Sex differences in clinical presentation, management and outcome in emergency department patients with chest pain

EM Advances

CJEM 2010;12(5):405-413

Abstract

Résumé

Introduction

Chest pain is the second most common chief complaint in emergency departments (EDs) in North America, accounting for more than 6 million patient visits annually.1 When evaluating a patient with acute chest pain, clinicians use readily available information obtained from the history, physical examination, electrocardiogram (ECG) and basic laboratory tests to identify noncardiac etiologies and determine the likelihood of acute coronary syndrome (ACS). Clinicians often base their decision on whether to pursue additional cardiac testing such as stress testing or coronary angiography on an unstructured assessment of the pretest probability of disease.

Previous studies have documented sex differences in the evaluation and management of acute chest pain. Other investigators have reported lower rates of cardiac catheterization in women, even after adjusting for baseline risk and other potential confounding factors.2–5 Silbergleit and McNamara6 reported lower rates of hospital admission in women with nontraumatic chest pain. Kaul and colleagues,7 in a large administrative database study of more than 54  00 patients, reported that women presenting to the ED with ACS were less likely than men to be admitted and to undergo coronary revascularization. Despite receiving less aggressive management, women in this study had similar outcomes compared with men at 1 year.

Most prior studies on sex differences in ACS have been conducted in the inpatient setting or have used large administrative databases to assess potential sex differences in management and outcome. Relatively few ED-based studies have been published. We hypothesized that sex differences in clinical presentation and pretest probability for ACS would account for management differences in ED patients with chest pain.

Methods

Study design and setting

We conducted a prospective cohort study enrolling consecutive eligible patients presenting with chest pain to the ED of a university-affiliated urban medical centre, with an annual ED census of 60 000 patient visits. The institution’s research ethics board approved the study without the need for written informed consent. Patients provided verbal consent during a telephone interview conducted by a study nurse.

Population

We designed the study to include patients at low to moderate risk for ACS, whose care often poses the greatest diagnostic challenge for clinicians. The study population consisted of patients over 24 years of age who presented to the ED with a primary complaint of chest pain. Exclusion criteria were as follows: acute STsegment elevation in at least 2 contiguous leads, hemodynamic instability or tachycardia (systolic blood pressure < 90 mm Hg; heart rate < 50 or > 100 beats/min), a history of cocaine use or positive test for cocaine, communication or language problems such that a reliable history could not be obtained, a clear traumatic etiology of pain, a terminal noncardiac illness or prior enrolment within 30 days.

Data collection

We identified variables to be collected based on literature review and consensus agreement from the investigation committee, comprised of the study authors. We designed standardized data collection forms to prospectively collect data on cardiac risk factors, cardiovascular history, characteristics of the chest pain history and physical examination, and outcomes according to standardized reporting guidelines for studies evaluating ED patients with potential ACS.8 Before data collection began, the primary investigator trained physician assessors to ensure unambiguous interpretation of data collection forms and uniform collection of data. We conducted a 2-monthrun-inphase during which the data collection forms and variable definitions were refined as necessary.

On patient arrival, registration clerks or triage nurses attached a standardized data collection form to the ED record of treatment for all patients with chest pain. Onduty attending emergency physicians certified in emergency medicine or supervised emergency medicine residents assessed patient eligibility, completed data collection forms and ordered diagnostic investigations as appropriate. Physicians completed data collection forms immediately after patient evaluation and before ordering diagnostic investigations to ensure that assessment of the clinical variables was not biased by knowledge of the outcome. We specifically instructed physicians to assess patients’ pretest probability for ACS after the ECG was performed but before obtaining the results of cardiac troponin T testing. Cardiac troponin T levels were measured on patients’ arrival at the ED and 6 hours or longer after the onset of pain, with at least 4 hours between samples. We used the Elecsys troponin T assay by Roche Diagnostics. The 99th percentile of the reference range for this assay is less than 0.01 µg/L and the 10% coefficient of variation is 0.035 µg/L.

After patient discharge, a study nurse attached the ED record of treatment to the standardized data collection form along with a copy of the first interpretable ECG and results of laboratory testing, cardiac stress testing and coronary angiography, when available. The study nurse collected additional data from the medical record of eligible enrolled patients and recorded it on a designated case record form. To determine the number of eligible patients who were missed, a study nurse reviewed the log of ED patients for all visits with a primary complaint of chest pain, and completed a separate case record form for missed eligible patients. The primary investigator, unaware of both predictor variables and patient outcome, interpreted ECGs of all enrolled patients according to current standardized reporting guidelines.8 We also reviewed the medical record for all patients starting at 1 month for the occurrence of outcomes. The electronic medical record at our institution contains information from both inpatient visits to the 4 major hospitals in our area and outpatient visits to clinics affiliated with the Ottawa Hospital. A study nurse conducted structured telephone followup 1 month from the ED visit for all enrolled patients to obtain information on any outcomes not documented in the medical record.

Outcome measures

We defined ACS as acute myocardial infarction (AMI), revascularization (percutaneous or surgical), death from cardiac or unknown cause, a new perfusion defect on radionuclide stress imaging, or a stenosis of 70% or greater in at least 1 of the major epicardial coronary arteries.9,10 We included all outcomes that occurred after patient assessment, whether in the ED, in the hospital or after ED discharge.

We defined AMI as either of the following: a cardiac troponin T level of 0.01 µg/L or greater with a rising or falling pattern (defined as a change of ω 0.03 µg/L for values that were initially < 0.20 µg/L; for levels ω 0.20 µg/L, a positive cardiac troponin T was defined as a change of ω 20% between samples);11,12 or development of pathologic Q waves on the ECG or ECG evolution consistent with AMI. We defined revascularization as reestablishment of coronary artery patency by percutaneous coronary intervention or coronary artery bypass graft surgery. We defined significant coronary disease as stenosis of 70% or greater in any of the major epicardial coronary arteries.9

All positive and 10% of randomly selected negative outcomes were confirmed by a second co-investigator blinded to the standardized data collection forms. Disagreements were resolved by consensus. If a consensus could not be reached between 2 co-investigators, a third co-investigator resolved discordances.

Statistical analysis

Univariate analysis techniques were used to determine the statistical significance of differences observed between men and women appropriate for the type of data: for nominal data, the χ2 test with continuity correction; for ordinal variables, the Mann–Whitney U test; for continuous variables, the unpaired 2tailed t test, using pooled or separate variance estimates, as appropriate. Receiver operating characteristic curve analysis was performed to determine the diagnostic accuracy of physicians’ pretest probability assessment for ACS by sex. Multiple logistic regression was performed to measure the association between female sex and coronary angiography within 30 days while controlling for predetermined confounders. To ensure stability of the regression coefficients, the number of variables entered into the multiple logistic regression model was restricted to maintain an eventpervariable ratio of at least 10:1.13 MedCalc version 10.4.0.0 (MedCalc Software) was used for receiver operating characteristic curve analysis and SAS software (SAS Institute, Inc.) version 9.1 TS Level 1M3 for all other analyses.

Results

The total ED census from Jul. 1, 2007, to Apr. 1, 2008, was 45 874 patient visits. During this period, 1527 (3.3%) patients were assessed for eligibility (Fig. 1). Of the 1415 patients eligible for enrolment, physicians prospectively completed data collection forms for 1017 (71.9%). We were unable to contact 47 patients by telephone at 30 days; the remaining 970 (95.4%) patients were contacted and included in the final analysis. Baseline characteristics of patients eligible for inclusion who were enrolled and missed were similar in all respects (Table 1).

The mean age of the patients was 59.5 (standard deviation 13.8) years (Table 2). Compared with male patients, a lower proportion of female patients were admitted to the hospital, had a history of previous myocardial infarction or had known coronary artery disease.

A lower proportion of women described their pain as worse with exertion or similar to previously diagnosed ischemia (Table 3). Clinicians considered the chest pain syndrome to be typical for ACS less frequently in women. Physicians classified a greater proportion of women as having a low (< 10%) pretest probability for ACS.

Figure 2 shows the results of physicians’ pretest probability assessments, by sex. On the whole, a greater proportion of women were classified in the lower pretest probability categories and a greater proportion of men were classified in the higher pretest probability categories.

Figure 3 shows the diagnostic accuracy of pretest probability assessments by sex. There was no significant difference in the area under the receiver operating characteristic curve (AUC) between women and men, respectively (AUC = 0.82, 95% confidence interval [CI] 0.76–0.83; AUC = 0.80, 95% CI 0.78–0.86; p = 0.73 for difference).

Physicians referred similar proportions of men and women for cardiac stress testing (Table 4); however, a lower proportion of stress tests were positive for ischemia in women. Women were subsequently referred for coronary angiography less frequently and had a lower rate of significant coronary artery disease. Among those referred for coronary angiography, the rate of significant coronary disease (80.6% v. 81.8%, p = 0.87) and the rate of revascularization (58.3% v. 63.6%, p = 0.57) were similar between sexes. The rate of revascularization among those with significant coronary disease on angiography was also similar (72.4% v. 77.8%, p = 0.55). There was a lower rate of AMI and no deaths in women within 30 days of the ED visit.

The unadjusted odds ratio (OR) for coronary angiography in women was 0.44 (95% CI 0.30–0.66). After controlling for predetermined confounders (e.g., age, thrombolysis in myocardial infarction risk score, elevated cardiac troponin T level, new ischemic changes on ECG, total number of cardiac risk factors, pretest probability for ACS and typical pain), the association between female sex and coronary angiography was no longer significant (OR 0.63, 95% CI 0.37–1.10). Table 5 shows the ORs and respective 95% CIs for each predictor in the adjusted multiple logistic regression model.

Discussion

In this prospective cohort study we observed that, compared with men, women had a lower prevalence of

known coronary artery disease and less frequently presented with typical chest pain. Physicians classified a greater proportion of women as having a low (< 10%) pretest probability for ACS. Despite similar rates of ECG, troponin T and stress testing between sexes, there was a lower rate of AMI and positive stress tests in women. Although a lower proportion of women were referred for coronary angiography, the adjusted association between female sex and coronary angiography was not significant. These data suggest that sex differences in clinical presentation and pretest probability likely account for the lower rate of coronary angiography in women and that care was appropriate for the probability of disease.

Our findings differ from another ED-based study in patients with potential ACS. Chang and coauthors3 observed that men received more cardiac catheterizations

and more stress tests than women, even after adjusting for potential confounding factors. We similarly found that men received more cardiac catheterizations then women; however, after adjusting for potential confounders, the association between female sex and coronary angiography was not significant. What are some potential explanations for these differences? One possibility is that our study collected data on physicians’ assessment of pretest probability for ACS and adjusted for it in the multiple logistic regression model. It is also possible that sociocultural differences between Ottawa, Ont., and Pittsburgh, Pa., may be associated with different patterns of the management of patient care. Finally, residual confounding may be present in both investigations. One potentially important confounder that was not assessed in either study was the impact of patient preference. It is possible that women, in concert with their physicians, less frequently opted for coronary angiography. As neither study collected data on patient preference, the degree to which this may have influenced results is uncertain.

nother study on ED patients reported findings consistent with our observations. Kaul and colleagues7 collected data on 54 134 ED patients in Alberta. These investigators identified ED patients admitted for AMI, unstable angina, stable angina and chest pain by merging data from 2 large databases in Alberta — the Ambulatory Care Classification System database and a hospital discharge database.5 They observed that women with each diagnosis were less likely than men to undergo revascularization within 1 year. In addition, these management differences were not associated with

sex differences in mortality at 1 year, suggesting that the lower rates of investigation and intervention in women did not result in worse outcomes.

Other studies that explore sex differences in clinical presentation in patients with ACS may put our observations in perspective. In their systematic review of studies comparing symptoms of ACS in men and women, Patel and coauthors14 found that women with ACS more frequently experienced back, jaw and neck pain, nausea and/or vomiting, dyspnea, palpitations and dizziness, whereas men more frequently presented with chest pain and diaphoresis. Similarly, Milner and colleagues15 in their study of 2073 patients admitted to hospital for AMI found that women were less likely than men to have a chief complaint of chest pain associated with their AMI. As most patients present to the ED with a primary symptom or complaint before diagnosis, much of ED-based research is based on chief complaints. In our study and another recent ED-based study3 patients with a primary complaint of chest pain were enrolled. If women with ACS are less likely to present with chest pain, it would therefore not be unexpected to observe a lower rate of ACS in our cohort. In this context, our study is consistent with other literature that suggests women with ACS present differently than men.16,17 However, among those who present with chest pain, women may have a lower rate of ACS.

One may question whether our observations were potentially influenced by workup or verification bias (e.g., women who underwent less intensive investigation

before ED presentation were considered to have a lower pretest probability for ACS by emergency physicians, underwent less intensive investigation and were therefore less frequently diagnosed with ACS). Although we considered this possibility, this explanation does not appear to be consistent with our observations. In our cohort ECGs were obtained in 100% of patients and cardiac troponin T levels in 99%. This suggests that the workup for AMI was not biased between sexes. In addition, similar proportions of men and women were referred for cardiac stress testing, and stress tests were less frequently positive for ischemia in women. Of those who were referred for angiography, there was a similar rate of significant coronary artery disease between sexes, and we observed no significant sex differences in revascularization among those diagnosed with significant coronary disease. These observations suggest that differences in the probability of ACS are a more likely explanation for management differences than bias.

Limitations

Our study had several limitations. We only included patients who presented with chest pain. Patients at risk for ACS who presented with non–chest pain syndromes such as shortness of breath, nausea, back pain, palpitations or generalized fatigue were not included. This limits the generalizability of these findings to those patients who present to the ED with a presenting symptom of chest pain. The patient sample was recruited from a single Canadian ED and findings may vary in other regions or countries with different ethnic and sociocultural characteristics.

Only 72% of eligible patients were enrolled. This is likely because physicians less reliably completed data collection forms at night when the ED was particularly busy. We collected demographic and cardiovascular history characteristics for all eligible patients who were missed and included, and observed no appreciable differences between groups. This decreases the risk of selection bias in our cohort.

Conclusion

Compared with men, women presenting to the ED with chest pain less frequently had typical features of chest pain, were more frequently classified as having a low pretest probability for ACS, had a lower rate of stress tests positive for ischemia and had a lower rate of AMI. These data suggest that sex differences in management were likely appropriate for the probability of disease. Future studies evaluating sex differences in patients with possible ACS should explore the impact of patient preference on investigation and intervention.