A total of 311 consecutive patients who underwent open surgical thoracoabdominal repair at our institution between 2000 and 2024 were included in the study. Among these, 68.2% (n = 212) were male and 31.8% (n = 99) were female. Preoperative patient characteristics, comorbidity burden, and extent of disease, as well as intraoperative parameters and operation times, postoperative complications, length of intensive care unit and hospital stay, and short-term mortality, were collected prospectively in our institutional database and analyzed retrospectively. If missing values were discovered during the retrospective evaluation, they were added using the clinic's internal documentation and archiving system to ensure completeness of the data. Long-term follow- up was performed via our aortic outpatient clinic with regularly scheduled follow-up visits. Data collection and analysis were conducted in accordance with ethics board approval by the Ethics Commission of Hannover Medical School (Approval No. 11461_BO_K).

This retrospective single-center analysis with prospective data collection and follow up was performed using a two-step approach (Figure 1). In the first step, one-to-one propensity score matching of female and male patients undergoing thoracoabdominal aortic repair was conducted. For this, groups were adjusted for discrepancies in preoperative characteristics such as age at operation and body mass index, as well as comorbidity burden including cardiovascular, respiratory, renal, and neurological comorbidities. The full list of preoperative parameters used for propensity score matching is provided in Supplementary Table S1. Patients’ sex was defined as the group indicator for propensity score matching. A logistic regression was performed on the group indicator, and the resulting propensity variable was used to select controls for the demander cases. The caliper was set to 0.05. A 1:1 propensity score matching without replacement was performed. Unmatched cases were excluded from outcome analyses in the matched cohort. Following this, both the initial study population and the matched groups were compared for pre-, intra-, and postoperative characteristics and outcome parameters. Comparative analyses between groups and survival analyses were performed for both the unmatched and the matched cohorts in parallel to account for matching and potential differences resulting from unequal baseline values.

In the second step, a multivariate binomial logistic regression model for in-hospital mortality was created separately for female and male thoracoabdominal aortic repair patients. Preoperative patient characteristics, comorbidities, and urgency of the operation were used as predicting variables. The full list of tested predicting variables is provided in Supplementary Table S2. Linearity was tested using the Box–Tidwell method with Bonferroni correction for all continuous variables (9). Goodness of fit of the logistic regression model was analyzed using the Hosmer–Lemeshow Test. To investigate possible interactions among the predicting variables, a combined model of the original study population was generated, incorporating the key predictors identified in the separate models. A correlation matrix was then constructed to identify possible interactions between predictors.

The surgical technique applied for open surgical thoracoabdominal aortic replacement did not differ between female and male patients. In both groups, thoracoabdominal repair was performed as described previously (10). Briefly, patients were placed in hemi-right-sided position, and a left-sided thoracotomy starting in the 5th to 7th intercostal space was performed and extended depending on the location and extension of the aortic pathology. The incision was then extended toward a paramedian laparotomy to access the abdominal aorta. After establishing cardiopulmonary bypass and cooling, the aorta was clamped proximally and distally, thus isolating the aneurysmatic segment, and graft implantation was performed according to the extent of the aortic disease. Intercostal, lumbar, and visceral arteries were reimplanted when required and suitable. After rewarming, deairing, weaning from cardiopulmonary bypass, and achieving hemostasis, patients were transferred to the surgical intensive care unit for further stabilization. To identify potential bias due to practice changes over the inclusion period, time frame-stratified sex distribution between male and female patients operated over time was analyzed.

The extent of thoracoabdominal aortic repair was categorized according to the Crawford classification of aortic diseases, considering the proximal and distal prosthesio-aortic anastomosis for sorting in the corresponding group (11). Cerebrovascular disease was defined as any radiographically diagnosed flow-limiting intra- or extracranial cerebrovascular stenosis or occlusion. Likewise, peripheral vascular disease was defined as any clinically radiographically diagnosed flow-limiting peripheral vascular stenosis or occlusion. Coronary artery disease was defined as any treated or untreated flow-limiting coronary artery obstruction diagnosed by coronary angiography. Chronic kidney disease was defined as a decreased glomerular filtration rate of less than 60 mL/min (1.73 m²) for at least 3 months (12). Acute kidney failure was defined as a threefold increase in serum creatinine or urine output of less than 0.5 mL per kg body weight per hour for 24 h (13). Sepsis was defined according to the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) (14). Low cardiac output syndrome was defined as a cardiac index <2.2 L/min and body surface area in m² with systolic blood pressure of <90 mmHg or recurrent catecholamine dependency. Respiratory failure was defined as the necessity of re-intubation or non-invasive ventilation after previous spontaneous breathing. Symptoms and complications resolving prior to discharge were defined as temporary, while those persisting at discharge were defined as permanent. Operations were categorized as urgent if the time between first admission and operation was under one week and as emergent if the time between first admission and operation was under 24 h.

Statistical analyses were performed using IBM SPSS Statistics 29 (IBM Corp., Armonk, NY, USA, 1989, 2021). Normal distribution of continuous variables was tested using the Shapiro–Wilk test. Non-normally distributed data are presented as median and interquartile range. The Mann–Whitney test was used for comparison of non-normally distributed data. A p-value of <0.05 was considered significant. Long-term survival analysis was performed for female and male patients in both the matched and unmatched cohort using Kaplan–Meier survival estimates and log-rank testing.

A total of 311 consecutive patients, including 99 women (31.8%), were included in the initial study population. Prior to propensity score matching, male patients had significantly higher body mass indices (BMI) with an index of 26.3 (IQR 24.3–28.9) in men compared to 23.1 (IQR 20.2–27.4) in women (Table 1). Furthermore, the male patients had higher rates of metabolic disorders, with significant differences in the preoperative prevalence of hyperlipidemia [12.12% (n = 12) vs. 27.83% (n = 59), p = 0.002] and coronary artery disease [21.21% (n = 21) vs. 37.74% (n = 80), p = 0.004] between male and female patients at the time of open thoracoabdominal aortic repair. In the time-stratified sex distribution analysis, the proportion of female and male patients operated on over time remained stable across all periods examined, ranging between 25% and 36% (Supplementary Figure S1).

Propensity score matching identified 82 one-to-one pairs of men and women from the initial study cohort, yielding a total of 164 patients, representing 52.7% of the initial study population and including 82.8% of the female patients in the study. The indication for surgery for the vast majority of patients, in both the male and female groups, was aneurysm of the aorta. Chronic dissection was present in 37.8% (n = 31) of the male patients and 30.5% (n = 25) of the female patients in the matched patient cohort (p = 0.410). In addition, one female patient (1.2%) was operated upon for late-diagnosed coarctation of the aorta. Through propensity score matching for the preoperative characteristics listed in Supplementary Table S1, adjustment for the differences in BMI and burden of disease between men and women in the unmatched cohort was achieved so that no statistically significant differences were found between the two groups in the propensity score matched cohort (Table 1). In the matched cohort, the standardized mean differences (SMD) between female and male patients were smaller than in the unmatched study population for all preoperative variables, except cerebrovascular disease.

Operative urgency did not differ between male and female patients in either the unmatched or the matched cohort (Table 2). In the unmatched cohort, a significantly higher incidence of Crawford Type 1 operations was performed in the female subgroup [18.37% (n = 18) vs. 9.48% (n = 20), p = 0.027], while no significant differences in the Crawford extents of repair were found after propensity score matching. Likewise, operation time, bypass time, and cross-clamp time did not differ significantly between the two groups.

Postoperative outcome parameters are summarized in Table 3. In this, female patients exhibited a higher rate for postoperative wound infections with an incidence of 12.12% (n = 12) compared to 4.25% (n = 9) in men (p = 0.01). However, no significant differences concerning the observed postoperative outcome parameters were found in the matched cohort. Likewise, the ICU length of stay of 3 days was similar in both groups [3 days (IQR 2–9 days) vs. 3 days (IQR 2–8 days, p = 0.979)], and the groups did not differ in the length of postoperative hospital stay.

The 30-day mortality rate of female patients was 13.3% (n = 13) in the unmatched cohort and 12.2% (n = 10) in the matched cohort compared to 16.5% (n = 35) and 17.1% (n = 14) in the unmatched and matched male cohort, respectively. Long-term follow-up was achieved in 98.4% of the original study population and 98.8% of the matched cohort. Median follow-up time was 6.9 years for the original population and 8.2 years for the matched cohort. No significant differences were found concerning 30-day mortality between men and women in both the unmatched cohort (p = 0.463) and the matched cohort (p = 0.377). Similarly, no differences were found concerning long-term survival after open thoracoabdominal aortic repair between men and women (Figure 2). The log-rank test revealed p = 0.750 for the unmatched cohort and p = 0.967 for the matched cohort.

Risk factor analysis was performed using multivariate binomial logistic regression models for both female and male patients of the initial study population. Hosmer–Lemeshow testing showed sufficient goodness of fit for the models of the male and female cohorts with p > 0.05 for both groups. Both models were statistically significant with p < 0.001 for each model. With Nagelkerke's R² of 0.534, a high amount of explained variance was found for the model of the female subgroup, and with Nagelkerke's R² of 0.257, an acceptable amount of explained variance was observed in the model for the male subgroup. In the combined model of the initial study population, the correlation between the key predictor variables and patients sex was <0.8, indicating that confounding interactions between sex and covariates were unlikely.

Concerning the predicting variables of in-hospital mortality of women or men undergoing open thoracoabdominal repair, different parameters were identified as significant predictors for early mortality in female versus male patients (Tables 4A,B). In male patients, cardiac preoperations and the need for urgent repair were identified as risk factors with odds ratios of 2.9 (p = 0.022) and 7.3 (p = 0.001), respectively. Of the patient inherent characteristics and comorbidity burden, only hyperlipidemia was a significant risk factor with an odds ratio of 1.5 (p = 0.006). In contrast, female patients undergoing thoracoabdominal aortic repair exhibited a completely different risk profile compared to males. While the predicting parameters of the male subgroup did not reach statistical significance for prediction of early mortality in the female subgroup, patient inherent comorbidity burden was more predictive for in-hospital mortality among women. Here, hypertension, chronic kidney disease, and the history of coronary artery disease were found to be predictive factors for in-hospital mortality. In contrast to the male subgroup, the age at operation was also found to be a risk factor for early mortality in females.

As the underlying study is a retrospective single-center analysis of a highly specialized operation, some inherent limitations apply. The absolute numbers of events per variable in the sex-specific risk factor analysis was relatively low, although the relative event rates per variable were comparatively high. The limited absolute numbers of events per variable result from the limited number of cases included in the study. This is due to the generally very selective and highly specialized nature of the operation, which is performed relatively infrequently, even at a high-volume center such as ours. Moreover, the cause of death in long-term follow-up, especially aortic versus non-aortic death, the and possible contribution of the covariates on overall mortality are unknown in a relevant proportion of the study population, since the vast majority of death occurred in the patients’ homes or peripheral hospitals, where these data are often not obtained or are unavailable.