase in SCD risk. This is in contrast to Hahad et al.,³⁰ who found education and occupational status to be more strongly associated with CVD and ACM than income. Income, even when averaged across 10 years, may depict a more present-day indication of SEP. Conversely, education mirrors the early life circumstances’ effect on adult health and will be a determinant of future income.¹¹ Therefore, it could be argued that adjusting for comorbidities in the education analysis may excessively attenuate the association between SEP and SCD/ACM risk, as comorbidities may mediate this relation to a higher degree when compared with the effect of income. This is seen in the larger difference between the results of the univariate and multivariable Cox models on education than income (Table 4 and Supplementary material online, Table S8).We chose to separate the income and education analysis, as both measure SEP in different aspects and may therefore provide a better understanding of the association between SEP and SCD risk while avoiding overadjustment.¹¹ Furthermore, we had more extensive coverage of income data than education data.

The increase in ACM and SCD risk was found despite access to universal healthcare and education in Denmark. However, drugs, while subsidized, are not free of charge, and there still exists a social gradient in healthcare outcomes.⁶ The association between SEP and SCD risk may vary between countries with different social security and healthcare systems. An article from Choi et al.³¹ compared a wide array of health outcomes between US and UK adults between the ages of 55 and 64 years. They found a significantly higher disparity of outcome between low- and high-income persons in the US population when compared with the UK population. Following this, results from Landon et al.³² described how low-income patients from six different countries presenting with acute myocardial infarction had poorer survival outcomes and readmission rates than high-income patients while also being less likely to undergo revascularization procedures. In contrast, they found no significant difference in disparity of outcome when comparing the included countries, one of them being the USA, despite them having varying degrees of public healthcare service. Their results suggest that large income-based disparities in health exist even in countries with universal healthcare and a strong social safety network. It should be noted that the US patients included were older, making them eligible for Medicare, which may make their healthcare coverage more similar to the other countries included. As of now, studies comparing the association of SEP and SCD risk between countries are lacking.

Societal factors, such as unequal access to healthcare, are known to disproportionately impact persons with a low SEP, leading to poorer health-related outcomes.⁷ Conversely, multiple studies have described how individual factors, such as health-related behaviours are also important in driving inequity in rates of SCD. Warming et al.⁶ described how ∼20% of the difference in SCD risk between low and high SEP cases was mediated by differences in the modifiable risk factors of smoking, BMI, and physical activity. Further, Zhang et al.² found that lifestyle only accounted for 12% of the association of socio-economic inequality and CVD and mortality in the US and UK adults of differing SEP. In contrast, a study from the Netherlands by Méjean et al.¹² found that diet, smoking, and alcohol consumption explained over 60% of the difference in coronary artery disease (CAD) across education levels.¹² It is possible that health behaviours may mediate the effect of SEP on SCD to a lesser extent than CAD despite it being the major cause of SCD.³³

While the literature presents varying estimates of the effect of modifiable risk factors on SCD, there is consensus in the fact that behavioural factors are important effect modifiers between SEP and SCD. Furthermore, it has been shown that persons with a low SEP are more susceptible to the detrimental effects of modifiable risk factors (e.g. smoking, poor diet, and physical inactivity) compared with persons with a high SEP, despite being equally exposed.^34,35 Similarly, our Cox analysis shows that SCD risk was significantly higher in low SEP groups even when adjusting for comorbidities. This suggests that SEP affects the risk of SCD not merely as a product of an increased comorbidity burden and may be a beneficial inclusion in future risk stratification.

Our results suggest that the risk of SCD may be more affected by differences in SEP when compared with the risk of ACM. Previous data on OHCA in Denmark has shown that low-income groups are less likely to receive cardiopulmonary resuscitation and have a pre-hospital shockable rhythm.³⁶ Additionally, Møller et al.³⁷ found that low-income patients presenting with OHCA were less likely to have coronary angiographies performed, while only the lowest income patients had lower rates of revascularization. These results may partly explain the difference in risk of SCD compared with ACM in low SEP patients. Additionally, health literacy, which describes the ability to function in a healthcare environment, may in part mediate the relationship between low SEP and CVD as it affects persons with a low SEP disproportionally.³⁸ Consequently, while reduction of risk behaviour will reduce mortality in low SEP groups, the socio-economic inequity in SCD risk may not be attenuated through healthy lifestyle alone. Further investigation is necessary to clarify the multifactorial mechanisms that mediate the association between SEP and SCD risk.

Clinical implications

Adjusting modifiable risk factors remains important in the prevention of CVD, and thus also in preventing SCD.¹⁸ Additionally, patient-centred interventions with a focus on low SEP patients or regions may be an efficient use of healthcare resources in the prevention of SCD.⁷ These may include economic policies aiming at making healthy choices affordable,³⁹ as well as taxation to disincentivise tobacco usage.³⁸ Furthermore, it may be useful to include SEP as a part of SCD risk assessment, as at-risk patients of lower SEP may not be identified by traditional risk factors alone.⁷ This may be due to SEP not usually being emphasized as a true risk factor for SCD though its emerging importance has been noted.⁴⁰ Socioeconomic position may be important to consider as SCD risk stratification continues to be a challenge in the general population.⁴¹ Patient-level analyses have sometimes shown conflicting results, which highlights the difficulties within understanding the relationship between SEP and SCD and the mechanisms that mediate it.⁴²

Our results also raise the question of equality vs. equity in a publicly funded healthcare system. Namely, whether more resources should be allocated towards targeted interventions among low SEP groups in order to reduce disparities in healthcare outcomes. Following this, outreach programmes, including health education and self-management of cardiovascular risk factors for low SEP groups, may be particularly important early in life. This is because cardiovascular health in children is predictive of cardiovascular health in adulthood³⁹ and disparities in health outcomes may be the result of accumulated exposure throughout life.⁴³ As the 2022 ESC Guidelines highlight the importance of adequate availability of public access defibrillation and community training in basic life support, our results indicate that it may be a priority to target low SEP communities.⁴¹ Furthermore, political initiatives that ensure affordable healthy diets may reduce the risk of SCD in low-income groups, as diet seems to be a determining factor in the association between SEP and CAD,⁴⁴ while CAD continues to account for the majority of SCD cases.⁴²

Strengths

A major strength of this study is the large sample size in the Danish national registries, which results in minimal risk of selection bias and loss to follow-up. It is imperative to have data on the whole population when investigating a topic such as SEP, as regional variations may exist. Additionally, it is a benefit to use equivalized annual income based on a 10-year average, as this reduces the influence of acute illness while accounting for yearly variation. Moreover, the quality of our income data is very robust, providing extensive coverage with almost no missing data of household income.

We also chose to forego using a compound SEP variable in favour of standardized economic and educational markers, as this should increase ease of comparison between research articles. This may also partly explain why we found household income, and not education, to be more strongly associated with ACM and SCD risk, though both were strongly associated. However, this may also be due to having fewer persons in the education analysis compared with the income analysis.

Limitations

The primary limitation of this study was the potential for unmeasured confounding due to the observational nature of the study. Retrospective studies have previously overestimated the burden of SCD.⁴⁵ However, this might also be related to the data quality utilized in retrospective studies. A prospective study would allow for a more standardized data collection.

The registers utilized in this study do not provide data on behavioural risk factors such as smoking status, alcohol consumption, BMI, exercise, and diet habits. Similarly, it was not possible to measure psychological risk factors, including stress, mental health, social network, and health literacy. Therefore, it is uncertain whether some of the observed outcome differences are due to a higher burden of unmeasured risk factors.

It should be noted that some cases of SCD included in this study were possibly cases of non-sudden death that, due to inadequate information on the death certificate, were misclassified as SCDs. Conversely, some cases of possible SCD were likely misclassified as non-sudden due to victims being found >24 h after death. To reduce the risk of misclassification, terminal illness and obvious non-sudden cases were excluded. Finally, all data were recorded in 1 calendar year, and therefore, it is not possible to report on temporal differences in SCD incidence.

Conclusions

In this nationwide study of SCD across SEP in Denmark in 2010, persons with a low SEP had significantly higher incidences of ACM and SCD compared with high SEP groups. The largest difference in incidence rates was found between persons with a low and high level of education. The difference in ACM and SCD risk persisted independently of age, sex, and comorbidity burden and was most pronounced between low- and high-income patients. Further investigation is needed to uncover the mechanisms that mediate the difference in SCD risk between groups of differing SEP.

Supplementary material

Supplementary material is available at Europace online.

Author contributions

T.H.L. and J.T.-H. have reviewed the death certificates and are joint senior authors of this manuscript. T.S.J. wrote the first draft of this manuscript. All authors have been involved in the further development of the manuscript, discussed the data, and prepared the manuscript for submission. T.S.J. acted as the guarantor of this manuscript. All authors have approved the final draft of the manuscript before submission.

Ethics approval

This study complies with the Declaration of Helsinki and was approved by the data-responsible institute in the Capital Region of Denmark. In Denmark, register-based studies that are conducted for statistical and scientific purposes do not require patient consent or ethical approval.

Disclosures

J.T.-H. is an editor of EP Europace and was not involved in the peer review process or publication decision.

Funding

T.S. was supported by a research grant from the Danish Cardiovascular Academy, which was funded by the Novo Nordisk Foundation (grant no. NNF20SA0067242) and The Danish Heart Foundation. J.T.-H. has received funding from the John and Birthe Meyer Foundation. B.G.W. has received funding from the Novo Nordisk Foundation.

Data availability

All data relevant to the study are included in the article or uploaded as supplementary information. Data from the national registries cannot be shared publicly due to privacy laws.