Heart disease is the leading cause of death for women in North America, accounting for more than 500 000 deaths per year.1 Several studies have shown differences in the outcomes of men and women after cardiac surgery.2^–4 However, most of these studies have focused on perioperative outcomes only, and the majority of research has been limited to gender-specific results after isolated coronary artery bypass graft surgery (CABG). These studies indicated that women undergoing CABG have higher early mortality and cardiac morbidity than men.4^–9 Furthermore, women undergoing CABG need more hospital resources than men in terms of blood transfusion, mechanical ventilation and length of time in intensive care unit and overall hospital stay.2

Heightened interest on the topic of cardiovascular disease in women has led to several recent publications on the subject, including a systematic review of gender-specific cardiovascular issues.10 To date, however, little is known regarding the gender-specific differences in the long-term outcomes of patients undergoing heart valve replacement surgery. Therefore, the purpose of the present study was to compare the long-term outcomes of women and men after valve replacement surgery, focusing on the long-term complications of reoperation and stroke, as well as overall survival after valve replacement.

METHODS

Patient population

We examined a cohort of 3118 adult patients who were prospectively followed after aortic valve replacement (AVR) or mitral valve replacement (MVR). All surgeries were performed at the University of Ottawa Heart Institute between 1976 and 2006. There were 1261 women and 1857 men; 2255 patients had AVR and 863 had MVR; 1177 patients underwent concurrent CABG. Patients who did not survive to hospital discharge after the operation were excluded from the study population because the focus of this study is on the long-term outcomes after surgery, and much has already been written on the subject of gender and perioperative mortality rates.6 11^–14 We also excluded patients who received prostheses that are no longer commercially available (for example, Ionescu-Shiley valve) to improve the generalisability and interpretability of the study. Patients who underwent replacement of more than one heart valve at the same operation were similarly excluded. Valve prostheses were implanted and oriented according to the manufacturer’s instructions, and their type and size were recorded for all patients.

Follow-up

After surgery, all patients were given appointments to attend a dedicated valve clinic where they were assessed by a physician 6–12 months postoperatively and thereafter on an annual basis. All patients in the cohort had at least one visit, and 73.6% had two or more visits. Additional follow-up data were collected through patient and family telephone interviews and contact with patients’ family physicians. At each visit, patients underwent a medical history focused on the occurrence of valve-related complications, a physical examination, electrocardiogram, chest radiograph, complete blood count, serum chemistries and international normalised ratio (INR) determinations when applicable. Patients received anticoagulation treatment according to guidelines in effect at the time, as previously described.15 The total follow-up of this cohort was 17 535 patient-years (mean 5.6 (4.5) years; maximum 29.1 years). The methods of the valve clinic follow-up and data collection have been reviewed and approved by the University of Ottawa Heart Institute human research ethics board, and this study complies with the Declaration of Helsinki.

Outcomes

The long-term outcomes of women after heart valve replacement were compared to the outcomes of men. Long-term prosthesis-related complications were recorded according to the guidelines for reporting morbidity and mortality after cardiac valvular operations.16 Complications arising at the time of the index valve replacement operation (that is, perioperative complications) were excluded from this long-term analysis. Briefly, stroke was defined as a thromboembolic event associated with the presence of a neurological deficit lasting more than three weeks or causing death, and was confirmed with computed tomography of the head.15 16 Reoperation was defined as any operation that repaired, altered or replaced a previously operated valve.16 17

Statistical analyses

Data were analysed in Intercooled Stata 9.2. Continuous data are presented as a mean (SD) and were compared between women and men by using a two-sided Student t test. Categorical data are presented as proportions and were compared between groups by using a Fisher’s exact test. Statistical significance was set at a p value of <0.05. In order to assess whether women more often received bioprosthetic valves after adjustment for age, multivariate logistic regression analysis was performed separately for AVR and MVR patients by including the variables of age and gender. Odds ratios (OR) are reported along with 95% confidence intervals (CI).

Actuarial analyses of the various outcomes in this study were performed by using the Kaplan-Meier method and are reported as mean (SE). Patients were censored at the time of their last follow-up visit or at the time of death if the outcome of interest had not occurred, and censoring was assumed to be independent of predictors and outcomes. Causes of death are summarised in the appendix (table AI). Actuarial long-term survival of the study cohort was compared to the expected long-term survival of a gender-matched and age-matched Canadian population by using life-expectancy data from Statistics Canada.18

Potential univariate predictors of outcomes were individually tested for equality with a log-rank test. In order to account for positive or negative confounding, multivariate Cox proportional hazards models were developed by incorporating all variables that had a p value of 0.20 or less on log-rank testing. Stepwise forward selection and backward elimination techniques were employed with p = 0.20 for entry and removal criteria. Risk factors for adverse outcomes after valve replacement that have been previously identified15 17 were considered in each model, including female gender, age, hypertension, coronary disease requiring CABG, diabetes mellitus, non-elective surgery, preoperative New York Heart Association (NYHA) class III/IV, preoperative left ventricular dysfunction, atrial fibrillation, and cigarette smoking (see appendix, table AII). In order to account for the changes in valve surgery practice over the time period of the study, the year of surgery was also considered in each model. Scaled Schoenfeld residuals were employed to assess the Cox proportional hazard assumption for each variable. A stratified Cox analysis was used if the proportional hazard assumption was not met for a particular factor. Hazard ratios (HR) are reported along with 95% CI. Unadjusted hazard ratios from univariate log-rank analysis comparing outcomes of women and men are described first, followed by hazard ratios adjusted for confounders, as determined by multivariate Cox proportional hazards models.

RESULTS

Patient and operative characteristics

The preoperative characteristics of the study cohort are presented in table 1. In the group of patients who underwent AVR, women were significantly older than men (p<0.0001), and men were more often smokers (p<0.0001). Among men, the indications for AVR were aortic stenosis in 1012 (68%), aortic regurgitation in 298 (20%), and mixed aortic stenosis and regurgitation in 179 (12%). For women, the indications for AVR were 613 (80%) aortic stenosis, 84 (11%) aortic regurgitation and 69 (9%) mixed aortic stenosis and regurgitation. Women more often had aortic stenosis and less often aortic regurgitation as their indication for operation compared to men (p<0.0001).

In the group who underwent MVR, there was a higher incidence of atrial fibrillation (p = 0.006) and hypertension (p = 0.04) in women than men. Among men, indications for MVR were 85 (23%) mitral stenosis, 220 (60%) mitral regurgitation and 63 (17%) mixed mitral stenosis and regurgitation. For women, indications for MVR were 163 (33%) mitral stenosis, 238 (48%) mitral regurgitation and 94 (19%) mixed mitral stenosis and regurgitation. Women more often had mitral stenosis and less often mitral regurgitation as their indication for operation than men (p = 0.03).

Table 1 additionally summarises the operative characteristics of the study cohort. At the time of operation, men more often underwent concomitant CABG than women (p<0.0001 for AVR and MVR), with significantly longer cardiopulmonary bypass times and aortic cross-clamp times (both p<0.0001 for AVR and MVR). On univariate analysis, women received bioprostheses more often than men at the time of AVR (p<0.0001). However, after adjusting for age, there was only a trend towards more bioprostheses implanted in women at the time of AVR (OR 1.2; CI 1.0 to 1.4; p = 0.10). Women did not have a higher likelihood of receiving a bioprosthesis at the time of MVR, either on univariate analysis (p = 0.25), or after adjusting for age (OR 0.9; CI 0.6 to 1.2; p = 0.38).

Reoperation

Figure 1 displays the long-term freedom from reoperation in the patient cohort. After bioprosthetic AVR, the 10-year freedom from reoperation was 95.8% (1.8%) in women, and 79.4% (4.4%) in men, respectively (unadjusted HR, women vs men 0.3; CI 0.2 to 0.7; p = 0.003). Indications for reoperation after bioprosthetic AVR in women were structural valve deterioration (SVD) in four, prosthetic valve endocarditis in four, and other conditions in one. Indications for reoperation after bioprosthetic AVR in men were SVD in 27, prosthetic valve endocarditis in seven, thromboembolism in two and other conditions in one. Following mechanical AVR, the 10-year freedom from reoperation was 94.2% (1.8%) in women and 95.0% (1.0%) in men, respectively (unadjusted HR, women vs men 0.7; CI 0.4 to 1.3; p = 0.28). Indications for reoperation after mechanical AVR in women were SVD in four, prosthetic valve endocarditis in one, valve thrombosis in two and other conditions in two. Indications for reoperation after mechanical AVR in men were SVD in nine, prosthetic valve endocarditis in 12, and other conditions in eight. After accounting for age and other co-morbidities, female gender was independently associated with better freedom from reoperation following bioprosthetic AVR (adjusted HR 0.4; CI 0.2 to 0.9; p = 0.04), but not following mechanical AVR (adjusted HR 0.8; CI 0.4 to 1.5; p = 0.52).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Freedom from reoperation after aortic valve replacement (AVR) (A) or mitral valve replacement (MVR) (B) in women and men. Hazard ratios (HR) adjusted with multivariate modelling (see text). After adjusting for potential confounders, there was a significantly lower reoperation rate in female patients after bioprosthetic AVR.

After bioprosthetic MVR, the 10-year freedom from reoperation was 69.1% (6.4%) in women, and 78.1% (6.6%) in men, respectively (unadjusted HR, women vs men 1.4; CI 0.8 to 2.4; p = 0.25). Indications for reoperation after bioprosthetic MVR in women were SVD in 25, prosthetic valve endocarditis in three and other conditions in eight. Indications for reoperation after bioprosthetic MVR in men were SVD in eight, prosthetic valve endocarditis in four and other conditions in seven. Following mechanical MVR, the 10-year freedom from reoperation was 91.1% (2.0%) in women, and 92.1% (2.6%) in men, respectively (unadjusted HR, women vs men 1.1; CI 0.5, 2.1; p = 0.86). Indications for reoperation after mechanical MVR in women were SVD in nine, prosthetic valve endocarditis in three, valve thrombosis in one and other conditions in four. Indications for reoperation after mechanical MVR in men were SVD in four, prosthetic valve endocarditis in two and valve thrombosis in two. After accounting for co-morbidities, female gender was not independently associated with better freedom from reoperation following bioprosthetic MVR (adjusted HR 1.0; CI 0.5, 1.7; p = 0.86) or mechanical MVR (adjusted HR 1.0; CI 0.5, 2.0; p = 0.95).

Stroke

Figure 2 displays the long-term freedom from stroke or stroke-related death in the patient cohort. After bioprosthetic AVR, 233 patients required long-term anticoagulation (incidence of anticoagulation women vs men, 17.1% vs 18.4%, p = 0.60; mean INR, women vs men 2.6 (0.7) vs 2.6 (0.7), p = 0.80). The 10-year freedom from stroke following bioprosthetic AVR was 90.9% (2.9%) in women, and 90.0% (4.4%) in men, respectively (unadjusted HR, women vs men 1.6; CI 0.7, 3.5; p = 0.26). Following mechanical AVR, the 10-year freedom from stroke was 86.4% (3.0%) in women, and 90.3% (1.8%) in men, respectively (unadjusted HR, women vs men 1.4; CI 0.8 to 2.4; p = 0.19). Women and men had equivalent levels of long-term anticoagulation after mechanical AVR (mean INR, women vs men 2.6 (0.7) vs 2.6 (0.8), p = 0.97). Three women less than 45 years of age with mechanical aortic prostheses had strokes. After accounting for age and other co-morbidities, female gender was independently associated with a greater risk of stroke following mechanical AVR (adjusted HR 1.7; CI 1.1, 2.7; p = 0.02), but not following bioprosthetic AVR (adjusted HR 1.5; CI 0.7, 3.5; p = 0.32).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Freedom from stroke or stroke-related death after aortic valve replacement (AVR) (A) or mitral valve replacement (MVR) (B) in women and men. Hazard ratios (HR) adjusted with multivariate modelling (see text). After adjusting for potential confounders, there were higher stroke rates in female patients with mechanical aortic and mitral prostheses.

After bioprosthetic MVR, 133 patients required long-term anticoagulation (incidence of anticoagulation women vs men, 48.2% vs 39.4%, p = 0.13; mean INR, women vs men 2.6 (0.7) vs 2.5 (0.6), p = 0.64). The 10-year freedom from stroke following bioprosthetic MVR was 90.3% (5.0%) in women, and 98.0% (2.0%) in men, respectively (unadjusted HR, women vs men 2.3; CI 0.5 to 11.0; p = 0.29). Following mechanical MVR, the 10-year freedom from stroke was 80.7% (3.9%) in women, and 88.9% (4.3%) in men, respectively (unadjusted HR, women vs men 2.1; CI 1.0 to 4.6; p = 0.06). Women and men had equivalent levels of long-term anticoagulation after mechanical MVR (mean INR, women vs men 2.7 (0.8) vs 2.7 (0.8), p = 0.46). Three women less than 45 years of age with mechanical mitral prostheses had strokes. After accounting for co-morbidities, there were trends towards more late strokes for women following bioprosthetic MVR (adjusted HR 2.9; CI 0.6 to 15.2; p = 0.20) and mechanical MVR (adjusted HR 2.1; CI 1.0 to 4.8; p = 0.06).

Late survival

Figure 3 displays the long-term survival of AVR patients. After bioprosthetic AVR, the 10-year actuarial survival was 70.0% (4.4%) in women (expected 10-year survival in age-matched Canadian population 85.6% (1.6%), p<0.0001) and 55.9% (3.5%) in men (expected survival 71.1% (1.6%), p<0.0001). On univariate analysis, women had significantly better long-term actuarial survival after bioprosthetic AVR (unadjusted HR, women vs men 0.6; CI 0.4 to 0.7; p<0.0001). Following mechanical AVR, the 10-year survival was 79.1% (3.0%) in women (expected survival 99.6% (0.4%), p<0.0001), and 73.3% (2.3%) in men (expected survival 95.5% (0.8%), p<0.0001). On univariate analysis, there was no gender difference in actuarial long-term survival after mechanical AVR (unadjusted HR, women vs men 0.9; CI 0.7, 1.3; p = 0.74). After accounting for age and other co-morbidities, female gender was independently associated with better actuarial survival following bioprosthetic AVR (adjusted HR 0.5; CI 0.3 to 0.6; p<0.0001), but not following mechanical AVR (adjusted HR 1.0; CI 0.7 to 1.4; p = 0.79).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Actuarial survival after mechanical (A) or bioprosthetic aortic valve replacement (AVR) (B) in women and men, with comparisons to the expected survival of gender-matched and aged-matched general populations. Hazard ratios (HR) adjusted with multivariate modelling (see text). After adjusting for potential confounders, actuarial survival was better in female patients after bioprosthetic AVR.

Figure 4 displays the long-term survival of MVR patients. After bioprosthetic MVR, the 10-year survival was 72.6% (5.2%) in women (expected survival 95.2% (1.7%), p = 0.10), and 55.1% (7.0%) in men (expected survival 71.3% (3.9%), p = 0.03). On univariate analysis, there was a trend towards better actuarial long-term survival in women after bioprosthetic MVR (unadjusted HR, women vs men 0.6; CI 0.4 to 1.0; p = 0.06). Following mechanical MVR, the 10-year survival was 70.9% (3.3%) in women (expected survival 99.7% (0.3%), p<0.0001), and 71.7% (3.8%) in men (expected survival 95.2% (1.4%), p<0.0001). On univariate analysis, there was no gender difference in actuarial long-term survival after mechanical MVR (unadjusted HR, women vs men 0.8; CI 0.6 to 1.2; p = 0.25). However, after accounting for co-morbidities, there were trends towards better survival for women following bioprosthetic MVR (adjusted HR 0.6; CI 0.4 to 1.0; p = 0.06) and mechanical MVR (adjusted HR 0.8; CI 0.5 to 1.1; p = 0.15).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4 Actuarial survival after mechanical (A) or bioprosthetic mitral valve replacement (MVR) (B) in women and men, with comparisons to the expected survival of gender-matched and aged-matched general populations. Hazard ratios (HR) adjusted with multivariate modelling (see text). After adjusting for potential confounders, there were trends towards better actuarial survival in female patients after mechanical and bioprosthetic MVR.

DISCUSSION

Women represent an increasing proportion of the cardiac surgery population.19 Over 200 000 American women undergo cardiac surgery each year, including 47 000 isolated valve operations.1 Several studies have documented differences in the outcomes of women and men after CABG, mostly with respect to perioperative and short-term outcomes.2^–9 Notably, very little research has been published comparing gender-specific outcomes after valve replacement surgery.10

In this follow-up study of 3118 patients (1261 women; 1857 men) who underwent valve replacement with contemporary prostheses, we found that: (1) women more often undergo valve replacement surgery for stenotic valve pathology than men who more often have regurgitant or mixed valvular disease; (2) women have a significantly lower rate of reoperation after bioprosthetic AVR; (3) women have more late strokes after mechanical AVR, and tend to have more late strokes after bioprosthetic and mechanical MVR; and (4) long-term survival following bioprosthetic AVR is significantly better for women than men. Additionally, we observed trends towards better survival for women after bioprosthetic and mechanical MVR. Thus, the long-term outcomes after valve replacement surgery differ between women and men, with better long-term survival favouring women.

Reoperation following valve replacement surgery is a common event, particularly in patients with bioprosthetic heart valves. Freedom from reoperation for currently available mechanical valves is generally greater than 95% at 10 years. In contrast, bioprostheses have a significantly higher rate of reoperation secondary to structural valve deterioration. Several investigators have previously suggested that female gender is a risk factor for increased bioprosthetic failure.20 21 However, in a recent analysis of the determinants of reoperation in 3233 patients with prosthetic heart valves, we reported a lower rate of reoperation in women following bioprosthetic AVR (HR 0.7, CI 0.5 to 1.0, p = 0.04).17 This finding was confirmed in the current study, illustrating a greater freedom from reoperation for women following bioprosthetic AVR (adjusted HR: 0.4; CI 0.2 to 0.9; p = 0.04), even after adjusting for age and other co-morbidities. While it is unclear why women had a lower rate of reoperation following bioprosthetic AVR, we speculate that hormonal milieu may contribute to slower bioprosthesis structural failure in women. Alternatively, a bias may exist on the part of cardiologists or cardiac surgeons against the consideration of valve reoperation in women with failing aortic bioprostheses. A similar referral bias against women has been identified among patients with native aortic valve disease.22 Moreover, female gender is a known risk factor for higher mortality at the time of aortic valve reoperation.23

Stroke is a devastating complication that results from embolism or intracranial haemorrhage,16 and appears to be more common in patients with mechanical prostheses.24 Following cardiac surgery, female gender has been demonstrated to be associated with a significantly higher risk of stroke perioperatively25 and late after surgery.11 Previous work from our group showed that women with prosthetic valves had approximately 1.7 times the embolic stroke risk of men, even after adjustment for other risk factors.15 These results were confirmed in the current study, with focus on a larger cohort with longer follow-up data. Female gender was independently associated with a greater risk for stroke following mechanical AVR (adjusted HR: 1.7; CI 1.1 to 2.7; p = 0.02), and there were trends towards more strokes in women following mechanical MVR (adjusted HR: 2.1; CI 1.0 to 4.8; p = 0.06) and bioprosthetic MVR (adjusted HR: 2.9; CI 0.6 to 15.2; p = 0.20). Notably, a large number of patients who received bioprostheses in this study required long-term anticoagulation (that is, atrial fibrillation), especially after MVR, although the incidence rates were similar between women and men. Overall, anticoagulation levels appeared adequate in all subgroups. The greater risk of stroke in women demonstrated in this analysis may relate to the higher incidence of hypertension among women in the general population26 (also seen in the current study), or may allude to the challenges of anticoagulation in young women (less than 45 years) with mechanical valves, particularly during pregnancy.27

While some controversy exists on the subject,28 29 the majority of studies examining gender-specific outcomes after valve replacement surgery have demonstrated higher perioperative mortality rates in women,11^–13 including two large registry analyses.6 14 In contrast, very little information was available until now regarding the long-term survival of men and women after valve replacement surgery. Previous investigations with shorter follow-up periods have suggested that long-term survival after valve replacement surgery is similar between men and women. In a Mayo Clinic study of 1012 consecutive AVR patients from 1983 to 1990 (329 women and 683 men), five-year postoperative survival was 77% for women and 83% for men (p = NS).29 Similarly, in a study from Toronto, female gender was not an independent predictor of long-term survival following combined valve and CABG surgery among 1567 patients (10-year survival: 56% men, 50% women, p = 0.06).11 In contrast, the current study demonstrated that female gender was independently associated with better long-term survival after bioprosthetic AVR, and trends existed towards better survival after bioprosthetic and mechanical MVR. This survival advantage may be a reflection of the longer life expectancy in women since, on average, women live 5.2 years longer than men in North America.10 30 Nevertheless, in all subgroups analysed in this study, long-term actuarial survival was worse compared to expected survival data from gender-matched and age-matched population controls.18 Overall, it appears that patients are not able to reach their full survival potential after seemingly successful valve replacement surgery. This may be related to progressive myocardial dysfunction, coronary artery disease, generalised atherosclerosis and atrial fibrillation late after surgery, as well as the development of prosthetic-related and anticoagulation-related complications that limit long-term survival.

Limitations

To our knowledge, this study is the largest to date to compare gender-specific outcomes after valve replacement surgery. However, the data presented must be interpreted in the context of the study design. An observational study, group differences and known confounders were controlled for in the multivariate analysis. Despite the statistical adjustments applied, however, unmeasured and unknown confounders may have influenced the results. It is possible that patients lost to follow-up after a number of visits may have had subsequent outcomes that were not accounted for in the analyses. Furthermore, the findings of this study may not necessarily be generalisable to all patients with prosthetic valves because the study represents a single institution’s experience. Finally, this study was limited by the lack of data regarding socioeconomic status31 as well as hormonal contraceptive use as a potential risk factor for thromboembolism in women.32

CONCLUSION

In conclusion, our analysis of the gender-specific long-term outcomes after valve replacement surgery reveals that women have more late strokes after valve replacement. In contrast, women undergo fewer reoperations and have better long-term survival than men. Further investigation to verify and explore the aetiology of these differences is warranted. We believe the present study will help to more accurately describe the long-term results after valve replacement surgery for women and men, especially in an era where more women than ever are being referred for cardiac surgery.