Differences in life expectancy between men and women after aortic valve replacement

Abstract

 

Open in new tabDownload slide

OBJECTIVES

Some researchers have observed an increased number of deaths during the follow-up of young patients who undergo aortic valve replacement due to severe aortic stenosis, suggesting that this procedure does not restore their life expectancy. Our goal was to confirm these findings and explore sex-based differences.

METHODS

All patients between 50 and 65 years of age who underwent isolated aortic valve replacement in 27 Spanish centres during an 18-year period were included. We compared observed and expected survival at 15 years of follow-up and estimated the cumulative incidence of death from a competing risks point of view. We stratified by sex and analysed if being a woman was an independent risk factor for death.

RESULTS

For men, the observed survival at 10 and 15 years of follow-up was 85% [95% confidence interval (CI) 83.6%–86.4%] and 72.3% (95% CI 69.7%–74.7%), respectively whereas the expected survival was 88.1% and 78.8%. For women, the observed survival at 10 and 15 years was 85% (95% CI 82.8%–86.9%) and 73% (95% CI 69.1%–76.4%), whereas the expected survival was 94.6% and 89.4%. At 15 years of follow-up, the cumulative incidence of death due to the disease in men and women was 8.2% and 16.7%, respectively. In addition, being a woman was an independent risk factor for death (hazard ratio = 1.23 (95% CI 1.02–1.48; P = 0.03).

CONCLUSIONS

After the aortic valve replacement, men and women do not have their life expectancy restored, but this loss is much higher in women than in men. In addition, being a woman is a risk factor for long-term death. Reasons for these findings are unknown and must be investigated.

INTRODUCTION

Aortic valve replacement (AVR) has been shown to modify the natural history of symptomatic severe aortic stenosis (SAS). Life expectancy after AVR has been historically considered similar to that of the general population. However, results of recent studies have observed a loss in life expectancy after this procedure, suggesting that surgery is not completely effective in solving the problem [1–3]. The observed life expectancy was slightly lower than expected in older patients but was much lower than expected in young individuals [1–3]. Given that SAS is a disease of older persons, the reported samples of young patients were small. In addition, around 70% of patients who undergo AVR are men; thus, conclusions could only be representative for men [4, 5].

Whether patients with SAS have their life expectancy restored after AVR is of utmost importance for patient information, informed consent and medical decision-making. Physicians, surgeons, patients and experts [1, 2] have recently requested an analysis of larger cohorts in other nations to decide if there really is a reduction in life expectancy and to investigate possible reasons.

We used a nationwide cohort to analyse the excess deaths among younger patients (50–65 years) who underwent isolated AVR and to compare their long-term survival with that of the general population of the same age, sex, year of surgery and nation. To further understand the differences between men and women, we stratified by sex and analysed the role of sex as a risk factor for long-term death.

METHODS

Selection of the sample

We used a national multicentre registry (SPAVALVE, SPAnish Aortic VALVE) including all patients from 50 to 65 years of age who underwent isolated AVR in 27 Spanish centres from January 2000 through September 2018. Endocarditis, nonelective operations and operations for isolated aortic regurgitation were excluded. Each centre had 2 researchers responsible for collecting the pre-, intra- and postoperative data and for the follow-up of all patients. The variables collected in this registry were reported for another study at ClinicalTrials.gov: NCT03595423.

We compared the long-term observed survival of those patients who underwent isolated AVR with the expected survival of the general population from Spain matched by age, sex and year of surgery. To match for these factors, we used the aggregate data freely available and provided by the National Institute of Statistics [6]. This method has been described previously [1, 7].

To determine if there were differences between the sexes, we stratified all the analyses by sex. In addition, we investigated what the situation was if patients survived the postoperative period. The postoperative period was considered to be 30 days after the operation or until hospital discharge if it was after 30 days.

This study complies with the Declaration of Helsinki and was approved by the corresponding institutional review boards (reference 2806, date: 7 February 2018).

Endpoints

Our endpoints were as follows (i) to compare the observed with the expected survival of men and women 50–65 years of age who underwent isolated AVR; (ii) to compare the observed with the expected survival of men and women 50–65 years of age who underwent isolated AVR and survived the postoperative period; and (iii) to determine if being a woman is a risk factor for long-term mortality among patients 50–65 years old who underwent isolated AVR.

Statistical analyses

Categorical and quantitative variables were described as n (%) and mean ± SD, respectively. Categorical variables were compared using the Fisher’s exact test. Quantitative variables were compared using the t-test, with equal or unequal variances, as appropriate. Because the P-value depends on sample size, we also calculated the standardized differences [8, 9].

Stratifying by sex, we compared the observed survival with the expected survival, which was calculated using the Ederer II method. This method, considered the best for this purpose [10], is able to estimate the expected survival of the sample using the theorical assumption that they did not have the disease. This method was previously used in other studies [1, 7]. If the 95% confidence interval (CI) of the observed survival included the expected survival, no difference could be shown. For the calculations, we used the aggregate data provided by the National Institute of Statistics [6].

Finally, we calculated the probability of death in a competing risk framework using the method proposed by Cronin and Feuer [11]. With this method, we calculated the cumulative incidence of death of SAS or of any related factor, the cumulative incidence of death of other causes and the overall cumulative incidence of death. All analyses were stratified by sex. The strs command was used to calculate these estimations [12]. One of the main advantages of this method is that it permits us to know the mortality related to the disease under study, without the need to know the causes of death [1, 7, 10, 12].

To determine the impact of being a woman on long-term mortality, we prepared a propensity score-matched analysis. The propensity score was calculated using a logistic regression with all baseline characteristics and the type of the prosthesis (biological/mechanical). Each woman was matched with 1 man using the nearest neighbour strategy without replacement. Balance diagnostics included evaluation of the standardized differences and visual graphics. Standardized differences >0.10 were considered important [8, 9]. After assessing the balance between the groups, we graphed the survival curves and compared them using a log-rank test. A hazard ratio (HR) was calculated using a univariate Cox regression analysis in the matched sample. The influence of sex was calculated using men as the reference, and the HR was presented with the 95% CI. Hazards proportionality was assessed by the visual assessment of the ln-minus-ln probability of survival.

All analyses were performed with STATA v.15^® (StataCorp, College Station, TX, USA).

RESULTS

Baseline characteristics

Five thousand and eighty-four (5084) patients underwent isolated AVR during the study period. Of these, 3466 (68.2%) were men and 1618 (31.8%) were women. The women were older than the men (59.1 ± 4.3 vs 59.9 ± 4.2 years and had body mass indexes of 28.6 kg/m² ± 4.4 vs 29.3 kg/m² ± 5.4, respectively). Conversely, men had more peripheral artery disease [235 (6.8%) vs 67 (4.1%)], chronic pulmonary disease [554 (15.9%) vs 103 (6.4%)], previous myocardial infarction [230 (6.6%) vs 46 (2.8%)], renal impairment [350 (10.1%) vs 89 (5.5%)], history of smoking [746 (21.5%) vs 145 (8.9%)] and history of alcoholism [321 (9.3%) vs 31 (1.9%)]. The left ventricular ejection fraction was higher in women (57.9% ± 11.8 vs 61.1% ± 10.1). Table 1 shows all baseline characteristics and the type of the prosthesis.

Observed and expected survival

The mean follow-up for the censored observations was 105.3 ± 54.3 months; 152 (3%) patients died during the postoperative period.

For men, the observed survival at 5, 10 and 15 years of follow-up was 92.2% (95% CI 91.2%–93.1%), 85% (95% CI 83.6%–86.4%) and 72.3% (95% CI 69.7%–74.7%), whereas the expected survival for these years was 94.9%, 88.1% and 78.8% (Fig. 1A).

Figure 1:

Global sample. (A) Observed and expected survival in men. (B) Observed and expected survival in women. (C) Incidence of death because of the disease in men. (D) Incidence of death because of the disease in women. All vertical axes are cut.

Open in new tabDownload slide

For women, the observed survival at 5, 10 and 15 years was 92.1% (95% CI 90.6%–93.4%), 85% (95% CI 82.8%–86.9%) and 73% (95% CI 69.1%–76.4%), whereas the expected survival for these years was 97.8%, 94.6% and 89.4% (Fig. 1B).

In the competing risk framework, the cumulative incidence of death of the disease in men at 5, 10 and 15 years of follow-up was 3%, 4.3% and 8.2% (Fig. 1C). In women, the cumulative incidence of death due to the disease was 5.9%, 10.2% and 16.7% (Fig. 1D).

For patients who survived the postoperative period, the observed survival in men at 5, 10 and 15 years of follow-up was 94.8% (95% CI 94%–95.6%), 87.5% (95% CI 86%–88.8%) and 74.4% (95% CI 71.7%–76.8%). The expected survival was 94.9%, 88.1% and 78.8% (Fig. 2A).

Figure 2:

Patients who survived the postoperative period. (A) Observed and expected survival in men. (B) Observed and expected survival in women. (C) Incidence of death because of the disease in men. (D) Incidence of death because of the disease in women. All vertical axes are cut.

Open in new tabDownload slide

In women, the observed survival at 5, 10 and 15 years was 95.3% (95% CI 94%–96.3%), 87.9% (95% CI 85.8%–89.8%) and 75.5% (95% CI 71.5%–78.9%). The expected survival was 97.8%, 94.6% and 89.4% (Fig. 2B).

Among patients who survived the postoperative period, we calculated the cumulative incidence of death due to the disease. In men, the incidence of death due to the disease at 5, 10 and 15 years of follow-up was 03%, 16% and 56% (Fig. 2C). In women, the cumulative incidence of death due to the disease was 2.6%, 7.2% and 13.9% (Fig. 2D).

Sex as an independent risk factor for death

Before matching, sex was not an independent risk factor for death (log-rank test, P = 0.63). The univariate Cox regression analysis revealed an HR of 0.96 (95% CI 0.83–1.12; P = 0.63). Survival curves can be observed in Fig. 3A.

Figure 3:

(A) Observed survival in men and women before the matching. (B) Survival after the matching. Vertical axes are cut.

Open in new tabDownload slide

After matching, 1618 pairs were analysed. All baseline characteristics were balanced between groups. Balance diagnostics are shown in Table 2 and Supplementary Material, Figs S1–S3. Survival curves between men and women can be observed in Fig. 3B. Being a woman was an independent risk factor for long-term death (log-rank test, P = 0.03). Cox regression analysis showed an HR = 1.23 (95% CI 1.02–1.48; P = 0.03).

DISCUSSION

The main finding of this study is that (i) the survival of young patients who underwent AVR was lower than the survival of the general population of the same age and sex. (ii) This loss in life expectancy was much higher in women than in men. (iii) Being a woman was an independent risk factor for long-term death.

Life expectancy after aortic valve replacement

Many studies have reported the long-term survival of patients undergoing AVR [13]. But this survival, without comparing it with that of the general population of the same geographical region, provides little information because life expectancy differs substantially among different nations. Health systems, gross domestic products, temperatures or food habits have been associated with life expectancy [7, 14]. So, important differences, even among industrialized countries, can be observed. Whereas a 65-year-old woman in the USA has a life expectancy of 20.6 years, a 65-year-old woman in Japan has a life expectancy of 24.4 years [15].

Recently, Glaser et al. [1] analysed the life expectancy of 2000 patients between 50 and 60 years of age who underwent AVR (isolated or concomitant with coronary artery bypass surgery) in their country. They showed that there was a loss in life expectancy, especially in young patients. Results of young patients stratified by sex were not reported. In our work, using a nationwide cohort, we analysed almost 5000 patients between 50 and 65 years of age who underwent isolated AVR. Consistent with the findings of these authors, men and women who underwent AVR had a loss in life expectancy compared with the general population of the same age, sex, year and nation. Although the observed long-term survival between both groups was similar, the loss in life expectancy was much higher in women. This difference occurs because women in the general population usually live between 4 and 9 years longer than men [16]. So, although 8% of men died at 15 years of follow-up of the disease or of related factors, around 17% of women died of these reasons. These reasons can be pooled into 3 groups: (i) factors related to the operation such as perioperative mortality or complications; (ii) late complications of the prosthesis such as endocarditis, degeneration or thrombosis; or (iii) patient-related factors such as cardiovascular risk factors or comorbidities that usually accompany this disease. The reasons why this loss in life expectancy occurs in both groups and why it affects more women than men are unknown and beyond the scope of this study. Nevertheless, the life expectancy of men who survive the postoperative period is almost restored, whereas the life expectancy of women is greatly reduced.

Because our sample comprised young patients, 80% of them had a mechanical prosthesis. The observed loss in life expectancy could be due to the inherent complications of that kind of prosthesis. However, a biological prosthesis can deteriorate rapidly and lead to a life-threatening situation [17]. A recent meta-analysis showed that mechanical prostheses yielded a slightly greater life expectancy in patients between 50 and 70 years of age [13]. So, we think that our results can be extrapolated to patients in that age range, regardless of whether they have a mechanical or a biological prosthesis.

Female sex as a risk factor for long-term death

As we mentioned previously, the observed survival of men and women was similar. This result occurred despite the fact that more men had peripheral artery disease, chronic pulmonary disease, history of smoking, history of alcoholism, previous myocardial infarctions, renal impairment and lower left ventricular ejection fractions. When we adjusted for all covariates after the propensity score matching, we found that being a woman was a risk factor for long-term mortality with 20% higher risk than men. There must be underlying biological or social reasons, but they are unknown to us and deserve further investigation.

Strengths and limitations

In this study, we present the largest sample of relatively young patients who underwent AVR and compared their long-term survival with the long-term survival of the general population matched by age, sex, year of surgery and nation. The patients in our sample had only isolated AVR; therefore, our sample is more homogeneous than the samples presented in the other published studies, in which the patients had concomitant coronary procedures. In addition, we stratified by sex and investigated the impact of this factor on long-term survival. People do not only die of 1 cause. Therefore, the excess of deaths of disease calculated in the setting of competing risks should give the real estimate of the situation [11]. The rest of the studies on this issue did not estimate this competing risks scenario [1–3].

Moreover, we acknowledge several limitations. This is a retrospective work and is therefore subject to intrinsic biases of this type of study. Some conditions such as pulmonary hypertension were not present in the database. In addition, causes of death during the follow-up period were not clear in many cases so we did not report them. Nevertheless, the method used permits us to know which deaths were related to the disease under study without the need to know the causes of deaths [1, 7, 10, 12].

In conclusion, women from the general population usually live longer than men. However, after AVR, both groups have the same life expectancy. After this operation, men and women do not have their life expectancy restored but this loss is much higher in women than in men. This difference occurs despite the fact that women with SAS have fewer associated comorbidities. In addition, being a woman is a risk factor for long-term death. Reasons for these findings are unknown and must be investigated. The results of this study suggest that men and women undergoing AVR should be studied separately from now on.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

Conflict of interest: none declared.

Author contributions

Daniel Hernandez-Vaquero: Conceptualization; Formal analysis; Writing—original draft; Writing—review & editing. Emiliano Rodriguez-Caulo: Conceptualization; Data curation; Writing—original draft; Writing—review & editing. Carlota Vigil-Escalera: Conceptualization; Investigation; Methodology. Oscar Blanco-Herrera: Data curation; Resources; Validation. Elisabet Berastegui: Data curation; Investigation. Javier Arias-Dachary: Data curation; Investigation. Souhayla Souaf: Investigation. Gertrudis Parody: Investigation. Gregorio Laguna: Investigation. Alejandro Adsuar: Data curation; Investigation. Manel Castellá: Data curation; Investigation. José F. Valderrama: Data curation; Investigation. Ivana Pulitani: Data curation; Investigation. Sergio Cánovas: Investigation; Validation. Andrea Ferreiro: Investigation; Validation. Antonio García-Valentín: Formal analysis; Investigation; Methodology. Manuel Carnero: Methodology. Pilar Pareja: Investigation. José A. Corrales: Data curation; Investigation. José A. Blázquez: Formal analysis; Investigation. Diego Macías: Investigation. Delfina Fletcher-Sanfeliu: Investigation. Daniel Martínez: Data curation. Elio Martín: Investigation; Methodology. Miren Martín: Data curation; Investigation. Juan Margarit: Investigation; Methodology. Rafael Hernández-Estefanía: Investigation. Emilio Monguió: Formal analysis; Investigation. Juan Otero: Data curation; Investigation. Jacobo Silva: Conceptualization; Supervision; Writing—original draft.

Reviewer information

European Journal of Cardio-Thoracic Surgery thanks the anonymous reviewers for their contribution to the peer review process of this article.

REFERENCES

ABBREVIATIONS

 
• AVR

  Aortic valve replacement

 
• CI

  Confidence interval

 
• HR

  Hazard ratio

 
• SAS

  Severe aortic stenosis

Author notes