Overview

STEMI is caused by the complete occlusion of an epicardial artery, leading to possible transmural infarction of the heart muscle. Treatment for patients with STEMI consists of reperfusion therapy (either pharmacological or catheter-based) to restore blood flow promptly in the occluded epicardial infarct-related artery. Pharmacological therapy consists of fibrinolysis or facilitated antithrombotic medications.²⁹ In general, patients with STEMI are not treated with CABG (unless emergent from PCI complications) but do receive optimal medical therapy in addition to treatment directed at removing the clot. Studies assessing the effectiveness of immediate PCI compared with fibrinolytics or immediate PCI compared with conservative/supportive therapy were evaluated for KQ 1.

Key Points

• Description of included studies: Seven studies (6 good quality, 1 fair) evaluated PCI with or without supportive pharmacologic therapy versus fibrinolysis or other routine medical care for women with STEMI and included a total of 4,527 patients, of which 1,174 (26%) were women.
• Effectiveness of interventions: A meta-analysis of five studies (all good quality) reporting 30-day composite outcomes (primarily death/MI/stroke) showed that PCI was better than fibrinolysis in women (OR 0.50; 95% CI, 0.36 to 0.72) and men (OR 0.54; CI, 0.42 to 0.70). Strength of evidence favoring PCI over fibrinolysis was high at 30-day followup. However, there was insufficient evidence for assessing outcomes at 1 year. These findings also are limited in that all the studies were conducted with either balloon angioplasty or bare-metal stents. The current use of drug-eluting stents may lead to different practice patterns and, potentially, increase the effectiveness of PCI. Individual outcomes by sex were rarely reported for heart failure, repeat hospitalization, length of hospital stay, angina relief, quality of life, or cognitive effects.
• Modifiers of effectiveness: Two studies (1 good quality, 1 fair) reported subgroup analyses of demographic or clinical factors in women and included a total of 395 patients, of which 167 (32%) were women. Both studies assessed the influence of age on in-hospital mortality or composite clinical outcomes (death/heart failure/MI/stroke) and showed no age-related differences in PCI compared with optimal medical therapy. Therefore, there was insufficient evidence of the comparative effectiveness of treatment strategies among subgroups of women with STEMI, which precludes any meaningful conclusions.
• Safety concerns: Two good-quality studies reported safety concerns in women with STEMI and included a total of 1,532 patients, of which 367 (24%) were women. One study reported a lower nadir hematocrit in women receiving PCI versus fibrinolysis but no statistically significant differences in the requirement for blood transfusion. Another study reported the proportion of women with intracranial hemorrhage who received PCI versus accelerated tissue plasminogen activator (t-PA) (0% vs. 4.1%). No studies systematically collected radiation exposure, contrast reactions, access site complications, or stent thrombosis in women with STEMI undergoing PCI. Strength of evidence for safety concerns in STEMI studies was insufficient.

Detailed Synthesis

We identified seven studies^36–42 that evaluated PCI with or without supportive pharmacologic therapy versus fibrinolysis or other routine medical care for women with STEMI. Of these seven studies, six were good quality, and one was fair quality. Table 3 presents a general description of these seven studies, including the study name, author, year, and related articles (i.e., study design and secondary papers); treatment comparisons evaluated; study population; and overall quality rating. Table 4 summarizes the women-specific outcomes (composite and individual) reported in these studies. Appendix F contains summary tables with sex-specific clinical outcomes for all followup time points.

Table 3

KQ 1: Study characteristics of RCTs evaluating women with STEMI.

Table 4

KQ 1: Outcomes reported in RCTs evaluating women with STEMI.

KQ 1a. Effectiveness of Interventions

A meta-analysis was performed on studies with similar composite outcomes measured at similar time points. This meta-analysis was divided into followup intervals of short term (≤30 days) and long term (≥1 year). The SHOCK study,³⁸ evaluating early revascularization versus medical stabilization, did not report 1-year data by sex, except for noting the lack of a treatment-by-sex interaction, and therefore was not included in the meta-analysis. Similarly, in the Minai study⁴¹ evaluating PCI versus no PCI, 3-year data reported no sex effect in a multivariate analysis; however, since the data were not reported by sex, this study also was excluded from the meta-analysis.

Short-Term Followup Studies

Five studies—CARESS-in-AMI,³⁶ DANAMI-2,³⁷ Dobrzycki,⁴⁰ GUSTO II-B,³⁹ and PAMI⁴²—were included in the meta-analysis based on comparable composite outcomes (primarily death/MI/stroke) and followup time points of 30 days or in-hospital. The published results from Dobrzycki et al. were inverted to change the reference arm to fibrinolysis. The PAMI study event rates by treatment group and sex were converted into odds ratios. Table 5 presents the outcomes, odds ratios, and confidence intervals for the meta-analysis.

Table 5

Sex results for STEMI on composite outcomes (short-term).

Forest plots for the random-effects model are shown in Figure 3. The summary odds ratio in women was 0.50 (95% CI, 0.36 to 0.72) and in men was 0.54 (CI, 0.42 to 0.70). The test for heterogeneity was nonsignificant. These results show that PCI was better than fibrinolysis in reducing death/MI/stroke in both sexes (p=0.0001 women, p<0.00001 men) at 30 days.

Figure 3

STEMI short-term (30 days) PCI versus fibrinolysis.

Intermediate-Term Followup Studies

Two studies—DANAMI-2⁴⁴ and Dobrzycki⁴⁰—reported sex-specific clinical outcomes at 1 year. Table 6 presents the outcomes, odds ratios, and confidence intervals for PCI compared with fibrinolysis by sex. Due to the heterogeneous clinical outcomes, a meta-analysis was not conducted.

Table 6

Sex results for STEMI on clinical outcomes (intermediate-term).

PCI Versus Fibrinolysis in High-Risk Groups

We identified two studies that evaluated a PCI or CABG strategy versus conservative/supportive medical therapy in high-risk groups.^38,41 These studies were not included in the meta-analysis since the actual results by sex were not reported in the manuscript; instead both papers report the results of a multivariate analysis with sex as a covariate in the model. The SHOCK study³⁸ was considered good quality and evaluated patients with cardiogenic shock and STEMI with a strategy of PCI or CABG within 6 hours versus initial medical stabilization that included fibrinolysis or insertion of an intra-aortic balloon pump (IABP). This study found that the early revascularization strategy was associated with a lower relative risk of death when compared with medical stabilization (risk ratio 0.72; 95% CI, 0.54 to 0.95). Analysis by sex did not identify any significant interaction by treatment arm. The study by Minai et al.⁴¹ was considered fair quality and evaluated PCI versus routine medical therapy without reperfusion in patients 80 years of age and older. There was no difference in the number of patients with the composite outcome of death/heart failure/repeat MI/stroke at 3 years between the treatment arms. No analysis by sex was done; however, in a multivariate analysis to evaluate factors associated with the composite outcome, sex was not found to be significantly associated with the outcome in the overall study population. This study was limited by the small sample size of 120 patients enrolled.

KQ 1b. Modifiers of Effectiveness

Two studies^41,42 evaluating women with STEMI assessed the characteristics of interest and included a total of 515 patients, of which 167 (32%) were women. The PAMI study⁴² was considered good quality and evaluated patients randomized to PCI versus fibrinolysis with t-PA. No difference was found in in-hospital mortality among women receiving PCI versus t-PA who were under 65 years of age (0% vs. 4%; p=0.42) nor among women 65 and older (5.9% vs. 21.9%; p=0.58).

The study by Minai et al.⁴¹ evaluated PCI versus routine medical therapy without reperfusion in patients 80 years of age and older. The results are noted above. Appendix G contains a summary table with study data related to modifiers of effectiveness (subgroup analyses).

KQ 1c. Safety Concerns

Two good-quality studies reported safety concerns in women with STEMI and included a total of 1,532 patients, of which 367 (24%) were women. In the PAMI study⁴² evaluating PCI versus t-PA in STEMI patients, the mean nadir hematocrit in women with PCI was 30 ± 5 percent versus 33 ± 5 percent in women with t-PA (p=0.0002). However, there was no statistically significant difference in the requirement for blood transfusion in women with PCI versus t-PA (18% vs. 8.8%; p=0.16). In the GUSTO II-B study³⁹ the proportion of women with intracranial hemorrhage was reported in women who received PCI versus accelerated t-PA (0% vs. 4.1%), but statistical analysis for this comparison was not done. Appendix H contains a summary table with study data related to safety concerns (harms).

Overview

Unstable angina is caused by reversible ischemia of the epicardial arteries, whereas NSTEMI is caused by the partial obstruction of the epicardial arteries and results in myocardial tissue damage. Patients with UA/NSTEMI are not candidates for immediate pharmacological reperfusion (i.e., fibrinolysis). The optimal management of UA/NSTEMI has the twin goals of immediate relief of ischemia and the prevention of serious adverse outcomes (i.e., death or MI). Optimal management is best accomplished with aggressive medical therapy that includes anti-ischemic therapy, antithrombotic therapy, ongoing risk stratification, and in some cases the use of invasive procedures.

Following initiation of aggressive medical therapy, two treatment pathways have emerged for treating patients without ST-segment elevation.²⁸ An “initial conservative strategy” calls for proceeding with an invasive evaluation only for those patients whose medical therapy fails (refractory angina or angina at rest or with minimal activity despite vigorous medical therapy) or in whom objective evidence of ischemia (dynamic electrocardiographic changes, high-risk stress test) is identified. The early invasive strategy triages patients to undergo an invasive diagnostic evaluation without first having a noninvasive stress test or having medical treatment fail. Patients treated with an early invasive strategy generally will undergo coronary angiography within 4 to 24 hours of admission; however, these patients also are treated with the usual UA/NSTEMI medications, including appropriate anti-ischemic, antiplatelet, and anticoagulant therapy. Therefore, studies assessing the effectiveness of early invasive versus initial conservative therapy were evaluated for KQ 2.

Key Points

• Description of included studies: Seven studies (6 good quality, 1 fair) compared early invasive therapy (PCI or CABG) with initial conservative therapy for women with UA/NSTEMI and included a total of 17,930 patients, of which 6,084 (34%) were women.
• Effectiveness of interventions: A meta-analysis of two good-quality studies reporting 6-month composite outcomes (death/MI) suggested a benefit of early invasive compared with initial conservative therapy in women (OR 0.77; 95% CI, 0.28 to 2.12), but early invasive therapy was superior to initial conservative therapy in men (OR 0.65; CI, 0.52 to 0.82; p=0.0002). At 1 year, a meta-analysis of five good-quality studies showed that composite outcomes (primarily death or MI) suggested a benefit for women who received early invasive therapy (OR 0.78; CI, 0.54 to 1.12) as well as for men (OR 0.88; CI, 0.64 to 1.20); however this benefit was not statistically significant. Finally, a meta-analysis of two good-quality studies with 5-year followup comparing early invasive and initial conservative therapy for the composite outcome of death or MI did not reach statistical significance in either sex. The summary odds ratio in women was 1.05 (CI, 0.81 to 1.35) and in men was 0.91 (CI, 0.53 to 1.56). The long-term analysis is limited by the low number of studies. Strength of evidence favoring an early invasive approach was low for women and high for men at 6-month followup; low for women and men at 1-year followup; and insufficient for women and low for men at 5-year followup. Similar to the STEMI studies, individual outcomes by sex were rarely reported for heart failure, repeat hospitalization, length of hospital stay, quality of life, or cognitive effects.
• Modifiers of effectiveness: Two good-quality studies comparing early invasive treatment with PCI with initial conservative treatment reported a subgroup analysis by risk stratification and included a total of 4,030 patients, of which 1,439 (36%) were women. These studies revealed conflicting results—one showed no difference in treatment outcomes in the intermediate- and high-risk groups; the other showed a higher event rate in women in the groups with moderate-to-high risk for thrombolysis in myocardial infarction (TIMI). Strength of evidence for modifiers of effectiveness for early invasive versus initial conservative treatments was insufficient.
• Safety concerns: One good-quality study (2,220 total patients, 757 women, 34% women) reported the harms associated with treatment of UA/NSTEMI by sex group but not the rates of events by treatment group. Bleeding in women undergoing PTCA was higher compared with men (adjusted OR 3.6; 95% CI, 1.6 to 8.3).⁵¹ However, bleeding related to CABG was similar in women and men with rates of 12.6 and 15 percent respectively. No studies systematically reported radiation exposure, contrast reactions, access site complications, stent thrombosis or infection, in women with UA/NSTEMI comparing initial conservative with early invasive therapy. Strength of evidence for safety concerns in these populations was insufficient.

Detailed Synthesis

We identified seven studies^22,52–57 that evaluated the effect of early invasive therapy compared with initial conservative therapy for UA/NSTEMI and reported results by sex. Of these seven studies, six were good quality, and one was fair. Table 7 presents a general description of these seven studies, including the study name, author, year, and related articles (i.e., study design and secondary papers); treatment comparisons evaluated; study population; and overall quality rating. Table 8 summarizes the women-specific outcomes (composite and individual) reported in these studies. Appendix F contains summary tables with sex-specific clinical outcomes for all followup time points.

Table 7

KQ 2: Study characteristics of RCTs evaluating women with UA/NSTEMI.

Table 8

Outcomes reported in RCTs evaluating women with UA/NSTEMI.

KQ 2a. Effectiveness of Interventions

A meta-analysis was performed on studies with similar composite outcomes measured at similar time points. This meta-analysis was divided into followup intervals of short term (6 months), intermediate term (1 year) and long term (5 years).

Short-Term Followup Studies

Two studies reporting 6-month outcomes—FRISC II⁶⁰ and TACTICS TIMI-18⁵¹—were included in the meta-analysis. The TIMI III-B study⁵⁷ (good quality) reported a shorter followup time point of 6 weeks and therefore was not included in this meta-analysis. In TIMI III-B, the proportion of women with the composite outcome of death/MI at 6 weeks was 6.1 percent in the early invasive arm and 8.9 percent in the initial conservative arm (p=0.24). The proportion of men with the same composite outcome was 7.8 percent and 7.3 percent, respectively (p=0.73). The RITA-2 study⁵⁴ (fair quality) reported outcomes for angina grade 2+ or exercise time and likewise was not included in the meta-analysis. In RITA-2, the proportion of women with angina graded 2 or higher at 6 months was 22.8 percent for the early invasive arm, and 39.8 percent for the initial conservative arm. In men, the proportion with angina graded 2 or higher was 20.5 percent and 31.4 percent, respectively. For the TACTICS TIMI-18 study, we used the adjusted odds ratio for the composite outcome of death/MI. Table 9 presents the outcomes, odds ratios, and confidence intervals for the meta-analysis.

Table 9

Sex results for UA/NSTEMI on composite outcomes (short-term).

Forest plots for the random-effects model are shown in Figure 4. The summary odds ratio in women was 0.77 (95% CI, 0.28 to 2.12) and in men was 0.65 (CI, 0.52 to 0.82). The test for heterogeneity was significant in women (p=0.01), but it was nonsignificant in men. These results for short-term outcomes suggested a non–statistically significant benefit of early invasive compared with initial conservative therapy in women but demonstrated that early invasive was superior to initial conservative therapy in men (p=0.0002). The two trials resulted in conflicting conclusions in women versus men, despite having similar results for the overall population.

Figure 4

UA/NSTEMI short-term (6 months) early invasive versus initial conservative.

Intermediate-Term Followup Studies

Five studies with 1-year data—FRISC II,⁵⁹ GUSTO IV-ACS,⁵³ ICTUS,⁵⁶ RITA-3,²² and TIMI III-B⁶⁴—were included in the meta-analysis. For the RITA-3 study, the adjusted odds ratio for the composite outcome of death/MI was selected for this analysis. For the ICTUS and TIMI III-B studies, event rates were converted to odds ratios. Table 10 presents the outcomes, odds ratios, and confidence intervals for the meta-analysis.

Table 10

Sex results for UA/NSTEMI on composite outcomes (intermediate-term).

Forest plots for the random-effects model are shown in Figure 5. The summary odds ratio in women was 0.78 (95% CI, 0.54 to 1.12) and in men was 0.88 (CI, 0.64 to 1.20). The test for heterogeneity was significant in women (p=0.01) and men (p=0.02). These results show trends favoring early invasive therapy in 1-year outcomes although these benefits were not statistically significant in either women or men.

Figure 5

UA/NSTEMI intermediate-term (1 year) early invasive versus initial conservative.

Long-Term Followup Studies

Two studies with 5-year, long-term data were included in the analysis: FRISC II⁵⁸ and ICTUS.⁶¹ Table 11 presents the outcomes, odds ratios, and confidence intervals for the meta-analysis.

Table 11

Sex results for UA/NSTEMI on composite outcomes (long-term).

Forest plots for the random-effects model are shown in Figure 6. The summary odds ratio in women was 1.05 (95% CI, 0.81 to 1.35) and in men was 0.91 (CI, 0.53 to 1.56). The test for heterogeneity was significant in men (p=0.005) but not in women. Although these results demonstrate a slight trend favoring initial conservative therapy in women, given the small suggested benefit at 5 years, the wide confidence interval crossing 1, and the trend favoring early invasive therapy suggested at earlier time points and across time points in men — we cannot support firm conclusions. There was no evidence of a sex effect. The analysis is limited by the low number of studies.

Figure 6

UA/NSTEMI long-term (5 years) early invasive versus initial conservative.

Some of the analyses above showed heterogeneity in clinical outcomes between the FRISC II and ICTUS studies. The major heterogeneity in these studies arises from the rates and threshold of invasive treatment in the conservative arm. In the FRISC II study, more conservative strategies were used, thus leading to lower rates of invasive treatment in the conservative groups. In the ICTUS study, the selective invasive group was more liberal with the rates of invasive therapy and almost as high as the invasive arms of the other studies, thus explaining some of the potential differences in the results.

KQ 2b. Modifiers of Effectiveness

We identified two good-quality studies^22,55 examining the effect of early invasive therapy compared with initial conservative therapy in women by subgroup; these included a total of 4030 patients, of which 1439 (36%) were women. The TACTICS TIMI-18 study⁵⁵ found that there was no significant benefit to the treatment of women with intermediate-risk (3 to 4) or high-risk (5 to 7) TIMI scores on the primary composite outcome of death/MI/rehospitalization for acute coronary syndrome with early invasive therapy (OR 0.72; 95% CI, 0.45 to 1.16) compared with initial conservative therapy (OR 0.56; CI, 0.23 to 1.32).⁵¹ There also was no significant benefit of early invasive therapy on the primary composite outcome for those presenting with ST-segment changes (OR 0.66; CI, 0.38 to 1.15). However, there did seem to be a reduced risk of the primary composite outcome among women who had an elevated troponin level and who underwent early invasive treatment compared with conservative treatment (OR 0.56; CI, 0.32 to 0.97). Men with ST-segment changes and elevated troponin levels also seemed to benefit from early invasive therapy but not those in intermediate- or high-risk groups based on TIMI risk scores.⁵¹

The RITA-3 study²² also examined the effect of early invasive therapy compared with initial conservative therapy in women by subgroup based on risk, which was derived from components of the TIMI risk score and a couple other aspects of the participants’ presentation at randomization, including aspirin use and angina severity. This study, unlike the TACTICS TIMI-18 study, found a higher event rate among women in moderate- and high-risk groups who underwent early invasive therapy compared with initial conservative therapy, with event rates of 13.4 percent versus 3.4 percent for those in the moderate-risk group and 11.7 percent versus 8.2 percent for those in the higher risk group. Men with moderate and higher risk had lower event rates if they were in the early invasive arm compared with the initial conservative arm, with 5.4 percent versus 9.4 percent for those in the moderate-risk group and 10.3 percent versus 17.9 percent for those in the higher risk group. This study also examined the effect of intervention group by body mass index group and found no effect of body mass index on treatment effect in either women or men.²² Appendix G contains a summary table with study data related to modifiers of effectiveness (subgroup analyses).

KQ 2c. Safety Concerns

We identified one good-quality study⁵⁵ (2,220 total patients, 757 women, 34% women) that reported the harms associated with treatment of UA/NSTEMI by sex group but not the rates of events by treatment group. The TACTICS TIMI-18 study, comparing early invasive therapy with initial conservative therapy, found that bleeding in women undergoing PTCA was higher than in men (adjusted OR 3.6; 95% CI, 1.6 to 8.3).⁵¹ They found, however, that the bleeding related to CABG was similar in women and men with rates of 12.6 and 15 percent, respectively. Appendix H contains a summary table with study data related to safety concerns (harms).

Overview

Angina resulting from progressive narrowing of the coronary arteries is the initial manifestation of ischemic heart disease in approximately 50 percent of patients.³⁰ Most angina is a sign of significant CAD, defined angiographically as a stenosis with ≥70 percent diameter in at least one major epicardial artery segment or with ≥50 percent diameter in the left main coronary artery. However, some angina is caused by stenotic lesions of lesser diameters, which have much less prognostic significance.³⁰

Chronic stable angina is classified as pain that classically occurs with moderate to severe exertion, is milder in nature, and relieved with rest or sublingual nitroglycerin. Unstable angina (UA) is defined as angina with at least one of three features: (1) it occurs at rest or with minimal exertion, (2) it is severe and of recent onset (within the past 4 to 6 weeks), and/or (3) it occurs in a crescendo pattern (i.e., more severe, more prolonged, or more frequent than previously experienced). UA and NSTEMI have a fairly similar pathophysiology, mortality rate, and management strategy when compared with STEMI; therefore they are often grouped together as UA/NSTEMI in clinical guidelines and trial populations.

The treatment of angina has two major purposes. The first is to prevent MI and death and thereby increase the quantity of life. The second is to reduce symptoms of angina and occurrence of ischemia, which should improve the quality of life.³⁰ All patients with stable or unstable angina are candidates for optimal medical therapy and also may be candidates for PCI or CABG based on findings from coronary angiography or if symptoms persist despite optimal medical therapy.

For KQ 3, we evaluated two sets of treatment strategies:

1. Revascularization (PCI or CABG) versus optimal medical therapy in women with stable angina
2. PCI versus CABG in women with either stable or unstable angina

Strategy 1: Revascularization Versus Optimal Medical Therapy in Stable Angina

Detailed Synthesis

We identified five studies^65–69 that reported outcomes for women with stable angina. Of these five studies, all were good quality. Two studies compared PCI with optimal medical therapy,^65,66 one compared CABG with optimal medical therapy,⁶⁷ and one compared medical management with transmyocardial revascularization (TMR).⁶⁹ Table 12 presents a general description of these five studies, including the study name, author, year, and related articles (i.e., study design and secondary papers); treatment comparisons evaluated; study population; and overall quality rating. Table 13 summarizes the women-specific outcomes (composite and individual) reported in these studies. Appendix F contains summary tables with sex-specific clinical outcomes for all followup time points.

Table 12

KQ 3 Strategy 1: Study characteristics of RCTs evaluating women with stable angina (PCI/CABG vs. optimal medical therapy).

Table 13

KQ 3 Strategy 1: Outcomes reported in RCTs evaluating women with stable angina (PCI/CABG vs. optimal medical therapy).

KQ 3a. Effectiveness of Interventions

Long-Term Followup Studies

Three studies were included in a meta-analysis: COURAGE,⁶⁵ MASS II,⁶⁸ and STICH.⁶⁷ These studies had similar followup times (4 to 5 years, except a 10-year followup in the MASS II study) and comparable outcomes (composite, death). No results were available for the short- or intermediate-term outcomes. The TMR study by Allen et al.⁶⁹ was excluded since the results were reported in a different fashion (i.e., whether sex had an impact on outcome for the TMR patients) and could not be incorporated into the meta-analysis. The OAT study⁶⁶ was excluded from the meta-analysis since the study subjects were enrolled ≥7 days after an acute MI with revascularization or medical therapy for an occluded artery; the patient population was deemed to be quite different from those enrolled in the other studies. The MASS II study numbers were inverted to place PCI in the comparison group. Table 14 presents the outcomes, odds ratios, and confidence intervals for the meta-analyses.

Table 14

Sex results for stable angina on composite outcomes (long-term).

Forest plots for the random-effects model are shown in Figures 7, 8, and 9. Separate analyses were created for (a) primarily PCI strategy, (b) primarily CABG strategy, and (c) either type of revascularization strategy compared with optimal medical therapy for the purposes of creating a sensitivity analysis given the subtle differences in patient populations and revascularization methods used in those studies.

Figure 7

Stable angina, revascularization (primarily PCI) versus medical therapy.

Figure 8

Stable angina, revascularization (primarily CABG) versus medical therapy.

Figure 9

Stable angina, revascularization (combined) versus medical therapy.

The primarily PCI strategy (Figure 7) studies included COURAGE and MASS II (PCI). The summary odds ratio in women was 0.64 (95% CI, 0.47 to 0.89) and in men was 1.03 (CI, 0.79 to 1.33). The test for heterogeneity was not significant for women and men. The results showed that PCI reduced death/MI in women but not in men.

The primarily CABG strategy (Figure 8) studies included MASS II (CABG) and STICH. The summary odds ratio in women was 0.56 (CI, 0.32 to 0.96) and in men was 0.62 (CI, 0.31 to 1.24). The test for heterogeneity was not significant for women (p=0.15) but was significant for men (p=0.0004). These results show that CABG was significantly better in reducing cardiovascular events than optimal medical therapy was in women (effect size p=0.04) but not in men.

We then combined all studies utilizing both types of revascularization strategies (PCI or CABG)—COURAGE, MASS II, and STICH. Figure 9 shows the forest plot for the random-effects model using the MASS II (CABG) cohort. The summary odds ratio in women was 0.59 (CI, 0.43 to 0.81) and in men was 0.71(CI, 0.49 to 1.02). The test for heterogeneity was not significant in women, but was significant in men. These results showed that revascularization was significantly better in reducing cardiovascular events than optimal medical therapy in women. Similar results were obtained if the model was run with the MASS II (PCI) cohort (figure not shown). For men, revascularization and optimal medical therapy were not statistically different.

KQ 3b. Modifiers of Effectiveness

No studies were identified that evaluated women presenting with stable angina related to modifiers of the effectiveness of revascularization versus optimal medical therapy.

KQ 3c. Safety Concerns

No studies were identified that evaluated women presenting with stable angina related to safety concerns for revascularization versus optimal medical therapy.

Strategy 2: PCI Versus CABG in Stable/Unstable Angina

Detailed Synthesis

We identified 10 studies^68,78–86 that evaluated PCI versus CABG for women presenting with unstable angina or NSTEMI. Of these 10 studies, 8 were good quality, and 2 were fair quality. Table 15 presents a general description of these 10 studies, including the study name, author, year, and related articles (i.e., study design and secondary papers); treatment comparisons evaluated; study population; and overall quality rating. Table 16 summarizes the women-specific outcomes (composite and individual) reported in these studies. Appendix F contains summary tables with sex-specific clinical outcomes for all followup time points.

Table 15

KQ 3 Strategy 2: Study characteristics of RCTs evaluating women with stable or unstable angina (PCI vs. CABG).

Table 16

KQ 3 Strategy 2: Outcomes reported in RCTs evaluating women with stable or unstable angina (PCI vs. CABG).

KQ 3a. Effectiveness of Interventions

A meta-analysis of studies was performed on those with similar composite outcomes measured at similar time points. This meta-analysis was divided into followup intervals of short term (≤30 days), intermediate term (1 year), and long term (≥2 years).

Short-Term Followup Studies

Two studies with short-term (30-day or in-hospital) outcomes—ARTS I⁷⁸ and BARI⁷⁹—were included in the meta-analysis. For both studies, the reported event rates/percentages were converted into odds ratios. Table 17 presents the outcomes, odds ratios, and confidence intervals for the meta-analysis.

Table 17

Sex results for stable/unstable angina on composite outcomes (short-term).

Forest plots for the random-effects model are shown in Figure 10. The summary odds ratio in women was 0.68 (95% CI, 0.24 to 1.93) and in men was 1.36 (CI, 0.44 to 4.24). The test for heterogeneity was nonsignificant. The findings were not statistically significant for demonstrating a benefit in PCI and CABG, although the odds ratios suggested a possible sex effect, with PCI showing more benefit in women and CABG showing more benefit in men, but the confidence intervals are too wide to support firm conclusions.

Figure 10

Stable/unstable angina short-term (30 days) PCI versus CABG.

Intermediate-Term Followup Studies

Two studies with 1-year outcomes—ARTS I⁷⁸ and CARDia⁸⁴—were included in the meta-analysis. In the SYNTAX study, ⁸³ there was no difference in the primary composite outcome (death/MI/stroke, or repeat revascularization) at 1 year between patients undergoing CABG and PCI (13.6% vs. 15.8%). No sex data by treatment in this fair-quality study were provided, but being female was a significant predictor of 1-year major adverse cardiovascular events (OR 0.50; 95% CI, 0.27 to 0.91; p=0.02 [interaction effect not reported]). The SYNTAX study was excluded from the meta-analysis since it did not report subgroup results by sex. The PRECOMBAT study⁸⁵ did not report sex-specific data at 1 year, so this study was not included in this intermediate-term followup analysis (but data reported at 2 years are included in the long-term followup analysis below). Event data from the ARTS I study were transformed into risk ratios. Table 18 presents the outcomes, odds ratios, and confidence intervals for the meta-analysis.

Table 18

Sex results for stable/unstable angina on composite outcomes (intermediate-term).

Forest plots for the random effects model are shown in Figure 11. The summary odds ratio in women was 1.30 (95% CI, 0.69 to 2.45) and in men was 1.19 (CI, 0.84 to 1.70). The test for heterogeneity was nonsignificant. These results show lower events in the CABG group for both sexes, but this benefit was not statistically significant.

Figure 11

Stable/unstable angina intermediate-term (1 year) PCI versus CABG.

Long-Term Followup Studies

To assess the long-term effect at ≥2 years, four studies were included in the meta-analysis: ARTS I (3-year),⁷⁸ BARI (5-year),⁷⁹ MASS II (10-year),⁶⁸ and PRECOMBAT (2-year).⁸⁵ Results of the ARTS I and BARI studies were transformed into risk ratios. For the BARI study, the survival rates were converted into death rates. For women, the death rates for CABG and PCI were 24 percent of 240 and 26 percent of 249, respectively. For men, survival rates by treatment group were not reported. The MASS II results were inverted to hazard ratios <1 favoring PCI and hazard ratios >1 favoring CABG. The GABI study⁸⁶ was excluded from the meta-analysis because it did not present data by sex. The GABI study randomized 359 patients (66 women) with angina CCS class II–IV, under age 75, and coronary multiple-vessel disease requiring revascularization of at least 2 major coronary vessels to either PCI or CABG. The authors report that the hazard ratio for death following PCI or CABG was not different between men and women at the 13-year followup. Table 19 presents the outcomes, odds ratios, and confidence intervals for the meta-analysis.

Table 19

Sex results for stable/unstable angina on composite outcomes (long-term).

Forest plots for the random effects model are shown in Figure 12. The summary odds ratio in women was 1.17 (95% CI, 0.90 to 1.54) and in men was 1.63 (CI, 1.20 to 2.23). The test for heterogeneity was nonsignificant. Similar to the intermediate-term outcomes, there was no definitive evidence of a sex effect. In men, CABG was significantly better than PCI (p=0.002); in women, results suggest lower events in the CABG group, but it was not statistically significant (p=0.24).

Figure 12

Stable/unstable angina long-term (≥2 years) PCI versus CABG.

KQ 3b. Modifiers of Effectiveness

We identified one good-quality study (915 total patients, 249 [27%] women) in which the long-term survival rate (7 years) in diabetic women was addressed. In the BARI study,⁷⁹ this subgroup analysis was not initially specified but was requested by the safety and monitoring board during the course of the trial on the basis of concurrent reports from another study. Survival rate at 7 years was significantly higher among diabetic patients (n=353) treated with CABG compared with those undergoing PCI (74.6% vs. 55.7%), a difference that remained significant in men (77.9% vs. 51.5%) but not in women (74.3% vs. 61.0%). Appendix G contains a summary table with study data related to modifiers of effectiveness (subgroup analyses).

KQ 3c. Safety Concerns

We identified one good-quality study⁷⁸ (1,205 total patients, 283 [23%] women) that addressed significant safety concerns and, specifically, risk of major bleeding in women (n=283). The ARTS I study was designed to compare CABG with PCI combined with stent implantation for the treatment of patients with multiple-vessel disease. Major bleeding was higher in women undergoing PCI compared with men (7.2% vs. 0.2%, p<0.001) but not among those assigned to CABG (1.4% vs. 2.8%). After adjusting for baseline characteristics, major bleeding complications remained higher among women in the PCI group (OR 29.4; 95% CI, 5.3 to 500; p=0.001 women vs. men) compared with the CABG group (OR 1.5; CI, 0.4 to 10.1, p=0.58 women vs. men). Appendix H contains a summary table with study data related to safety concerns (harms).