Article Text
PDF
Abstract
Objectives To delineate the clinical characteristics and investigate the determinants that may affect the prognosis of surgical patients with abdominal sepsis.
Design A case-control study was nested in a cohort of surgical patients with abdominal sepsis between 2008 and 2022. We extracted patient’ medical records to execute descriptive statistical analyses. Multiple logistic regression models and subgroup analysis were employed to elucidate the risk factors of in-hospital mortality.
Setting Two tertiary hospitals in China.
Participants 476 surgical patients diagnosed with abdominal sepsis between 2008 and 2022 were analysed.
Interventions None.
Outcome measures Descriptive statistics were used to examine pertinent patient information, including demographic details, laboratory findings, surgical interventions and anaesthetic records. Multivariate logistic regression was used to identify independent risk factors for in-hospital mortality. Subgroup analyses were conducted to explore the impact of specific clinical characteristics on outcomes.
Results 476 patients diagnosed with abdominal sepsis were analysed, exhibiting an in-hospital mortality rate of 7.56%. Advanced age (OR 6.77, 95% CI 2.46 to 18.66, p<0.001), the presence of diabetes (OR 2.61, 95% CI 1.04 to 6.56, p=0.041) and higher preoperative Sequential Organ Failure Assessment (SOFA) score (OR 3.48, 95% CI 1.16 to 10.43, p=0.026) were identified as significant predictors of increased in-hospital mortality risk through a multinomial logistic regression model. Conversely, individuals afflicted with biliary diseases (OR 0.15, 95% CI 0.04 to 0.64, p=0.010) demonstrated a reduced risk of in-hospital mortality. Subgroup analysis revealed that low serum albumin levels emerged as a risk factor for in-hospital mortality in the patients with gastrointestinal diseases (OR 20.23, 95% CI 2.21 to 184.84, p=0.008) or advanced age (OR 10.52, 95% CI 2.29 to 48.31, p=0.002) through multinomial logistic regression analysis.
Conclusion In this retrospective analysis, we delineated the clinical characteristics of surgical patients with abdominal sepsis and pinpointed risk factors associated with in-hospital mortality. These findings underscore the necessity for more tailored perioperative management strategies for patients with sepsis characterised by advanced age, diabetes, higher preoperative SOFA score and reduced preoperative albumin levels. Clinicians should prioritise early recognition and aggressive management of these high-risk individuals, including timely surgical intervention, optimisation of nutritional status and vigilant monitoring of organ function. These insights underscore the critical role of individualised care in enhancing the prognosis of surgical patients with abdominal sepsis.
Trial registration number ChiCTR2400081823.
- risk factors
- mortality
- adult anaesthesia
- gastrointestinal infections
- intensive & critical care
- surgery
Data availability statement
Data are available in a public, open access repository. The data supporting the findings of this study are available at (https://doi-org.ezproxy.u-pec.fr/10.6084/m9.figshare.25523737.v1).
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
The study employs a robust dataset encompassing 15 years from two tertiary hospitals, and comprehensive data collection spans perioperative periods, providing a holistic view of patient management.
Advanced statistical methods, including multivariate logistic regression and subgroup analyses, were employed to identify risk factors systematically.
The use of uniform diagnostic criteria based on Sepsis 3.0 enhances the reliability and comparability of the cohort.
As a retrospective nested case-control study, potential biases include missing data and incomplete follow-up, which may influence the reliability of certain variables.
Introduction
Sepsis, a complex syndrome characterised by life-threatening organ dysfunction due to a dysregulated host response to infection,1 remains a pervasive condition with significant global impact, accounting for approximately 11 million fatalities annually.2 It affects virtually all organs and tissues, leading to diminished tissue perfusion, dysregulated inflammatory responses, immune dysfunction, and coagulation anomalies, culminating in multiple organ injuries.3
Following pulmonary origins, abdominal sepsis ranks as the second most prevalent source of sepsis, presenting a particularly challenging clinical course with notably high prevalence and mortality rates.4 The infection sources in abdominal sepsis, typically classified by organ, encompass peritonitis, cholangitis, diverticulitis, pancreatitis, abdominal abscesses, intestinal perforations, abdominal trauma and pelvic inflammatory disease.5 Compared with other types of sepsis, abdominal sepsis more frequently precipitates a proinflammatory state, systemic vasodilation, hypotension and myocardial depression, thereby manifesting as severe sepsis or septic shock, even during the early phase.4 5
The restoration of adequate organ perfusion, alongside the administration of timely antimicrobial agents and the implementation of source control measures, constitutes the foundational triad in the management of sepsis. Surgical intervention is critically important for timely source control in cases of abdominal sepsis. Surgical strategies predominantly encompass the removal or repair of perforated organs, resection of diseased organs such as the appendix and gallbladder and debridement of necrotic or ischaemic tissues. Despite advancements in surgical techniques and methodologies, the morbidity and mortality rates associated with abdominal sepsis remain distressingly high.6 Research conducted by Sartelli et al has underscored the significant impact of abdominal sepsis on mortality rates, noting an increase from 1.2% in the absence of sepsis to 15.1% when sepsis is present.7 Against this concerning backdrop, this study aims to characterise the clinical profiles and identify prognostic factors influencing outcomes in surgical patients with abdominal sepsis within a dual-centre Chinese cohort.
Materials and methods
Study design and data collection
The clinical dataset for this study was derived from two prominent healthcare institutions: the First Medical Center of the Chinese People’s Liberation Army (PLA) General Hospital, which boasts 4000 patient beds, and the PLA Air Force Medical Center, equipped with 1500 patient beds. We conducted a retrospective case-control study that was nested within a cohort of patients with abdominal sepsis who underwent surgical interventions in the operating room under general anaesthesia between January 2008 and December 2022.8 Data collection was executed in a standardised fashion, adhering to stringent definitions to ascertain data integrity. The conduct and dissemination of this research adhered to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for cohort studies.
This nested design ensures temporal validity, as all exposures were recorded prior to outcome ascertainment, mitigating recall bias and preserving causal inference. Conditional logistic regression--the standard analytical method for nested case-control studies--was appropriately used to calculate ORs, reflecting our focus on differentiating risk profiles between outcome groups rather than estimating relative risks.
Patient and public involvement
Patients and/or the public were not involved in the design, conduct, reporting, or dissemination plans of this research.
Definitions
Cases were retrospectively assessed using the Sepsis 3.0 criteria, defined by concurrent infection and a change in baseline of the Sequential Organ Failure Assessment (SOFA) score of 2 points or more to represent organ dysfunction. The baseline score was presumed to be zero, except in instances where patients were documented to have pre-existing organ dysfunction prior to infection onset.1 Concurrent infection was determined through an evaluation of patient symptoms and signs, diagnostic imaging, microbiological studies and host factors. In ambiguous situations, cases were meticulously reviewed to ascertain the presence of abdominal infection, with physicians re-evaluating diagnoses based on the clinical data at hand. The perioperative conditions of patients with abdominal sepsis were meticulously assessed and analysed, drawing on medical records, radiological findings, blood analyses and other pertinent diagnostic evaluations.
Selection of patients with abdominal sepsis
Given the retrospective nature of this study, potential biases, such as selection bias, information bias and confounding, were considered. To minimise selection bias, we employed strict inclusion and exclusion criteria, ensuring that only patients who met the Sepsis 3.0 diagnostic criteria for abdominal sepsis and underwent surgical intervention were included.
Participants in the study were selected from the database based on the following criteria: (1) adult patients diagnosed with abdominal sepsis; (2) patients undergoing either emergency or elective abdominal surgery under general anaesthesia. The inclusion criteria were established through a search using the keywords: ‘septic shock, infectious peritonitis, abdominal infection, intestinal fistula, biliary fistula, pancreatic fistula, sepsis, digestive tract perforation, bowel perforation, gastric perforation, appendicular perforation, suppurative cholangitis, suppurative appendicitis, suppurative cholecystitis, suppurative peritonitis, anastomotic fistula, abdominal abscess, intestinal torsion, intestinal necrosis, gangrenous appendicitis or pancreatitis’. These keywords were used to filter through preoperative diagnoses in patients’ records. The timeframe for this retrieval extended from 1 January 2008 to 31 December 2022.
Exclusion criteria include the presence of severe vital organ dysfunction in the medical history, individuals under the age of 18 years or patients presenting with infections outside the abdominal region prior to the surgical procedure. Patients with severe pre-existing organ dysfunction, such as end-stage heart, liver or renal failure, were excluded to ensure the primary outcomes could be attributed to the surgical management of abdominal sepsis rather than complications arising from terminal organ failure. This exclusion also aimed to enhance the homogeneity of the study population, as patients with severe organ dysfunction often require distinct management strategies that differ substantially from those with potentially reversible sepsis-related organ dysfunction. In cases where patients underwent multiple surgeries during their hospital stay, only the initial surgery’s data were considered for analysis to maintain consistency across the cohort and to avoid confounding factors that might arise from subsequent surgeries, such as changes in clinical status or the impact of prior interventions.
The examination and extraction of pertinent patient information were meticulously conducted from the database, including demographic details, laboratory findings, surgical interventions and anaesthetic records. The anaesthetic information retrieved comprised preoperative anaesthesia assessments, agents used for the induction and maintenance of anaesthesia, instances of hypotension, intraoperative fluid therapy, blood transfusions, urine output and administration of vasoactive medications.
We conducted an inter-rater reliability assessment using the kappa statistic to quantify the agreement between the data extractors at both institutions. Any discrepancies identified during this process were resolved through discussion and consultation with the study team to ensure data integrity. Additionally, we harmonised data definitions and excluded variables with inconsistent reporting to further enhance the reliability of our findings. Variables with >15% missing data were excluded from the analysis to minimise bias and <15% of missing data were handled using multiple imputation techniques to minimise bias and preserve the sample size. Five imputed datasets were generated, and the results were pooled using Rubin’s rules to obtain final estimates. In alignment with the established selection criteria, the process of data extraction was executed independently by two researchers (YS and TT). Any discrepancies or uncertainties encountered were adjudicated by a third investigator (JL).
Statistical analysis
Patients were stratified according to their in-hospital survival status, resulting in the formation of two distinct cohorts: non-survival group and survival group. Initial investigations employed univariate analysis to ascertain risk factors contributing to in-hospital mortality. Variables that achieved statistical significance in the univariate analyses (p<0.05) were subsequently included in a multivariate logistic regression model to calculate ORs and 95% CIs. Continuous variables, such as age and SOFA scores, were initially analysed as continuous variables to maximise statistical power and avoid information loss. However, to enhance clinical interpretability and enable subgroup analyses, these variables were also dichotomised using clinically relevant thresholds or median values. For instance, age was dichotomised at 60 years, a commonly used threshold to define elderly patients in clinical practice, while SOFA scores were dichotomised at the median value. Sensitivity analyses comparing models using continuous versus dichotomised variables revealed no significant differences in the direction or magnitude of the observed associations. Subgroup analyses were conducted to elucidate the prognostic implications for patients exhibiting diverse clinical characteristics.
To evaluate and mitigate potential multicollinearity in the regression models, we calculated variance inflation factors (VIFs) for each predictor variable. A VIF value exceeding 10 is typically considered indicative of significant multicollinearity. In our models, all VIF values were below 5, indicating that multicollinearity was not a substantial concern. Furthermore, we examined the correlation matrix for predictor variables and identified no strong correlations (r>0.7) among the variables included in the multivariate models. These steps confirm that multicollinearity did not adversely affect the regression models, thereby strengthening the reliability of the estimated coefficients and ORs.
Statistical analyses were conducted using SPSS Statistics V.25.0 (IBM SPSS, Armonk, New York, USA) along with R V.4.3.0. Dichotomous variables were represented as frequencies (percentages). The distribution of continuous variables was evaluated with the Shapiro-Wilk test to determine the suitability of parametric or non-parametric statistical methods. Variables with Shapiro-Wilk p values >0.05 were considered normally distributed and were summarised as mean±SD. Group comparisons for these variables were conducted using independent sample t-tests. Variables that deviated from normality (p<0.05) were summarised as median and IQR, with comparisons conducted using the Mann-Whitney U test. For dichotomous variables, differences were assessed using the Pearson’s χ2 test or Fisher’s exact test, contingent on the data’s distribution characteristics. All statistical tests were conducted as two-sided, with a p value threshold of <0.05 set to denote statistical significance.
Results
Demographic information and characteristics of patients
In the initial search, a total of 5165 patients were identified. From this population, 2437 individuals were excluded due to the absence of abdominal or pelvic surgical procedures or patients under 18 years of age. Following a thorough review of the medical records, a cohort of 476 patients, who met the Sepsis 3.0 diagnostic criteria, was selected for the final analysis (figure 1).9
Flow diagram of the study participants.
The demographic information and clinical characteristics of the study population are depicted in table 1. The mean age of patients was 56.00±16.80 years, with 44.12% aged over 60 years. The cohort predominantly comprised male patients, accounting for 71.22% (339 patients) of the sample. The median SOFA scores preoperatively and postoperatively were recorded at 3.00 (IQR 2.00–4.00) and 2.00 (IQR 1.00–3.00), respectively, indicating a median reduction in SOFA score of 1.00 (IQR 0.00–2.00). The predominant preoperative diagnosis was gastrointestinal (GI) disease, identified in 241 patients (50.63%), followed by biliary disease in 118 patients (24.79%), pancreatic disease in 41 patients (8.61%) and trauma in 19 patients (3.99%). 302 patients (63.45%) were admitted to the intensive care unit (ICU) following surgery, with a median ICU stay duration of 4.32 days (IQR 2.00–10.83). Furthermore, the median duration of hospital stay was documented as 15.00 days (IQR 8.00–26.25). Within the hospital setting, a total of 36 deaths were recorded, resulting in an in-hospital mortality rate of 7.56% and a 1-year mortality rate of 8.19%. 59 patients developed sepsis as a consequence of tumour-induced obstruction or perforation, representing 12.39% of the study population. The most frequent underlying malignancy was colon cancer (17 patients), followed by rectal cancer (13 patients), cholangiocarcinoma (9 patients), stomach cancer (6 patients), pancreatic cancer (5 patients), duodenal cancer (4 patients), hepatocellular carcinoma (3 patients) and gallbladder cancer (2 patients). These patients exhibited a significantly higher mortality rate, with an in-hospital mortality rate of 18.64%, which is more than triple the rate observed in non-oncological patients (6.00%). This differential outcome underscores the increased vulnerability of oncological patients to adverse prognoses.
Baseline characteristics and clinical outcomes of patients
The mean body mass index (BMI) of the cohort was 24.83±3.65, with a statistically significant difference observed between survivors (24.73±3.55) and non-survivors (26.13±4.52, p=0.027). However, the clinical significance of this difference remains uncertain. Obesity (BMI≥30) was present in 31.93% of the cohort, with no significant difference between survivors and non-survivors (p=0.851).
Table 2 delineates the anaesthetic and intraoperative data for the patient cohort. Within the group of 476 patients, anaesthesia induction was achieved using propofol, etomidate and ketamine for 456 (95.80%), 26 (5.46%) and 9 (1.89%) patients, respectively. A combined approach of intravenous and inhalational anaesthesia was administered to a majority of the patients (85.50%, 407 patients). Furthermore, 228 (47.90%) patients were administered at least one type of vasoactive agent via continuous infusion, encompassing phenylephrine (167 patients), norepinephrine (115 patients) and dopamine (55 patients). The median total fluid intake during the operative period was quantified at 16.81 mL/kg/hour (IQR 11.91–22.94). This total comprised a median crystalline fluid intake of 11.34 mL/kg/hour (IQR 8.02–15.75) and a colloidal fluid intake of 3.97 mL/kg/hour (IQR 0.00–6.37). Blood products were administered to 115 patients (24.16%). Intraoperative hypotension was observed in 355 patients (74.58%), with a median of 10.00 hypotensive episodes recorded (IQR 3.00–22.00). On the first postoperative day, extubation was unsuccessful in 58 patients (12.18%), and 113 patients (23.74%) received vasoactive drug infusions.
Anaesthetic and surgical information of patients
Univariate and multivariate logistic regression analysis of risk factors associated with in-hospital mortality
We observed that the age of the non-survival group was significantly higher than that of the survival group (67.58±15.92 vs 55.05±16.53 years, p<0.001). The mortality rate among female patients with sepsis appears to be higher than that among male patients (13.14% vs 5.31%, p=0.003); however, this difference did not maintain statistical significance after adjustment for covariates (p=0.072). Patients classified under the American Society of Anesthesiologists (ASA) physical status classes IV–V demonstrated a significantly increased mortality rate in comparison with those classified within ASA grades II–III (12.88% vs 5.52%, p=0.007).
Additionally, the non-survival group exhibited a higher prevalence of coronary heart disease (CHD) (22.22% vs 6.82%, p=0.003) and diabetes (33.33% vs 15.00%, p=0.004), along with a higher incidence of septic shock prior to surgery (58.33% vs 34.77%, p=0.005), and elevated preoperative and postoperative SOFA scores compared with the survival group. GI disease emerged as the predominant preoperative diagnosis across all patients, with an in-hospital mortality rate of 9.54% (23 patients). Among the 41 patients who underwent surgical intervention for pancreatic diseases, three succumbed postsurgery, resulting in a survival rate of 92.68%. The ICU stay of the non-survival patients (9.62 days, IQR 2.37–14.49) was more than two times that of survival patients (3.87 days, IQR 1.94–9.95).
Table 2 delineates the intraoperative anaesthetic and surgical information. We observed that the administration of vasoactive agents was more prevalent in the non-survival group, with notable differences in the usage of phenylephrine (52.78% vs 33.64%, p=0.021) and norepinephrine (61.11% vs 21.14%, p<0.001). Furthermore, the volume of crystalloid solution infused was significantly higher in the survival group (11.42 mL/kg/hour, IQR 8.08–15.75 vs 9.93 mL/kg/hour, IQR 6.85–15.31, p=0.348), whereas the volume of colloid solution infused was greater in the non-survival group (4.46 mL/kg/hour, IQR 3.43–6.51 vs 3.87 mL/kg/hour, IQR 0.00–6.35, p=0.167). Additionally, lactate levels, intraoperative bleeding volume and operative duration were significantly higher in the non-survival group, exhibiting increases of approximately 43.82%, 100% and 28.46%, respectively, compared with the survival group.
Preoperative and postoperative laboratory parameters are delineated in table 3. The summarised data illustrate that the non-survival group exhibited significantly lower levels of preoperative albumin (29.85±6.30 vs 34.21±6.79 g/L, p<0.001), platelets (121.50×109/L, IQR 79.00–174.50 vs 160.50×109/L, IQR 109.00–230.25, p=0.018) and haemoglobin (109.75±24.23 vs 123.07±31.28 g/L, p=0.003) in comparison with the survival group.
Preoperative and postoperative laboratory parameters of patients
Variables that demonstrated statistical significance in the univariate analysis with p values <0.05 were subsequently incorporated into a multivariate analysis. The univariate regression identified 21 factors of statistical relevance. Following this, a multivariate logistic regression analysis was conducted. The reference categories for categorical variables were chosen based on clinical relevance and common practice in the literature. The outcomes of the definitive model revealed that age, with an OR of 6.77 (95% CI 2.46 to 18.66, p<0.001), diabetes, with an OR of 2.61 (95% CI 1.04 to 6.56, p=0.041) and higher preoperative SOFA score, with an OR of 3.48 (95% CI 1.16 to 10.43, p=0.026), constituted significant risk factors for in-hospital mortality. Conversely, biliary disease was associated with a reduced risk of in-hospital mortality, exhibiting an OR of 0.15 (95% CI 0.04 to 0.64, p=0.010) (figure 2).
Forest plot based on multivariate logistic regression analysis of all the patients. Significant predictors are highlighted with asterisks (*p<0.05). ASA, American Society of Anesthesiologists; BMI, body mass index; CHD, coronary heart disease; pre-ALB, preoperative albumin; pre-Hb, preoperative haemoglobin; pre-PLT, preoperative platelet; pre-SOFA, preoperative Sequential Organ Failure Assessment score.
Subgroup analysis
Subgroup analysis of GI diseases
Patients diagnosed with GI diseases constituted the majority of the study cohort, representing 50.63% (241 patients) of the total patient population. Within this group, digestive tract perforation (DTP) emerged as the most common condition, accounting for 49.79% of cases. 15 of these patients succumbed to their condition in the hospital, resulting in a mortality rate of 12.50%. Appendicular diseases were diagnosed in 30.29% (73 patients) of the cases, with a notable absence of in-hospital mortality reported among these patients. Appendicitis was associated with the lowest incidence of septic shock at 13.70%, whereas patients presenting with bowel necrosis exhibited the highest incidence at 71.43% (online supplemental table 1).
Supplemental material
A total of 35 cases were admitted due to GI obstruction, with these patients being significantly older (63.17±16.40 years) compared with other patients. The median intraoperative blood loss for patients with GI obstruction was the highest recorded at 200 mL (IQR 100–300), and a higher proportion of these patients (34.29%, 12 patients) required the administration of vasoactive drugs on the first postoperative day. Moreover, patients suffering from GI bleeding had prolonged operative times (189.50 min, IQR 129.75–214.00), and extended hospital stays (23.50 days, IQR 13.00–28.00) compared with other patients.
The prevalence of pre-existing CHD was significantly higher in the non-survival group (30.43%, 7/23 patients) than in the survival group (6.88%, 15/218 patients), and the non-survival group also exhibited significantly elevated postoperative SOFA score (3.00, IQR 1.50–4.00 vs 1.00, IQR 0.00–2.00, p=0.001) (online supplemental table 2). Variables demonstrating significance (p<0.05) in the univariate analysis were subsequently incorporated into a multivariate logistic regression analysis. This analysis identified advanced age as a risk factor for in-hospital mortality (OR 42.83, 95% CI 4.13 to 444.34, p=0.002), and a lower preoperative serum albumin level was also found to increase the risk of in-hospital mortality (OR 20.23, 95% CI 2.21 to 184.84, p=0.008) (online supplemental figure 1).
Supplemental material
Subgroup analysis of elderly patients
The patients aged over 60 years comprised 44.12% (210 patients) of the study population, with a mortality rate of 12.86% (27 patients). Of these, 109 patients (22.90%) were aged over 70 years, with an in-hospital mortality rate of 14.68% (16 patients). Furthermore, the cohort aged 80 years and older, representing 8.19% (39 cases) of the study group, experienced a notably higher in-hospital death rate of 23.08% (nine patients) (online supplemental table 3).
The likelihood of possessing preoperative comorbidities and encountering septic shock was elevated among the older demographic, with occurrences documented in 51.43% (108 patients) and 36.67% (77 patients), respectively. Elderly patients developing septic shock exhibited a significantly increased risk of in-hospital mortality compared with their counterparts without septic shock (22.08% vs 7.52%, p=0.002).
GI diseases emerged as the predominant preoperative diagnosis within this elderly cohort, accounting for 50.95% (107 patients), notably DTP (29.05%, 61 patients) and GI obstruction (9.52%, 20 patients) (online supplemental table 4). Biliary disease is the second leading cause of sepsis in elderly patients (32.86%, 69 patients). Elderly patients afflicted with sepsis suffering from DTP demonstrated a higher propensity for in-hospital mortality (22.95%, 14/61 patients) (online supplemental table 4).
Multivariate analysis identified that GI disease (OR 4.66, 95% CI 1.19 to 18.25, p=0.027) and a lower preoperative serum albumin level (OR 10.52, 95% CI 2.29 to 48.31, p=0.002) significantly correlated with an increased risk of in-hospital mortality among the elderly population, underscoring the critical impact of disease types and nutritional status on patient outcomes (online supplemental figure 2).
Supplemental material
Subgroup analysis of septic shock
In this cohort, 174 patients, constituting 36.55% of the study population, developed septic shock prior to undergoing surgery. Of these, 21 patients died in the hospital, resulting in an in-hospital mortality rate of 12.07%. Furthermore, these patients exhibited a prolonged hospital stay—approximately 32.14% longer—compared with those who did not experience shock (online supplemental table 5).
Patients presenting with bowel necrosis demonstrated the highest predisposition towards septic shock, at a rate of 71.43% (five patients), followed by those with GI obstruction at 57.14% (20 patients), pancreatic diseases at 56.10% (23 patients) and DTP at 45.83% (55 patients). 124 patients received at least one type of vasoactive agent for continuous infusion, predominantly phenylephrine (86 patients, 49.43%) and norepinephrine (83 patients, 47.70%).
Elevated lactate levels were observed in patients who developed septic shock, averaging 2.52±2.32 mmol/L, which is approximately 73.79% higher than that of other patients (1.45±1.06 mmol/L). Additionally, intraoperative urine output was markedly reduced by >20% in patients experiencing shock (1.72 mL/kg/hour, IQR 0.72–2.81) compared with those without shock (2.09 mL/kg/hour, IQR 1.07–2.37).
The patients with septic shock who died were significantly older than those who did not (70.43±15.82 vs 54.22±16.80 years, p<0.001), and they more commonly had comorbidities, including CHD and diabetes (online supplemental table 6). And the multivariate analysis indicated that age (OR 3.44, 95% CI 1.01 to 11.82, p=0.050) and CHD (OR 3.83, 95% CI 1.02 to 14.42, p=0.047) emerged as significant risk factors for in-hospital mortality among patients with septic shock, highlighting the importance of age and cardiac function as determinants in the prognosis of septic shock outcomes (online supplemental figure 3).
Supplemental material
Discussion
While previous studies have extensively explored the risk factors for sepsis in general, there remains a significant gap in the literature regarding surgical patients with abdominal sepsis, particularly in the context of perioperative management and outcomes. Our study addresses this gap by providing a comprehensive analysis of the clinical characteristics and risk factors associated with in-hospital mortality in this specific patient population. Leveraging a large, multicentre dataset spanning 15 years, we identified key predictors, such as advanced age, diabetes, higher preoperative SOFA scores and reduced preoperative albumin levels. These findings have important implications for perioperative care, particularly in tailoring interventions for high-risk subgroups, including elderly patients and those with GI diseases. By corroborating existing knowledge and offering new insights, our study advances the understanding of how targeted management strategies can improve outcomes in this vulnerable population group.
The rationale for focusing solely on abdominal sepsis managed surgically is twofold. First, abdominal sepsis represents a distinct clinical entity with unique pathophysiological mechanisms and management challenges compared with sepsis originating from other sources. Surgical intervention is often critical for source control in abdominal sepsis, as it directly addresses the underlying infection through procedures such as the removal or repair of perforated organs, resection of diseased tissues and debridement of necrotic or ischaemic areas. Second, surgical patients with abdominal sepsis represent a high-risk population because of the complexity of their conditions, which often involve multiple organ systems and require coordinated perioperative care. By limiting the study to surgical patients, we aimed to provide a more targeted analysis of how perioperative factors influence the prognosis of those undergoing surgery, distinguishing them from non-surgical cases in which other management strategies might apply.
While advanced age, diabetes and preoperative SOFA scores were identified as significant predictors of mortality, it is noteworthy that certain variables, such as sex, did not retain statistical significance in the multivariate analysis. Although univariate analysis indicated a higher mortality rate among female patients (13.14% vs 5.31%, p=0.003), this difference was not significant after adjusting for covariates (p=0.072). This finding contrasts with some previous studies that have reported sex-based differences in sepsis outcomes,10 potentially due to variations in study populations, sample sizes or the influence of confounding factors. The lack of significance in our study may be attributed to the relatively small proportion of female patients (28.78%) or the predominance of other risk factors, such as age and comorbidities, which may overshadow the impact of sex on mortality. Further research involving larger and more balanced cohorts is needed to comprehensively explore the role of sex in sepsis outcomes.
GI diseases constitute the principal origin of abdominal sepsis and significantly contribute to mortality rates.11 These conditions compromise the epithelial integrity of the GI tract and disrupt the equilibrium of intestinal flora. This disruption facilitates the translocation of bacteria through the intestinal wall into the bloodstream or their direct dissemination to various locations within the abdominal cavity, potentially culminating in abdominal sepsis or septic shock.12 Within the scope of GI diseases, this study identified DTP as exhibiting a notably high incidence, accounting for approximately 50% of all patient cases. The majority of DTP instances involved upper GI perforations, with chiefly gastroduodenal perforations induced by peptic ulcers. Additionally, nearly 25% of patients experienced lower GI perforation, predominantly due to colonic perforations. The leading causes of lower GI perforation were neoplasms or malignancies (41.38%), followed by traumatic injury (34.48%). The primary surgical interventions employed in clinical settings include perforation repair, subtotal gastrectomy and bowel resection, performed through either open or laparoscopic methods. Laparoscopic repair of perforated ulcers is being increasingly adopted, with 16 patients in this study successfully managed using this technique.
Elderly patients were inherently at a greater risk of developing GI malignancies, with 11 (31.43%) of the total patients presenting with GI obstruction related to GI cancer. Of these, five patients succumbed during their hospital stay, four of whom were diagnosed with GI cancer. These findings underscore that the selection between palliative stoma creation or primary tumour resection should be tailored according to the individual clinical presentations of patients with GI obstruction induced by cancers.
The majority of biliary infections were attributed to biliary stones, with the common bile duct (50.85%), gallbladder (33.90%) and intrahepatic bile duct (25.42%) identified as the predominant sites of affliction, of which 15 patients (12.72%) had co-infections. It was observed that patients with sepsis diagnosed with biliary diseases exhibited a reduced risk of in-hospital mortality in comparison with those with GI diseases. Among the 118 patients with biliary diseases, three fatalities were recorded, characterised by advanced ages of 73, 81 and 82 years, respectively. The deceased were diagnosed with cholangiocarcinoma, cholangitis and cholecystitis, respectively. Hence, although patients with biliary sepsis exhibit a lower mortality rate, it remains imperative to prioritise care for elderly individuals.
The cohort comprised 41 patients with pancreatic diseases, among whom 36 had pancreatitis or its complications, while five were diagnosed with pancreatic cancer. The management strategy emphasises the importance of controlling the systemic inflammatory response syndrome at an early stage and, subsequently, ensuring the comprehensive removal of necrotic infected tissues. Surgical intervention, alongside broad-spectrum antibiotics, plays a crucial role in the excision of infected necrotic tissue. In this study, nearly 70% (24 patients) underwent percutaneous nephroscopic necrosectomy for the removal of pancreatic necrotic tissues, while the remainder underwent exploratory laparotomy or endoscopic retrograde cholangiopancreatography. Previous research highlighted that in patients with severe necrotising pancreatitis, the mortality rate was significantly higher in patients undergoing early surgery (within 24–48 hours after onset) at 58%, compared with 27% in those undergoing delayed surgery (12 days postonset), thereby questioning the clinical value of early surgical intervention.13 In the current study, the median time from admission to surgery was recorded at 323.74 hours (approximately 13.49 days) for patients with pancreatitis. Only four patients (11.11%) with pancreatitis received surgical treatment within 48 hours of admission, and no fatalities were observed. The decision to delay surgery is made to alleviate early systemic inflammatory response, facilitate complete liquefaction of pancreatic necrotic tissue, thus achieving more effective debridement. It is imperative to recognise the rapid progression and narrow therapeutic window associated with severe acute pancreatitis, as delayed surgical intervention may preclude the opportunity for curative treatment.
It is important to note that some subgroup analyses, such as those involving patients with pancreatic diseases (n=41), were limited by small sample sizes. This limitation reduces the statistical power of the analyses and may affect the generalisability of the findings. While the results provide valuable insights, they should be interpreted with caution.
Postponing surgical intervention for >24 hours after the onset of infection is a significant risk factor for treatment failure.14 Furthermore, a delay in achieving source control exceeding 3–6 hours has been linked to heightened case fatality rates.15 Hence, prompt surgical intervention is paramount to enhance prognosis in this specific patient cohort. In this investigation, over 60% of individuals with abdominal sepsis were subjected to emergency surgery. No discernible disparity in mortality rates was observed between the emergency and elective surgery cohorts. The median duration from admission to the surgical procedure in this study was recorded at 15.47 hours (IQR 3.70–124.89), with no variation noted between the survivor and non-survivor groups.
The Surviving Sepsis Campaign guideline recommends that critically ill patients with sepsis be admitted to the ICU within 6 hours.16 Previous research has demonstrated that patients with haemodynamic instability requiring vasopressor administration face a higher risk of mortality when ICU admission is delayed beyond 6 hours,17 highlighting the critical importance of timely escalation of care in managing patients with sepsis. Early ICU admission enables prompt implementation of advanced monitoring, aggressive resuscitation and organ support, all of which are essential to mitigating the systemic effects of sepsis.
Patients presenting with abdominal sepsis frequently exhibit preoperative circulatory and respiratory system dysfunctions, compounded by rapid disease progression and the complexity of various invasive procedures, thereby presenting challenges to anaesthesiologists. Guidelines recommend the initiation of fluid resuscitation on the diagnosis of ‘sepsis’, advocating for the administration of crystalloid at 30 mL/kg to be completed within 1 hour and the employment of vasoactive medications to sustain mean arterial pressure >65 mm Hg.16 The findings from our study indicated that the median intraoperative fluid administration was 16.81 mL/kg/hour, likely attributed to the majority of patients receiving fluid resuscitation prior to surgical intervention.
The careful titration of anaesthetic drugs during both induction and maintenance phases is crucial to mitigate intraoperative haemodynamic instability and minimise adverse effects. In our research, the choice of anaesthetic agents for induction or maintenance did not appear to significantly impact clinical outcomes. Dexmedetomidine, in particular, has attracted attention for its anti-inflammatory and sedative properties, as well as its capacity to shorten the duration of delirium among critically ill patients.18 However, our study indicated a higher risk of in-hospital mortality in patients with sepsis receiving dexmedetomidine (22.22%, 8/36 patients vs 8.64%, 38/440 patients, p=0.018). Similarly, the Sedation Practice in Intensive Care Evaluation (SPICE) III trial reported higher mortality rates associated with dexmedetomidine use in patients ≤65 years of age, highlighting the need for caution when administering this drug to patients with sepsis undergoing surgery.19 The contradictory findings may be due to differences in patient populations, dosing regimens or the timing of dexmedetomidine administration. However, the specific role of dexmedetomidine in the management of patients with sepsis within this study could not be ascertained due to the limited number of patients administered with this drug. Consequently, further evidence, particularly from large-scale clinical trials, is essential to definitively ascertain the influence of anaesthetic drugs in this medical context.
The early detection of sepsis poses considerable challenges due to the heterogeneity of its clinical manifestations and the lack of specificity associated with sepsis. Consequently, there is a pressing demand for biomarkers that can accurately guide the diagnosis and prognosis assessment. Elevated blood lactate concentrations in patients with septic shock are indicative of compromised tissue perfusion and are correlated with organ dysfunction and increased mortality rates.20 In this study, lactate levels in the septic shock cohort were observed to be higher than those in the non-shock group, with particularly elevated levels noted in non-surviving patients. Previous research has demonstrated that postoperative monitoring of albumin levels can identify patients at a high risk of postoperative complications and predict the likelihood of postoperative infections.21 22 Albumin uniquely benefits the survival rate and prognosis in the management of sepsis.23 Our findings indicated that reduced serum albumin levels were associated with increased mortality among elderly patients or those patients with GI diseases. Therefore, vigilant monitoring and prompt supplementation of albumin should be considered in these patients.
Neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio and systemic immune-inflammation index offer significant diagnostic and prognostic insights for patients with sepsis.24–26 However, this study did not identify the correlation between these indicators and the prognosis of patients with sepsis. Given the complex pathophysiology of sepsis and the heterogeneity of the patient population affected by sepsis, reliance on a single biomarker is insufficient for early diagnosis and prognostic evaluation in sepsis. Therefore, integrating multiple biomarkers represents a promising direction for future research.
Age was identified as an independent predictive factor for in-hospital mortality among patients with sepsis in this study, with an OR of 6.77 and a 95% CI ranging from 2.46 to 18.66. Senior patients are particularly susceptible to infections, attributable to diminished physical and immune functionalities. And ageing is associated with decreased GI function, which increases the risk for GI diseases in elderly patients. Notably, the early clinical manifestations of sepsis may be subtle in older patients, potentially leading to delays in the timely identification and initiation of treatment. Therefore, elderly patients are at an increased risk of developing septic shock and facing a poorer prognosis, thus requiring special attention in clinical management.27
CHD was associated with increased risk of in-hospital mortality in patients with septic shock in this study. During septic shock, patients with CHD experience an exacerbated imbalance between oxygen supply and demand, which further aggravates myocardial ischaemia and hypoxia.28 And myocardial depression induced by sepsis has a profound effect on depressing cardiac function, thereby worsening patient prognosis.29
Furthermore, diabetes emerged as a potential risk factor for mortality in this cohort. This association may stem from compromised bodily defenses, diminished phagocytic activity of white blood cells, increased vulnerability to infections and impaired wound healing. Additionally, the physiological stresses of surgery, blood loss and anaesthesia can precipitate cardiac and renal dysfunctions in patients with diabetes, thereby augmenting perioperative mortality risks.30 Therefore, perioperative blood glucose monitoring and management for patients with diabetes with sepsis may require increased attention.
Drawing on the findings of this study, we propose targeted strategies to optimise perioperative management and improve outcomes in surgical patients with abdominal sepsis. Early identification of high-risk patients—particularly those of advanced age, with diabetes, or elevated preoperative SOFA scores—is essential. These individuals may benefit from more aggressive monitoring and intervention strategies, including early admission to the ICU and the use of advanced haemodynamic monitoring to guide resuscitation efforts. Optimisation of nutritional status is essential, particularly for elderly patients and those with GI diseases. Early nutritional support and albumin supplementation should be prioritised, and preoperative assessment of nutritional status should be conducted to facilitate timely interventions. For patients with diabetes, intensified perioperative blood glucose management is critical to mitigate complications such as cardiac and renal dysfunction. Perioperative monitoring and strict glycaemic control should be emphasised to reduce mortality risks. By implementing these strategies, clinicians can improve perioperative care, reduce mortality and enhance outcomes in patients with abdominal sepsis. Further research is needed to validate and refine these approaches for optimal patient management.
Limitations
This investigation is subject to several limitations. First, despite spanning a 15-year period and incorporating data from two major medical centres, the sample size remained relatively modest. Moreover, the study population was recruited from two tertiary hospitals in China, which may limit the generalisability of the findings to other regions or healthcare systems. Future multicentre studies involving diverse populations and healthcare systems are necessary to validate and extend these findings.
Furthermore, this study was nested in a large cohort; some unknown confounding factors could have introduced a bias. One significant concern is the potential for information bias, as the data relied on medical records that may contain incomplete or inconsistent information. While we employed multivariate logistic regression to adjust for known confounders, unmeasured or unknown confounders may still have influenced the observed associations. Additionally, the retrospective design of this study limits our ability to establish causal relationships between the identified risk factors and in-hospital mortality. To address these limitations, future research should prioritise prospective, multicentre studies involving diverse populations and healthcare systems.
Conclusion
In this retrospective analysis, we delineated the clinical characteristics of patients afflicted with abdominal sepsis who underwent surgical interventions and pinpointed risk factors associated with in-hospital mortality. The findings underscore the necessity for enhanced focus on the perioperative management of patients with sepsis, particularly those characterised by advanced age, diabetes and diminished preoperative albumin levels. However, given the retrospective nature of this study, further prospective, multicentre studies are needed to validate the identified predictors and confirm their generalisability across diverse populations and healthcare settings.
Data availability statement
Data are available in a public, open access repository. The data supporting the findings of this study are available at (https://doi-org.ezproxy.u-pec.fr/10.6084/m9.figshare.25523737.v1).
Ethics statements
Patient consent for publication
Ethics approval
This study was approved by the Ethics Committee Board of the First Medical Center of the Chinese PLA General Hospital (approval number: S2023-614-01) and the Ethics Committee Board of the PLA Air Force Medical Center (approval number: 2023-15-S01). The requirement for patient consent was waived due to the retrospective design of the study.
Acknowledgments
We acknowledge the Chinese People's Liberation Army (PLA) General Hospital and the PLA Air Force Medical Center for providing the data used in the current study. Additionally, we extend our gratitude to all individuals who contributed to this work, both directly and indirectly.
References
Footnotes
YS and TT contributed equally.
Contributors YS: investigation, data curation, software, writing—original draft. TT: conceptualisation, investigation, methodology, funding acquisition, writing—review and editing. XY: investigation, methodology, software. BY: data curation. WM: conceptualisation, funding acquisition, writing—review and editing. JL: conceptualisation, investigation, methodology, data curation, writing—review and editing. JL is designated as the guarantor of this study and is responsible for the integrity of the data and the accuracy of the analysis.
Funding This work was supported by the National Key Research and Development Programme of China (2018YFC2001901) and the PhD Booster Programme of Air Force Medical Center (2021ZT020).
Disclaimer The study sponsors were not involved in the design of the study; the collection, analysis and interpretation of data; writing the report or the decision to submit the report for publication.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.