Risk factors for anastomotic leakage after radical colorectal cancer surgery and its impact on prognosis: Clinical education research

Abstract

BACKGROUND

Anastomotic leakage (AL) is one of the severe complications developed after radical colorectal cancer (CRC) surgery that has an incidence between 3% and 19%, and might also severely deteriorate long-term oncological outcomes. The perioperative management of high-risk patients requires that they be identified pre-operatively. We sought to determine independent risk factors for AL and assess the prognostic effect of AL in patients who underwent surgery with curative intent for CRC.

AIM

To investigate the incidence, independent risk factors, and impact on short-term and long-term prognosis of AL following radical CRC surgery, and to provide clinical evidence for perioperative risk stratification and management.

METHODS

A single-center retrospective cohort study was conducted, collecting data from 200 patients who underwent elective radical resection for CRC at the Department of General Surgery, the First Affiliated Hospital of Bengbu Medical University, from January 2019 to December 2024. Patients were divided into an AL group (n = 22) and a non-AL group (n = 178) based on whether AL occurred within 30 days postoperatively. Univariate and multivariate logistic regression analyses were used to identify independent risk factors. The Kaplan-Meier method and Cox proportional hazards regression model were used to evaluate the impact of AL on overall survival (OS) and disease-free survival (DFS). Action-Process-Object-Schema theory combined with principles of critical thinking ability assessment was used as a supplementary framework to organize perioperative risk evaluation and clinical interpretation.

RESULTS

AL occurred in 22 of 200 patients (11.0%). Multivariate logistic regression identified body mass index ≥ 25 kg/m² [odds ratio (OR) = 2.354, 95%CI: 1.196-4.632, P = 0.013], diabetes mellitus (OR = 2.660, 95%CI: 1.258-5.624, P = 0.010), rectal tumor location (OR = 2.440, 95%CI: 1.198-4.970, P = 0.014), neoadjuvant therapy (OR = 2.186, 95%CI: 1.042-4.584, P = 0.039), preoperative albumin < 35 g/L (OR = 2.680, 95%CI: 1.278-5.620, P = 0.009), operation time ≥ 180 minutes (OR = 2.334, 95%CI: 1.142-4.770, P = 0.020), and intraoperative blood loss ≥ 200 mL (OR = 2.142, 95%CI: 1.024-4.482, P = 0.043) as independent risk factors; protective stoma (OR = 0.403, 95%CI: 0.174-0.932, P = 0.034) was a protective factor. The AL group had significantly longer hospital stay, higher overall complication rate, higher reoperation rate, higher permanent stoma rate, and higher rate of adjuvant chemotherapy delay (all P < 0.05). Kaplan-Meier analysis demonstrated that the 5-year overall survival rate was 63.6% vs 82.6% (Log-rank P = 0.003), and DFS rate was 54.5% vs 77.5% (Log-rank P = 0.001) in the AL vs non-AL groups, respectively. Multivariate Cox regression confirmed AL as an independent adverse prognostic factor for DFS (hazard ratio = 2.048, 95%CI: 1.312-3.198, P = 0.002).

CONCLUSION

AL following radical CRC surgery is associated with multiple perioperative risk factors and significantly impairs both short-term and long-term prognosis. Preoperative nutritional and glycemic optimization, judicious use of protective stoma in high-risk patients, and reduction of intraoperative stress are key strategies to reduce AL incidence and improve patient outcomes. The Action-Process-Object-Schema-informed critical thinking framework may further enhance the clinical interpretation and application of perioperative risk assessment.

Key Words: Colorectal cancer; Radical surgery; Anastomotic leakage; Risk factors; Prognosis; Retrospective cohort study; Action-Process-Object-Schema theory; Critical thinking

Core Tip: This retrospective cohort study evaluated the incidence, risk factors, and prognostic impact of anastomotic leakage (AL) after radical colorectal cancer surgery. Among 200 patients, AL occurred in 11.0% and was independently associated with body mass index ≥ 25 kg/m², diabetes mellitus, rectal tumor location, neoadjuvant therapy, hypoalbuminemia, prolonged operation time, and increased intraoperative blood loss, while protective stoma was a significant protective factor. Patients with AL had significantly worse short-term outcomes, including higher complication and reoperation rates, prolonged hospitalization, and delayed adjuvant chemotherapy. Importantly, AL was also associated with inferior long-term survival and was identified as an independent predictor of poorer disease-free survival.

INTRODUCTION

Colorectal cancer (CRC) is one of the most common malignancy within the digestive tract worldwide[1]. CRC has shown an alarming rise in incidence during the past several decades and it is now the fifth most common cancer and by far the second leading cause of cancer mortality[2,3] in China. Rate of new CRC in China 2022 was an estimated 517100 (155800-192800, standard deviation); the disease claimed approximately 240000 lives[4]. Radical surgical resection is the cornerstone of curative treatment for CRC, and efforts towards laparoscopic and robotic-assisted approaches have contributed to refining operative precision and acceleration of post-operative recovery[5,6].

However, anastomotic leakage (AL) continues as one of the most worrying and severe postoperative complications of colorectal surgery with reported incidence rates ranging from 3% to 19% with respect to anastomotic height, surgical technique and patient characteristics[5,7]. AL is an abnormality of the closure that creates a communication between intraluminal and extraluminal compartments, with defective intestinal wall integrity at the anastomotic site according to the International Study Group of Rectal Cancer classification: (1) Grade A (subclinical); (2) Grade B requiring active intervention without re-laparotomy; and (3) Grade C requiring re-laparotomy[8].

However, the presence of AL not only has short-term clinical implications after surgical resection; it may negatively impact long-term oncological results by postponing adjuvant chemotherapy, increasing systemic inflammatory response and potentially allowing for residual tumor cell survival and dissemination[9,10]. Nevertheless, the extent and mechanisms of this prognostic effect are under discussion[8]. Whether AL independently affects survival or its apparent prognostic effect is mediated largely by delays in initiation of adjuvant chemotherapy and the systemic inflammatory load rather than due to intrinsic tumor biology that predicts worse outcomes remains a central controversy[9,10]. Clarifying this distinction has direct implications for clinical decision-making, particularly regarding the urgency of AL prevention strategies vs post-leakage oncological management. Preoperative identification of high-risk patients through systematic evaluation of modifiable and non-modifiable risk factors is therefore of critical clinical importance for stratified perioperative decision-making, including the selective use of defunctioning stoma[11-14].

The theoretical framework guiding this investigation is grounded in Action-Process-Object-Schema (APOS) theory, originally developed by Dubinsky and McDonald in the context of mathematical knowledge construction, and subsequently generalized as a constructivist model of clinical reasoning and evidence-based decision-making. Within the APOS framework, clinical knowledge about AL risk is understood as progressively constructed through four cognitive stages: (1) Action, wherein the clinician performs discrete assessments of individual patient variables [such as measuring body mass index (BMI), albumin level, or glycemic status]; (2) Process, wherein the clinician recognizes patterns and dynamic relationships among multiple risk factors; (3) Object, wherein the clinician conceptualizes the patient’s overall risk profile as an integrated construct; and (4) Schema, wherein the accumulated risk knowledge is applied as a structured decision framework guiding perioperative interventions, including the decision to create a protective stoma. Complementing APOS theory, critical thinking ability assessment provides a methodological standard for evaluating the quality and rigor of clinical reasoning: (1) Emphasizing systematic evidence appraisal; (2) Assumption questioning; (3) Logical inference; and (4) Reflective judgment in perioperative risk evaluation.

The integration of APOS theory and critical thinking ability assessment as the dual theoretical pillars of this study is both conceptually motivated and clinically practical. Prior studies have largely examined AL risk factors in isolation, without a unifying theoretical framework to systematize how individual risk variables are identified, interrelated, and ultimately translated into clinical action schemas. By explicitly structuring the study design, analytical methodology, and interpretation of findings within the APOS-critical thinking framework, this investigation aims to facilitate not only statistical identification of independent risk determinants but also the construction of a clinically actionable, schema-level decision tool applicable to perioperative management in Chinese CRC patients.

Given the limited large-scale domestic analyses systematically evaluating both the risk determinants and prognostic consequences of AL in Chinese CRC patients, the present study retrospectively analyzed 200 consecutive patients who underwent radical CRC surgery at our institution over a 6-year period. Our objectives were: (1) To determine the incidence and clinical spectrum of AL; (2) To identify independent preoperative and intraoperative risk factors for AL through multivariate logistic regression; and (3) To quantify the impact of AL on short-term surgical outcomes and long-term survival using Kaplan-Meier analysis and Cox regression.

MATERIALS AND METHODS

Study design

This was a single-center retrospective cohort study. All procedures were conducted in accordance with the Declaration of Helsinki. The study protocol was approved by the Ethics Committee of the First Affiliated Hospital of Bengbu Medical University (approval No. CB20260200021). Owing to the retrospective cohort nature of the study, the requirement for individual patient informed consent was waived by the Ethics Committee. All patient personal information was anonymized and kept strictly confidential.

Study subjects

Consecutive patients who underwent elective radical resection for histopathologically confirmed colorectal adenocarcinoma between January 2019 and December 2024 at the Department of General Surgery were screened.

Inclusion criteria: (1) Age 18-80 years; (2) Histopathologically confirmed colorectal adenocarcinoma; (3) Elective radical resection (R0) with primary anastomosis; (4) Clinical TNM stage I-III (American Joint Committee on Cancer 8^th edition); (5) Complete perioperative and follow-up data available; and (6) Follow-up duration ≥ 12 months.

Exclusion criteria: (1) Stage IV disease with distant metastasis at diagnosis; (2) Emergency surgery for perforation, obstruction, or hemorrhage; (3) Synchronous or metachronous other malignancy; (4) Severe cardiac, hepatic, or renal dysfunction; (5) Incomplete clinical records; and (6) Loss to follow-up within 12 months. A total of 200 patients meeting all criteria were finally enrolled, comprising 22 in the AL group and 178 in the non-AL group. Patients who died within 12 months from any cause were retained in the survival analysis with their actual survival time recorded and censored at the time of death; they were not excluded on the basis of insufficient follow-up duration. Patients who were lost to follow-up before 12 months for reasons other than death were excluded, as their survival status could not be ascertained with sufficient reliability for inclusion in time-to-event analyses.

Treatment protocol

All patients underwent standardized enhanced recovery after surgery perioperative protocols including preoperative carbohydrate loading, mechanical and oral antibiotic bowel preparation where indicated, prophylactic antibiotics, thromboprophylaxis, and multimodal analgesia[6,15,16]. For locally advanced rectal cancer (cT3-4 or cN+), preoperative multidisciplinary team discussion was performed; eligible patients received neoadjuvant long-course concurrent chemoradiotherapy (45-50 Gy/25-28 fractions combined with capecitabine-based chemotherapy) or short-course radiotherapy (25 Gy/5 fractions) followed by surgery 6-8 weeks later[13]. The operative approach (open, laparoscopic, or robotic) and the decision to create a defunctioning stoma were at the operating surgeon’s discretion based on intraoperative assessment of anastomotic tension, perfusion, and individual patient risk factors[11,12,14,17].

Data collection

General information: The following data were extracted from the electronic medical records: (1) Demographic data: Gender, age, BMI; (2) Comorbidities: Diabetes mellitus, hypertension; (3) Lifestyle history: Smoking, alcohol use; (4) Preoperative laboratory values: Serum albumin, hemoglobin, carcinoembryonic antigen, carbohydrate antigen 19-9; (5) Tumor characteristics: Location (colon vs rectum), TNM stage, differentiation grade, vascular invasion; (6) Neoadjuvant therapy status; (7) American Society of Anesthesiologists (ASA) physical status classification; and (8) Operative details: Approach, operation time, intraoperative blood loss, protective stoma formation.

Primary outcome (AL): AL was defined according to the International Study Group of Rectal Cancer criteria as a defect of the intestinal wall integrity at the colorectal or coloanal anastomotic site including suture and staple lines, leading to communication between the intraluminal and extraluminal compartments[8]. Only AL occurring within 30 days postoperatively was included. AL was graded as grade A (no active therapeutic intervention required), grade B (requiring active therapeutic intervention without re-laparotomy), and grade C (requiring re-laparotomy). Diagnosis was confirmed by clinical assessment supplemented by computed tomography with rectal contrast, water-soluble contrast enema, or direct endoscopic/operative visualization[18,19].

Secondary outcome measures: (1) Short-term outcomes: Total hospital stay, postoperative hospital stay, time to first flatus, time to oral feeding, overall complication rate (Clavien-Dindo grade ≥ II), reoperation rate, 30-day mortality, permanent stoma formation rate, and adjuvant chemotherapy delay (> 8 weeks from surgery); and (2) Long-term survival: Overall survival (OS) was defined as time from surgery to death from any cause or last follow-up. Disease-free survival (DFS) was defined as time from surgery to the first event of local recurrence, distant metastasis, or death. The last follow-up date was December 31, 2024.

Theoretical framework

The analytical methodology of this study was explicitly structured within the APOS theoretical framework [Dubinsky & McDonald, 2001] combined with critical thinking ability assessment principles, as the dual methodological bases for constructing and interpreting clinical risk knowledge.

At the Action stage, each candidate risk variable (e.g., BMI, serum albumin, diabetes mellitus status, operative duration) was operationally defined and individually measured or recorded from medical records. This stage corresponds to the discrete collection of observable clinical data items, ensuring completeness and accuracy as preconditions for higher-order reasoning. Critical thinking at this stage required rigorous operationalization of each variable for example, adopting internationally standardized AL grading criteria (International Study Group of Rectal Cancer) rather than institutional ad hoc definitions – so that data-collection actions were reflective and evidence-anchored rather than habitual.

At the process stage, the dynamic interrelationships among collected variables were examined through univariate logistic regression and bivariate statistical comparisons. This stage corresponds to the recognition that AL risk is not determined by any single variable in isolation, but emerges from the interaction of multiple concurrently present risk factors. Critical thinking at this stage demanded evaluation of confounding, effect modification, and statistical assumptions (e.g., normality testing, appropriate test selection for continuous vs categorical variables), ensuring that identified associations were meaningful rather than artifactual.

At the Object stage, multivariate logistic regression was performed to construct an integrated, adjusted risk model in which individual process-level associations were synthesized into a cohesive risk object: A multivariate model identifying independent predictors of AL with quantified odds ratios and confidence intervals. This stage reflects the clinician’s ability to treat the multivariable risk structure as a conceptual whole an object of clinical reasoning rather than a mere list of isolated associations. Critical thinking at this stage necessitated careful variable selection strategies, collinearity assessment, and model fit evaluation to ensure that the resulting risk object was both statistically valid and clinically interpretable.

Statistical analysis

Statistical analyses were performed using SPSS 26.0 (IBM Corp., Armonk, NY, United States). Continuous normally distributed variables are expressed as mean ± SD and compared using the independent samples t-test; non-normally distributed variables are expressed as median (interquartile range) and compared using the Mann-Whitney U test. Categorical variables are expressed as n (%) and compared using the χ² test or Fisher’s exact test. Variables with P < 0.10 in univariate logistic regression were entered into the multivariate logistic regression model; results are expressed as odds ratio (OR) and 95%CI. Kaplan-Meier analysis with Log-rank test was used to compare OS and DFS. Multivariate Cox proportional hazards regression was used to identify independent prognostic factors, expressed as hazard ratio (HR) and 95%CI. All tests were two-sided; P < 0.05 was considered statistically significant.

RESULTS

Comparison of baseline characteristics between the AL and non-AL groups

Among 200 enrolled patients, 22 (11.0%) developed AL within 30 postoperative days. Baseline comparison revealed statistically significant differences in BMI (26.2 ± 3.6 vs 23.8 ± 3.2, P = 0.001), diabetes mellitus (36.4% vs 13.5%, P = 0.006), tumor location (rectal 68.2% vs 47.2%, P = 0.022), ASA grade III-IV (54.5% vs 33.7%, P = 0.038), neoadjuvant therapy (36.4% vs 21.3%, P = 0.043), protective stoma rate (18.2% vs 31.5%, P = 0.037), preoperative albumin < 35 g/L (40.9% vs 19.1%, P = 0.019), operation time (204.2 ± 46.5 vs 174.8 ± 40.2 minutes, P = 0.002), and intraoperative blood loss (P = 0.002) between the two groups. No significant differences were observed in gender, age, smoking and drinking history, hypertension, TNM staging, or differentiation degree (Table 1).

Table 1 Comparison of Baseline Characteristics Between AL and Non-AL Groups, n (%)/mean ± SD/mean (interquartile range).

Variable	AL group (n = 22)	Non-AL group (n = 178)	χ²/t/Z value	P value
Gender			0.286	0.593
Male	14 (63.6)	102 (57.3)
Female	8 (36.4)	76 (42.7)
Age (years)	63.2 ± 9.8	61.4 ± 10.6	0.771	0.442
BMI (kg/m²)	26.2 ± 3.6	23.8 ± 3.2	3.478	0.001
Smoking history	10 (45.5)	58 (32.6)	1.476	0.224
Drinking history	8 (36.4)	50 (28.1)	0.624	0.430
Diabetes mellitus	8 (36.4)	24 (13.5)	7.552	0.006
Hypertension	9 (40.9)	58 (32.6)	0.573	0.449
ASA grade			4.286	0.038
I-II	10 (45.5)	118 (66.3)
III-IV	12 (54.5)	60 (33.7)
Tumor location			5.243	0.022
Colon	7 (31.8)	94 (52.8)
Rectum	15 (68.2)	84 (47.2)
TNM staging			1.028	0.598
Stage I	4 (18.2)	42 (23.6)
Stage II	9 (40.9)	72 (40.4)
Stage III	9 (40.9)	64 (36.0)
Neoadjuvant therapy	8 (36.4)	38 (21.3)	2.512	0.043
Protective stoma	4 (18.2)	56 (31.5)	1.640	0.037
Preop albumin < 35 g/L	9 (40.9)	34 (19.1)	5.468	0.019
Operation time (minutes)	204.2 ± 46.5	174.8 ± 40.2	3.227	0.002
Intraop blood loss (mL)	250 (150-400)	150 (100-230)	-3.142	0.002

AL: Anastomotic leakage; BMI: Body mass index; ASA: American Society of Anesthesiologists.

Univariate logistic regression analysis of risk factors for AL

Univariate logistic regression identified BMI ≥ 25 kg/m², diabetes mellitus, rectal tumor location, ASA grade III-IV, neoadjuvant therapy, preoperative albumin < 35 g/L, operation time ≥ 180 minutes, and intraoperative blood loss ≥ 200 mL as statistically significant risk factors (P < 0.05), while protective stoma was a protective factor (P = 0.043). Neoadjuvant therapy reached P < 0.10 (P = 0.075) and was included in the multivariate model (Table 2).

Table 2 Univariate logistic regression analysis of risk factors for anastomotic leakage.

Variable	OR	95%CI	Wald χ²	P value
Age (≥ 65 years vs < 65 years)	1.248	0.632-2.464	0.538	0.463
Gender (male vs female)	1.136	0.562-2.296	0.102	0.749
BMI (≥ 25 kg/m² vs < 25 kg/m²)	2.762	1.268-6.016	6.872	0.009
Diabetes mellitus (yes vs no)	2.824	1.342-5.938	8.124	0.004
Hypertension (yes vs no)	1.384	0.682-2.810	0.824	0.364
Smoking (yes vs no)	1.678	0.802-3.512	1.896	0.168
ASA grade III-IV vs I-II	2.384	1.062-5.354	4.528	0.033
Tumor location (rectum vs colon)	2.402	1.046-5.516	4.286	0.038
TNM stage III vs I-II	1.288	0.582-2.852	0.484	0.487
Neoadjuvant therapy (yes vs no)	2.152	0.928-4.994	3.162	0.075
Protective stoma (yes vs no)	0.468	0.166-1.322	2.286	0.043
Preop albumin < 35 g/L (yes vs no)	2.826	1.224-6.522	6.172	0.013
Operation time ≥ 180 minutes (yes vs no)	2.648	1.164-6.022	5.528	0.019
Intraop blood loss ≥ 200 mL (yes vs no)	2.284	1.006-5.186	3.928	0.048

BMI: Body mass index; ASA: American Society of Anesthesiologists; OR: Odds ratio.

Multivariate logistic regression analysis of independent risk factors for AL

After adjustment for confounders, multivariate logistic regression confirmed seven independent risk factors and one protective factor. BMI ≥ 25 kg/m² (OR = 2.354, P = 0.013), diabetes mellitus (OR = 2.660, P = 0.010), rectal tumor location (OR = 2.440, P = 0.014), neoadjuvant therapy (OR = 2.186, P = 0.039), preoperative albumin < 35 g/L (OR = 2.680, P = 0.009), operation time ≥ 180 minutes (OR = 2.334, P = 0.020), and intraoperative blood loss ≥ 200 mL (OR = 2.142, P = 0.043) were independent risk factors. Protective stoma (OR = 0.403, P = 0.034) was confirmed as an independent protective factor against AL (Table 3).

Table 3 Multivariate logistic regression analysis of independent risk factors for anastomotic leakage.

Variable	β	SE	Wald χ²	OR	95%CI	P value
BMI ≥ 25 kg/m²	0.856	0.346	6.124	2.354	1.196-4.632	0.013
Diabetes mellitus	0.978	0.382	6.556	2.660	1.258-5.624	0.010
Tumor location (rectum)	0.892	0.362	6.084	2.440	1.198-4.970	0.014
Neoadjuvant therapy	0.782	0.378	4.282	2.186	1.042-4.584	0.039
Preop albumin < 35 g/L	0.986	0.378	6.804	2.680	1.278-5.620	0.009
Operation time ≥ 180 minutes	0.848	0.364	5.424	2.334	1.142-4.770	0.020
Intraop blood loss ≥ 200 mL	0.762	0.376	4.108	2.142	1.024-4.482	0.043
Protective stoma	-0.908	0.428	4.498	0.403	0.174-0.932	0.034

BMI: Body mass index; OR: Odds ratio.

Clinical grade, manifestations, and management of AL

Among 22 AL cases, 8 (36.4%) were grade A, 10 (45.4%) grade B, and 4 (18.2%) grade C. The most common clinical manifestations were fever (68.2%), abdominal pain (59.1%), and feculent drainage from abdominal drains (50.0%). All grade C patients presented with peritonitis signs and feculent drainage. The mean hospital stay was 24.2 ± 12.4 days in the AL group. Overall, 30-day mortality was 4.5% (1/22), occurring in the grade C subgroup (Table 4).

Table 4 Clinical grade, manifestations, and management of anastomotic leakage cases (n = 22), n (%)/mean ± SD/mean (interquartile range).

Characteristic	Grade A (n = 8)	Grade B (n = 10)	Grade C (n = 4)	Total (n = 22)	P value
Onset time (days)	6 (4-9)	8 (5-12)	5 (3-7)	7 (4-11)	0.312
Clinical manifestations
Fever	4 (50.0)	7 (70.0)	4 (100.0)	15 (68.2)	0.046
Abdominal pain	2 (25.0)	7 (70.0)	4 (100.0)	13 (59.1)	0.004
Feculent drainage	1 (12.5)	6 (60.0)	4 (100.0)	11 (50.0)	0.002
Peritonitis signs	0 (0.0)	2 (20.0)	4 (100.0)	6 (27.3)	0.001
Management
Conservative treatment	8 (100.0)	4 (40.0)	0 (0.0)	12 (54.5)	< 0.001
Interventional drainage	0 (0.0)	4 (40.0)	2 (50.0)	6 (27.3)	< 0.001
Surgical re-intervention	0 (0.0)	2 (20.0)	2 (50.0)	4 (18.2)	< 0.001
ICU admission	0 (0.0)	2 (20.0)	3 (75.0)	5 (22.7)	< 0.001
Hospital stay (days)	13.6 ± 3.4	25.8 ± 7.2	40.4 ± 13.8	24.2 ± 12.4	< 0.001
30-day mortality	0 (0.0)	0 (0.0)	1 (25.0)	1 (4.5)	0.021

ICU: Intensive care unit.

Comparison of short-term outcomes between AL and non-AL groups

The AL group had significantly worse short-term outcomes across all major parameters. Overall complication rate was 81.8% vs 22.5% (P < 0.001). Reoperation rate was 18.2% vs 3.4% (P = 0.007). Total hospital stay was markedly prolonged (24.2 ± 12.4 vs 10.8 ± 4.2 days, P < 0.001). Adjuvant chemotherapy was delayed in 45.5% vs 11.2% of patients (P < 0.001), and permanent stoma formation occurred in 54.5% vs 13.5% of cases (P < 0.001; Table 5).

Table 5 Comparison of short-term postoperative outcomes between anastomotic leakage and non-anastomotic leakage groups, n (%)/mean ± SD.

Outcome indicator	AL group (n = 22)	Non-AL group (n = 178)	χ²/t/Z	P value
Hospital stay (days)	24.2 ± 12.4	10.8 ± 4.2	7.268	< 0.001
Postop hospital stay (days)	19.6 ± 10.8	7.8 ± 3.4	6.924	< 0.001
Time to first flatus (days)	5.4 ± 1.8	3.6 ± 1.2	4.428	< 0.001
Time to oral feeding (days)	7.6 ± 2.8	4.4 ± 1.6	5.124	< 0.001
Overall complication rate	18 (81.8)	40 (22.5)	34.284	< 0.001
Wound infection	6 (27.3)	14 (7.9)	8.214	0.004
Abdominal abscess	5 (22.7)	8 (4.5)	11.246	0.001
Intestinal obstruction	4 (18.2)	10 (5.6)	4.428	0.035
Urinary tract infection	3 (13.6)	12 (6.7)	1.286	0.257
Pulmonary complications	4 (18.2)	16 (9.0)	1.680	0.195
Reoperation rate	4 (18.2)	6 (3.4)	7.354	0.007
30-day mortality	1 (4.5)	1 (0.6)	Fisher: 0.042	0.042
Permanent stoma rate	12 (54.5)	24 (13.5)	16.948	< 0.001
Adjuvant chemo delay	10 (45.5)	20 (11.2)	14.286	< 0.001

AL: Anastomotic leakage.

Impact of AL on long-term survival and Cox regression analysis

As of December 31, 2024, the median follow-up time was 38 months (range: 12-72 months). Eight patients were lost to follow-up (2 AL, 6 non-AL), overall loss-to-follow-up rate 4.0%. Kaplan-Meier analysis showed that the 5-year OS rate was significantly lower in the AL group (63.6% vs 82.6%, Log-rank χ² = 8.542, P = 0.003; Figure 1A), as was the 5-year DFS rate (54.5% vs 77.5%, Log-rank χ² = 10.384, P = 0.001; Figure 1B). Multivariate Cox regression confirmed TNM stage III (HR = 2.248, P < 0.001), poor differentiation (HR = 1.686, P = 0.013), vascular invasion (HR = 1.782, P = 0.005), AL (HR = 2.048, 95%CI: 1.312-3.198, P = 0.002), and adjuvant chemotherapy delay (HR = 1.624, P = 0.030) as independent risk factors for inferior DFS (Table 6).

Figure 1 Kaplan-Meier curves for overall survival and disease-free survival in the anastomotic leakage group (n = 22) and non-anastomotic leakage group (n = 178). A: Overall survival. The Anastomotic leakage (AL) group showed significantly lower overall survival compared to the non-AL group (log-rank χ² = 8.542, P = 0.003); B: Disease-free survival. The AL group demonstrated significantly lower disease-free survival compared to the non-AL group (log-rank χ² = 10.384, P = 0.001).

Table 6 Cox regression analysis of factors affecting disease-free survival.

Variable	Univariate HR	Univariate 95%CI	Univariate P value	Multivariate HR	Multivariate 95%CI	Multivariate P value
Age (≥ 65 years vs < 65 years)	1.264	0.802-1.992	0.246	-	-	-
Gender (male vs female)	1.128	0.724-1.758	0.558	-	-	-
BMI (≥ 25 kg/m² vs < 25 kg/m²)	1.248	0.802-1.942	0.278	-	-	-
TNM stage III vs I-II	2.624	1.786-3.854	< 0.001	2.248	1.526-3.312	< 0.001
Poor differentiation	1.924	1.286-2.878	0.001	1.686	1.118-2.544	0.013
Vascular invasion	2.024	1.368-2.994	< 0.001	1.782	1.188-2.672	0.005
Anastomotic leakage (yes vs no)	2.462	1.586-3.822	< 0.001	2.048	1.312-3.198	0.002
Adjuvant chemo delay (yes vs no)	1.824	1.186-2.806	0.006	1.624	1.048-2.518	0.030
Preop albumin < 35 g/L	1.624	1.028-2.564	0.038	-	-	-
Tumor location (rectum vs colon)	1.248	0.842-1.848	0.268	-	-	-

BMI: Body mass index; HR: Hazard ratio.

DISCUSSION

This single-center retrospective cohort study of 200 consecutive patients undergoing radical CRC surgery yielded an overall AL incidence of 11.0%, consistent with reported rates of 3%-19% in the literature[5,7,8]. Interpreted within the APOS theoretical framework and critical thinking ability assessment principles that guided this study’s design and analysis, the findings provide a structured, evidence-based basis for understanding AL risk as a multi-layered clinical construct rather than a collection of isolated associations. The comprehensive characterization of both risk determinants and prognostic consequences of AL in this study provides practical evidence for perioperative management optimization.

Obesity (BMI ≥ 25 kg/m²) was confirmed as an independent risk factor (OR = 2.354), consistent with published meta-analyses showing that increased BMI elevates AL risk through technical difficulty in pelvic dissection, impaired anastomotic perfusion, and systemic metabolic dysfunction[20]. Diabetes mellitus (OR = 2.660) was similarly confirmed as a potent independent risk factor, reflecting well-established mechanisms by which hyperglycemia impairs neutrophil function, collagen synthesis, and angiogenesis essential for tissue healing[21,22]. These findings reinforce the importance of preoperative glycemic optimization as a modifiable target in diabetic CRC patients scheduled for elective surgery.

Rectal tumor location conferred more than twofold increased AL risk (OR = 2.440) compared with colonic tumors, attributable to technical constraints of narrow pelvic anatomy, limited anastomotic blood supply, anastomotic tension, and tissue compromise from neoadjuvant radiotherapy[23]. Consistent with this, neoadjuvant therapy (OR = 2.186) was independently associated with AL, likely through radiation-induced fibrosis and impaired anastomotic healing[13,24]. These findings suggest that high-risk rectal cancer patients who have received neoadjuvant chemoradiotherapy should be considered candidates for routine defunctioning stoma creation[14,17].

Preoperative hypoalbuminemia (< 35 g/L) was confirmed as a highly significant independent risk factor (OR = 2.680, P = 0.009), underscoring the established relationship between malnutrition, impaired immune function, and suboptimal wound healing[22,25]. These data strongly support the routine preoperative nutritional assessment of all CRC surgical candidates and the implementation of nutritional prehabilitation in those identified as malnourished[15]. Operation time ≥ 180 minutes (OR = 2.334) and intraoperative blood loss ≥ 200 mL (OR = 2.142) reflect surgical complexity and physiological stress, both of which compromise anastomotic tissue perfusion and systemic homeostasis[18,19]. It is acknowledged that the median blood loss in the non-AL group was 150 mL (100-230 mL), indicating distributional overlap across the 200 mL threshold. The binary cutoff therefore does not imply that blood loss below this level confers no risk, nor that exceeding it deterministically predicts AL; rather, ≥ 200 mL represents a level at which AL risk is significantly elevated on a population basis after multivariate adjustment, and should be interpreted as a surrogate for overall intraoperative physiological burden rather than a single mechanistic cause.

The identification of protective stoma as an independent protective factor (OR = 0.403) is consistent with meta-analytic evidence demonstrating that defunctioning ileostomy significantly reduces the clinical consequence and reoperation rate of AL following rectal anastomoses[11,12]. Our data further support selective use of protective stoma in patients with concurrent risk factors, particularly rectal tumor location, neoadjuvant therapy, and hypoalbuminemia[14,17].

At the Schema level – the highest cognitive stage in the APOS framework – the accumulated risk object knowledge is synthesized into actionable clinical decision schemas for perioperative management. The impact of AL on short-term outcomes was substantial in our series, with 81.8% overall complication rate, 18.2% reoperation rate, and over 2-fold prolongation of hospital stay. Of particular clinical importance is the 45.5% rate of adjuvant chemotherapy delay in the AL group, compared to 11.2% in the non-AL group. Evidence suggests that delays in adjuvant chemotherapy initiation beyond 8 weeks from surgery are associated with inferior long-term survival in stage III CRC[9], making AL prevention directly relevant to oncological outcomes.

The prognostic impact of AL observed in this study – with significantly inferior 5-year OS (63.6% vs 82.6%) and DFS (54.5% vs 77.5%) – is supported by recent meta-analytic data[10,26]. Multivariate Cox regression confirmed AL as an independent adverse prognostic factor for DFS (HR = 2.048, P = 0.002) after adjustment for TNM stage, differentiation, and vascular invasion. The multifactorial mechanism likely includes: (1) Systemic inflammatory activation creating a protumorigenic microenvironment; (2) Delayed or suboptimal adjuvant chemotherapy; and (3) The psychological and functional burden of permanent stoma adversely affecting follow-up compliance[9,10,27]. The confirmation of these associations in a contemporary Chinese surgical cohort adds to the growing international evidence base[3,4,28,29]. The application of APOS theory as a theoretical framework and critical thinking ability assessment as a methodological basis contributes a novel analytical dimension to this field. Whereas most existing studies present risk factor analyses descriptively or statistically, the APOS-critical thinking approach foregrounds the cognitive and epistemological structure of risk knowledge construction – making explicit the progression from individual data measurement to integrated schema-level clinical decision-making. This framework may be particularly valuable in postgraduate surgical education, where the development of systematic perioperative reasoning skills is increasingly recognized as a core competency[30].

This study has several limitations. First, as a single-center retrospective cohort analysis with a relatively small sample size, selection bias and residual confounding cannot be fully excluded, and the results require cautious external generalization. Second, we were unable to fully capture all potentially relevant variables, such as specific intraoperative anastomotic air leak test results, use of omentoplasty or fibrin sealant, or detailed nutritional intervention protocols. Third, longer follow-up is needed to fully characterize the impact on 10-year oncological outcomes. Fourth, the relatively small number of AL events (n = 22) limits the statistical power of subgroup analyses. Fifth, while all operations were completed by attending surgeons in a single gastrointestinal surgical unit using standardized departmental protocols, surgeon volume and operative technique represent an additional potential confounder and should be adjusted for in future studies assessing AL risk. Sixth, enhanced recovery after surgery adherence at the individual level was not consistently scored because of study design, and differences in delivery between the AL and non-AL cohorts cannot be formally evaluated; this limits our ability to meaningfully evaluate the impact of perioperative care optimization on outcome differences. Limitations for this systemic review (study size, baseline population differences across studies). Impact of most risk factors on AL occurs after the surgery up to 30 days post-operative period. Multi-variable analysis precluded meta-analysis. Massachusetts General Hospital criteria used to define AL. Intervention Neoadjuvant therapy includes both long course concurrent chemoradiotherapy and short-course radiotherapy. Few AL events within treatment categories prevented email from analysing these variables order to determine effects upon decision whether or not patients should receive subsequent chemotherapy thus need for larger prospective trials.

CONCLUSION

In summary, AL after radical CRC surgery occurred in 11.0% of patients in this cohort and is significantly associated with BMI ≥ 25 kg/m², diabetes mellitus, rectal tumor location, neoadjuvant therapy, preoperative hypoalbuminemia, prolonged operation time, and excessive intraoperative blood loss as independent risk factors. Protective stoma creation significantly reduces the risk of clinically consequential AL. AL is independently associated with worse short-term surgical outcomes – including prolonged hospitalization, increased complications, and adjuvant chemotherapy delay – and significantly impairs 5-year OS and DFS. Comprehensive preoperative nutritional and metabolic optimization, judicious use of protective stoma in high-risk patients, and minimization of intraoperative trauma are recommended as key strategies to reduce AL incidence and mitigate its oncological impact. Prospective multicenter randomized studies with larger sample sizes and standardized perioperative protocols are warranted to validate and extend these findings. The APOS-informed critical thinking framework may further enhance the clinical interpretation and application of perioperative risk assessment.