Discussion
The observed male-female differences in presentation, procedural characteristics and treatment outcomes in isolated AV or concomitant CABG/AV surgery illustrate that male-female differences in cardiac surgery extend beyond isolated CABG and are also apparent in AV surgery. This study also accentuates the complexity of addressing male-female differences in cardiac surgery, as differences are apparent in presentation and treatment, and in outcome and risk stratification approaches.
In this discussion, we will address key findings for the two surgical treatments separately, the limitations of standard risk models and potential further studies that may help optimise cardiac surgery treatment for both men and women.
AV surgery
Our analysis showed that in-hospital mortality for AV surgery was comparable between male and female patients. The AUC for the logistic EuroSCORE 1 was larger in men than in women, indicating that this model is a better predictor of outcome for men. Our findings confirm the results of earlier studies where no male-female differences in mortality outcome and non-cardiac morbidity were found.19–23 In contrast, Andrei et al
24 found that women with bicuspid AVs undergoing AV surgery were at a higher risk of in-hospital mortality. Short-term (30 days) and long-term mortality, however, appeared to be similar.
CABG/AV surgery
For concomitant CABG/AV surgery, in-hospital morality was significantly higher in women than in men in the present study. Again, the AUC for the logistic EuroSCORE 1 was larger in men than in women, indicating that this model is a better predictor of outcome for men. Doenst et al
25 reported a higher perioperative risk profile in women undergoing concomitant valve and CABG surgery and found the following risk factors to be associated with higher perioperative mortality in women: hypertension, diabetes mellitus, congestive heart failure, atrium fibrillation and stroke. None of these factors appeared to show a significant association with higher female in-hospital mortality in our study. Similar to our findings, Flameng et al
26 reported female sex, in addition to patient age, to be an independent preoperative risk factor of early mortality in concomitant valve and CABG surgery. This observation was not found by Saxena et al.27 The authors report a mean age of 76 in women in their cohort and argue that their results may reflect the impact of advanced age in reducing male-female bias.
Interestingly, in line with the current literature,24 25 27–29 we also found female patients to be older in both surgical groups, which might explain why in both groups, women received significantly less often a mechanical valve and more often stented and stentless bioprosthesis as compared with men. This age difference in male versus female patients could also fuel speculations concerning later presentation of female patients in the disease process or even higher levels of frailty at time of presentation, potentially explaining the worse outcomes in female patients after CABG/AV surgery.
Based on our results, we could additionally speculate whether the observed differences in male-female outcomes after CABG/AV surgery are due to other possible risk factors not included in the current study, including genetic or hormonal risk factors specific for women. These potential risk factors form a separate challenge altogether given the complexity of the potential underlying mechanisms. Therefore, future studies in these domains are warranted to unravel other potential female-specific risk factors after cardiac surgery.
EuroSCORE
Although EuroSCORE 1 is one of the most widely used cardiac surgical risk models, several studies have shown during the last decade that it overpredicts patient mortality and is no longer appropriately calibrated. In our study for both groups, the performance of the EuroSCORE 1 in men and women, both in terms of accuracy and calibration, is suboptimal. In addition, the accuracy of the model was inferior in female patients in both subgroups. This observation might be explained by the fact that 72% of the original EuroSCORE population was male. Although EuroSCORE 1 includes woman versus man as a separate risk factor, it has been shown that a female-specific weighing of the other 16 risk factors that are included in EuroSCORE 1 is needed for adequate prediction of outcome in female patients.30
Gender-specific risk prediction models
In our patient population there were only a few risk factors, which are included in EuroSCORE 1, predictive of short-term mortality in both male and female patients. In this regard, male and female-specific risk prediction models are needed for patients undergoing cardiac surgery. More recently, the new EuroSCORE 2 has been developed.14 Although improvements have been made with this updated model, the risk score was again built using a mixed population of male and female patients who underwent predominantly CABG surgery. Future studies need to clarify whether the performance of this updated model is also adequate in female patients.
The observed differences in (the weight of) risk factors between men and women in the predictive in-hospital mortality models that we constructed for men and women separately underline the need for further refinement of current prediction models in cardiac surgery. Probably, the best approach will be to develop, next to generic models such as EuroSCORE 2 that are mainly useful for benchmarking purposes, separate risk models for men and women in order to improve predictive ability for both sexes, but also to further stimulate research into the causes of the observed differences between men and women, be it sex, gender or a combination of the two.
Future recommendations
Further studies need to focus on exploration of the underlying mechanisms of the observed male-female differences in presentation, procedure and early mortality in order to optimise treatment for both male and female AV and combined CABG/AV surgery.
Given the significantly worse outcome in women compared with men for combined CABG/AV surgery, underlying factors contributing to worse disease outcome should be further investigated. Especially interesting are the effect of possible referral bias, atypical female presentation and/or differences in procedural characteristics for women on female outcomes as compared with men.
Finally, given the literature, there seems to be an age effect in the observed male-female differences, further studies zooming in on the intersection of age and male-female differences, especially the treatment of younger female patients, are warranted.
Strengths and limitations
The main strength of this study is that we were able to use a large contemporary national database with high-quality information (99% completeness) about all cardiac surgical interventions from all hospitals in the Netherlands. The major limitation of the current study, inherent to the type of database used, is the absence of detailed information regarding the disease aetiology, the absence of information on survival beyond hospital stay.