Quantifying abdominal abscess burden using area under the curve methods inspired by dehiscence, infection/inflammation, seroma, hematoma classification

Abstract

BACKGROUND

Postoperative intra-abdominal abscesses remain a common complication of abdominal surgery and are associated with prolonged treatment and delayed recovery. Although the Clavien-Dindo (CD) classification is widely used to grade postoperative complications, patients with the same CD grade may experience substantially different clinical courses. Inspired by the intervention framework of the dehiscence, infection/inflammation, seroma, hematoma classification system, we developed an area under the curve (AUC)-based approach to quantify the cumulative treatment burden.

AIM

To evaluate the utility of an AUC-based measure of the cumulative burden and factors associated with postoperative intra-abdominal abscesses abscess treatment.

METHODS

We retrospectively analyzed 59 patients with postoperative intra-abdominal abscesses. Treatment intensity was graded according to the level of intervention required for abscess management and integrated over time using an AUC-based approach. Patients were stratified into high- and low-AUC groups using the upper quartile (19.5) as the cutoff.

RESULTS

Diabetes mellitus (P = 0.016), greater intraoperative blood loss (P = 0.007), and abscess size ≥ 5 cm (P = 0.037) were associated with higher AUC values. CD grade III complications were more frequent in the high-AUC group than in the low-AUC group (P < 0.001). Despite this association, substantial variation in the treatment burden remained within individual CD grades, with patients in the high-AUC group requiring significantly longer treatment courses and hospital stays.

CONCLUSION

This AUC-based approach identifies clinically meaningful variations in treatment burden that are not captured by CD grade alone.

Key Words: Clavien-Dindo classification; Postoperative intra-abdominal abscesses; Treatment burden; Abscess drainage; Comprehensive complication index

Core Tip: Patients with the same Clavien-Dindo grade may undergo markedly different treatment courses for postoperative intra-abdominal abscesses. Inspired by the intervention framework of the dehiscence, infection/inflammation, seroma, hematoma classification, we developed an area under the curve based measure that incorporated both treatment intensity and treatment duration. This approach may help quantify the overall burden of infection management and identify postoperative infections that are difficult to treat.

INTRODUCTION

Postoperative intra-abdominal abscesses are common complications of abdominal surgery associated with prolonged hospitalization, increased healthcare costs, prolonged antibiotic exposure, and impaired postoperative recovery. Despite advances in perioperative management and minimally invasive surgical techniques, postoperative intra-abdominal abscesses remain clinically important because some patients require prolonged treatment or additional interventions before complete resolution[1].

Several classification systems have been developed to evaluate postoperative complications. The Clavien-Dindo (CD) classification is currently the most widely accepted system and grades complications according to the most invasive therapeutic intervention required during the clinical course[2]. Although the CD classification provides a simple and reproducible framework for reporting complications, it primarily reflects the highest level of intervention and does not account for the duration of treatment or the longitudinal clinical course.

More recently, the dehiscence, infection/inflammation, seroma, hematoma (DISH) classification has been proposed as a procedure-specific grading system for surgical site infections. Unlike the CD classification, the DISH classification recommends documenting individual infection-related events and therapeutic interventions throughout the treatment course, thereby providing a more detailed description of infectious complications[3]. However, because the DISH classification is relatively new, data regarding its practical application in clinical research remain limited, and no standardized method has been established for integrating serial DISH assessments into a quantitative measure of the overall disease burden.

In daily clinical practice, treatment difficulty is determined not only by the invasiveness of a single intervention but also by the cumulative burden of management over time. For example, a patient requiring prolonged antibiotic therapy may experience a greater treatment burden than a patient who undergoes a single drainage procedure followed by rapid recovery despite receiving a lower CD grade.

To address this, we adapted the infection component of the DISH classification and transformed serial treatment-related events into a time-integrated quantitative metric using area under the curve (AUC) approach. By incorporating both treatment intensity and duration, this method allows dynamic assessment of the entire treatment course of postoperative intra-abdominal abscesses.

This study aimed to evaluate the clinical utility of an AUC-based adaptation of the DISH infection score and to investigate the factors associated with a high cumulative treatment burden in patients with postoperative intra-abdominal abscesses.

MATERIALS AND METHODS

Study design and patient selection

This retrospective single-center study included consecutive patients diagnosed with postoperative intra-abdominal abscesses at the Department of Gastroenterological Surgery of Itami City Hospital between December 2019 and November 2025.

Eligible patients were adults (≥ 18 years) with postoperative intra-abdominal abscesses confirmed by computed tomography (CT), together with clinical and/or laboratory evidence of infection. Patients diagnosed after hospital discharge and subsequently readmitted for abscess treatment were included.

Patients were excluded if the infectious process was distinct from that of a postoperative intra-abdominal abscess. Patients with a postoperative pancreatic fistula, anastomotic leakage, acute cholecystitis, acute pancreatitis, or fever of unknown origin were excluded from the study. Patients without radiological evidence of an intra-abdominal fluid collection on CT were also excluded.

Patients with postoperative pancreatic fistula and anastomotic leakage were excluded because these conditions represent distinct pathological entities with treatment strategies and clinical courses that differ from those of postoperative intra-abdominal abscesses.

Clinical management

The indications for intra-abdominal drainage and the selection of the drain type were determined at the operating surgeon’s discretion. Postoperative intra-abdominal abscesses were diagnosed based on clinical findings, laboratory evidence of infection, and CT findings. When an abscess was identified, treatment options included antibiotic therapy alone, image-guided drainage under local anesthesia (ultrasound- or CT-guided drainage), and surgical drainage under general anesthesia. Antibiotic administration included both oral and intravenous routes. The treatment choice was determined by the attending surgeon based on the patient’s clinical condition and radiological findings. The maximum diameter of the abscess was measured on the CT images and recorded for analysis. All patients were hospitalized at the time abscess treatment was initiated.

Construction of the AUC-based DISH infection score

To quantify the cumulative treatment burden associated with postoperative intra-abdominal abscesses, we adapted the Inflammation/Infection Management Scale of the DISH classification into a time-integrated scoring system.

The DISH classification evaluates surgical-site complications using both clinical presentation and clinical intervention/management scales. Within the inflammation/infection domain, the management intensity was categorized according to the level of therapeutic intervention required during the clinical course. Based on this concept, we developed a time-dependent intervention score for post-operative intra-abdominal abscesses.

For this study, the four intervention levels were defined as follows: Level 1, clinical observation without therapeutic intervention; level 2, administration of systemic antibiotics; level 3, image-guided drainage under local anesthesia; and level 4, surgical drainage under general anesthesia.

The dates of abscess diagnosis, initiation of antibiotic therapy, drainage procedures, reoperation, and completion of treatment were retrospectively identified for each patient using electronic medical records. At the time of abscess diagnosis, the patients entered the scoring system at level 1, and the intervention score subsequently changed according to the treatment provided during the clinical course. The day of surgery was defined as the baseline for the graph, with all patients having an initial score of zero.

The treatment course was plotted with postoperative days on the x-axis and the intervention score on the y-axis. The observation period began on the day of surgery and continued until the completion of the abscess treatment, which was defined as the discontinuation of antibiotic therapy. The AUC was calculated using the trapezoidal rule (Figure 1). A larger AUC reflects a longer treatment duration, greater management intensity, or both.

Figure 1 Examples of area under the curve. A: The patient was diagnosed with an abscess on postoperative day 14, was started on antibiotics on postoperative day 15, and completed antibiotic therapy on postoperative day 31. The area under the curve (AUC) was 24.5. (0 + 1)/2 × 14 + (1 + 2)/2 × (15 - 14) + (2 + 0)/2 × (31 - 15) = 24.5. This patient was classified into the high-AUC group; B: On postoperative day 6, the patient was diagnosed with an abscess and simultaneously started on antibiotics. The patient underwent abscess drainage on postoperative day 7, the drain was removed on postoperative day 13, and antibiotic therapy was completed on postoperative day 14. The AUC was 19.0. (0 + 1)/2 × 6 + (2 + 3)/2 × (7 - 6) + (3 + 2)/2 × 1 + (2 + 2)/2 × (13 – 7 - 1) + (2 + 0)/2 × (14 - 13) = 19. This patient was classified into the low-AUC group.

For example, patients treated with antibiotics alone typically followed an intervention score transition of 0→1→2→0. In contrast, patients requiring image-guided drainage or surgical drainage demonstrated temporary escalation to level 3 or level 4 during treatment, resulting in transitions such as 0→1→2→3→2→0 or 0→1→2→4→2→0.

The distribution of the AUC values was examined, and the patients were subsequently stratified into high- and low-AUC groups using the upper quartile of the AUC distribution as the cutoff value.

Statistical analysis

Continuous variables were presented as medians and interquartile ranges. Categorical variables were presented as numbers and percentages. Comparisons were performed using the Mann-Whitney U test for continuous variables and the χ² test or the Fisher's exact test for categorical variables, as appropriate. Associations were assessed using univariable logistic regression and expressed as odds ratios with 95% confidence intervals. Statistical significance was set at P < 0.05. JMP 17.0 (SAS Institute, Cary, NC, United States) was used for all statistical analyses. Missing data were excluded from the analysis.

RESULTS

Patient characteristics

A total of 59 patients with postoperative intra-abdominal abscesses were included in this study. The median age was 65 years [interquartile range (IQR): 59-77 years], and 36 patients (61.0%) were male. Forty-two patients were treated with antibiotic therapy alone, whereas 17 patients required image-guided or surgical drainage. Baseline patient characteristics are summarized in Table 1. Malnutrition was assessed based on the criteria of the Global Leadership Initiative on Malnutrition[4]. A total of 19 patients (32.2%) had malignancies. Emergency surgery was performed in 36 (61.0%) patients. Regarding the surgical approach, 12 patients (20.3%) underwent open surgery and the rest underwent laparoscopic or robotic surgery. The median operative time and estimated blood loss were 168 minutes (IQR: 93-275 minutes) and 15 mL (IQR: 5-100 mL), respectively. In addition, 33 patients (55.9%) had intraoperative drains.

Table 1 Baseline patient characteristics.

Characteristics	Total (n)	n (%)
Age, years, median (IQR)	59	65 (59-77)
Male	59	36 (61.0)
ASA classification	59
I		7 (11.9)
II		39 (66.1)
III		12 (20.3)
IV		1 (1.7)
Smoker	59	32 (54.2)
Diabetes mellitus	59	15 (25.4)
BMI, kg/m2, median (IQR)	57	22.2 (20.4-24.5)
Malnutrition (GLIM criteria)	59	28 (47.5)
Malignancy	59	19 (32.2)
Emergency surgery	59	36 (61.0)
Laparotomy	59	12 (20.3)
Operative time, minutes, median (IQR)	59	168 (93-275)
Intraoperative blood loss (mL)	59	15 (5-100)
Intraoperative drain placement	59	33 (55.9)
Wound class	59
Clean operation		1 (1.7)
Clean-contaminated operation		19 (32.2)
Contaminated operation		5 (8.5)
Dirty/infected operation		34 (57.6)
Procedure	59
Appendectomy		19 (32.2)
Colorectal surgery		24 (40.7)
Upper gastrointestinal surgery		6 (10.2)
Hepatobiliary and pancreatic surgery		7 (11.9)
Others		3 (5.1)

ASA: American Society of Anesthesiologists; BMI: Body mass index; GLIM: Global leadership initiative on malnutrition; IQR: Interquartile range.

Distribution of AUC values

The median AUC was 13.0 (IQR: 9.75 -19.5; range, 3.0-91.0). The distribution of the AUC values is shown in Figure 2. The upper quartile of the AUC distribution was 19.5. Accordingly, 16 and 43 patients were classified into high-and low-AUC groups, respectively.

Figure 2  Distribution of the area under the curve values.

Factors associated with a high AUC

The comparisons between the high- and low-AUC groups are summarized in Table 2. Diabetes mellitus was significantly more common in the high-AUC group than in the low-AUC group (50.0% vs 16.3%, P = 0.016). Patients in the high-AUC group had significantly greater intraoperative blood loss (80 mL vs 15 mL, P = 0.007) and were more likely to have abscesses measuring ≥ 5 cm in diameter (68.8% vs 34.9%, P = 0.037). Emergency surgery was less frequent in the high AUC group than in the low AUC group (37.5% vs 69.8%, P = 0.036).

Table 2 Comparisons between the high- and low-area-under-the-curve groups, n (%).

Characteristics	Variable coding	High-AUC (n = 16)	Low-AUC (n = 43)	Odds ratio (95%CI)	P value
Age, years, median (IQR)	Per 1-year	74 (69.3-76.3)	68 (55.0-75.5)	1.04 (1.00-1.10)	0.0653
Male	Yes/No	9 (56.3)	27 (62.8)	1.31 (0.40-4.22)	0.7661
ASA score	III, IV/I, II	5 (31.3)	8 (18.6)	1.99 (0.51-7.31)	0.3106
Smoker	Yes/No	6 (37.5)	26 (60.5)	0.39 (0.11-1.25)	0.1473
Diabetes mellitus	Yes/No	8 (50.0)	7 (16.3)	5.14 (1.46-19.2)	0.0161
BMI, kg/m2, median (IQR)	Per 1-unit	22.3 (20.3-24.7)	22.2 (20.5-24.3)	0.94 (0.77-1.10)	0.8036
Malnutrition (GLIM criteria)	Yes/No	2 (12.5)	5 (11.6)	1.08 (0.14-5.70)	1.0000
Malignancy	Yes/No	9 (56.3)	16 (37.21)	2.17 (0.68-7.19)	0.2410
Emergency	Yes/No	6 (37.5)	30 (69.77)	0.26 (0.07-0.84)	0.0358
Laparotomy	Yes/No	5 (31.3)	7 (16.3)	2.34 (0.59-8.89)	0.2766
Operative time, minutes, median (IQR)	Per 1-minute	198 (122.8-266.8)	144 (78.5-271.5)	1.00 (0.97-1.01)	0.2228
Intraoperative blood loss, mL, median (IQR)	Per 1-mL	80 (25-115)	15 (5-40)	1.00 (1.00-1.01)	0.0074
Intraoperative drain placement	Yes/No	9 (56.3)	24 (55.8)	0.90 (0.28-3.01)	1.0000
Infected operation	Yes/No	9 (56.3)	25 (58.1)	0.93 (0.29-3.03)	1.0000
Time to abscess diagnosis, days, meidan (IQR)	Per 1-day	9.5 (5.8-12.5)	7 (5.0-9.0)	1.09 (1.00-1.22)	0.0915
Diameter of abscess > 5 cm	Yes/No	11 (68.8)	15 (34.9)	4.11 (1.25-15.2)	0.0371
Multiple abscesses	Yes/No	2 (12.5)	10 (23.3)	0.47 (0.07-2.09)	0.4822
Postoperative hospital stays, days, median (IQR)	Per 1-day	24 (17.8-32.0)	17 (11.5-21.5)	1.07 (1.01-1.14)	0.0070
Duration of antibiotics, days, median (IQR)	Per 1-day	17 (13.3-21.3)	9 (7.0-13.0)	1.26 (1.11-1.51)	0.0003
Abscess to discharge, days, median (IQR)	Per 1-day	16 (13.8-25.3)	11 (8.5-16.0)	1.08 (1.02-1.16)	0.0074

Time to abscess diagnosis: Number of days from surgery to abscess diagnosis; abscess to discharge: Number of days from abscess diagnosis to hospital discharge; AUC: Area under the curve; ASA: American Society of Anesthesiologists; BMI: Body mass index; GLIM: Global Leadership Initiative on Malnutrition; IQR: Interquartile range.

There were no significant differences between the groups with respect to age, sex, body mass index, smoking status, nutritional status, operative approach, drain placement, or presence of multiple abscesses. Age tended to be higher in the high-AUC group than in the low-AUC group, although the difference was not statistically significant (74 years vs 68 years, P = 0.065). The post-diagnosis length of stay tended to be longer in the high-AUC group (16 days vs 11 days, P = 0.074).

Relationship between AUC and the CD classification

Among the 59 patients, 42 were classified as having CD grade II and 17 as having CD grade III. CD grade III complications were significantly more frequent in the high-AUC group than in the low-AUC group (68.8% vs 14.0%, P < 0.001; Figure 3A).

Figure 3 Comparison of high- and low-area under the curve groups. A: The proportion of patients with Clavien-Dindo (CD) grade II and III complications; B: Postoperative hospital stays of patients with CD grade II complications (26.2 days vs 17 days, P = 0.026); C: Abscess treatment durations of patients with CD grade II complications (18.4 days vs 7 days, P = 0.0004); D: Abscess treatment durations of patients with CD grade III complications (18 days vs 8 days, P = 0.044). AUC: Area under the curve; CD: Clavien-Dindo.

Despite this association, substantial heterogeneity in the treatment burden remained within individual CD grades. Among patients with CD grade II complications, those in the high-AUC group had significantly longer postoperative hospital stays (26.2 days vs 17 days, P = 0.026) and longer abscess treatment durations (18.4 days vs 7 days, P = 0.0004) than those in the low-AUC group (Figure 3B and C).

Similarly, among patients with CD grade III complications, the abscess treatment duration was significantly longer in the high-AUC group than in the low-AUC group (18 days vs 8 days, P = 0.044; Figure 3D), whereas the postoperative hospital stay did not differ significantly between the groups (28 days vs 28 days, P = 0.960).

DISCUSSION

In this study, we developed an AUC-based measure of the cumulative treatment burden based on the intervention framework of the DISH classification. The principal finding was that patients with the same CD grade often underwent substantially different treatment courses. Even within identical CD grades, patients with higher AUC values required longer treatment and hospitalization times than those with lower AUC. These findings suggest that the proposed approach may provide additional information regarding the clinical burden of postoperative intra-abdominal abscesses.

The intervention framework of the DISH classification provided a conceptual basis for this model. Postoperative intra-abdominal abscesses are typically managed through a stepwise escalation of treatment intensity, ranging from observation and antibiotic therapy to image-guided drainage and surgical reintervention[3]. Inspired by this framework, we assigned numerical values to different intervention levels and integrated them over time using the AUC approach. This enabled the simultaneous assessment of treatment intensity and duration throughout the clinical course.

The CD classification remains the most widely accepted system for reporting postoperative complications because of its simplicity and reproducibility[5]. However, patients with the same CD grades may undergo markedly different treatment courses[6]. As illustrated in Figure 3B-D, considerable heterogeneity remained within individual CD grades in our study, suggesting that the proposed AUC-based approach may complement conventional complication grading by capturing the cumulative treatment burden[7].

Several factors were associated with a higher AUC, including diabetes mellitus, greater intraoperative blood loss, and larger abscess size. These findings are clinically plausible because impaired immunity, delayed wound healing, and a greater infectious burden may contribute to prolonged treatment and recovery[8,9]. Although age was not statistically significant, patients with higher AUC values tended to be older, suggesting that reduced physiological reserves may influence the duration of infection management[10]. The high-AUC group had a lower rate of emergency surgery. Possible explanations for this finding include the poor general patient status at baseline and the high technical difficulty of the surgery. Our findings suggest a longer length of hospital stay after abscess diagnosis in the high-AUC group. This finding supports the clinical utility of AUC. For reference, the highest AUC value was observed in a patient who underwent three separate abscess drainages at different sites, followed by a patient who underwent abscess drainage during a reoperation.

An important feature of the proposed approach is its ability to integrate treatment intensity and duration into a single quantitative measure. A comprehensive complication index was developed to summarize the cumulative impact of postoperative complications rather than solely focusing on the most severe events[6]. Similarly, our AUC-based approach aimed to quantify the overall burden of postoperative intra-abdominal abscess management during the treatment course. Because the score can be calculated using routinely collected clinical data, it may provide a practical tool for identifying difficult-to-treat infections, comparing treatment strategies, and evaluating healthcare resource utilization. In this study, we retrospectively reviewed the clinical course of patients during hospitalization using electronic medical records. However, drains are sometimes managed in outpatient settings in other countries, which raises concerns data. If routine scoring is consistently implemented without data loss, it can serve as an objective evaluation criterion. We also considered the duration between the date of surgery and the date of abscess diagnosis. This was based on the assumption that the abscess had gradually formed during this period. However, the exact onset of abscess formation could not be definitively determined.

This study has limitations. First, the proposed score was inspired by the intervention framework of the DISH classification rather than representing a direct application of the DISH system. Second, data collection is challenging in settings without access to electronic medical records. Third, the weighting assigned to each intervention level was not formally validated, and may not fully reflect the true clinical burden of individual treatments. The model assumed equal intervals between the intervention levels, although the burdens associated with observation, antibiotic therapy, drainage, and reoperation may not be linear. Therefore, it may be necessary to assign different score to oral and intravenous antibiotics. Weighting of the assigned values according to the level of each intervention may better reflect the patient burden. Although abscesses may have negatively impacted patient-reported outcomes and quality of life, such patient burdens were not incorporated into the current score. In addition, there is certainly room for refinement in how we evaluate pre-diagnostic patient burden in future studies. Finally, this was a retrospective single-center study with a relatively small sample size, and external validation was not performed. The sample size was insufficient to perform a multivariate analysis. As this study is strictly exploratory in nature, we did not perform corrections for multiple comparisons, and the results should be interpreted with caution. Further studies with larger sample sizes are warranted. Future multicenter studies are needed to optimize the score weighting and validate the reproducibility and clinical utility of this approach.

CONCLUSION

The proposed AUC-based approach quantified the cumulative treatment burden in patients with postoperative intra-abdominal abscesses. Incorporating both treatment intensity and duration, it may complement conventional complication grading and help identify difficult-to-treat infections.