Sepsis and septic shock after breast cancer surgery: risk modeling and machine learning validation across 328,292 cases
Abstract
Postoperative sepsis is a rare but life-threatening complication after breast cancer surgery. Although infection risk in breast surgery is generally low, cancer-related immune dysregulation combined with surgical stress may increase susceptibility, yet large-scale data on incidence, risk factors, and outcomes remain limited. We conducted a retrospective analysis of the ACS-NSQIP database (2008–2022). Adult female patients undergoing breast cancer surgery were included. The primary outcome was ACS-NSQIP-coded 30-day postoperative sepsis; septic shock was analyzed separately and descriptively. Multivariable logistic regression was used to evaluate factors associated with postoperative sepsis. Model performance was assessed by area under the receiver operating characteristic curve (AUC) with bootstrap correction for optimism and compared with machine-learning approaches trained with explicit handling of class imbalance. Among 328,292 patients (47.8% partial, 38.6% simple, 13.6% radical mastectomy), sepsis occurred in 838 (0.26%), with incidence increasing across procedure categories (0.08% vs. 0.40% vs. 0.46%; p < 0.001). Patients with sepsis had markedly higher rates of reoperation (46.5% vs. 4.0%) and unplanned readmission (73.7% vs. 2.1%; both p < 0.001). The strongest adjusted associations with postoperative sepsis were observed for prior sepsis (aOR 4.51), functional dependence (aOR 3.80), and ASA class III (aOR 3.16; all p < 0.001). Reconstruction was also associated with higher adjusted odds of sepsis, with aORs of 2.15 for autologous and 1.59 for implant-based reconstruction. The model showed good discrimination (apparent AUC 0.752; optimism-corrected AUC 0.746) and was well calibrated (bootstrap-corrected calibration slope 0.967). No machine-learning approach outperformed logistic regression on the held-out test set (DeLong p = 0.077–0.409). Sepsis after breast cancer surgery is uncommon but clinically significant, with incidence increasing across procedure categories. Reconstruction, particularly autologous reconstruction, was associated with sepsis after adjustment for measured covariates. Because oncologic and treatment-related variables are not captured in ACS-NSQIP, this association should be interpreted as observational rather than causal. These findings may inform individualized preoperative risk assessment and counseling.
Related
- Development and validation of a machine learning model for preoperative prediction of intraoperative hypothermia in gynecological laparoscopic surgery: a two-center cohort study
- Development and external validation of interpretable machine learning models for 7-, 14-, and 28-day mortality in sepsis-associated liver injury
- An interpretable machine learning model for predicting intensive care unit mortality risk in elderly patients with concomitant cardiogenic shock and sepsis: Development and external validation using two large critical care databases
- Identification and validation of biomarkers related to oxidative stress in pediatric sepsis utilizing WGCNA and multiple machine learning algorithms
- Development and validation of a prognostic prediction model for elderly patients with sepsis-associated acute kidney injury based on interpretable machine learning
BibTeX
@article{jiang2026sepsis,
title = {{Sepsis and septic shock after breast cancer surgery: risk modeling and machine learning validation across 328,292 cases}},
author = {Jun Jiang and Julius M. Wirtz and Merih Gizlenci and Victoria Kong and Kavitha Däster and Constanze Elfgen and Christoph Tausch and Anh Thy Nguyen and Jennifer A. Watson and Lukas Prantl and Philipp Moog and Hans-Guenther Machens and Syena Moltaji and Stefan O. P. Hofer and Florian Falkner and Samuel A. Knoedler},
journal = {Scientific Reports},
year = {2026},
month = oct,
doi = {10.1038/s41598-026-73577-6},
url = {https://www.nature.com/articles/s41598-026-73577-6},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Sepsis and septic shock after breast cancer surgery: risk modeling and machine learning validation across 328,292 cases
AU - Jun Jiang
AU - Julius M. Wirtz
AU - Merih Gizlenci
AU - Victoria Kong
AU - Kavitha Däster
AU - Constanze Elfgen
AU - Christoph Tausch
AU - Anh Thy Nguyen
AU - Jennifer A. Watson
AU - Lukas Prantl
AU - Philipp Moog
AU - Hans-Guenther Machens
AU - Syena Moltaji
AU - Stefan O. P. Hofer
AU - Florian Falkner
AU - Samuel A. Knoedler
PY - 2026
DA - 2026/10/02
JO - Scientific Reports
DO - 10.1038/s41598-026-73577-6
UR - https://www.nature.com/articles/s41598-026-73577-6
AB - Postoperative sepsis is a rare but life-threatening complication after breast cancer surgery. Although infection risk in breast surgery is generally low, cancer-related immune dysregulation combined with surgical stress may increase susceptibility, yet large-scale data on incidence, risk factors, and outcomes remain limited. We conducted a retrospective analysis of the ACS-NSQIP database (2008–2022). Adult female patients undergoing breast cancer surgery were included. The primary outcome was ACS-NSQIP-coded 30-day postoperative sepsis; septic shock was analyzed separately and descriptively. Multivariable logistic regression was used to evaluate factors associated with postoperative sepsis. Model performance was assessed by area under the receiver operating characteristic curve (AUC) with bootstrap correction for optimism and compared with machine-learning approaches trained with explicit handling of class imbalance. Among 328,292 patients (47.8% partial, 38.6% simple, 13.6% radical mastectomy), sepsis occurred in 838 (0.26%), with incidence increasing across procedure categories (0.08% vs. 0.40% vs. 0.46%; p < 0.001). Patients with sepsis had markedly higher rates of reoperation (46.5% vs. 4.0%) and unplanned readmission (73.7% vs. 2.1%; both p < 0.001). The strongest adjusted associations with postoperative sepsis were observed for prior sepsis (aOR 4.51), functional dependence (aOR 3.80), and ASA class III (aOR 3.16; all p < 0.001). Reconstruction was also associated with higher adjusted odds of sepsis, with aORs of 2.15 for autologous and 1.59 for implant-based reconstruction. The model showed good discrimination (apparent AUC 0.752; optimism-corrected AUC 0.746) and was well calibrated (bootstrap-corrected calibration slope 0.967). No machine-learning approach outperformed logistic regression on the held-out test set (DeLong p = 0.077–0.409). Sepsis after breast cancer surgery is uncommon but clinically significant, with incidence increasing across procedure categories. Reconstruction, particularly autologous reconstruction, was associated with sepsis after adjustment for measured covariates. Because oncologic and treatment-related variables are not captured in ACS-NSQIP, this association should be interpreted as observational rather than causal. These findings may inform individualized preoperative risk assessment and counseling.
SN - 2045-2322
ER -