Risk factors for out-of-hospital cardiac arrest: the Reykjavik Study

Abstract

Aims To examine risk factors for out-of-hospital cardiac arrest in the Reykjavik Study, a long-term, prospective, population-based cohort study that started in 1967.

Methods and results From 1987 to 1996, 137 men and 44 women out of the 8006 men and 9435 women in the study sustained out-of-hospital cardiac arrest due to cardiac causes. Determinants included coronary artery disease (CAD), its classical risk factors, and age, body mass index (BMI), heart rate, cardiomegaly, and erythrocyte sedimentation rate. Electrocardiograms (ECGs) were examined for various abnormalities. Significance was determined by Cox regression analysis. In multivariable analysis, the risk in men was significantly associated with age, diastolic blood pressure, cholesterol, current smoking, and previous diagnosis of myocardial infarction (MI). In women, the risk was associated with diastolic blood pressure, elevated levels of cholesterol and triglycerides, and increased voltage on ECG. Increased BMI was inversely related to women's risk of out-of-hospital cardiac arrest.

Conclusion In this prospective, population-based cohort study previous MI and the classical risk factors for CAD significantly increased the risk of out-of-hospital cardiac arrest, the endpoint of this study. Increased voltage on ECG additionally increased women's risk.

Out-of-hospital cardiac arrest, Myocardial infarction, Smoking, Cholesterol, Hypertension

Introduction

Sudden cardiac death has been defined as an unexpected natural death within 1 h of the onset of acute symptoms due to cardiac causes.^1,2 A high proportion of sudden cardiac deaths are of arrhythmic nature and occur out-of-hospital, nevertheless some do occur after >1 h of symptoms.³ In this respect, the out-of-hospital cardiac arrest individuals are a special group. In the Western world, the incidence of sudden cardiac death is largely a function of the prevalence of coronary artery disease (CAD). In the United States, more than 950 000 deaths each year are due to CAD and of these 225 000 are caused by out-of-hospital cardiac arrest.⁴ In Maastricht, The Netherlands, out-of-hospital cardiac arrests are 1 per 1000 inhabitants per year.⁵ In the Albany–Framingham Study, ∼50% of all coronary heart disease deaths were sudden and unexpected, occurring instantaneously or up to 1 h after the onset of symptoms.⁶ Sudden cardiac death and fatal myocardial infarction (MI) share many risk factors. However, in some studies, certain differences have emerged, and in the Paris Prospective Study I circulating non-esterified fatty acid was an independent risk factor for sudden death, but not for fatal MI.⁷ In Reykjavik and four other European cities, the number of out-of-hospital cardiac arrests in whom resuscitation was attempted ranged 50–66 per year per 100 000 inhabitants, with overall survival between 15 and 23%.⁸ The majority of these cardiac arrests are due to CAD.⁹ Although numerous large cohort studies have been carried out on sudden death, epidemiological information on the more narrowly defined arrhythmic death presenting as out-of-hospital cardiac arrest is limited.³ We present results on risk factors for on-scene electrocardiogram (ECG) verified out-of-hospital cardiac arrests that were determined prospectively in the large population-based cohort of the Reykjavik Study.^10,11

Methods

The Reykjavik Study

The design of the Reykjavik Study has been described in detail elsewhere.^10,11 The selected study subjects were all residing in the Reykjavik area on 1 December 1966, born 1907–34 (males) and 1908–35 (females). This cohort was divided by birth dates into six groups that were invited to participate in the different stages of the study. The study was conducted in five stages: 1967–69, 1970–72, 1974–79, 1979–84, and 1985–91. At the first stage, one group was invited, which at later stages was always invited together with a new group. The number of people invited was 24 933 and by the end of 1991, the overall response rate was 76%, with 18 912 participants at various stages from 64 to 77%. The history of angina was assessed from the Rose Chest Pain Questionnaire.¹² Each participant was examined by a physician, had a chest X-ray taken in two planes, and a standardized 12-lead ECG recorded and evaluated according to the Minnesota Code.¹³ On each visit height, weight, and arterial blood pressure were recorded, serum total cholesterol and triglycerides measured, and a glucose tolerance test performed. Several other biochemical and haematological blood tests were also done. External and internal laboratory quality controls were used, including duplicated assays and pooled serum for interassay variation as well as control-serum from the Hyland Normal Clinical Control and Seronorm from Nyegaard A/S, Oslo.

The laboratory participated in the glucose standardization in September 1974. The World Health Organization Lipid Reference Centre in Prague performed external quality control of lipid measurements. The diagnosis of MI was established by review of hospital records, laboratory results, and ECG changes for admitted patients and classified according to the WHO criteria applied in the MONICA registry.¹⁴ For those who were not admitted or died before reaching hospital, diagnosis was mainly obtained from autopsy reports and death certificates. Blood pressure was measured with a mercury sphygmomanometer, and the mean value of two blood pressure measurements on separate occasions used for reference. Hypertension was defined as systolic blood pressure >160 mmHg, diastolic blood pressure >95 mmHg, or taking antihypertensive medication. Cardiomegaly was defined as relative heart volume >550 mL/m² for men and >500 mL/m² for women.

Cardiac arrest database

Information about all out-of-hospital cardiac arrests in the Reykjavik area attended to by ambulance crews has, since 1982, been entered into a central database and later modified according to the Utstein guidelines.¹⁵ The documentation includes call-response interval, witnesses to arrest, initial rhythm, basic cardiopulmonary resuscitation by bystanders, advanced cardiac life support, any return of spontaneous circulation, admission to intensive care, discharge from hospital, and underlying etiology of the cardiac arrest. With approval from the Bioethics Committee of Landspítali University Hospital in Reykjavik, the Icelandic Data Protection Authority, and the National Bioethics Committee, we specifically investigated individuals that were participants in the cohort of the Reykjavik Study and sustained cardiac arrest outside hospital in the Reykjavik area from 1987 to 1996. Non-cardiac causes excluded, 503 individuals in the Reykjavik area sustained out-of-hospital cardiac arrest during the 10 year period from 1987 to 1996 and 181 (36%) of those, 137 men and 44 women, had participated in the Reykjavik Study. Although all deaths and major events are monitored in the Reykjavik Study, we checked the MONICA registry for out-of-hospital cardiac arrests and no additional cases were found. The cause for cardiac arrest was obtained from hospital records and from death certificates and autopsy records of those who died before reaching hospital.

Statistical analysis

Cox regression was applied to assess the predictive power of potential risk factors on cardiac arrest. The risk period used was from the beginning of 1987 until cardiac arrest, death, or end of 1996, whichever first occurred. The risk factor values used for analysis were the last measurement before 1987 or the first measurement in 1987 or later for the 276 men and 1597 women enrolled in the fifth stage of the Reykjavik Study. Subjects enrolled in the fifth stage were investigated between 1985 and 1991. Analyses were conducted separately for men and women. As the values of triglycerides and erythrocyte sedimentation rate (ESR) were skewed to the right and their logarithmic transforms showed more predictive power, the transforms were used [a value of 1 was added to ESR as it can take a value of 0]. Multivariable Cox regression was performed using variables found significant in bivariable analysis (age always included) and stepwise backward elimination performed for those variables not found to have significant independent prediction power. Calibration of the models was performed using the Hosmer–Lemeshow χ² statistic.¹⁶ Cross-validation was performed in the following manner.¹⁷ The group of each gender was randomly divided into 10 groups of equal size. For each of the 10 different combinations of 9 groups, Cox regression was performed, beta coefficients estimated, and survivor curves computed. The 10th group was used for obtaining observed and expected numbers of cardiac arrest. For those who died without having suffered cardiac arrest, their cumulative risk at death was used. For others, cumulative risk after 10 years was used. Receiver operating characteristic (ROC) curves were computed for the mean values from the 10 results for each gender. Tests were two-sided. The significance level chosen was 0.05 and the programme package used was SPIDA. The essential assumption made for the Cox models is that the hazard ratios (HRs) associated with different levels of each covariate are constant and do not depend on the duration of survival. A P(PH)-value less than 0.05 for any covariate would indicate a significant departure from this assumption for that covariate. In SPIDA, the values of P(PH) are calculated according to the methods detailed by Schoenfeld¹⁸ and Harrell and Lee,¹⁹ and no significant values were obtained. Squares of the continuous variables were added in the regression to test non-linearities and were dropped if non-significant. The sequentially rejective Bonferroni test was used to account for multiple testing.²⁰

Results

Table 1 shows the clinical characteristics of the 8006 men and 9435 women alive and participating in the Reykjavik study on 1 January 1987. In addition, the characteristics of those sustaining out-of-hospital cardiac arrest are also shown. The frequency of various ECG abnormalities is shown in Table 2.

From 1987 to 1996, 137 men and 44 women in the Reykjavik Study sustained out-of-hospital cardiac arrest due to cardiac disease. Table 3 lists arrhythmias documented on the first rhythm strips taken by the crew of the emergency ambulance on scene confirming the arrhythmic mode of death. In men, ventricular fibrillation is the most common arrhythmia on the first rhythm strip, whereas asystole is most common in women. Of the three types of arrhythmias on the first rhythm strip, pulseless electrical activity (PEA) was the least common. There was inadequate information of initial rhythm in five men. Defibrillation was not delivered in these five cases, so asystole or PEA is the most likely arrhythmia. In 80 (58.4%) of the 137 men, the underlying immediate cause was an acute MI, 26 (18.9%) had chronic CAD, and another 4 (2.9%) probable CAD. The remaining subjects had mostly various other cardiac diagnoses. In 24 (54.5%) of the 44 women, an acute MI was the immediate underlying cause and 4 (9.1%) had chronic CAD. Various other cardiac diseases were the cause of cardiac arrest in 16 women.

Risk factors for out-of-hospital cardiac arrest

In men, the following were significant risk factors in age-adjusted analysis (Tables 4 and 5): age, body mass index (BMI), hypertension, diastolic blood pressure, antihypertensive medication, cholesterol, triglycerides, current smoking, previous history of MI, ESR and Q-waves, ST-depression and T-wave abnormalities on ECG. Systolic blood pressure, diabetes, heart rate, family history, history of angina pectoris, and cardiomegaly were not significantly associated with out-of-hospital cardiac arrest in age-adjusted analysis. In women, the following variables were significant in age-adjusted analysis: systolic and diastolic blood pressure, cholesterol, triglycerides, current smoking, angina pectoris, heart rate, high voltage on ECG, as well as ST-depression and T-wave abnormalities. BMI was inversely associated with women's risk of sustaining out-of-hospital cardiac arrest. Although the relative risk associated with previous MI in women was 3.27, it was not statistically significant, even in age-adjusted analysis, probably owing to the low number of cases.

In multivariable analysis (Table 6), the following variables were associated with increased risk in men: age, diastolic blood pressure, cholesterol, current smoking, and previous MI. Women's risk of sustaining out-of-hospital cardiac arrest was, in multivariable analysis, associated with the following: diastolic blood pressure, total cholesterol, triglycerides, and increased voltage on ECG. There was no collinearity between systolic and diastolic blood pressure. BMI was inversely associated with women's risk of out-of-hospital cardiac arrest.

Risk factors for out-of-hospital cardiac arrest in individuals with CAD

Of the 8006 men, a total of 717 had history of either MI or angina and of these, 31 suffered cardiac arrest. After the onset of overt CAD, none of the risk factors was predictive of out-of-hospital cardiac arrest in men in multivariable analysis except cholesterol with a relative risk of 1.40 (95% CI: 1.00–1.95; P=0.04). Of the 9435 women, 674 had previous manifestations of CAD and of these, only eight sustained cardiac arrest. In these women, cholesterol was significantly associated with cardiac arrest, with a relative risk of 1.86 (95% CI: 1.22–2.85; P=0.004) and current smoking was a significant risk factor, with a relative risk of 6.97 (95% CI: 1.40–34.60; P=0.02).

Calibration of the models showed Hosmer–Lemeshow χ² values of 29.94 and 5.89 for men and women, respectively. For eight degrees of freedom, this corresponds to P-values of <0.001 and 0.66, which means good fit for men, but poor for women. Cross-validation gave values of an area under the ROC curves of 68.6% for men and 62.7% for women.

Discussion

Owing to the comprehensive registry of out-of-hospital cardiac arrest in the Reykjavik area since 1982, we were able to enter cardiac arrest on the basis of on-scene ECG records as an endpoint into the Reykjavik Study database, the long-term population-based cohort study of the same area in Iceland. We examined which baseline characteristics of this cohort are risk factors for out-of-hospital cardiac arrest. Several studies have been carried out on the epidemiology of sudden death.^{22,23,25,26–28,30–34} However, studies on the risk factors of arrhythmic death presenting as out-of-hospital cardiac arrest are sparse. Although related and overlapping, the terms sudden cardiac death and out-of-hospital cardiac arrest are not synonymous. In a recent prospective study of sudden cardiac death among women in the United States, Albert et al.³ reviewed 154 cases defined by witnessed symptom onset within 1 h of death. Of those, 135 (88%) fulfilled the criteria for arrhythmic death. In addition, some out-of-hospital cardiac arrests occur after >1 h of symptoms. It is of importance to understand the main risks associated with out-of-hospital cardiac arrest, which usually is unexpected and lays a heavy burden on local emergency systems.

CAD is the most common cause of out-of-hospital cardiac arrest. In the Reykjavik Study, ∼80% of the cardiac arrest incidents in men were due to CAD, the majority caused by an acute MI. Acute MI triggered 64% of women's cardiac arrests. The increased risk of cardiac arrest for men in this study was about four-fold if they had prior diagnosis of MI. History of MI more than doubled women's risk, not, however, reaching statistical significance. The infrequent condition in women and the relatively few events decreased the power to determine the long-term significance of this variable.

Of the classical risk factors for CAD, smoking at entry into the Reykjavik Study had the strongest association with out-of-hospital cardiac arrest in men, increasing their risk by ∼70%. Smoking was a risk factor for women in age-adjusted analysis, but did not reach statistical significance in multivariable analysis except in women with previous CAD. In our study, having smoked in the past did not increase the risk of sustaining out-of-hospital cardiac arrest. In the CAST study, it was shown that smoking cessation was accompanied by marked reduction in arrhythmic death and overall mortality in a high-risk cohort of patients with advanced CAD.²¹ In the Paris Prospective Study I, average tobacco consumption 5 years prior to screening significantly increased the risk of sudden death, but not of fatal MI.²² In a study comparing sudden and non-sudden coronary deaths in the United States, current cigarette smoking was more liable to result in sudden than in non-sudden coronary death with a significant odds ratio (OR) of 1.3 for risk of sudden coronary death.²³

Total cholesterol significantly increased the risk of cardiac arrest in this study by ∼40% in both sexes per 1 mmol/L. Triglycerides were a significant risk factor in age-adjusted analysis in both sexes, but only significant in multivariable analysis in women. However, the independent contribution of triglycerides is not certain because HDL levels were not measured. In the Paris Prospective Study I, triglycerides increased the risk for both sudden death and fatal MI in age-adjusted analysis, but not significantly in multivariable analysis.²² In the prospective cardiovascular Munster Study, among 4849 men, elevated triglycerides was an independent risk factor for both an early MI and cardiac death.²⁴

Of the blood pressure variables, diastolic blood pressure was the most significant risk factor in both sexes and highly significant after adjustment for other risk factors. Left ventricular hypertrophy (LVH) has been identified as one of the strongest blood pressure independent risk factors for sudden death²⁵ and already in the early stages of the Framingham Study, it became clear that annual rate of sudden death rose progressively with the blood pressure and men with prior evidence of LVH on ECG had a highly significant five-fold increase in risk of sudden death.²⁶ In our analyses, we used voltage criteria on ECG, ST- and T-wave abnormalities, and axis deviation independently but not together as the combined variable LVH. In the age-adjusted analysis voltage criteria on ECG reached statistical significance in women, ST-depression and T-wave abnormality in both men and women, but axis deviation in neither sex.

Increased BMI was associated with out-of-hospital cardiac arrest in men in age-adjusted analysis but did not reach statistical significance after adjustment for other risk factors. Increased BMI was inversely associated with cardiac arrest in women after adjustment for other variables. These results should be interpreted with caution because hypertension and diabetes, the mediators in the causal pathway of obesity, were adjusted for in the analysis.^3,27 Although some controversy remains in the literature about BMI as a risk factor,^28,29 it emerged as a significant risk factor for sudden death in both sexes in the Paris Prospective Study I²² and in the Framingham Study.²⁶ However, in a recent study, it is suggested that it is the presence of the metabolic syndrome and not BMI that predicts future cardiovascular risk in women.³⁰ Further study is needed, especially on the relation among BMI and smoking, diet, physical activity, and markers of inflammation.

It has been suggested that there may be an association between inflammation in the vasculature and sudden cardiac death.³¹ In our study, C-reactive protein was not measured. ESR in men was associated with cardiac arrest in age-adjusted analysis; however, it did not reach statistical significance after adjustment for other risk factors. In a nested case–control study involving 97 cases of sudden cardiac death among apparently healthy men enrolled in the Physician's Health Study, C-reactive protein levels at baseline were significantly associated with sudden cardiac death controlled for multiple cardiac risk factors.³² In this study, neither homocysteine nor lipid level was significantly associated with risk of sudden cardiac death.

Heart rate in women was significantly associated with out-of-hospital cardiac arrest corrected for age, but did not reach statistical significance after adjustment for other risk factors. The incidence of diabetes mellitus is low in the Reykjavik Study. Power to determine its significance as a risk factor is therefore limited. Cardiomegaly was not associated with cardiac arrest in this study. Q-waves on ECG were strongly associated with cardiac arrest in age-adjusted analysis for men, and in multivariable analysis MI was the strongest risk factor for men. After adjustment for other risk factors, increased voltage on ECG was significant in women. ST- and T-wave abnormalities were significantly associated with cardiac arrest in age-adjusted analysis in both sexes, but did not reach statistical significance after adjustment for other risk factors. Silent ST-segment and T-wave changes have been previously reported as independent predictors of reduced survival in the Reykjavik Study.³³ A strong independent relation has been identified between sudden death in patients with various health problems and left bundle branch block, right bundle branch block combined with a left anterior fascicular block, and extrasystoles.³⁴ The only ECG-related sudden death risk factors in the Framingham Study, which were significant after adjustment for other risk factors, were ECG-MI and intraventricular conduction abnormality in those with overt CAD, LVH and tachycardia in men and in women without CAD, and non-specific ST-T abnormalities in men.³⁵

The results of this study on risk factors for out-of-hospital cardiac arrest agree in general, with epidemiological studies on the related and overlapping but not identical endpoint of sudden cardiac death. In the Framingham Study³⁶ and the British Study on risk factors in middle-aged men,³⁷ pre-existent CAD and most of the classical cardiovascular risk factors were identified as risk factors for sudden death. However, in the Framingham Study, none of the major modifiable risk factors was predictive of sudden death after onset of CAD. In our study, cholesterol is still a significant risk factor in both sexes with CAD, and current smoking increased the risk in women with CAD of having out-of-hospital cardiac arrest.

Strengths and limitations

In this large population-based study, the risk factors for on-scene verified out-of-hospital cardiac arrest could be prospectively studied in the 8006 men and 9435 women with a mean follow-up time of 10 years. During the study period, 137 men, but only 44 women, in the Reykjavik Study cohort sustained out-of-hospital cardiac arrest. This limits the power of our study to detect and reject possible risk factors for out-of-hospital cardiac arrest, especially in women. With the sequentially rejective Bonferroni adjustment for multiple testing, triglycerides and diastolic blood pressure would not be significant in women and current smoking not significant in men. The age distribution of the participants was restricted such that individuals born in 1907–35 were studied in 1987–96, clearly limiting the information on the age distribution of cardiac arrests. In the Reykjavik Study, total cholesterol was measured but not HDL cholesterol. High levels of triglycerides may reflect low levels of HDL. Therefore, it is possible that the significance of triglycerides is overestimated in the study as an independent risk factor for cardiac arrest. Numerous other parameters with potential impact on the future risk of cardiac arrest were not assessed, including left ventricular function and the presence of spontaneous arrhythmias. However, this is a limitation shared by all large epidemiologic studies. C-reactive protein, which recently was found to be predictive of sudden cardiac death in men, was not carried out when baseline levels were obtained in this part of the Reykjavik Study. ESR was not independently predictive of out-of-hospital cardiac arrest, but the contribution of inflammation may be underestimated by this less specific marker of inflammation as by the dynamic nature of inflammatory markers.

Conclusions

In this prospective, population-based cohort study with out-of-hospital cardiac arrest as the endpoint, established CAD, current smoking, high blood pressure, and hypercholesterolaemia were the main risk factors. Increased voltage on ECG increased women's risk as well. After adjustment for all other variables, BMI did not increase men's risk and was inversely associated with out-of-hospital cardiac arrest in women. Out-of-hospital cardiac arrest is a dramatic event that usually leads to immediate activation of emergency services in the community. Understanding the epidemiology of out-of-hospital cardiac arrest increases our ability to predict who, in the general population, are at risk for cardiac arrest. Although more specific markers of risk for cardiac arrest are needed, our results emphasize that modification of coronary risk, both at the community level (population strategy) and at the individual level (high risk strategy), will also confer risk reduction with respect to cardiac arrest. History or signs of MI and its classical risk factors identify those individuals in the community exposed to the greatest risk of suffering cardiac arrest.

Acknowledgement

This study was supported by a grant from the Scientific Fund of Landspítali University Hospital, Reykjavik, Iceland.

Table 1

Characteristics of participants of the Reykjavik Study, the total cohort and those who sustained cardiac arrest, obtained at first visit

	Men				Women
	Total (8006)		Cardiac arrest (137)		Total (9435)		Cardiac arrest (44)
	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Age in 1987 (years)	64.3	7.3	66.8	6.3	63.9	7.5	65.8	6.5
BMI (kg/m2)	25.9	3.5	26.8	3.4	25.4	4.3	24.1	4.1
Hypertension (%)	31.0		46.0		26.7		40.9
Systolic blood pressure (mmHg)	140	19	144	20	137	20	147	24
Diastolic blood pressure (mmHg)	87	10	90	11	83	10	87	11
Antihypertensive medication (%)	11.6		20.4		15.3		15.9
Cholesterol (mmol/L)	6.3	1.0	6.7	1.0	6.6	1.2	7.2	1.7
Triglycerides (mmol/L)	1.3	0.8	1.7	1.6	1.1	0.6	1.3	0.6
Current smoker (%)	45.1		52.6		36.0		50.0
Diabetes (%)	2.6		2.9		2.0		0.0
Heart rate (b.p.m.)	67.4	9.7	67.1	7.9	70.6	10.6	74.3	9.0
Family history of MI (%)	16.3		18.3		22.6		31.8
Diagnosis MI (%)	4.2		15.3		1.4		4.6
Diagnosis angina pectoris (%)	4.8		7.3		5.7		13.6
ESR (mm/h)	5.8	7.8	7.1	7.4	10.9	11.6	12.8	9.2
Cardiomegaly (%)	8.0		12.8		5.4		7.9

Table 2

Frequency of ECG diagnoses (Minnesota Code) in men and women in the Reykjavik Study, obtained at first visit

	Men		Women
	Total (8006) (%)	Cardiac arrest (137) (%)	Total (9435) (%)	Cardiac arrest (44) (%)
ECG—Q-waves	3.0	8.8	1.3	2.3
ECG—QRS axis deviation	1.6	2.9	1.0	2.3
ECG—increased voltage	4.0	3.6	0.7	11.4
ECG—ST-depression	4.0	8.8	2.9	9.1
ECG—T-wave abnormality	8.7	17.5	7.7	18.2
ECG—A–V block	1.5	2.2	0.7	0.0
ECG—bundle branch block	4.9	5.8	4.0	4.6
ECG—atrial fibrillation/flutter	1.2	2.2	0.4	0.0

The ECG Minnesota Code; 111–136: Q‐wave abnormality; 210–250: QRS axis deviation; 310–340: high amplitude R-waves (increased voltage); 410–440: ST-junction and segment depression; 510–540: T-wave items (abnormalities); 610–660: A–V conduction defect; 710–780: ventricular conduction defect (bundle branch block); 830: atrial fibrillation or flutter, persistent or intermittent.

Table 3

Initial rhythm of 181 individuals who sustained out-of-hospital cardiac arrest, recorded by emergency ambulance personnel

	Ventricular fibrillation	Asystole	PEA	Other
Men	77	41	14	5a
Women	14	21	9
Total	91	62	23	5a

^aInadequate information on initial rhythm.

Table 4

HR associated with out-of-hospital cardiac arrest

	Men		Women
	HR (95% CI)	P-value	HR (95% CI)	P-value
Age (years)	1.05 (1.02–1.07)	<0.001	1.04 (1.00–1.08)	0.09
BMI (kg/m2)	1.06 (1.02–1.11)	0.004	0.92 (0.85–1.00)	0.04
Hypertension (%)	1.71 (1.22–2.41)	0.002	1.72 (0.92–3.20)	0.09
Systolic blood pressure (mmHg)	1.01 (1.00–1.02)	0.10	1.02 (1.00–1.03)	0.008
Diastolic blood pressure (mmHg)	1.02 (1.01–1.04)	0.002	1.04 (1.01–1.07)	0.003
Antihypertensive medication (%)	1.79 (1.18–2.72)	0.006	0.94 (0.42–2.13)	0.88
Cholesterol (mmol/L)	1.44 (1.25–1.66)	<0.001	1.43 (1.18–1.74)	<0.001
Triglycerides (ln)	2.21 (1.56–3.14)	<0.001	2.13 (1.13–4.00)	0.02
Current smoker (%)	1.48 (1.05–2.07)	0.02	1.88 (1.04–3.40)	0.04
Diabetes (%)	0.96 (0.35–2.60)	0.94	—
Heart rate (b.p.m.)	1.00 (0.98–1.01)	0.66	1.03 (1.00–1.05)	0.02
Family history (%)	1.37 (0.87–2.16)	0.18	1.83 (0.92–3.65)	0.09
MI (%)	3.78 (2.35–6.06)	<0.001	3.27 (0.78–13.7)	0.1
Angina pectoris (%)	1.66 (0.87–3.18)	0.13	2.59−(1.09–6.16)	0.03
ESR (ln)	1.26 (1.05–1.51)	0.01	1.27 (0.86–1.85)	0.23
Cardiomegaly (%)	1.00 (0.88–2.49)	0.14	1.33 (0.40–4.45)	0.64

Results from bivariable Cox regression adjusted for age.

Table 5

HR associated with out-of-hospital cardiac arrest

	Men		Women
	HR (95% CI)	P-values	HR (95% CI)	P-values
ECG—Q-waves	2.79 (1.54–5.06)	<0.001	1.62 (0.22–11.8)	0.63
ECG—QRS axis deviation	1.48 (0.55–4.03)	0.44	1.95 (0.27–14.13)	0.51
ECG—increased voltage	0.81 (0.33–1.97)	0.64	16.1 (6.3–41.4)	<0.001
ECG—ST-depression	1.86 (1.01–3.40)	0.045	2.98 (1.05–8.42)	0.04
ECG—T-wave abnormality	1.88 (1.20–2.95)	0.006	2.39 (1.10–5.23)	0.03
ECG—A–V block	1.35 (0.43–4.26)	0.60
ECG—bundle branch block	1.07 (0.52–2.20)	0.84	1.14 (0.27–4.69)	0.86
ECG—atrial fibrillation/flutter	1.53 (0.49–4.83)	0.47

Results from Cox regression adjusted for age. ECG diagnoses by Minnesota Code.

Table 6

HR of out-of-hospital cardiac arrest^a

	Men		Women
	HR (95% CI)	P-value	HR (95% CI)	P-value
Age (years)	1.05 (1.02–1.07)	<0.001	1.00 (0.96–1.04)	0.99
BMI (kg/m2)	1.04 (0.99–1.09)	0.11	0.88 (0.81–0.96)	0.003
Diastolic blood pressure (mmHg)	1.02 (1.01–1.04)	0.01	1.04 (1.02–1.07)	0.002
Cholesterol (mmol/L)	1.41 (1.21–1.63)	<0.001	1.42 (1.14–1.76)	0.001
Triglycerides (ln)	1.44 (0.97–2.14)	0.07	2.11 (1.05–4.26)	0.04
Current smoker	1.69 (1.20–2.39)	0.003	1.64 (0.89–3.05)	0.11
CHD, diagnosis MI	3.85 (2.23–6.23)	<0.001	2.63 (0.62–11.0)	0.19
ECG (increased voltage)	0.90 (0.37–2.23)	0.82	14.5 (5.5–38.2)	<0.001

^aVariables were included in a multivariable model when they were significant in age-adjusted analyses in at least one gender.