Mortality Differences Between Men and Women Following First Myocardial Infarction FREE

DIFFERENCES in early mortality between men and women who experience acute myocardial infarction (AMI) are related to comorbidity and demographic (ie, age) dissimilarities.^{1 - 2} The issue of sex differential in mortality after AMI has been addressed in the context of clinical trials and other investigations that sometimes were not aimed specifically at answering the question of whether some relevant differences remain after controlling for confounding factors, and some authors have reported significant increases even after adjustment for some or all of these factors.^{3 - 8} A wide heterogeneity in methods, inclusion criteria, and adjustment procedures as well as exclusion of subgroups of patients from analyses precludes straightforward comparison of results among studies.²

Although population-based registries of AMI that include out-of-hospital deaths should provide a reliable framework for the assessment of mortality differences between men and women, hospital AMI registries include better assessment of severity of illness and underlying comorbidity. In this study, we used data from the RESCATE (Recursos Empleados en el Síndrome Coronario Agudo y Tiempos de Espera [Resources Used in Acute Coronary Syndrome and Delays in Treatment]) study, which is a prospective, multicenter registry of consecutive patients with first AMI admitted to 4 teaching hospitals in Catalonia, Spain, to assess the influence of sex on survival following AMI.

The RESCATE study was aimed at determining whether the availability of tertiary resources in the same center to which patients were admitted with acute coronary syndromes (first AMI) affected the number of and delay in diagnostic and therapeutic procedures performed.⁹ Participants in this comprehensive registry were followed up for 6 months to determine mortality and morbidity related to the index event.

From 1992 through 1994, consecutive patients aged 80 years or younger with a diagnosis of AMI within 72 hours after the onset of AMI symptoms reported in the emergency department or during hospital stay were enrolled. Acute myocardial infarction was diagnosed based on international World Health Organization criteria¹⁰ when abnormal new Q waves appeared in serial electrocardiograms or 2 of the following criteria were present: sequential ischemic ST-T changes on the electrocardiogram; increase in cardiac enzyme levels (creatine phosphokinase greater than twice the upper normal value, myocardial fraction of total creatine phosphokinase >10%, or both); and typical chest pain lasting more than 20 minutes. Patients with residence outside the study catchment areas, with life-threatening diseases other than the index event, previous coronary artery bypass grafting (CABG), percutaneous transluminal coronary angioplasty (PTCA), or coronary angiography in the last 6 months were excluded. Patients enrolled in ongoing clinical trials were not excluded.

A composite primary end point was predefined that included mortality or readmission within 6 months following the onset of AMI for any of the following reasons: reinfarction, ventricular fibrillation or tachycardia, congestive heart failure, or unstable angina. The following definitions were used: reinfarction, considered when a new infarction occurred more than 28 days after the onset of the initial event; ventricular fibrillation or tachycardia, considered as an end point only when sustained and prompted hospital admission; congestive heart failure, diagnosed on clinical grounds according to standard diagnostic criteria^{11 - 12} ; and unstable angina, progressive and rest angina considered unstable and requiring hospital admission, according to Braunwald criteria.¹³

Patients were followed up in an outpatient clinic. Those who did not attend their appointment were contacted by telephone. Medical records were reviewed to determine whether the cause for readmissions constituted an end point for the study. The cause of out-of-hospital deaths was investigated by interviews with the physician who signed the death certificate and with the patient's relatives, if necessary.

Each hospital followed its own treatment protocols and no attempt was made to standardize patient management in the participating institutions. However, all 4 hospitals had written AMI protocols that were in accordance with international guidelines.^{14 - 16}

The following variables were prospectively recorded: demographic characteristics, smoking status, history of hypertension, diabetes, chronic obstructive pulmonary disease, and peripheral vascular disease, AMI location, presence of Q waves on electrocardiogram, development of acute pulmonary edema or cardiogenic shock, presence of severe arrhythmia (defined as the occurrence of at least 1 episode of ventricular fibrillation or sustained ventricular tachycardia requiring immediate medical intervention) within the first 72 hours, delay from onset of symptoms to time of first inhospital monitoring, length of coronary care unit stay and hospital stay, and use of thrombolytic agents, exercise test, echocardiography, coronary angiography, PTCA, and CABG.

The sample size permitted us to obtain a statistical power of 0.80 in a 2-tailed test with an α level of .05 if a difference of 8% or more in the 6-month event rates was observed between men and women (20% and 28%, respectively, of primary end points). Because we allowed for a 10% increase in sample size to compensate for patients lost to follow-up, we estimated that 1300 patients were required, at least 325 of whom had to be women, to detect an adjusted odds ratio (OR) of 1.5 or more as statistically significant (P<.05) for women.

Differences between men and women were assessed with the χ² test for categorical variables and with the Student t test or Mann-Whitney U test as appropriate for continuous variables. Survival curves were estimated with the Kaplan-Meier method. Adjusted ORs for 28-day case fatality and 6-month mortality and morbidity were estimated using unconditional logistic regression.¹⁷ To control for differential characteristics of women and men also related to prognosis, all variables that met confounding factor criteria (ie, factors that statistically differed in bivariate analysis between men and women and were further associated with mortality, both at an α level of .10, but could not be considered as mechanisms of death) were included in the models.

Of the 2397 patients registered in the 4 participating hospitals, 1460 (60.8%) met the inclusion criteria (331 women and 1129 men). Reasons for exclusion were previous AMI (18.4% overall, 19.7% in men, and 14.7% in women), age 80 years or older (9.6% overall, 5.9% in men, and 20.7% in women), residence outside the catchment area (4.5% overall, 4.8% in men, and 3.5% in women), revascularization or angiography or both in the previous 6 months (1.3% overall, 1.6% in men, and 0.7% in women), and administrative reasons or terminal disease other than coronary artery disease (5.8% overall, 5.7% in men, and 5.8% in women). Only 3 patients were lost to overall follow-up.

Compared with men, women were older, more frequently had a history of diabetes, hypertension, or previous angina, and developed acute pulmonary edema or cardiogenic shock (Table 1). In contrast, women developed severe ventricular arrhythmias less often, smoked less, and had a history of chronic obstructive pulmonary disease less frequently. Median time between symptoms and initial presentation to the emergency department was 1 hour longer among women than men (P<.001). Delay between the emergency department and admission to the coronary care unit was also longer among women (9.3 vs 7 hours; P<.001). Thrombolysis, early coronary angiography, PTCA, and CABG were used more frequently among men than women at 28 days, although PTCA was more frequently used among women during the 6-month follow-up (Table 2). However, the difference in PTCA use-rate was not statistically significant when patients were analyzed by 3 age groups (Table 3). The rates of urgent and elective use of these diagnostic and therapeutic procedures did not differ between men and women either at 28 days or between 28 days and 6 months (data not shown).

Other treatments such as antiplatelet drugs, heparin, β-blockers, and warfarin sodium were similarly administered in the acute phase of AMI to men and women in a subsample of 370 patients in whom these therapies were recorded: 94.9% and 96.2%, 34.6% and 35.5%, 6.3% and 5.3%, and 22.5% and 17.1%, respectively. Mortality rates were significantly higher among women than men at 28 days (18.5% vs 8.3%) and at 6 months (28.5% vs 10.8%) (Figure 1), as were readmission rates at 6 months (23.3% vs 12.2%).

Compared with deceased patients, survivors of 6-month follow-up were younger, had a history of diabetes or angina less frequently, and less often presented with severe ventricular arrhythmia, anterior AMI, acute pulmonary edema, or cardiogenic shock. The survivor group also had a greater proportion of smokers and patients with non–Q-wave AMI (Table 4). Survivors also more frequently received thrombolytic agents and PTCA both at 28 days and at 6 months. Diagnostic procedures were performed more often among those who survived than those who were deceased or were readmitted at 6 months (Table 4).

The influence of female sex on 28-day and 6-month case fatality was independent of comorbidity and coronary risk factors (model 1) (Table 5) and was also independent of use of thrombolytic therapy (model 2). However, with further adjustment for severity-related clinical variables (ie, pulmonary edema or cardiogenic shock and ventricular arrhythmias) (model 3), the excess risks of women decreased for 28-day case fatality and 6-month mortality, respectively, and were no longer statistically significant. In contrast, female sex was independently associated with 6-month mortality or readmission even in the fully adjusted model (Table 5, models 1, 2, and 3).

This study shows that women are at increased risk for 28-day case fatality and 6-month mortality or readmission after a first myocardial infarction.

Several studies have found that women fare worse than men after myocardial infarction.^{3 - 8} However, agreement does not exist as to which variables are to be adjusted for in the statistical analyses. Most important, variables related to the previous risk status require adjustment because they determine the response capacity of a patient with AMI. However, use of thrombolytic agents, which are administered early in the acute phase of AMI, also needs to be adjusted for since it dramatically modifies prognosis.

In our study, to assess whether the worse prognosis of women is related to severity of AMI, further adjustment for clinical variables (ie, development of acute pulmonary edema or cardiogenic shock and presence of severe ventricular arrhythmias) was performed. Since the 2 types of complications may be considered mechanisms of death, these variables failed to meet the criteria for being considered confounding factors. Consequently, these models (Table 5) could be formally regarded as inappropriate, but they aid in understanding the mechanisms by which women may die of AMI more frequently than men. We found that both greater 28-day mortality and 6-month mortality are largely explained by more severe AMI profiles in women (Table 5, model 3) but not by their history of cardiovascular risk factors, comorbidity, or even by thrombolytic treatment. However, the 6-month mortality or readmission rate was greater in women than in men independent of adjustment for these covariates (Table 5, model 3).

Our 6-month mortality rate is similar to the 1-year mortality rate reported by Greenland et al.³ Apart from this study, it is difficult to compare our findings with other studies owing to variations in patient characteristics, study designs, and follow-up durations.

We found that men were more likely than women to have angiograms, PTCA, and CABG procedures performed during the initial 28-day period. This has also been found in other studies.^{18 - 19} However, after that 28-day period, the rates favored women for PTCA and marginally for cardiac catheterization. This finding may indicate that women develop more severe AMI than men and require more therapy after discharge, which resembles the observations of other studies.²⁰ This observation is further supported by greater readmission rates for reinfarction, congestive heart failure, ventricular fibrillation or tachycardia, or unstable angina in women following discharge after their first AMI. When stratified by age group, differences in PTCA use disappeared, indicating that they were mainly because of this factor. In each age subgroup, a greater proportion of women did not undergo an exercise test. However, an echocardiogram was obtained in a greater proportion of women younger than 75 years, which may suggest that women are more physically limited for exercise testing and efforts were being made to obtain clinical information by other diagnostic procedures. Since women with AMI are more severely ill than men, similarities in the rate of diagnostic and therapeutic invasive procedures between sexes could be interpreted as undertreatment for women who have proved to benefit from more intensive therapeutic (eg, PTCA²¹ ) and diagnostic effort; however, this issue was beyond the scope of the present study. Other routine treatments such as antiplatelet drugs, heparin, and β-blockers were similarly provided to a subsample of men and women during the acute phase of myocardial infarction.

To some extent, our study failed to measure some initial cardiac frailty of women who develop AMI. Such a condition could be related to an unnoticed degree of previous cardiac failure or to other comorbid factors that determine the worse prognosis that we observed. In accordance with this hypothesis is the finding that the increased risk of 28-day mortality in 35- to 69-year-old women was explained by previous treatment with diuretics or inotropic medication.²²

In our study, women presented to the emergency department approximately 1 hour later than men, which might explain their lower thrombolysis rate. An explanation for this delay could be the fact that women were older than men and may have been less sensitive to symptoms than younger people. However, this is unlikely since the median delay of men older than 65 years (2.8 hours) was significantly lower than that of women of the same age (3 hours) (P=.01), although the magnitude of the difference was lower than the overall.

Population AMI registries suggest that the differences observed in case fatality^{23 - 25} between hospitalized men and women are balanced by a larger prehospital case fatality rate among men. These population registries include patients with previous AMI, thereby making it difficult to compare our results with their data. A population registry in the area where the present study took place suggests similar findings.^{26 - 27} However, as pointed out by Sonke et al,²³ only hospital registries in which severity of AMI is measured allow assessment of whether increased mortality in women is related to AMI severity. Management of the out-of-hospital acute phase of myocardial infarction needs a public health approach, whereas the different case fatality of men and women among admitted patients is a clinical problem that requires clarification. The available clinical information in population registries is limited because many out-of-hospital deaths are classified as unclassifiable or insufficient data .^{24 - 25} This makes it difficult to ascertain whether death was due to AMI and to determine the patient's clinical features and history characteristics. Conversely, this information is consistently available in most hospital registries.²³

In our study, we included only patients with first AMI to prevent the need for more complex model adjustments (eg, time since last event, number of previous events, or severity of these events). This selection provides more homogeneity than when all cases of AMI are included. Moreover, this method eliminates the possibility of weighting more severely ill patients readmitted for recurrent events who enter the AMI registery on several occasions. As previously suggested,²⁸ the use of some restrictive criteria such as upper-age limit (80 years in our study) and only first AMI may help to improve the accuracy of excess risk estimates, since generalization of the results, although not complete, is sufficiently wide, precise, and easily comparable.

We reported all-cause mortality in the study, but cardiac deaths accounted for the vast majority of deaths in this cohort. Only 11 (5.4%) of the 205 deaths that occurred in the study were of noncardiac origin. Exclusion of these 11 cases did not change the results. The follow-up period of 6 months provides a window in which most cardiac complications related to the index event occur.

Another source of variation in the results of different studies is the inclusion or exclusion of patients who die early after admission (eg, in the emergency department). In our study, all patients in whom a diagnosis of AMI was established were included. This issue is particularly relevant given that early mortality in the general population may have accounted for more than 30% of 28-day AMI mortality in 25- to 74-year-old patients.^{26 - 27}

In conclusion, our data indicate that both higher 28-day case fatality and 6-month morbidity and mortality among women with a first AMI compared with men are mainly due to greater severity, but are independent of age and related comorbidity. The 6-month rates for readmission or death were higher in women, independent of any factor considered including severity.