Sudden cardiac arrest (SCA) remains a major public health problem, claiming at least 250,000 lives in the U.S. on an annual basis (1- 2). There are significant sex-related differences in prevalence and manifestation of SCA. The condition manifests more commonly and at younger ages in men, implying a protection conferred by female sex (1,3). Prevalence of coronary artery disease (CAD), the diagnosis most commonly associated with SCA, is significantly lower in female SCA (4- 5). A recent autopsy study in a group 35 to 44 years of age found significantly higher rates of unexplained sudden deaths in women compared with men (6). Female SCA cases are more likely to present with pulseless electrical activity (PEA) or asystole compared with male cases (1,7). Since rates of successful resuscitation are significantly lower in PEA/asystole versus ventricular tachycardia/fibrillation (2% vs. 25%, respectively [1,8]), there are implications for survival from SCA in women.

Due to these significant sex-based differences, the possibility that women may have a different SCA risk profile from men merits consideration. Available comparisons of SCA risk predictors between the sexes would suggest that coronary disease-related risk factors are common to both men and women (4). However, an earlier analysis of the Ore-SUDS (Oregon Sudden Unexpected Death Study) performed in a small number of patients suggested that women who suffer SCA may have a lower prevalence of severe left ventricular (LV) dysfunction (9), the major risk stratification variable currently utilized in clinical practice. Given the reported lower rates of CAD and higher rates of unexplained sudden death among female SCA cases, we hypothesized that sex-specific risk factors would be identified among cases of SCA in the general population. As part of the ongoing prospective, population-based Ore-SUDS study, we performed a larger, comprehensive comparison between women and men of all available clinical risk predictors of SCA.

Detailed methods regarding the ongoing Ore-SUDS study have been published earlier (1,3,9). Cases of SCA in the Portland, Oregon, metropolitan area (population approximately 1 million) were identified between February 1, 2002, and January 31, 2007, using multiple sources. During 2002 to 2005, cases of out-of-hospital SCA were identified from emergency medical services (approximately 70%), the medical examiner's office (approximately 25%), and local emergency rooms (approximately 5%). During 2005 to 2007, ascertainment was limited to SCA cases with attempted resuscitation or those that were investigated by the medical examiner.

SCA was defined as a sudden unexpected pulseless condition occurring within 1 h of symptom onset (witnessed events) or within 24 h of having last been observed alive and symptom-free (unwitnessed) (10). Furthermore, the location of SCA had to be in an out-of-hospital or emergency room setting. A diagnosis of SCA was assigned after in-house adjudication by 3 physicians who evaluated circumstances of cardiac arrest and all available clinical data. Survivors of SCA were included. Subjects with chronic terminal illnesses (e.g., cancer), known noncardiac causes of sudden death (e.g., pulmonary embolism, cerebrovascular accident), traumatic deaths, and overdoses were excluded. This study was approved by the Institutional Review Board of Oregon Health and Science University as well as all other participating hospitals and health systems.

SCA cases included in the analysis of demographics and arrest circumstances were at least 35 years old at the time of event; analysis of clinical characteristics was restricted to the majority subset with medical records available. Records were retrospectively reviewed for demographic information as well as detailed clinical history of CAD, left ventricular mass (LVM)/LV hypertrophy, hypertension, smoking, diabetes, dyslipidemia, obesity, and other comorbidities. When available and unrelated to the SCA event, electrocardiograms (ECGs), echocardiograms, coronary angiograms, and laboratory data were reviewed as well as all available autopsy data. Information regarding LV systolic function (left ventricular ejection fraction [LVEF]) was obtained from an echocardiogram, LV angiogram, or multigated acquisition scan performed before and unrelated to SCA.

LV systolic function was classified using LVEF as follows: normal (LVEF ≥55%), mild-moderate LV dysfunction (LVEF 36% to 54%), and severe LV dysfunction (LVEF ≤35%). LVM was calculated from echocardiograms before arrest using the American Society of Echocardiography modified equation (11), and LVM was indexed to body surface area. LV hypertrophy was defined as LVM/body surface area >134 g/m² for men and >110 g/m² for women (12). Definite CAD was defined as ≥50% stenosis of a major coronary artery on angiogram or post-mortem examination; history of percutaneous coronary intervention or coronary artery bypass grafting; physician report of myocardial infarction; pathologic Q waves on ECG; or myocardial infarction history determined by any 2 of the following: ischemic symptoms, ECG changes, or positive troponins/creatinine kinase-MB. The QT interval was obtained from a standard 12-lead pre-arrest ECG, with measurements conducted manually using calipers, and corrected using Bazett's formula (QTc). Sex-specific categories of QTc were also used: men ≤430 ms (normal), 431 to 450 ms (borderline), and >450 ms (abnormal); women ≤450 ms (normal), 451 to 470 ms (borderline), and >470 ms (abnormal).

Hypertension was defined as clinical history of hypertension and diabetes mellitus (DM) as clinical history of DM or use of antidiabetic medications. Dyslipidemia was defined as clinical history or use of lipid-lowering medications. Response time was defined as the time from dispatch of emergency medical personnel to their arrival on the scene. Return of spontaneous circulation after resuscitation was defined as a return of a palpable pulse in conjunction with a systolic blood pressure of at least 60 mm Hg. Survival to hospital discharge was defined as discharge to home or a noncritical care facility.

Statistical analyses were performed using SAS version 9.1 (SAS Institute Inc., Cary, North Carolina). Demographic and clinical characteristics of male and female SCA cases were summarized using frequencies for categorical variables and means with SDs for continuous variables, and sex differences were examined using Pearson's chi-square tests for categorical variables or t tests for continuous variables. Sex differences were also evaluated and adjusted for age using the Cochran-Mantel-Haenszel test for categorical variables and analysis of covariance for continuous variables. Multiple logistic regression was performed using variables found to be significant in univariate comparisons. For all analyses, a value of p < 0.05 was considered statistically significant.

During the 5-year period of ascertainment, 1,568 SCA cases were identified (Table 1). The majority of cases were male subjects (1,012 men, 64% vs. 556 women, 36%; p < 0.0001), and women were older (mean age 71 ± 14 years vs. 65 ± 14 years, p < 0.0001) (Table 1). Information on presenting arrhythmia at the time of arrest was available in 706 (70%) men and 356 (64%) women (Table 1). Ventricular fibrillation or ventricular tachycardia was more commonly observed among men and PEA/asystole among women (p < 0.0001). Sex differences observed in the presenting rhythm remained significant after adjusting for age (p = 0.0004). Women were more likely to have a return of spontaneous circulation during resuscitation (women 25% and men 21%, p = 0.04). However, there were no significant sex differences in the proportion of arrests that were witnessed, in response time, or in rates of survival to hospital discharge (p ≥ 0.25) (Table 1).

Of the total 1,568 cases, 1,258 had medical records available (805 men [80%] and 453 women [81%]). Men were more likely to be active or former smokers compared with women (42% vs. 36% and 36% vs. 24%, respectively; p < 0.0001) (Table 2). Previously documented CAD was lower in women versus men (40% vs. 49%; p = 0.001), but there were no differences with respect to history of myocardial infarction (29% vs. 26%; p = 0.19) (Table 2). Hypertension and DM were more common among women (73% vs. 62%, p = 0.0001 and 40% vs. 34%, p = 0.03, respectively). Only 18 subjects (1.1% of the total, n = 1,568) had implantable cardioverter-defibrillators (ICDs) implanted before the event, 14 (1.4%) men and 4 (0.7%) women (p = 0.23). There were no sex differences in the prevalence of history of prior cardiac arrest, dyslipidemia, obesity, history of chronic obstructive pulmonary disease or asthma, history of seizure, or presence of LV hypertrophy (p ≥ 0.10).

Records documenting LV function were available in 307 (38%) men and 178 (39%) women (Table 2). Women were significantly more likely to have normal LV systolic function compared with men (49% vs. 35%), and men were more likely to have severe LV systolic dysfunction than women (35% vs. 21%, p = 0.001). QTc was significantly longer in women compared with men (457 ± 39 ms vs. 444 ± 45 ms; p = 0.003).

After adjustment for age, all clinical comparisons in (Table 2) remained consistent, except history of myocardial infarction, which became significant (p = 0.05), and the sex-specific QTc categories, which lost significance (p = 0.16).

In the multivariate logistic regression analysis, all 1,258 subjects had information on age, CAD, hypertension, history of myocardial infarction and DM. LV systolic function was available for 485 of the 1,258 cases. (Table 3) presents the results of the logistic model (n = 485). Women were significantly less likely to have CAD (odds ratio: 0.34; 95% confidence interval [CI]: 0.20 to 0.60) and severe LV dysfunction (odds ratio: 0.51; 95% CI: 0.31 to 0.84) compared with men (Table 3). Hypertension, history of myocardial infarction, and DM were no longer significantly different by sex. In a model including QTc (n = 223 subjects with complete data), longer QTc remained associated with female sex, and the other associations remained consistent but the CIs were wider.

In this population-based analysis, sex-based comparisons of potential risk factors for SCA identified 2 distinct clinical factors that were unique to women who experience SCA. Women were significantly less likely than men to have severe LV dysfunction as a precursor to SCA. In addition, there was a significantly lower likelihood of women having an established diagnosis of CAD before SCA. These findings coincided with trends toward higher prevalence of DM and hypertension in women. There were no significant sex differences in the mean body mass index or prevalence of obesity, dyslipidemia, history of chronic obstructive pulmonary disease/asthma, or LV hypertrophy. After correction for age, significant sex differences disappeared for the sex-specific QTc categories that account for underlying male-female differences in the QTc interval, but remained for QTc measured continuously. This may imply that when the underlying longer QTc in women is taken into account, QTc does not appear to differentiate female from male SCA patients. With regard to presenting arrhythmia at the time of cardiac arrest, ventricular fibrillation or ventricular tachycardia was more commonly observed among men and PEA/asystole among women.

The lower rate of severe systolic LV dysfunction in women who experience SCA is a finding with potentially important implications. Risk stratification based on degree of LV systolic dysfunction remains the major current and clinically utilized method of SCA risk stratification. The vast majority of patients who undergo ICD implantation have some form of cardiomyopathy with evidence of severe LV systolic dysfunction (LVEF ≤35%). From community-based studies, we and others have previously reported that only 25% to 30% of SCAs occur in subjects who have severely reduced LV systolic function (1,3,9,13- 14). Using the existing guidelines for ICD implantation, we are likely to miss the opportunity to effectively risk stratify almost 70% to 75% of individuals who will experience SCA, and the current findings suggest that this number could be even higher for women.

The significantly lower prevalence of recognized CAD among women who experience SCA is a novel finding in a population-based setting. There are several possible explanations. The fact that these results are consistent with published findings among hospitalized survivors of SCA (15) would suggest that this reflects a lower prevalence of significant CAD among female SCA cases. However, the possibility that women without CAD are more likely to be successfully resuscitated from SCA cannot be ruled out. Other potential explanations are that consideration of traditional risk factors may lead to under-recognition of CAD in women (16), that physicians pursue a less aggressive management approach to CAD in women than in men (17), and/or that there are higher rates of microvascular CAD (as opposed to epicardial CAD) among women (18). While this phenomenon clearly warrants further investigation, CAD remains the condition most commonly associated with SCA. Symptoms related to CAD often prompt health care provider visits and an opportunity for risk stratification and initiation of drug therapy. Therefore, it is also likely that the lower prevalence of recognized CAD among women impedes effective risk stratification and prevention of SCA in women.

It is of interest that a recent analysis of a Medicare population sample (1991 to 2005) reported that men were significantly more likely to undergo ICD implantation for both primary and secondary prevention of SCA (hazard ratio: 3.15, 95% CI: 2.86 to 3.47, and hazard ratio: 2.44, 95% CI: 2.30 to 2.59, respectively) (19). This same sex disparity has also been reported in 2 other database analysis studies (20- 21). A subgroup analysis of MADIT-II (Multicenter Automatic Defibrillator Implantation Trial II) (22) also reported lower rates of enrollment in women. However, women and men experienced a similar survival benefit from ICD therapy. Therefore, development of methods that enhance SCA risk stratification in women should become an even higher priority. This need is underscored by our additional findings that the proportion of female SCA cases was higher than expected. In 2004, in contrast to earlier studies that have reported a 25:75 female/male ratio (23- 24), we observed that 40% of all cases were women (1,3). Over a period of 5 years, these findings have been consistent in the Ore-SUDS study with an average prevalence of 36%. While the reasons for this changing trend need to be evaluated in detail, one possibility is that it corresponds to the altered sex distribution in prevalence of and mortality from CAD (25).

If prevalence of severe LV systolic dysfunction in women is low and corresponds to lower rates of diagnosed coronary disease, what are the mechanisms of SCA in the majority of women? The established association between CAD risk factors such as smoking, hypertension, and diabetes is common to both women and men (4). However, there appear to be distinct differences in how SCA manifests in women. In a post-mortem study, a subgroup of younger women tended to have plaque erosion with relatively little coronary arterial narrowing and less plaque calcium (26). Furthermore, the majority of women (69% in an earlier prospective study [4]) tend to have no history of cardiac disease before SCA (5). This could be consistent with higher rates of vulnerable plaque rupture without pre-existing severe stenosis. There are other potential mechanisms. Based on higher rates of diastolic heart failure among women in the community, it has been suggested that women who experience SCA could have a higher prevalence of LV diastolic dysfunction (with preserved LV systolic dysfunction) than men (27- 28). Our findings that hypertension and DM were more common in women compared with men (73% vs. 62%, p = 0.0001 and 39% vs. 34%, p = 0.03, respectively) would support the diastolic dysfunction hypothesis. There is evidence that common as well as rare gene variants may contribute to SCA and that at least some of these may be distributed more commonly among women. In community-based studies, common variants in the SCN5A gene can increase susceptibility to SCA among women (29), but not in men (30). The majority (approximately 70%) of all reported cases of the familial or acquired long QT syndromes are women. Consistent with prior observations, the QTc was significantly longer in women (31). Women were more likely to manifest with PEA compared with men and had higher rates of return of spontaneous circulation, but there were no differences in rates of survival to hospital discharge.

Based on their design, population-based studies of SCA have some limitations that relate mostly to the fact that all subjects analyzed may not have uniform information available. This is directly attributable to the fact that a large proportion of patients may have SCA without prior warnings or health care provider visits for cardiac testing. While prospective cohort studies could circumvent this limitation, the incidence of SCA is such that the large size of cohort required renders this approach unfeasible. However, in the present study, availability of medical records were comparable in men and women (80% vs. 81%). Also, documentation of LV function was available in 307 (38%) men and 178 (39%) women, further decreasing the likelihood of sex bias in this regard. Nonetheless, it remains possible that if the entire population were screened with echocardiograms, the overall prevalence of LV dysfunction may be different. Also, while subjects included in our analysis constituted the vast majority of SCA cases in the region, limiting ascertainment to individuals that underwent attempted resuscitation or were investigated by the medical examiner during the last 2 years of this 5-year study may impact the generalizability of the findings to all SCA cases.

In this population-based analysis, the proportion of women with SCA was higher than anticipated (36% women, 64% men). However, prevalence of severe LV systolic dysfunction and previously established CAD were significantly lower in women versus men. This could, in part, explain the lower rates of ICD implantation in women. Although the underlying pathophysiology and risk factors for SCA in women are generally assumed to be similar to men, these findings suggest that women may have unique risk predictors. Since fewer women may be eligible for ICD implantation based on LVEF criteria alone, the identification of novel SCA risk predictors for women becomes an important priority.

The authors would like to acknowledge the significant contribution of American Medical Response; Portland/Gresham fire departments; the Multnomah County Medical Examiner's office; and the emergency medicine, cardiology, and primary care physicians and allied health personnel of the 16 area hospitals.