Sex Differences in 1-Year Outcomes After Percutaneous Coronary Intervention in the Veterans Health Administration

Abstract

Background: Advancements in percutaneous coronary intervention (PCI) for treating obstructive coronary artery disease have reduced major adverse events, including mortality. Yet, evidence as to whether women and men experience similar outcomes is mixed. The objective was to examine sex differences in 1-year major adverse cardiac outcomes for the national population of patients undergoing PCI at Veterans Health Administration (VA) cardiac catheterization laboratories.

Methods: All Veterans undergoing PCI at VA hospitals between October 1, 2007 and September 30, 2013 (N = 64,757; Women = 1,040) were included. Cox proportional hazards models compared 1-year postprocedural outcomes [rehospitalization for myocardial infarction (MI), all-cause mortality, and major adverse cardiovascular events (MACE)] by sex.

Results: Women Veterans undergoing PCI were more likely to be younger, black, obese, and have chronic depression and less likely to have common cardiovascular risk factors and to have had prior cardiac events than Veteran men. One-year rates for women versus men were 2.1% and 2.5% for rehospitalization (p-value = 0.57); 3.5% and 4.9% for mortality (p-value = 0.14), and 5.4% and 6.9% for MACE (p-value = 0.18). There were no significant sex differences in any of the outcomes in Cox proportional hazards models.

Conclusions: Despite differences in clinical risk factors at the time of PCI, women and men Veterans treated at VA cardiac catheterization laboratories experienced comparable 1-year rehospitalization for MI, mortality, and MACE post-PCI. These results demonstrated similar 1-year post-PCI outcomes for men and women in a national population of patients who have more comorbidities and mental health issues than the general population.

Introduction

Cardiovascular disease is the leading cause of death in the United States accounting for over 610,000 deaths each year.¹ Despite the belief that heart disease is a “man's disease,” a similar number of women and men die annually from heart disease in the United States.¹ However, women and men experience cardiovascular disease very differently, with sex differences in certain risk factors, clinical symptoms, presentation for evaluation, and referral for appropriate cardiovascular treatment.^2–6 Moreover, compared to men, women have lower utilization of evidence-based treatments for cardiovascular disease and lower quality of care, especially younger women.^7–11

Percutaneous coronary intervention (PCI), or coronary angioplasty, is a nonsurgical procedure for treating obstructive coronary artery disease (CAD, including unstable angina, acute MI, and multivessel CAD). Over the last few decades, advancements in PCI have reduced inhospital and long-term major adverse cardiovascular events (MACE) and mortality in patients undergoing coronary interventions.^12,13 However, evidence on sex differences in outcomes after PCI has been mixed^14–19 and chiefly relies on studies based on local or regional samples or multisite registries rather than national samples or population-based data.

The Veterans Health Administration (VA) healthcare system is the largest national healthcare delivery system in the United States, with electronic medical record data and a cardiac catheterization laboratory clinical quality program called the VA Clinical Assessment Reporting and Tracking (CART) Program,²⁰ which is capable of elucidating sex differences in PCI outcomes on a national scale. As the leading cause of hospitalization for Veterans,²¹ cardiovascular disease is a serious issue for the VA. Veterans are at higher risk of heart disease than the general population with their risk compounded by their increased prevalence of comorbidities linked with cardiovascular disease, including diabetes, spinal cord injury, and posttraumatic stress disorder (PTSD).^21,22

With the twofold increase in the population of women Veterans utilizing VA services in the last decade alone,²³ the VA has made a concentrated effort to identify gender disparities in quality of care. In the VA, both women and men patients use any outpatient VA and primary care services comparably, but women Veterans tend to use more services than men: in FY12, women Veterans were more likely to be frequent users (six of more) and heavy users (12 or more) of outpatient VA care across all age groups, were more likely to be heavy users of primary care (six or more primary care visits), were more likely to use mental health or substance use disorder services, and were twice as likely to have utilized at least one day of non-VA (fee) medical care than men Veterans.^23,24 Prior research has shown that women Veterans have more risk factors for cardiovascular disease, including higher low-density lipoprotein (LDL) cholesterol levels than Veteran men after adjustments for age.²⁵ They also have higher rates for military sexual trauma, which is associated with a higher risk for PTSD, depression, and anxiety.²⁶ In terms of treatment, a prior study demonstrated that women Veterans undergoing diagnostic cardiac catheterization at the VA had lower rates of obstructive CAD than men Veterans and had lower 1-year mortality and all-cause rehospitalization rates.²⁷ However, whether women with obstructive CAD after PCI have similar outcomes to men with obstructive CAD is unknown.

The aim of this study was to examine sex differences in 1-year major adverse cardiac outcomes in a national population of patients undergoing PCI at VA facilities by assessing sex differences in 1-year postprocedural outcomes, including rehospitalization for myocardial infarction (MI), all-cause mortality, and MACE. This research advances the current literature by (1) capturing a population of all PCI patients in a national medical care system; (2) including patients with significant comorbidities; and (3) including a younger female population presenting with cardiovascular risks, a population that has been historically understudied due to limited inclusion in research.³ Because women Veterans are at increased risk and utilize VA services differently than their male counterparts,^23,24 we hypothesized that men and women Veterans had different outcomes after PCI.

Materials and Methods

Study design and setting

This is a retrospective examination of all Veterans nationally undergoing PCI at the VA cardiac catheterization laboratories between October 1, 2007 and September 30, 2013. The study utilizes the VA CART Program, a national clinical quality program for all VA cardiac catheterization laboratories launched in 2005 to support clinical care, promote the quality of care, and advance knowledge through research.²⁰ A key feature of the CART Program is a clinical software application designed to collect standardized data on all coronary angiograms and PCI completed in all VA cardiac catheterization laboratories nationwide. The software is embedded in the VA electronic medical record and allows providers to enter patient and procedural information (preprocedure assessment, coronary angiography, and PCI) as part of routine clinical workflow. We evaluated all Veterans undergoing PCI at the VA: the first PCI for all patients aged 18 and older undergoing treatment at any of the 77 VA catheterization laboratories with follow-up information was included.

Data and measures

The CART software was designed using standardized definitions which conform to the definitions and standards of the American College of Cardiology's National Cardiovascular Data Registry (ACC-NCDR) and incorporates features such as pull-down menus and automated clinical report generation to ensure uniformity of data entry by different providers and in different cardiac catheterization laboratories.²⁸ CART clinical data are combined with administrative VA data sources (National Care Patient Data, Pharmacy Benefits Management Database, the Vital Status File, VA Medicare Database, the Decision Support Network, the Corporate Data Warehouse, and the Planning Systems Support Group) to create a longitudinal data repository, which supports the quality assessment and quality improvement mission of the CART Program. CART captures mortality from the VA Vital Status database that pulls from VA administrative data, VA Beneficiary Identification Records Locator Subsystem (BIRLS), Medicare, and Social Security Administration death files. Biological sex was identified from the CART database and VA administrative files.²⁷

Outcomes: We assessed three outcomes at 1 year post-PCI: (1) rehospitalization for MI, (2) all-cause mortality, and (3) MACE (rehospitalization for MI, stroke, or death). We excluded MI codes occurring within the 14-day period after the PCI based on prior work demonstrating that these reflect the index presentation and coronary procedure, rather than representing a de novo MI.²⁹ Outcomes were assessed through from October 1, 2007 to September 30, 2013. We identified ICD-9 codes for rehospitalization for MI and stroke from the VA administrative data file (available upon request). All-cause mortality was determined by the VA Vital Status file.

Patient characteristics: Patient demographics (including age, race, and ethnicity), clinical risk factors, and comorbidities are presented in Table 1 and were determined from the VA electronic medical record using standard definitions.²⁸ We defined obesity as a body–mass index (BMI) ≥30. Framingham 10-year cardiovascular risk scores were calculated using age, sex, renal failure, diabetes, history of smoking, and average blood pressure and cholesterol data over the year before PCI (hypertension and hyperlipidemia were used if blood pressure or cholesterol was missing),³⁰ and were categorized as low: <10%, intermediate: 10%–20%, or high: >20%.^31,32 Because we cross-referenced CART clinical data with other VA administrative data, most variables were not missing. Race and cholesterol were the exceptions (both missing approximately 5%), and we imputed those missing values using regression methods in the SAS procedure PROC MI.³³

Table 1.

Characteristics of Veteran Population Who Underwent Percutaneous Coronary Intervention and Bivariate Comparisons by Sex

Variable	All Veterans, N and median/% (N = 64,757)	Women Veterans, N and median/% (N = 1,040)	Men Veterans, N and median/% (N = 63,717)	p-value
Demographics
Age [Median (IQR)]	64.2 (60.0–70.5)	60.4 (54.7–65.8)	64.2 (60.1–70.6)	<0.0001
Race				<0.0001
White	51,915 (80.2%)	702 (67.5%)	51,213 (80.4%)
Black	7,776 (12.0%)	240 (23.1%)	7,536 (11.8%)
Other	5,066 (7.8%)	98 (9.4%)	4,968 (7.8%)
Ethnicity: Hispanic	4,435 (6.8%)	63 (6.1%)	4,372 (6.9%)	0.31
Risk factors and Comorbidities
Tobacco use	41,345 (63.8%)	644 (61.9%)	40,701 (63.9%)	0.19
Obese (vs. nonobese)	31,079 (48.0%)	542 (52.1%)	30,537 (47.9%)	0.0073
Chronic Depression	23,267 (35.9%)	553 (53.2%)	22,714 (35.6%)	<0.0001
Hypertension	58,720 (90.7%)	913 (87.8%)	57,807 (90.7%)	0.0012
BPS [Median (IQR)]	134.3 (125.4–143.5)	134.6 (125.0–144.4)	134.3 (125.4–143.5)	0.72
BPD [Median (IQR)]	75.7 (69.7–82.1)	74.4 (68.4–80.9)	75.8 (69.7–82.1)	<0.0001
Hyperlipidemia	58,473 (90.3%)	898 (86.3%)	57,575 (90.4%)	<0.0001
Cholesterol [Median (IQR)]	162.7 (139.0–192.0)	186.0 (157.3–221.0)	162.3 (139.0–191.7)	<0.0001
LDL [Median (IQR)]	91.0 (71.3–115.7)	105.5 (82.3–134.0)	90.6 (71.0–115.3)	<0.0001
HDL (Median (IQR))	37.0 (31.3–44.0)	43.3 (36.7–52.6)	37.0 (31.3–44.0)	<0.0001
Diabetes	31,741 (49.0%)	495 (47.6%)	31,246 (49.0%)	0.36
Congestive heart failure	16,172 (25.0%)	220 (21.2%)	15,952 (25.0%)	0.0041
Chronic obstructive pulmonary disease	15,542 (24.0%)	228 (21.9%)	15,314 (24.0%)	0.11
Cerebrovascular disease	12,481 (19.3%)	207 (19.9%)	12,274 (19.3%)	0.6
Peripheral arterial disease	15,477 (23.9%)	204 (19.6%)	15,273 (24.0%)	0.0011
Framingham risk				<0.0001
High	18,410 (28.4%)	82 (7.9%)	18,328 (28.8%)
Medium	36,057 (55.7%)	549 (52.8%)	35,508 (55.7%)
Low	10,290 (15.9%)	409 (39.3%)	9,881 (15.5%)
Prior MI	24,641 (38.1%)	344 (33.1%)	24,297 (38.1%)	0.0009
Prior CABG	18,415 (28.4%)	187 (18.0%)	18,228 (28.6%)	<0.0001
Indication for PCI				0.0004
ACSa	32,624 (50.4%)	583 (56.1%)	32,041 (50.3%)
Stable angina	24,759 (38.2%)	365 (35.1%)	24,394 (38.3%)
Other/unknown	7,374 (11.4%)	92 (8.8%)	7,282 (11.4%)

Comparisons were conducted using chi-square tests for categorical variables and Mann–Whitney Wilcoxon nonparametric tests for continuous variables.

BPS, blood pressure systolic; BPD, blood pressure diastolic; LDL, low-density lipoprotein; HDL, high-density lipoprotein; MI, myocardial infarction; CABG, coronary artery bypass grafting; PCI, percutaneous coronary intervention; ACS, acute coronary syndrome.

^aACS Includes STEMI (ST segment elevation myocardial infarction), NSTEMI (Non ST segment elevation myocardial infarction), and unstable angina.

Statistical analyses

In descriptive analyses, we compared demographic and clinical characteristics among PCI patients by sex. We conducted comparisons using chi-square tests for categorical variables and Mann–Whitney Wilcoxon nonparametric tests for continuous variables. We compared the unadjusted rates of each outcome of interest using estimated 1-year cumulative incidence, cumulative incidence plots, and Gray's test,³⁴ accounting for the censoring of outcomes for some patients, as well as for death as a competing risk for MI.

To compare postprocedural outcomes by sex, we constructed a series of Cox proportional hazards models with sex as the primary independent predictor (male as reference group) and a robust estimator of the covariance matrix to account for clustering by hospital.³⁵ The null model was used to assess the unadjusted association between sex and the outcomes. To determine the impact of risk factors on this relationship, covariates were added sequentially to the previous model in the following order: (model 1) age, (model 2) Framingham risk category (low, medium, or high), (model 3) race (white, black, or other), and (model 4) other cardiac risk factors, including congestive heart failure (CHF), cerebrovascular disease, peripheral arterial disease (PAD), chronic obstructive pulmonary disease (COPD), obese (vs. not obese), and indication for PCI at presentation [(1) acute coronary syndrome (ST segment elevation myocardial infarction (STEMI), Non ST segment elevation myocardial infarction (NSTEMI), and unstable angina, (2) stable angina, or (3) other/unknown]. Due to low event rates among Hispanic women, ethnicity was not included in the models.

All tests for statistical significance were two tailed, and p-values <0.05 were considered statistically significant. The statistical analyses were performed by the CART Coordinating Center at the Denver VA Medical Center using SAS version 9.4 (SAS Institute, Inc., Cary, NC) and R 3.2.2 “cmprsk” package.^34,36 This study was approved by the Colorado Multiple Institutional Review Board and the VA Greater Los Angeles Institutional Review Board.

Results

The study population consisted of 64,757 Veterans who underwent PCI during the study period, including 1,040 women (1.6%). Patient characteristics at time of procedure are shown in Table 1 by sex. Female Veterans were younger than male Veterans (median age 60 vs. 64 years, respectively; p < 0.0001), more likely to be black (23% vs. 12%; p < 0.0001), obese (52% vs. 48%; p < 0.007), and have chronic depression (53% vs. 36%; p < 0.0001). Female Veterans were less likely to have common cardiovascular risk factors such as hypertension (88% vs. 91%; p = 0.0012) and hyperlipidemia (86% vs. 91%; p < 0.0001). For cholesterol, females had higher median overall cholesterol (186.0 vs. 162.3; p < 0.0001) and higher median LDL (105.5 vs. 90.6; p < 0.0001), but they also had higher median high-density lipoprotein (HDL 43.3 vs. 37.0; p < 0.0001) than the males. They also had significantly lower Framingham risk scores: only 8% were high risk compared to 29% of men (p < 0.0001). Furthermore, female Veterans were less likely to have cardiovascular disease such as CHF (21% vs. 25%; p = 0.0041), PAD (20% vs. 24%; p = 0.0011), prior MI (33% vs. 38%; p = 0.0009), and prior coronary artery bypass grafting (CABG; 18% vs. 29%; p < 0.0001) compared to male Veterans. For PCI indication, female Veterans were more likely to present with ACS, whereas male Veterans were more likely to present with stable angina and unknown/other reasons. There were no significant sex differences in tobacco use and diabetes.

One-year post-PCI outcomes in the overall population are shown in Table 2. Rates for rehospitalization for MI were 2.1% for females and 2.5% for males (p-value = 0.57). Mortality was 3.5% for females and 4.9% for males (p-value = 0.14), and MACE was 5.4% for females and 6.9% for males (p-value = 0.18). There were no significant sex differences in the base Cox proportional hazards models (model 1, sex and age only; results not shown) for any of the outcomes, and sex remained nonsignificant in all adjusted Cox proportional hazards models (model 4 shown on Table 3).

Table 2.

Estimated 1-Year Cumulative Incidence Rates for Rehospitalization, Mortality, and Major Adverse Cardiovascular Events by Sex

Event within 1 year	Women Veterans (%)	Men Veterans (%)
Rehospitalization	2.09 (1.17, 2.97)	2.45 (2.29, 2.54)
Mortality	3.49 (2.33, 4.64)	4.91 (4.73, 5.08)
MACE	5.38 (4.00, 6.75)	6.86 (6.67, 7.06)

MACE, major adverse cardiovascular events.

Table 3.

Sex Differences Using Cox Proportional Hazards Models for 1-Year Rehospitalization, Mortality, and Major Adverse Cardiovascular Events Controlling for Baseline Covariates

1-year outcome	Hazard ratio and CI Women versus men Veterans	p-value
Rehospitalization	0.97 (0.55, 1.70)	0.9074
Mortality	0.89 (0.62, 1.29)	0.5385
MACE	0.94 (0.70, 1.26)	0.6624

All models adjust for age, Framingham risk category, race, and other cardiac risk factors (including congestive heart failure, cerebrovascular disease, peripheral artery disease, and chronic obstructive pulmonary disease), obesity, and acuity of presentation (acute coronary syndrome, stable angina, or other/unknown).

Discussion

The objective of this study was to examine sex differences in 1-year major adverse cardiovascular outcomes, including mortality for the national population of patients undergoing PCI at VA cardiac catheterization laboratories. We found no significant sex differences in 1-year post-PCI outcomes for this national population: women and men treated at VA cardiac catheterization laboratories experienced comparable rates of rehospitalization for MI, mortality, and MACE one year post-PCI. There were no sex differences even before adjusting for Framingham risk, race, obesity, other cardiovascular diagnoses (CHF, CAD, PAD, and COPD), and indication for PCI indicating that although the men tended to be sicker and at higher risk on most cardiovascular indicators at presentation for PCI than women, both groups still experienced similar outcomes. Our results are consistent with the prior studies showing that men and women experience similar 1-year post-PCI mortality,^16,37–39 MI,³⁷ and MACE;³⁷ however, unlike the prior literature, we found no significant differences in outcomes by sex even before adjusting for sex differences in risk.

This research is one of the first studies of sex differences and PCI on a national population of patients in a healthcare system, including 1-year mortality, MI, and MACE. Unlike prior studies, where women were older and estrogen protection over the years may have been a factor in delaying the onset of cardiovascular disease,^16,37–42 the women in our study were younger on average than the men. Yet, we still found no significant sex differences. Consistent with prior research,^{16,37–40,42} women were more likely to have some comorbidities (i.e., depression and obesity) than the men, yet lower rates of prior cardiac events or traditional cardiac risk factors. Veterans have on average two more comorbidities than the general population^43,44 and have higher rates of depression and other mental health conditions.⁴⁵ Therefore, the Veteran population undergoing PCI was considerably sicker than cohorts in prior studies, yet comparable outcomes by sex remained.

Much of the inconsistency in the literature on sex differences in outcomes after PCI has focused on short-term outcomes showing women at increased risk for some complications^{14–19,37,46} and similar outcomes for others.^{14,16,38,40,47} Perhaps contemporary improvements in PCI^12,13 have alleviated previously seen gender differences in short-term MACE and mortality outcomes post-PCI. At the VA specifically, there may be more equitable use of bleeding avoidance strategies and other therapies, which were previously shown to be used less often with female patients in other cohorts,¹⁵ thus leading to more equitable outcomes. Another possibility is that women who survive the early complications of PCI have a survival advantage in the long term. Given that two studies with even longer term outcomes (30–36 months) found similar risk for MI⁴⁸ and actual lower risk of mortality for women,^40,48 future research should focus on the full spectrum of the cardiovascular healing process post-PCI that spans from inhospital complications through multiyear follow-up.

Our findings support comparable treatment outcomes for women and men Veterans post-PCI; however, our data were limited to those patients seen in the cardiac laboratory for PCI and cannot address any sex/gender-related referral bias. Because past research has demonstrated sex differences in diagnosis and referral for noninvasive testing in the general population,³ additional research is needed to evaluate referral patterns and timing to the cardiac procedures. Examinations of these patterns over time can also examine whether delays in presentation to emergency departments have changed for women and men with increased public education and awareness campaigns such as the American Heart Association's Go Red for Women.⁴⁹ As many cardiac emergencies are treated at the closest emergency room, which may or may not be VA, the focus on VA cardiac catheterization laboratories could have led to underreporting of events that were treated outside the VA and not recorded in the VA electronic medical record. Even though the CART data capture all procedures done outside of the VA that are paid for by the VA (fee basis files), future work should focus on understanding dual treatment both inside and outside the VA healthcare system. Finally, even though our analyses combined multiple years of national VA data and included over 1,000 women Veterans, the number of women Veterans undergoing PCI and experiencing postprocedural outcomes compared to men was small. This sample size restricted the number of covariates and interaction terms we could include in the final multivariable risk models, and we may have been underpowered to detect differences in the outcomes examined, particularly rehospitalization for MI. However, our estimates suggest that there are no clinically meaningful differences in these outcomes by sex. As the population of women Veterans continues to increase, there is an opportunity to continue to monitor sex differences in PCI outcomes.

Conclusions

The results are consistent with work from United States and international registries while expanding the broader literature to demonstrate similar 1-year post-PCI mortality, MI, and MACE rates for men and women in a national population of patients who are sicker and have more mental health issues than the general population. Despite the presence of these factors that usually reduce access to definitive care, we found no sex differences in outcomes: women and men Veterans experience similar post-PCI 1-year major adverse cardiac outcomes. Although the number of women Veterans at the VA has more than doubled in the last decade,²³ women are still a numerical minority at the VA. Given that the populations of men and women Veteran are different, especially in terms of age, race, cardiovascular risk factors, and indication for PCI, future research should examine interactive effects on sex and these factors in this growing population.

Acknowledgments

This research was supported by the Enhancing Mental and Physical health of Women through Engagement and Retention (EMPOWER) VA QUERI program (QUE 15-272; PI: Hamilton) and the VA Women's Health Research Network (SDR 10-012; PIs: Yano, Frayne and Hamilton). Dr. Joseph was funded by a VA Advanced Women's Health Fellowship, VA Greater Los Angeles Healthcare System (TWH 65-016). Dr. Daugherty was supported by Award Number K08HL103776 and RO1HL133343 from the National Heart, Lung and Blood Institute. Dr. Yano's effort was funded through a VA HSR&D Senior Research Career Scientist Award (RCS 05-195)

The views expressed in the manuscript are solely those of the authors, and do not necessarily represent the views of the U.S. Department of Veterans Affairs, the United States government or the National Heart, Lung and Blood Institute.

Author Disclosure Statement

All authors report that no competing financial interests exist.