Roger VL, Farkouh ME, Weston SA, Reeder GS, Jacobsen SJ, Zinsmeister AR, Yawn BP, Kopecky SL, Gabriel SE. Sex Differences in Evaluation and Outcome of Unstable Angina. JAMA. 2000;283(5):646-652. doi:10.1001/jama.283.5.646
Author Affiliations: Division of Cardiovascular Diseases and Internal Medicine (Drs Roger, Reeder, and Kopecky) and Sections of Biostatistics (Ms Weston and Dr Zinsmeister) and Epidemiology (Drs Jacobsen and Gabriel), Mayo Clinic and Mayo Foundation, and Department of Research, Olmsted Medical Center (Dr Yawn), Rochester, Minn; and Division of Cardiology, Cardiovascular Institute, Mount Sinai Medical Center, New York, NY (Dr Farkouh).
Clinical Cardiology Section Editors:
Bruce Brundage, MD, University of California, Los Angeles, School of
Medicine; Margaret A. Winker, MD, Deputy Editor, JAMA.
Context The existence of sex bias in the delivery of cardiac care is controversial,
and little is known about the association between sex and delivery of care
and outcomes at an early point in the diagnostic sequence, such as when patients
present for the evaluation of chest pain.
Objective To test the hypothesis that female sex is negatively associated with
care delivered to and outcomes of persons diagnosed as having unstable angina.
Design Inception population-based cohort study with an average of 6 years of
Setting Emergency departments (EDs) in Olmsted County, Minnesota.
Patients A total of 2271 Olmsted County residents (1306 men and 965 women) who
presented to the ED for the first time with symptoms meeting criteria for
unstable angina between 1985 and 1992.
Main Outcome Measures Use of cardiac procedures within 90 days of ED visit, overall mortality,
and cardiac events (cardiac death, nonfatal myocardial infarction, nonfatal
cardiac arrest, and congestive heart failure), compared by sex and Agency
for Health Care Policy and Research cardiovascular risk category (low, intermediate,
Results Women were older (P<.001), more likely to
have a history of hypertension (P = .001), and less
likely to present with typical angina (P = .004)
than men. Men were more likely than women to undergo noninvasive cardiac tests
(relative risk [RR], 1.27; 95% confidence interval [CI], 1.14-1.40) as well
as invasive cardiac procedures (RR, 1.72; 95% CI, 1.51-1.97). After adjustment,
male sex was associated with a 24% increase in the use of cardiac procedures.
Survival of both men and women in the high and intermediate risk categories
was significantly lower than expected per the general population (P<.001). Women had a worse outcome than men, but after multivariate
adjustment, male sex was associated with a trend toward an increase in the
risk of death (RR, 1.23; 95% CI, 0.99-1.54) and significantly associated with
increased risk of cardiac events (RR, 1.21; 95% CI, 1.03-1.42).
Conclusions Our population-based data indicate that after an ED visit for symptoms
of unstable angina, the use of cardiac procedures was lower in women, but
after taking into account baseline characteristics, men experienced worse
Sex differences in the delivery of cardiac care have been clearly documented.1- 11
Although considerable controversy remains concerning the appropriateness of
such differences, there is evidence to suggest that once the diagnosis of
coronary disease is established, such as after myocardial infarction (MI)
or after coronary angiography, the delivery of cardiac care is similar for
men and women.12 Thus, it is of interest to
examine sex differences at an earlier point in the diagnostic sequence, when
the presence of coronary disease is not established. A recently published
computerized survey of 720 primary care physicians indicated that sex independently
influenced the management of chest pain,13
a frequent symptom in women and a frequent cause of emergency department (ED)
visits. It is unknown whether a similar difference exists for patients receiving
a diagnosis of unstable angina, and whether it affects outcomes.
This study tested the hypothesis that female sex was negatively associated
with the care delivered to and the outcome of persons diagnosed in the ED
with unstable angina.
Epidemiologic research in Olmsted County, Minnesota, is possible because
the county is relatively isolated from other urban centers and nearly all
medical care is delivered to local residents by a handful of providers. Other
than that a higher proportion of the working population is employed in the
health care industry, the characteristics of the population of Olmsted County
are similar to those of US whites in general.
Potential epidemiologic studies in the community are enhanced by the
fact that each provider uses a medical record system, whereby all data collected
on an individual are assembled in 1 place. Thus, the details of every encounter,
including visits to the ED, and all laboratory results, pathology reports,
and correspondence concerning each patient, can be accessed. The result is
the linkage of medical records from all sources of medical care used by the
Olmsted County population; these are easily retrievable because the Mayo Clinic
has maintained extensive indices based on clinical and histologic diagnoses
and surgical procedures since the early 1900s.14,15
Since 1966, similar indices have been developed for non-Mayo providers under
the aegis of the Rochester Epidemiology Project.
The Rochester Epidemiology Project indices, augmented by the log books
of the EDs (St Mary's, Rochester Methodist, and Olmsted Community hospitals),
were used to identify a population-based cohort of Olmsted County residents
who had an initial ED visit for symptoms consistent with unstable angina between
January 1, 1985, and December 31, 1992. The study was approved by the Mayo
Clinic Institutional Review Board. Minnesota State law requires that patients
seen after January 1997 must authorize use of their medical records for research
studies, and patients who refused research authorization were excluded.
Residency in Olmsted County was verified with information from birth
certificates and city and county directories.
Eligible patients included all residents presenting for the first time
to the ED with chest pain meeting criteria for unstable angina during the
study period. Eligibility was determined based on evidence available in the
ED. Unstable angina was defined as 1 of the following: symptoms of angina
at rest lasting longer than 20 minutes, new onset exertional angina that met
the Canadian Cardiovascular Society criteria for class 3 or higher, variant
angina, or postmyocardial infarction (MI) angina.16
Patients were excluded if they presented to the ED with criteria for ongoing
acute MI (ST-segment elevation ≥1 mm in 2 or more leads) or with definite
other cause of chest pain, including pleuritic chest pain, pulmonary embolism,
musculoskeletal cause, trauma, pneumonia, pericarditis, and dissecting aortic
The medical records were reviewed by a team of physicians and nurse
abstractors to collect data with which to characterize patient demographics,
clinical signs and symptoms, and cardiovascular disease risk factors (history
of familial coronary disease, smoking, diabetes mellitus, hypertension, and
hyperlipidemia defined by a total serum cholesterol level ≥5.7 mmol/L [220
mg/dL]). Elevated creatine phosphokinase (CPK) was defined as peak CPK greater
than 1.5 times the upper limit of normal. The evaluation of symptoms was made
from ED assessment or the physician's initial admission note and the information
used to make this assessment always preceded test results. Patients were classified
in 3 risk categories according to the guidelines for initial disposition of
patients with unstable angina published by the Agency for Health Care Policy
and Research (AHCPR, now the Agency for Healthcare Research and Quality).16 All electrocardiograms were read by cardiologists
or ED physicians blinded to the hypothesis of the study and sex of the patient.
All end points were ascertained through record review. Disposition from
the ED was categorized as intensive care unit admission, admission to telemetry
or to general ward bed, or home dismissal. The use of cardiac and gastrointestinal
procedures during the 90 days following the ED visit was examined. Cardiac
procedures were categorized as noninvasive, including procedures performed
at rest (echocardiography, resting radionuclide angiography, resting sestamibi
study, and ultrafast computed tomography) and stress tests (exercise and pharmacologic
stress testing with or without nuclear or ultrasound imaging), or invasive,
consisting of diagnostic coronary angiography. Gastrointestinal procedures
included upper gastrointestinal radiography and upper gastrointestinal endoscopy,
esophageal manometry, and reflux studies.
The Rochester Epidemiology Project obtains copies of death certificates
for all Olmsted County residents known to have died. These are linked to the
medical record of all Olmsted County residents. This ensures virtually 100%
complete follow-up for the mortality end point.
Cardiac events included cardiac deaths, nonfatal MI, nonfatal cardiac
arrest, and congestive heart failure. The definition of MI was based on the
documentation in the medical record of the occurrence of chest pain typical
for an ischemic origin, or characteristic changes in the electrocardiogram
and/or cardiac enzymes. Congestive heart failure was defined as the occurrence
of exertional or paroxysmal nocturnal dyspnea responding symptomatically to
digitalis, diuretics, or afterload-reducing agents. The definition of cardiac
arrest was based on documentation of this diagnosis by the clinician in the
Associations between sex and baseline characteristics with sex were
tested by means of proportional hazards. The outcome measures examined were
disposition from the ED, use of cardiac procedures within 90 days after ED
visit, overall survival, and survival free of cardiac events (defined as cardiac
death or nonfatal cardiac event, including nonfatal MI, nonfatal cardiac arrest,
and congestive heart failure).
Among hospitalized patients, disposition from the ED (intensive care
unit, telemetry, or nonmonitored hospital ward vs intensive care) was analyzed
using logistic regression analysis adjusting for sex and AHCPR risk category.
Use of cardiac procedures within 90 days after the ED visit was analyzed
by means of proportional hazards modeling. Association between sex and use
of any cardiac procedure was examined and proportional hazards modeling within
the group undergoing cardiac procedures was used to examine the association
between sex and use of invasive procedure. Survival was analyzed by means
of the Kaplan-Meier method. Survival observed in each sex and AHCPR risk category
was compared with the age- and sex-specific expected survival of the 1990
Minnesota population using the log-rank statistic. Proportional hazards modeling
was used to examine the association between sex and time to all-cause death
and between sex and time to first cardiac event.
Variables in the models included age; type of chest pain (typical or
atypical for angina pectoris); AHCPR risk category; number of coronary disease
risk factors; history of MI; ST-segment depression on an electrocardiogram;
elevated CPK; and the exposure variable of interest, sex. Tests for first-order
interactions between sex and other variables were examined.
Between 1985 and 1992, 6812 Olmsted County residents were evaluated
in 1 of the Olmsted County EDs for chest pain. Screening of the records of
these visits by a nurse abstractor under supervision of 1 of the authors (M.E.F.)
resulted in the identification of individuals eligible for study entry. The
reasons for exclusion from the study were residency outside Olmsted County
(799), electrocardiogram criteria for acute MI with ST elevation ≥1 mm
in 2 or more leads (247), nonincident nature of the episode of unstable angina
during the study period (1082), noncardiac or noncoronary cause of chest pain
(1647), angina pectoris but with no AHCPR criteria for unstable angina (47),
refusal to be included in studies (82), and not meeting physician criteria
Thus, during the study period, 2271 residents of Olmsted County (1306
men and 965 women) were seen in the ED for chest pain meeting criteria for
unstable angina and constitute the study cohort.
At index ED visit, women were approximately 7 years older and less likely
to be smokers than men and were more likely to have a history of hypertension
and hypercholesterolemia (Table 1).
Prevalence of history of familial coronary artery disease or MI was the same
in both sexes. In terms of symptom status at the time of the ED visit, the
type of chest pain was more frequently described as atypical for angina pectoris
in women, although similar percentages of men and women had prolonged pain.
Women were more likely to have electrocardiogram abnormalities than men.
According to the AHCPR guidelines, the majority of patients (69%) were
classified as intermediate risk, qualifying for admission to the intensive
care unit or for telemetry; 19% as high risk, justifying admission to the
intensive care unit; and 12% as low risk, potential candidates for outpatient
observation. There was no association between risk category and sex, and the
association of sex with baseline characteristics described above remained
unchanged after adjustment for AHCPR risk category. The frequency of patients
with elevated CPK was 18% for both men and women.
ED Disposition. Few patients were dismissed to home (men, 4%; women, 6%) or admitted
to a nonmonitored hospital ward (men, 1%; women, 2%). The majority of the
patients were admitted to a monitored or intensive care unit. After adjustment
for age and AHCPR risk category, there was no association between sex and
the type of hospital unit to which patients were admitted, telemetry vs intensive
care (odds ratio [OR], 1.13; 95% confidence interval [CI], 0.93-1.36).
Procedures Used Within 90 Days After ED Visit. A total of 1408 procedures were used in the 965 women and 2576 procedures
in the 1306 men. The majority of procedures were cardiac, mainly stress tests
and coronary angiography. The overall use of procedures was high (men, 87%;
women, 76%) and men were 30% more likely to undergo any procedure than women
(relative risk [RR] for men vs women, 1.31 [95% CI, 1.19-1.43]). Eighty-five
percent of men and 72% of women underwent a cardiac procedure (RR, 1.34; 95%
CI, 1.22-1.47). Women were more likely to undergo gastrointestinal procedures
than men (RR, 0.74; 95% CI, 0.60-0.93). The percentage of men and women undergoing
various types of cardiac procedures is shown in Table 2 along with the RRs for men vs women. Both the crude RRs
and the RRs adjusted for age and AHCPR risk category are presented. Men were
more likely to undergo noninvasive cardiac procedures, which reflects a greater
use of stress testing in men. There was no sex difference in the use of echocardiography
and other resting procedures. The use of coronary angiography was higher in
men. These crude associations were minimally attenuated by adjustment for
age and AHCPR risk category.
The intensity of procedure use was greater in men than in women. Twenty-four
percent of women and 13% of men underwent no procedure, whereas 18% of women
and 29% of men underwent more than 2 procedures (P
for trend, .001).
Table 3 shows the model
used to examine the association between sex and cardiac procedure use. After
adjustment for other predictor variables including age, typical nature of
the chest pain, presence of ST-segment depression, elevated CPK, and AHCPR
risk category, men were 24% more likely to undergo cardiac procedures (RR,
1.24; 95% CI, 1.11-1.37). This association remained similar when the analysis
was restricted to the use of procedures performed within 30 days of the index
event (RR, 1.22; 95% CI, 1.01-1.36).
Further analysis examined the association between sex and type of cardiac
procedures used among the 1803 persons undergoing any cardiac procedure. After
adjustment for other covariates (including age, AHCPR risk category, type
of chest pain, elevated CPK, and cardiac risk factors), men were more likely
to undergo coronary angiography than women were (RR for male sex, 1.40; 95%
CI, 1.21-1.61; P<.001).
The mean (SD) length of stay in the hospital was 3.02 (2.80) days for
the 198 patients (118 men and 80 women) who did not undergo any testing and
5.86 (8.81) days for the 1974 patients (1130 men and 844 women) who underwent
at least 1 test.
The mean (SD) duration of follow-up was 6.0 (3.1) years. In men, 358
deaths and 1050 cardiac events (179 cardiac deaths, 50 nonfatal ventricular
arrhythmias, 240 nonfatal MIs, and 581 cases of congestive heart failure)
occurred during 8054 person-years of follow-up. In women, 351 deaths and 881
cardiac events (171 cardiac deaths, 26 nonfatal ventricular arrhythmias, 187
nonfatal MIs, and 497 cases of congestive heart failure) occurred during 5502
person-years of follow-up.
All-Cause Mortality. The 6-year overall survival rate was 78% for men and 71% for women (P<.001). Observed survival was not different from expected
survival for men and women in the low-risk category. In the intermediate-
and high-risk categories, an excess mortality was observed with significant
differences between observed and expected survival for both men and women
(P<.001). For women in the AHCPR high-risk category,
the 3-year survival was 68% (95% CI, 61%-74%) vs an expected 3-year survival
of 86%. Likewise for men in the high-risk category, the 3-year survival was
75% (95% CI, 69%-80%) vs an expected 3-year survival of 88%.
Associations between clinical variables and time to death are shown
in Table 4. Women fared worse
than men in that male sex was univariately associated with a 31% decrease
in the risk of death.
In addition to sex, age, AHCPR risk category, and cardiac risk factors,
history of MI and elevated CPK and ST-segment depression were univariate predictors
of time to death. After adding these variables to the model, there was a trend
toward an excess risk of death in men. The analyses were repeated while examining
the outcome during the first 30 days of follow-up and consistently indicated,
by the direction of the association, an excess risk of death in men (adjusted
RR for male sex, 2.10; 95% CI, 0.81-5.44; P = .14).
Cardiac Events. The univariate 6-year survival free of cardiac events was 70% for men
and 63% for women (P = .001). Associations between
baseline characteristics and time to cardiac events are shown in Table 5.
Women fared worse than men, but, in addition to sex, age, AHCPR risk
category, and cardiac risk factors, history of prior MI, ST-segment depression
and elevated CPK were univariate predictors of time to cardiac event. After
adding these variables in the model, there was a 21% excess risk of cardiac
events in men.
There was no interaction between sex and the dependent variables included.
The analyses were repeated while examining the outcome during the first 30
days of follow-up and indicated a 44% increase in the risk of cardiac events
in men at 30 days (adjusted RR for male sex, 1.44; 95% CI, 1.02-2.05; P = .04).
This study tested the hypothesis that, among adults receiving an ED
diagnosis of unstable angina, female sex is negatively associated with the
delivery of cardiac care and subsequent outcome. We found an association between
female sex and lesser use of cardiac procedures that could not be explained
by measured sex differences in baseline characteristics. The majority (88%)
of men and women receiving an ED diagnosis of unstable angina were classified
as intermediate or high risk according to the AHCPR guidelines, and their
survival was significantly worse than the expected survival. This indicates
that unstable angina portends a poor prognosis irrespective of sex. In survival
analyses and after adjustment for age and other baseline characteristics,
men were at increased risk for cardiac events and death.
Each year in the United States, approximately 5 million persons are
evaluated in EDs for chest pain,17 underscoring
the importance of chest pain as a public health problem. Of these patients,
those with unstable angina represent the greatest challenge to clinicians
because of the risk, albeit small, of adverse cardiac outcomes.16
Extensive efforts have been made to define better triage mechanisms that would
allow reliable identification of high-risk patients while minimizing the number
of hospital admissions for chest pain syndromes in low-risk patients,18- 25
and practice guidelines have been published for the disposition of patients
presenting to the ED with unstable angina.16
In the current series, a small percentage of the whole population of both
sexes was classified as low risk, consistent with previously reported aggregate
data.26 The lack of sex difference in risk
category and the small percentage of women in the low-risk group underscore
the continuing challenge that the evaluation and management of chest pain
syndromes continues to represent in both sexes.
Little information is available on the association between sex and the
delivery of care for unstable angina. A secondary analysis from the Acute
Cardiac Ischemia Time-Insensitive Predictive Instrument (ACI-TIPI) clinical
trial indicates that women with angina pectoris seen in the ED were less likely
to undergo cardiac catheterization or revascularization.27
This sex difference in the delivery of cardiac care was not observed with
MIs.27 Even though these data, which support
the "Yentl syndrome," whereby to be treated as men, women would have to present
like men, described by Healy,12 are provocative,
several points must be kept in mind. First, our study was not designed to
investigate sex differences in the delivery of cardiac care as a primary hypothesis,
and the observed differences may have been related to unmeasured confounding
variables. Second, the delivery of care could conceivably have been influenced
by the trial setting, thus compromising external validity.
Recently, Schulman et al13 reported the
results of a survey of primary care physicians that examined referral to cardiac
catheterization according to race and sex among case vignettes of patients
presenting with chest pain. The authors reported that race and sex influenced
physicians' recommendations for the management of chest pain, with women being
less likely to be referred to cardiac catheterization. While the study design
was free of some of the biases related to retrospective analyses, it addressed
the management of chest pain but not unstable angina, a syndrome associated
with the perception of a greater likelihood of adverse events. In addition,
because hypothetical examples were used, the study could not examine the impact
of the documented sex differences in care on outcome.
The current study extends the findings of previous studies by indicating
a strong positive association between male sex and use of cardiac procedures
in a geographically defined population diagnosed in the ED with unstable angina.
One could hypothesize that reported sex differences in the sensitivity and
specificity of noninvasive procedures, in particular, stress testing, could
play a role in the sex differences in utilization that were noted in our study.28- 32
The data presented herein indicate that the majority of women and men
receiving an ED diagnosis of unstable angina are classified in the high and
intermediate AHCPR risk categories and experience excess mortality compared
with the general population. The public health implications of these findings
are that women with unstable angina constitute a relatively high-risk population.
After accounting for age and other baseline characteristics, men experience
worse outcomes than women do both at 30 days and throughout the entire follow-up,
with an excess risk of cardiac events and a trend toward excess risk of death.
The clinical implications of these data in terms of health care delivery are
challenging. In particular, while women in this cohort were subjected to fewer
cardiac procedures than men, one cannot infer from these data that more aggressive
management of women with unstable angina is warranted. Indeed, opposite inferences
can be drawn from these analyses: that procedures are overutilized in men
or underutilized in women. It is uncertain that performing more invasive cardiac
procedures in women, who present with unstable angina at an older age than
men, would result in better outcomes. This hypothesis would be best tested
in a randomized trial, the relevance of which is underscored by the large
proportion of persons who receive an ED diagnosis of unstable angina and belong
to AHCPR risk categories associated with excess mortality.
Underascertainment of persons presenting to the ED for the evaluation
of unstable angina is unlikely because the resources of the Rochester Epidemiology
Project provide access to all medical records for the care provided to Olmsted
County residents. However, ascertainment of all patients seeking medical care
for chest pain may have not been complete. Other settings in which such evaluations
may occur theoretically could include office visits to primary care physicians
or cardiologists. However, the intent of our study was to capture the cohort
of patients with chest pain who meet criteria for unstable angina, and the
likelihood that such symptoms would not result in a referral to the ED is
small. Reliance on the medical records for measurement of potential confounding
variables should be considered. The care delivered to the present cohort reflects
practices in 1985-1992, which raises the issue of generalizability to more
contemporary practices. While no data suggest that patterns of care have changed,
this should be the subject of future studies.
As in any observational study, residual confounding should be considered
as an alternate explanation of the findings reported herein; in particular,
patient preference could not be captured.
Because this study was population-based, the potential effects of referral
bias are greatly diminished; however, the underrepresentation of minorities
in this cohort, which was cared for essentially by 1 group of providers, compromises
the generalizability to different racial and ethnic groups or to different
health care environments. However, no single population or health care system
can be expected to be completely representative of the general population;
thus, our findings should be further examined in different populations and
These population-based data indicate that after an ED visit for unstable
angina, the utilization of cardiac procedures is lower in women. Men experience
worse outcomes than women, with an increased risk of death and cardiac events.
Further studies are needed to examine whether such adverse outcomes can be
altered by interventions.