Abstract
Background
Sex-based differences in outcomes after oncological surgery remain underexplored despite growing interest in personalized and equitable cancer care. Sex differences in in-hospital mortality and failure-to-rescue rates after major gastrointestinal cancer surgery were analysed using nationwide real-world data.
Methods
German Diagnosis Related Groups discharge data from 2010–2023 were analysed, including adult patients undergoing curative-intent resections for colorectal, gastric, pancreatic, and oesophageal cancers. Multivariable generalized linear models were applied to estimate adjusted sex-specific risks of in-hospital mortality and failure to rescue.
Results
Among 870 754 patients (376 883, 43.3% women), the overall in-hospital mortality rate was 5.9%. Women had lower adjusted odds of death (20 343 (5.4%) versus 31 265 (6.3%); adjusted odds ratio 0.82, 95% confidence interval 0.81 to 0.84) and failure to rescue (18 097 of 151 616 (11.9%) versus 28 583 of 231 245 (12.4%); adjusted odds ratio 0.92, 0.89 to 0.94) compared with men. The mortality advantage for women was consistent across cancer sites except for oesophageal resection. Temporal analyses showed a continuous decline in mortality and failure to rescue for both sexes over the 14-year study period.
Conclusion
Female sex was independently associated with lower in-hospital mortality and failure-to-rescue rates after major gastrointestinal cancer surgery in routine clinical practice. Male sex was associated with a higher risk of adverse short-term outcomes, identifying men as a higher-risk group in routine oncological surgical care. Although causal mechanisms cannot be inferred, the results suggest the need for sex-informed perioperative management and rescue strategies to improve the quality of oncological surgical care.
Keywords: oncological surgery, quality of health care, health inequities, sex factors, digestive system surgical procedures, treatment outcome
This population-based study analysed sex differences in mortality and failure-to-rescue rates among patients undergoing gastrointestinal cancer surgery in Germany. In this cohort of 870 754 patients, women had a significantly lower adjusted in-hospital mortality rate (5.4 versus 6.3%) and failure-to-rescue rate (11.9% versus 12.4%) than men, except in oesophageal resections.
Introduction
Sex-based differences in cancer incidence, biology, treatment, and outcomes are increasingly being recognized across multiple tumour entities1–4. Little is known about how sex influences postoperative outcomes after major cancer surgery, a critical component of multidisciplinary oncological care. As surgical resection remains the cornerstone of curative treatment for most gastrointestinal malignancies, understanding sex-related risk patterns in perioperative mortality and failure to rescue (FTR) as one possible underlying mechanism is essential for improving quality of care and patient safety.
Although biological mechanisms such as hormonal modulation, immune response, and cardiometabolic profiles have been proposed to explain sex differences in cancer outcomes, system-level factors, including care pathways, complication recognition, and rescue processes, may also contribute. The concept of FTR, defined as death following a major complication, provides a sensitive measure of hospital performance and healthcare system responsiveness5,6. Yet, evidence regarding sex-specific disparities in FTR within oncological surgery remains scarce, particularly from large, representative populations.
Real-world evidence derived from administrative and registry data enables the investigation of outcome disparities at the population level, complementing randomized trials that often under-represent women and older patients. The present nationwide study aimed to assess sex-based differences in in-hospital mortality and FTR after major gastrointestinal cancer surgery in Germany over a 14-year period (2010–2023), using comprehensive hospital discharge data and adjusting for co-morbidities, hospital volume, and temporal trends. It was hypothesized that women would experience lower mortality and FTR rates, reflecting both biological and systemic determinants of sex-specific vulnerability in oncological care.
Methods
Study design
A retrospective, population-based cohort study was conducted using anonymized nationwide administrative data from German Diagnosis Related Groups (DRG) statistics, covering all inpatient hospital admission between 1 January 2010 and 31 December 2023. The DRG data set includes mandatory reporting of patient demographics, admission characteristics, diagnoses (International Classification of Diseases, 10th revision—German Modification, ICD-10-GM), procedures (Operationen- und Prozedurenschlüssel (OPS) codes; German modification of the International Classification of Procedures in Medicine), co-morbidities, length of hospital stay, and discharge outcomes. All German hospitals are legally required to contribute to this system, ensuring complete national coverage and standardized coding (source: Research Data Centre of the Federal Statistical Office and Statistical Offices of the Federal States). For more information see the Digital Objective Identifier (DOI) linked with the data set (DOI 10.21242/23141.2010.00.00.6.1.0).
The study adhered to the European Society for Medical Oncology Guidance for Reporting Oncology real-World evidence, emphasizing transparent data provenance, population definition, confounding control, and reproducibility7. The analysis was registered at ClinicalTrials.gov (NCT07017712).
Study population
Adult patients (aged ≥ 18 years) who underwent major surgical resection with curative intent for one of the following gastrointestinal cancers were included: colon, rectal, gastric, pancreatic, or oesophageal malignancies. For oesophageal cancer, ‘complex oesophageal surgery’ referred to major oncological oesophagectomy procedures and excluded endoscopic resections, local ablative therapies, and palliative or bypass procedures. Relevant OPS procedure codes and ICD-10-GM diagnosis codes were used to identify eligible patients; detailed code lists are provided in Table S1. Patients with incomplete demographic information (sex or age), benign disease, procedures for metastatic disease alone, or secondary admissions were excluded to ensure cohort consistency. A procedure with curative intent was defined operationally as performance of a major oncological resection typically undertaken with curative intent for the respective gastrointestinal cancer, based on procedure codes. Patients undergoing non-resectional procedures, metastasis-only interventions, bypass procedures, palliative operations, or secondary admissions were excluded. As clinical intent, margin status, and multidisciplinary treatment decisions are not available in the administrative data set, whether the procedure was done with curative intent was inferred from the type of surgical procedure performed.
Variables and definitions
The primary outcomes were in-hospital mortality (defined as death occurring at any time during the index hospital admission) and FTR, defined as death following at least one major postoperative complication during the same hospital stay. FTR rates were calculated only among patients with recorded complications. Thirty-day postoperative mortality rates, complication rates, intensive care unit (ICU) admission, reoperation within the hospital stay, and patients for whom major complications resulted in extended medical treatment (ICU admission, reoperation, and prolonged hospital stay) were analysed ss secondary endpoints. These definitions were chosen because postdischarge information and cause-specific linkage are not available in the DRG data set.
Complications were identified using validated ICD-10-GM codes encompassing surgical, medical, and infectious adverse events (Table S1). Explanatory variables included sex, age, admission type (elective versus emergency), co-morbidity burden, procedure-specific hospital volume (categorized into quartiles), and calendar year. Sex was recorded as a binary biological variable (female versus male) as coded in the administrative data set. Patients with missing or non-binary sex entries were excluded from the analysis as there were very small numbers and lack of consistent coding. Information on gender identity, gender roles, transgender status, intersex variations, or other sociocultural factors was not available. Emergency admission reflects an unplanned or urgent hospital admission and does not necessarily imply immediate emergency surgery. Co-morbidity burden was assessed using the Elixhauser co-morbidity system, which comprises 31 predefined conditions identified from ICD-10-GM codes and has been widely validated for risk adjustment in administrative data by capturing a broad spectrum of chronic and acute diseases8. The Elixhauser category ‘liver disease’ includes both mild and severe hepatic disorders (for example chronic hepatitis, cirrhosis, hepatic failure), reflecting varying degrees of impaired liver function. Definitive oncological resection may occur after clinical stabilization during the same hospital stay. The DRG data set does not contain reliable information on tumour stage, histopathological characteristics, or neoadjuvant therapy. Diagnosis codes reflect the presence of malignancy but do not allow standardized staging across tumour entities. Prolonged length of stay was defined as a hospital stay exceeding that observed for 75% of patients with the same diagnosis and resection procedure.
Statistical analysis
Descriptive analyses were used to compare baseline characteristics between women and men. Categorical variables are summarized as counts and percentages, and continuous variables as median (interquartile range) or mean(standard deviation, s.d.). Differences were assessed using the χ2 or Mann–Whitney U test, as appropriate. Sex-related differences in the various endpoints were determined by bivariable analysis, and expressed as absolute and relative risk differences.
To take possible baseline differences between patients into account, risk-adjusted sex-specific differences were estimated using generalized linear models (GLMs) with binomial distribution and logit link, providing adjusted odds ratios (ORs) and 95% confidence intervals. Co-variables included age, admission type, individual Elixhauser co-morbidities, year of surgery, and hospital volume quartile. Adjustment for calendar year was included in all multivariable models to account, in part, for temporal trends in outcomes and potential changes in coding practices over the study period. Regression models were developed for the complete study cohort and stratified by gastrointestinal cancer type, separately for each outcome. Adjusted ORs and adjusted risk differences were calculated with 95% confidence intervals. In previous analyses of this data set, hospital-level clustering was assessed using intraclass correlation coefficients (ICCs) and found to be minimal (ICC < 0.05), which is why unnecessarily complicated statistical models that impair interpretability were avoided.
To mitigate the risk of bias, multiple subgroup analyses were undertaken by restricting the analysis to elective procedures, stratifying by hospital volume, and stratifying according to the number of co-morbidities (≤ 3 versus ≥ 4). Time trends in mortality and FTR were evaluated using calendar year as a continuous variable in separate GLMs to estimate annual percentage change. All analyses were performed using R version 4.3.0 (R Foundation for Statistical Computing, Vienna, Austria). Two-sided P < 0.050 was considered statistically significant.
Ethical and legal considerations
This study used fully anonymized secondary data provided by the German Federal Statistical Office under national data protection regulations (§21 KHEntgG). No patient consent or institutional ethics approval was required according to German law. The study followed the Declaration of Helsinki and complied with all relevant European data governance frameworks. The results were reported according to STROBE guidelines9.
Results
Patient characteristics
A total of 870 754 patients were included from January 2010 to December 2023. The proportion of patients excluded owing to missing data was < 0.1% (212 patients). More men underwent surgery for gastrointestinal cancer than women (493 871 (56.7%) versus 376 883 (43.3%). This trend was consistent across individual resection procedures, with men representing 51.6% of colon resections (233 822 patients) and up to 81.1% of oesophageal procedures (37 257 patients) (Table S2). Women had a shorter hospital stay (mean(s.d.) 20.5(14.7) versus 21.1(16.3) days) and were less likely to require postoperative ventilation (40 900 (10.9%) versus 76 039 (15.4%)). Further differences were observed between men and women in terms of type of admission, number of Elixhauser co-morbidities, and the most common co-morbidities (Table 1).
Table 1.
Patient characteristics
| Total (n = 870 754) | Sex | P† | ||
|---|---|---|---|---|
| Men (n = 493 871) | Women (n = 376 883) | |||
| Age (years) | ||||
| Mean(s.d.) | 70.0(11.9) | 69.0(11.4) | 71.2(12.4) | < 0.001‡ |
| ≥ 80 | 196 468 (22.6%) | 90 368 (18.3%) | 106 100 (28.2%) | < 0.001 |
| Procedure | < 0.001 | |||
| Complex oesophageal surgery | 45 917 (5.3%) | 37 257 (7.5%) | 8660 (2.3%) | |
| Colon resection | 452 796 (52.0%) | 233 822 (47.3%) | 218 974 (58.1%) | |
| Rectal resection | 219 316 (25.2%) | 137 309 (27.8%) | 82 007 (21.8%) | |
| Gastric resection | 67 077 (7.7%) | 40 295 (8.2%) | 26 782 (7.1%) | |
| Pancreatic resection | 85 648 (9.8%) | 45 188 (9.1%) | 40 460 (10.7%) | |
| Length of hospital stay (days), mean(s.d.) | 20.9(15.6) | 21.1(16.3) | 20.5(14.7) | < 0.001‡ |
| ICU ventilation | ||||
| No. with ventilation | 116 939 (13.4%) | 76 039 (15.4%) | 40 900 (10.9%) | < 0.001 |
| Duration (hours), median (i.q.r.) | 50 (13–203) | 57 (15–235) | 43.0 (11–157) | < 0.001‡ |
| Type of admission | < 0.001 | |||
| Elective | 644 567 (74.0%) | 372 031 (75.3%) | 272 536 (72.3%) | |
| Emergency | 226 187 (26.0%) | 121 840 (24.7%) | 104 347 (27.7%) | |
| No. of Elixhauser co-morbidities | < 0.001 | |||
| 0 | 73 366 (8.4%) | 42 835 (8.7%) | 30 531 (8.1%) | |
| 1 | 139 671 (16.0%) | 80 769 (16.4%) | 58 902 (15.6%) | |
| 2 | 167 915 (19.3%) | 95 213 (19.3%) | 72 702 (19.3%) | |
| > 2 | 489 802 (56.3%) | 275 054 (55.7%) | 214 748 (57.0%) | |
| Co-morbidity* | ||||
| Hypertension, uncomplicated | 464 778 (53.4%) | 264 062 (53.5%) | 200 716 (53.3%) | 0.051 |
| Fluid and electrolyte disorder | 379 746 (43.6%) | 198 886 (40.3%) | 180 860 (48.0%) | < 0.001 |
| Metastatic cancer | 298 507 (34.3%) | 168 331 (34.1%) | 130 176 (34.5%) | < 0.001 |
| Cardiac arrhythmia | 174 216 (20.0%) | 105 495 (21.4%) | 68 721 (18.2%) | < 0.001 |
| Diabetes, uncomplicated | 171 323 (19.7%) | 104 160 (21.1%) | 67 163 (17.8%) | < 0.001 |
| Coagulopathy | 112 918 (13.0%) | 69 806 (14.1%) | 43 112 (11.4%) | < 0.001 |
| Renal failure | 101 443 (11.7%) | 58 791 (11.9%) | 42 652 (11.3%) | < 0.001 |
| Hypothyroidism | 100 671 (11.6%) | 31 326 (6.3%) | 69 345 (18.4%) | < 0.001 |
| Congestive heart failure | 98 293 (11.3%) | 55 910 (11.3%) | 42 383 (11.2%) | 0.273 |
| Peripheral vascular disorder | 96 628 (11.1%) | 65 807 (13.3%) | 30 821 (8.2%) | < 0.001 |
Values are n (%) unless otherwise stated. *Only the ten most common co-morbid conditions in the patient cohort are reported. s.d., standard deviation; ICU, intensive care unit; i.q.r., interquartile range. †χ2 text, except ‡Mann–Whitney U test.
Surgical outcomes
The overall in-hospital mortality rate of the study cohort was 5.9% (n = 51 608). Mortality rates were higher among men for four of the five cancer types analysed, ranging from a 0.6% higher mortality rate in rectal resections (3.5% for women versus 4.0% for men; 95% confidence interval (c.i.) 0.39 to 0.71) up to a 1.8% higher mortality rate in pancreatic resections (7.4 versus 9.2%; 95% c.i. 1.5 to 2.2) (Fig. 1, Tables S3 and S4). Women had a higher mortality rate only in complex oesophageal surgery (mortality difference 0.7 (95% c.i. 0.1 to 1.3)%). Although the mortality differences appeared small, they became more apparent if set in relation to the absolute mortality rate of the comparator group and therefore expressed as relative risk (Fig. 1). The respective values for pancreatic surgery were 24.9 (19.4 to 30.6)% higher risk of in-hospital death for men compared with women. Women exhibited an increased risk of death specifically in oesophageal carcinoma resections. Overall mortality refers to in-hospital mortality and is therefore higher than 30-day mortality, reflecting prolonged postoperative hospital stays after major gastrointestinal cancer surgery.
Fig. 1.
Observed surgical outcome rates in gastrointestinal cancer surgery and relative risk in men and women for relevant complications and mortality (870 754 patients, 2010–2023)
Surgical outcome rates are shown separately for men and women. Values in parentheses are the relative risk for the group with the higher outcome rate (higher-risk group); relative risk values have been omitted when differences between groups were not statistically significant in univariable analysis. ICU, intensive care unit; > 75% LOS, length of hospital stay above 75th percentile for the given surgical procedure. FTR, failure to rescue.
Regarding the second primary outcome, mortality after major complication (FTR rate), the results roughly followed a similar pattern as observed for the first primary outcome. Among the 382 861 patients who developed postoperative complications, 47 680 (12.4%) died during the in-hospital stay, with lower rates in woman than in man (18 097 of 151 616 (11.9%) versus 28 583 of 231 245 (12.4%)). The highest sex difference was observed in gastric resections with an increased relative risk of death after a major complication of 12.7 (6.2 to 16.9)% in men.
The incidence of complications (all and major) was significantly higher among men for four of the five cancer conditions analysed (all P < 0.001), with the exception of oesophageal cancer (any complication, P = 0.765; major complication, P = 0.838). The occurrence of major complications was associated more frequently with extended medical treatment in men (ICU admission, reoperation, and prolonged hospital stay; all P < 0.001).
Death after major complications, with or without extended medical treatment, was more common in men who had gastric and pancreatic cancer resections (all P < 0.001), and, to a lesser degree, also in colon cancer resections (P = 0.024).
Multivariable logistic regression models for mortality and FTR
In multiple logistic regression analysis, sex (P < 0.001), age (P < 0.001), emergency admission (P < 0.001), type of gastrointestinal cancer resection (P < 0.001), and presence of liver disease (adjusted OR 15.01, 95% c.i. 14.37 to 15.69), coagulation disorder (adjusted OR 3.07, 3.00 to 3.14), heart failure (adjusted OR 1.90, 1.86 to 1.95), and neurological disorder (adjusted OR 1.83, 1.76 to 1.91), increased the risk of death (Table 2, Fig. 2, and Table S4). Surgery in high-volume hospitals decreased the odds of death (P < 0.001).
Table 2.
Multivariable logistic regression models for mortality and failure to rescue
| Mortality | Failure to rescue | |||
|---|---|---|---|---|
| Odds ratio | P | Odds ratio | P | |
| Sex | ||||
| Male | 1.00 (reference) | 1.00 (reference) | ||
| Female | 0.82 (0.81, 0.84) | < 0.001 | 0.92 (0.89, 0.94) | < 0.001 |
| Age (years) | ||||
| < 30 | 1.00 (reference) | 1.00 (reference) | ||
| 30–39 | 1.20 (0.69, 2.22) | 0.542 | 1.09 (0.62, 2.04) | 0.773 |
| 40–49 | 2.00 (1.22, 3.57) | 0.010 | 1.75 (1.05, 3.14) | 0.044 |
| 50–59 | 3.15 (1.93, 5.57) | < 0.001 | 2.69 (1.63, 4.80) | < 0.001 |
| 60–69 | 5.00 (3.07, 8.84) | < 0.001 | 4.19 (2.55, 7.48) | < 0.001 |
| 70–79 | 7.83 (4.81, 13.84) | < 0.001 | 6.37 (3.88, 11.37) | < 0.001 |
| > 79 | 14.87 (9.14, 26.30) | < 0.001 | 11.32 (6.89, 20.19) | < 0.001 |
| Type of admission | ||||
| Elective | 1.00 (reference) | 1.00 (reference) | ||
| Emergency | 1.61 (1.58, 1.65) | < 0.001 | 1.40 (1.37, 1.43) | < 0.001 |
| Procedure | ||||
| Rectal resection | 1.00 (reference) | 1.00 (reference) | ||
| Complex oesophageal surgery | 2.14 (2.05, 2.23) | < 0.001 | 1.96 (1.87, 2.05) | < 0.001 |
| Colon resection | 1.27 (1.23, 1.30) | < 0.001 | 1.35 (1.31, 1.39) | < 0.001 |
| Gastric resection | 1.57 (1.52, 1.63) | < 0.001 | 1.69 (1.63, 1.76) | < 0.001 |
| Pancreatic resection | 1.64 (1.59, 1.70) | < 0.001 | 1.62 (1.56, 1.68) | < 0.001 |
| Co-morbidity* | ||||
| Liver disease | 15.01 (14.37, 15.69) | < 0.001 | 12.41 (11.84, 13.02) | < 0.001 |
| Coagulopathy | 3.07 (3.00, 3.14) | < 0.001 | 2.38 (2.33, 2.44) | < 0.001 |
| Congestive heart failure | 1.90 (1.86, 1.95) | < 0.001 | 1.62 (1.58, 1.66) | < 0.001 |
| Other neurological disorder | 1.83 (1.76, 1.91) | < 0.001 | 1.66 (1.59, 1.74) | < 0.001 |
| Cardiac arrhythmia | 1.73 (1.69, 1.77) | < 0.001 | 1.59 (1.55, 1.63) | < 0.001 |
| Fluid and electrolyte disorder | 1.56 (1.53, 1.60) | < 0.001 | 1.21 (1.18, 1.24) | < 0.001 |
| Alcohol abuse | 1.55 (1.46, 1.66) | < 0.001 | 1.39 (1.30, 1.49) | < 0.001 |
| Psychoses | 1.53 (1.33, 1.74) | < 0.001 | 1.24 (1.07, 1.43) | 0.003 |
| Chronic pulmonary disease | 1.35 (1.31, 1.39) | < 0.001 | 1.23 (1.20, 1.27) | < 0.001 |
| Cerebrovascular disease | 1.33 (1.27, 1.39 | < 0.001 | 1.27 (1.21, 1.34 | < 0.001 |
| Mean annual hospital procedure volume | ||||
| < 20 | 1.00 (reference) | 1.00 (reference) | ||
| 20–49 | 0.92 (0.88, 0.96) | < 0.001 | 0.94 (0.89, 0.98) | < 0.001 |
| > 50 | 0.84 (0.80, 0.89) | < 0.001 | 0.88 (0.83, 0.92) | < 0.001 |
| Year | ||||
| 2010 | 1.00 (reference) | 1.00 (reference) | ||
| 2011 | 0.95 (0.91, 1.00) | 0.038 | 0.94 (0.89, 0.99) | 0.031 |
| 2012 | 0.85 (0.81, 0.89) | < 0.001 | 0.85 (0.80, 0.89) | < 0.001 |
| 2013 | 0.84 (0.80, 0.88) | < 0.001 | 0.81 (0.77, 0.86) | < 0.001 |
| 2014 | 0.78 (0.74, 0.82) | < 0.001 | 0.76 (0.72, 0.80) | < 0.001 |
| 2015 | 0.79 (0.75, 0.83) | < 0.001 | 0.77 (0.73, 0.81) | < 0.001 |
| 2016 | 0.77 (0.73, 0.80) | < 0.001 | 0.73 (0.69, 0.77) | < 0.001 |
| 2017 | 0.74 (0.70, 0.77) | < 0.001 | 0.70 (0.66, 0.74) | < 0.001 |
| 2018 | 0.72 (0.68, 0.76) | < 0.001 | 0.69 (0.65, 0.73) | < 0.001 |
| 2019 | 0.71 (0.68, 0.75) | < 0.001 | 0.67 (0.64, 0.71) | < 0.001 |
| 2020 | 0.68 (0.65, 0.72) | < 0.001 | 0.65 (0.61, 0.69) | < 0.001 |
| 2021 | 0.68 (0.64, 0.72) | < 0.001 | 0.65 (0.62, 0.69) | < 0.001 |
| 2022 | 0.72 (0.68, 0.76) | < 0.001 | 0.69 (0.65, 0.73) | < 0.001 |
| 2023 | 0.72 (0.69, 0.76) | < 0.001 | 0.69 (0.65, 0.73 | < 0.001 |
Values in parentheses are 95% confidence intervals. *Ten most relevant co-morbid conditions.
Fig. 2.
Risk-adjusted mortality and FTR rates in patients undergoing gastrointestinal cancer surgery, stratified by procedure and sex (870 754 patients, 2010–2023)
Risk adjustment was undertaken separately for each surgical procedure and accounted for age, co-morbidities, year of surgery, admission type, and hospital mean annual procedure volume. FTR, failure to rescue. *P < 0.050, †P < 0.001 (P values were derived from average marginal comparisons based on multivariable logistic regression models.).
The same risk and protective factors as for death were identified for the occurrence of FTR.
Adjusted sex-differences in surgical outcomes
After adjustment for age, co-morbidities, year of surgery, admission type, and hospital volume, the sex-related differences in postoperative outcomes largely persisted, but were moderately attenuated in magnitude (Fig. 3 and Table S4).
Fig. 3.
Adjusted surgical outcome differences between men and women in gastrointestinal cancer surgery for relevant complications and mortality (870 754 patients, 2010–2023)
Risk-adjusted differences between men and women are shown with 95% confidence intervals. Negative values indicate a lower proportion in women than men. The intensity of the colours indicates the significance (darker) or non-significance (lighter) of the risk-adjusted difference. Risk adjustment was undertaken separately for each procedure using a generalized linear model, incorporating age, individual co-morbidities, year of surgery, admission type, and hospital mean annual procedure volume. ICU, intensive care unit; > 75% LOS, length of hospital stay above 75th percentile for the given surgical procedure; FTR, failure to rescue.
Overall and 30-day mortality rates remained lower among women across most cancer types. Adjusted absolute mortality differences ranged from −0.8 (95% c.i. −1.0 to −0.7)% in rectal resections to −1.0 (−1.4 to −0.7)% in pancreatic resections. Women had higher mortality rates following complex oesophageal surgery, with an adjusted difference of +1.0 (0.4 to 1.6)%. Adjusted absolute FTR differences reached up to −1.0 (−1.6 to −0.5)% in pancreatic resections. Women exhibited a higher adjusted risk of death after major complications in oesophageal cancer surgery (adjusted difference +1.9 (0.8 to 3.0)%).
The adjusted differences for any complication ranged from −4.9% (pancreatic resection) to −7.3% (rectal resection), and for major complications from −4.9 to −6.7%, with no excess risk observed in women with oesophageal cancer. Extended medical treatment also remained less frequent among women (P < 0.001), although adjustment reduced the gap substantially.
Discussion
In this analysis, women consistently demonstrated lower in-hospital mortality and FTR rates than men for all major gastrointestinal cancer procedures, except oesophageal cancer resections. This finding persisted after comprehensive risk adjustment and across most tumour sites. Male sex remained associated with higher postoperative mortality and complication rates, whereas a disadvantage for women persisted in complex oesophageal surgery.
Only a limited number of studies have analysed sex differences in hospital mortality and FTR rates after gastrointestinal cancer surgery, with heterogeneous findings1,10. Several studies11,12 reported less favourable outcomes in men, including higher complication and mortality rates after minimally invasive gastrectomy. Raised mortality and reoperation rates among men were also observed after colorectal cancer surgery, but not procedures for pancreatic and oesophageal cancers13. Other studies14–16 described higher complication rates, increased 30-day mortality, and higher FTR rates in women. Similar patterns have been reported in cardiac and vascular surgery3.
The findings indicate that sex-based differences in surgical outcomes are procedure-specific and influenced by biological, behavioural, and healthcare-related factors. Lower mortality and FTR rates in rectal cancer resections may reflect lower co-morbidity burden, fewer emergency admissions, and younger patient age compared with colon cancer cohorts. In contrast, higher risks in oesophageal resections highlight heterogeneity across tumour types. The high proportion of emergency admissions for oesophageal and pancreatic resections likely reflects acute deterioration resulting from tumour-related complications (for example obstruction, bleeding, or sepsis), requiring urgent hospital admission, whereas definitive oncological resection often follows clinical stabilization during the same admission rather than immediate emergency surgery. Evidence from cardiac and vascular surgery suggests that complication rates are comparable between the sexes but that the FTR rate is higher among women3, whereas both complication and FTR rates were higher in men in the present cohort No consistent disadvantage for one sex across all procedures was observed.
Persistent differences after adjustment may reflect biological mechanisms. Sex-related variation in genetics, hormonal profiles, metabolism, immune response, and anatomy may confer a protective effect in women1,17. Experimental evidence indicates less pronounced immunosuppression after trauma in women, linked to hormonal regulation of immune function18. Anti-inflammatory hormonal effects may contribute to lower complication rates19. Behavioural and psychosocial factors are more accurately conceptualized as gender-related and cannot be assessed in the present data set20.
Higher FTR rates in men may reflect both biological vulnerability and differences in recognition and management of complications. As these were preoperative patient characteristics, this difference in the FTR model also reflects the additional risk associated with the occurrence of complications. Lower healthcare-seeking behaviour among men may contribute to more advanced tumour stages at presentation21,22 and this, along with the neoadjuvant therapy, is an important determinant of postoperative risk and may differ by sex. The existing literature reports heterogeneous findings. In bladder and rectal cancer, women often present with more advanced stages of disease than men17,23. For gastric cancer, the situation is reversed24.
FTR was defined using in-hospital mortality and may underestimate postoperative mortality when deaths occur after discharge. Sex differences in discharge patterns, hospice referral, and location of death may influence capture of postoperative deaths25,26. Directional consistency between in-hospital and 30-day mortality supports the robustness of the findings.
The signal towards worse outcomes in women undergoing oesophageal resections contrasts with otherwise favourable outcomes for women. The low proportion of women and lack of consistent statistical significance suggest that this finding is hypothesis-generating.
Co-morbidities such as liver disease, coagulopathy, and cardiac conditions were strongly associated with increased mortality and FTR, indicating the multifactorial nature of postoperative risk.
Strengths of this study include the large nationwide cohort and the ability to undertake procedure-specific risk adjustment. Differentiating between postoperative complications and FTR allows assessment of system response to adverse events. Limitations include potential coding inaccuracies and lack of clinical detail, including data on tumour stage, performance status, and treatment regimens. Hospital- and surgeon-level factors were not available. Residual confounding cannot be excluded, and variation in coding practices over time may have introduced bias.
Because FTR was based on in-hospital mortality, postoperative deaths occurring after discharge were not captured, so sex-based differences in discharge practices, hospice use, or location of death may have introduced differential misclassification. Because of the absence of postdischarge data, 90-day mortality could not be assessed. Although behavioural and healthcare system factors have been discussed as potential explanations for the observed differences, the data set captures biological sex only. The inability to identify transgender or intersex patients, gender-related variables, and sociocultural determinants represents an important limitation, as these populations may experience disproportionate healthcare disparities and adverse outcomes.
Whether surgery was undertaken with curative intent was inferred from procedure coding rather than clinical decision-making, so some degree of misclassification cannot be excluded. Causality cannot be inferred as in prospective randomized studies owing to the retrospective observational design and the absence of data on escalation-of-care processes and clinical decision-making pathways. Nonetheless, previously validated and established methods for risk adjustment were applied27,28.
The results highlight the importance of integrating sex as a critical variable in perioperative risk stratification and quality improvement efforts. The present findings may inform future hypothesis-driven research and the development of tailored quality improvement strategies to reduce FTR and mortality rates, without implying modifiable causal mechanisms. Future research should incorporate prospective data collection with detailed clinical variables, and examine institutional and sociocultural factors contributing to sex-based disparities. Interdisciplinary strategies, including tailored care pathways and early complication detection protocols, may help reduce outcome differences.
Supplementary Material
Acknowledgements
U.W. and R.H. are statisticians.
Contributor Information
Rene Mantke, Department of General Surgery, University Hospital Brandenburg, Brandenburg Medical School Theodor Fontane, Brandenburg, Germany; Faculty of Health Sciences Brandenburg, Brandenburg Medical School Theodor Fontane, Brandenburg, Germany.
Ulrike Weber, Department of General Surgery, University Hospital Brandenburg, Brandenburg Medical School Theodor Fontane, Brandenburg, Germany.
Markus W Büchler, Botton-Champalimaud Pancreatic Cancer Centre, Champalimaud Foundation, Lisbon, Portugal; Department of General, Visceral and Transplant Surgery, Heidelberg University Hospital, Heidelberg, Germany.
Richard Hunger, Department of General Surgery, University Hospital Brandenburg, Brandenburg Medical School Theodor Fontane, Brandenburg, Germany.
Funding
Funded by the Medical School Brandenburg (MHB) publication fund supported by German Research Foundation (DFG).
Author contributions
René Mantke (Conceptualization, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing—review & editing), Ulrike Weber (Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing—original draft), Markus Büchler (Conceptualization, Investigation, Methodology, Project administration, Resources, Supervision, Writing—review & editing), and Richard Hunger (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing—original draft, Writing—review & editing)
Disclosure
The authors declare no conflict of interest.
Supplementary material
Supplementary material is available at BJS Open online.
Data availability
The primary data are provided for research purposes by the German Federal Statistical Office and may not be made publicly available.
References
- 1. Baraibar I, Ros J, Saoudi N, Salvà F, García A, Castells MR et al. Sex and gender perspectives in colorectal cancer. ESMO Open 2023;8:101204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Haupt S, Caramia F, Klein SL, Rubin JB, Haupt Y. Sex disparities matter in cancer development and therapy. Nat Rev Cancer 2021;21:393–407 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Wagner CM, Joynt Maddox KE, Ailawadi G, Ibrahim AM. Failure to rescue female patients undergoing high-risk surgery. JAMA Surg 2025;160:29–36 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Yang W, Rubin JB. Treating sex and gender differences as a continuous variable can improve precision cancer treatments. Biol Sex Differ 2024;15:35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Ghaferi AA, Birkmeyer JD, Dimick JB. Complications, failure to rescue, and mortality with major inpatient surgery in Medicare patients. Ann Surg 2009;250:1029–1034 [DOI] [PubMed] [Google Scholar]
- 6. Wakeam E, Hyder JA, Lipsitz SR, Cohen ME, Orgill DP, Zinner MJ et al. Hospital-level variation in secondary complications after surgery. Ann Surg 2016;263:493–501 [DOI] [PubMed] [Google Scholar]
- 7. Castelo-Branco L, Pellat A, Martins-Branco D, Valachis A, Derksen JWG, Suijkerbuijk KPM et al. ESMO guidance for reporting oncology real-world evidence (GROW). ESMO Real World Data Digit Oncol 2023;1:100003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005;43:1130–1139 [DOI] [PubMed] [Google Scholar]
- 9. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 2007;335:806–808 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Damanakis AI, Toader J, Wahler I, Plum P, Quaas A, Ernst A et al. Influence of patient sex on outcomes after pancreatic surgery: multicentre study. Br J Surg 2022;109:746–753 [DOI] [PubMed] [Google Scholar]
- 11. Patel D, Sanepara A, Patel NV, Gadhiya D, Dhillon BK, Singh A et al. Sex-based differences in outcomes following minimally invasive gastrectomy for gastric cancer: a nationwide study. J Clin Oncol 2025;43:e16058 [Google Scholar]
- 12. Kommuru S, Ibrahim Y, Ashara YP, Singh K, Shah MP, Shah SD et al. Sex differences in all-cause inpatient mortality risk in gastric cancer: nationwide inpatient population-based study. Cureus 2022;14:e28602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Jochum F, Hamy AS, Gougis P, Dumas É, Grandal B, Sauzey M et al. Sex-related differences in oncological surgery and postoperative outcomes: comprehensive, nationwide study in France. Br J Surg 2024;111:znae179. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Busweiler LA, Henneman D, Dikken JL, Fiocco M, van Berge Henegouwen MI, Wijnhoven BP et al. Failure-to-rescue in patients undergoing surgery for esophageal or gastric cancer. Eur J Surg Oncol 2017;43:1962–1969 [DOI] [PubMed] [Google Scholar]
- 15. Sah BK, Zhu ZG, Wang XY, Yang QM, Chen MM, Xiang M et al. Post-operative complications of gastric cancer surgery: female gender at high risk. Eur J Cancer Care (Engl) 2009;18:202–208 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Yan Q, Kim J, Hall DE, Shinall MC, Reitz KM, Stitzenberg KB et al. Sex-related differences in acuity and postoperative complications, mortality and failure to rescue. J Surg Res 2023;282:34–46 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Doshi B, Athans SR, Woloszynska A. Biological differences underlying sex and gender disparities in bladder cancer: current synopsis and future directions. Oncogenesis 2023;12:44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Choudhry MA, Bland KI, Chaudry IH. Trauma and immune response—effect of gender differences. Injury 2007;38:1382–1391 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Mehdi SF, Pusapati S, Khenhrani RR, Farooqi MS, Sarwar S, Alnasarat A et al. Oxytocin and related peptide hormones: candidate anti-inflammatory therapy in early stages of sepsis. Front Immunol 2022;13:864007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Courtenay WH. Constructions of masculinity and their influence on men’s well-being: a theory of gender and health. Social Science & Medicine 2000;50:1385–1401 [DOI] [PubMed] [Google Scholar]
- 21. Ballering AV, Olde Hartman TC, Verheij R, Rosmalen JGM. Sex and gender differences in primary care help-seeking for common somatic symptoms: a longitudinal study. Scand J Prim Health Care 2023;41:132–139 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Thompson AE, Anisimowicz Y, Miedema B, Hogg W, Wodchis WP, Aubrey-Bassler K. The influence of gender and other patient characteristics on health care-seeking behaviour: a QUALICOPC study. BMC Fam Pract 2016;17:38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Sarasqueta C, Zunzunegui MV, Enríquez Navascues JM, Querejeta A, Placer C, Perales A et al. Gender differences in stage at diagnosis and preoperative radiotherapy in patients with rectal cancer. BMC Cancer 2020;20:759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Li H, Wei Z, Wang C, Chen W, He Y, Zhang C. Gender differences in gastric cancer survival: 99 922 cases based on the SEER database. J Gastrointest Surg 2020;24:1747–1757 [DOI] [PubMed] [Google Scholar]
- 25. Rashid Z, Khalil M, Khan MMM, Altaf A, Munir MM, Woldesenbet S et al. Upper gastrointestinal cancers: trends and determinants of location of death. Surgery 2025;179:108797. [DOI] [PubMed] [Google Scholar]
- 26. Lees CS, Seow H, Chan KKW, Gayowsky A, Merchant SJ, Sinnarajah A. Sex-based analysis of quality indicators of end-of-life care in gastrointestinal malignancies. Curr Oncol 2024;31:1170–1182 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Hunger R, Mantke R. Outcome quality beyond the mean—an analysis of 43 231 pancreatic surgical procedures related to hospital volume. Ann Surg 2022;276:159–166 [DOI] [PubMed] [Google Scholar]
- 28. Mantke R, Seliger B, Ogino S, Büchler MW, Hunger R. Not only caseload but also patient selection is predictive of mortality after pancreatic resection. Ann Surg Open 2025;6:e536. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The primary data are provided for research purposes by the German Federal Statistical Office and may not be made publicly available.