Gastroenterology

Diagnostic nomogram for closed-loop small bowel obstruction requiring emergency surgery

a b s t r a c t

Purpose: This study aimed to build a diagnostic model of closed-loop Small bowel obstruction (CL-SBO) using clinical information, blood test results, and computed tomography findings.

Methods: All patients who were diagnosed with small bowel obstruction (SBO) and underwent surgery between January 1, 2018, and October 31, 2021, in the affiliated hospital of Qingdao university were reviewed, and their relevant preoperative information was collected. All variables were selected using univariate analysis and back- ward stepwise regression to build a diagnostic nomogram model. K-fold cross-validation and bootstrap resam- pling techniques were used for internal validation, and data from Qingdao Central Hospital were used for external validation. We also evaluated the diagnostic performance of each CT finding and performed subgroup analysis according to Bowel ischemia in the closed-loop small bowel obstruction (CL-SBO) group.

Results: A total of 219 patients (95 in the CL-SBO group and 124 in the open-loop small bowel obstruction [OL-SBO] group) were included in our research. D-dimers (median 1085 vs. 690, P = 0.019), tenderness (77.9% vs. 59.7%, P = 0.004), more than one beak sign (65.3% vs. 30.6%, P < 0.001), radial distribution (18.9% vs. 6.5%, P = 0.005), whirl sign (35.8% vs. 8.9%, P < 0.001), and ascites (71.6% vs. 53.2%, P = 0.006) were selected as the predictive variables of the nomogram. This model’s Harrell’s C statistic was 0.786 (95% confidence interval (CI), 0.724-0.848), and the Brier Score was 0.182. The Harrell’s C statistic of external validation was 0.784 (95%CI, 0.664-0.905); the Brier score was 0.190. Regarding the CT findings, radial distribution, U/C-shaped loop, and whirl sign had high specificity (93.5%, 96.0%, and 91.1%, respectively), but low sensitivity (18.9%, 8.4%, and 35.8%, respectively). D-dimer levels and tenderness were also associated with bowel ischemia.

Conclusion: The nomogram accurately predicted CL-SBO in patients with SBO, and surgery should be considered when patients have a high risk for developing CL-SBO.

(C) 2022 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://

creativecommons.org/licenses/by-nc-nd/4.0/).

  1. Introduction

Small bowel obstruction (SBO) is one of the most common acute abdominal diseases. Known since the time of Hippocrates, the Operative treatment of this disorder was not accepted until the 19th century when the use of antibiotics and anesthetics became widespread [1]. Nowadays, about 3 million patients with SBO undergo surgery in the

Abbreviations: SBO, small bowel obstruction; CL-SBO, closed-loop small bowel ob- struction; CT, computed tomography; OL-SBO, open-loop small bowel obstruction; HIS, Hospital Information System; MBO, malignant bowel obstruction; WBC, white blood cell; CRP, C-reactive protein; Alb, albumin; PCT, Procalcitonin; ROC, receiver operating characteristic; PPV, positive predictive value; NPV, negative predictive value; CI, confi- dence interval.

* Corresponding author at: Department of Emergency General Surgery, The Affiliated Hospital of Qingdao University, Qingdao, Shandong 266003, China.

E-mail address: [email protected] (S. Li).

United States, accounting for 12-16% of all surgical patients [2-4]. According to the position of the obstruction, SBO can be divided into open-loop small bowel obstruction (OL-SBO) and closed-loop small bowel obstruction (CL-SBO). In OL-SBO, the bowel is obstructed at one point, while in CL-SBO, there are two related points of obstruction.

The Bologna guidelines show that a sign of a closed loop is one of the three indications for emergent surgery [5]. CL-SBO is more likely to cause ischemia and failure of non-operative treatment [6-9]. Patients are often treated surgically when suspected to have CL-SBO, whether they have bowel ischemia or not [8,9]. Therefore, early diagnosis of CL-SBO plays a crucial role in the treatment of SBO.

Computed tomography (CT) is the most important diagnostic tool for SBO, with a sensitivity and specificity of 91% and 89%, respectively, according to a meta-analysis in China [10]. However, CT is not an ideal tool for distinguishing between CL-SBO and OL-SBO. In a retrospective study by Makar et al. [11], five radiologists independently evaluated

https://doi.org/10.1016/j.ajem.2022.10.022

0735-6757/(C) 2022 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

patients’ preoperative images and showed poor to median sensitivity, proving that the subjective judgment of radiologists is not always reli- able. Currently, studies on CL-SBO have focused on specific CT signs, with a few investigations focusing on the diagnostic methods. There- fore, we aimed to develop and validate a diagnostic prediction model for CL-SBO. We hypothesized that CL-SBO can be diagnosed through a combination of clinical presentations and laboratory and imaging tests.

  1. Methods
    1. Data source and patients

The study was designed according to the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis statement for reporting [12] and was approved by the local ethics com- mittee of our institution. We gathered all the patients’ information from January 1, 2018, to October 31, 2021, using the Hospital Information System (HIS) of the affiliated hospital of Qingdao University. We then collected all patient information in Qingdao Central Hospital from Janu- ary 1, 2018, to December 31, 2021, as our external validation dataset. Patients were included if they met all three inclusion criteria: (1) diag- nosed with SBO by surgery; (2) patients 18 and older at the time of sur- gery; (3) and underwent a Preoperative CT scan and had the CT images available. The exclusion criteria were as follows: (1) not diagnosed with SBO; (2) not undergoing surgery from the point of presentation until discharge; (3) having malignant bowel obstruction (MBO), that is, SBO caused by tumors (in our opinion, once a patient had a tumor, the indications for surgery were not determined by whether a patient has CL-SBO); (4) having SBO caused by an external hernia, such as an abdominal wall hernia; (5) younger than 18 years of age at the time of surgery; (6) being diagnosed with intussusception intraoperatively; and (7) having inaccessible preoperative CT images.

    1. Predictor variables

Using surgical findings as the standard, our patients were divided into the CL-SBO and OL-SBO groups. The intraoperative reports were re- corded by >10 gastrointestinal surgeons from five different surgical groups in our hospital. The proportion of bowel ischemia in both groups was determined based on surgical findings. Patient characteristics (sex, age, smoking), symptoms (abdominal pain, abdominal distension, nausea, and vomiting), signs (tenderness and rebound tenderness), history of the surgery, and blood test results (white blood cell [WBC], neutrophil and lymphocyte counts, C-reactive protein [CRP], procalcitonin [PCT], lactic acid, albumin [Alb], and d-Dimer) were

collected. Symptoms and signs were recorded within 24 h after admis- sion. History of surgery only included abdominal and pelvic surgeries. All blood test results were the results obtained within three days postoperatively.

CT predictor variables included the radial distribution (Fig. 1), U/C-shaped dilated bowel loop, whirl sign (Fig. 2), beak sign (Fig. 3), number of beak signs (greater or <1), small bowel feces sign, mesen- teric haziness, small bowel wall thickening (Fig. 3), ascites, and the lon- gest diameter of the small bowel. Our hospital uses a somatom sensation dual-source spiral CT machine and a somatom sensation 64- slice spiral CT machine with a nominal section thickness of 0.5 mm, tube voltage of 120 kV, and tube rotation time of 0.25-0.35 s. Two gen- eral surgeons with four (YL Li) and six (Z Tian) years of experience in performing abdominal CT independently evaluated the images. The dif- ferent results were judged by one of the two senior radiologists (WQ Bi or QL Ji) and a senior surgeon (SK Li) with >35 years of experience.

    1. Univariate analysis

All corrected variables between the two groups were compared. The Shapiro-Wilk was used to test the normality of the distribution. Continuous variables are shown as mean +- standard deviation and an- alyzed using the Student’s t-test if they conform to normal distribution. Variables are presented as median (percentage 25 [P25], percentage 75 [P75]) and analyzed using the Wilcoxon Mann-Whitney U test. Missing values were replaced with multiple imputations. Categorical variables were analyzed using the chi-squared test. Differences were considered statistically significant at P < 0.05. IBM SPSS Statistics (Version 26.0, IBM Corporation, New York, US) was used for all primary data analyses.

    1. Derivation and validation of the model

Variables that had significant differences on univariate analysis were then selected by backward stepwise regression. Akaike information cri- terion (AIC) was used for evaluation during the process. Some variables were also added or deleted according to the suggestions of profes- sionals. A diagnostic model, shown as a nomogram, was constructed using all the selected variables. The model’s discriminative ability was quantified using the receiver operating characteristic (ROC) curve and Harrell’s C statistic, and the calibration was evaluated using a calibration plot [13]. K-fold cross-validation and bootstrap resampling were used for internal validation. Data from Qingdao central hospital dataset were analyzed for external validation. The final model obtained from the derivation group was applied to the external validation group. R (version 4.0.3) was used for the nomogram creation.

Image of Fig. 1

Fig. 1. Radial distribution in computed tomography (CT) and the intraoperative photograph. The patient was an Elderly woman. A radial distribution is seen on preoperative CT. During Exploratory laparotomy, the upper abdominal cavity has heavily adhered, and an adhesion zone has formed between the sigmoid colon and Small intestine mesentery, about 70 cm from the ligament of Treitz. The small intestine is herniated into it, resulting in Ischemic necrosis of the herniated small intestine and proximal intestinal obstruction. The small intestine is resected (about 80 cm). The asterisk shows the radial distribution. The arrow points to inter-intestinal ascites.

Image of Fig. 2

Fig. 2. Whirl sign on CT and the intraoperative photograph. The patient was an elderly man. Preoperative CT showing a whirl sign (white arrow). Intraoperative exploration of the whole small intestine showing clockwise torsion 360?. Small intestine is highly dilated, showing dark red color. Ascending colon connecting with the abdominal wall incision. Left side of the transverse colon adhering to the residual stomach.

    1. Performance of CT signs

To evaluate the diagnostic performance of each CT sign, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV), and their 95% confidence intervals (CIs) were calculated using IBM SPSS Statistics.

  1. Results
    1. Study population

Using the HIS of our hospital, 863 patients with SBO who had under- gone surgery were analyzed. A total of 477 patients had SBO secondary to tumors. Patients who underwent surgery for intussusception (n = 24), large bowel obstruction (n = 7), external hernia (n = 22), or others predictive model“>(n = 254) were excluded. Finally, 95 patients with CL-SBO and 124 with OL-SBO were confirmed surgically. In the CL-SBO and OL-SBO groups, 55 (57.9%) and 23 (18.5%) patients had bowel ischemia, respectively (Fig. 4).

    1. Predictive model variables

Table 1 presents the results of the predictive model variables on uni- variate analysis. There were no significant differences in baseline char- acteristics (age, sex, smoking status, and history of surgery) between the two groups. Regarding clinical features, there was less Abdominal distention (65.3% vs. 82.3%, P = 0.004) and more tenderness (77.9%

Image of Fig. 3

Fig. 3. Beak sign and Small bowel wall thickening on CT. The patient was an elderly man. Preoperative CT shows a beak sign (white arrow) and a small bowel wall thickening (as- terisk). During the surgery, a ring-shaped adhesion was formed between a diverticulum about 70 cm from the ileocecal area and the surrounding adhesions, causing a closed- loop small bowel obstruction. The affected small bowel was gray-green in color.

vs. 59.7%, P = 0.004) and rebound tenderness (30.5% vs. 15.3%, P = 0.007) in patients in the CL-SBO group compared with those in the OL-SBO group. No significant differences in nausea, vomiting, and ab- dominal pain were found between the groups. Blood tests showed that WBC (8.18 vs 6.42 x 109/L, P = 0.002), neutrophil (6.31 vs 4.32 x 109/L, P < 0.001), and D-dimer (1085 vs. 690 ng/ml, P = 0.019) levels were higher in the CL-SBO group, and there was no significant differ- ence in lymphocyte, CRP, PCT, lactic acid, and Alb levels between the two groups. Regarding CT findings, beak sign (78.9% vs. 50%, P < 0.001), >1 beak sign (65.3% vs. 30.6%, P < 0.001), radial distribution (18.9% vs. 6.5%, P = 0.005), whirl sign (35.8% vs. 8.9%, P < 0.001), mes-

enteric haziness (85.3% vs. 66.1%, P = 0.001), and ascites (71.6% vs. 53.2%, P = 0.006) were more in the CL-SBO group. U/C-shaped loop, signs of small bowel feces, and small bowel wall thickening were not significantly different between the two groups.

    1. Predictive model

Backward stepwise regression demonstrated that distention, D-dimer, tenderness, >1 beak sign, whirl sign, and ascites were included in the final predictive model. Since distention is a subjective indicator that may cause inaccurate diagnosis, it was omitted from the model. Radial distribution showed a significant difference on univariate analysis and is considered an important CT finding [2,6,11]. Therefore, radial distribution was in- cluded in the final model. The diagnostic model used as a nomogram is

Image of Fig. 4

Fig. 4. Flowchart of the patients.

Table 1

Univariate analysis of variables in the predictive model.

CL-SBO (n = 95)

OL-SBO (n = 124)

Mean +- SD

Mean +- SD

Difference value

95%CI

P value

Age

60.52 +- 13.294

59.18 +- 15.354

1.338

(-5.23 to 2.56)

0.499

Alb

37.354 +- 6.1143

35.660 +- 7.1026

1.694

(-0.13 to 3.52)

0.069

Longest diameter

36.002 +- 8.9569

38.484 +- 9.4792

-2.482

(-4.97 to 0.01)

0.051

Median (P25, P75)

Median (P25, P75)

Difference value

95%CI

P value

WBC

8.18 (4.31, 9.47)

6.42 (5.68, 12.4)

-1.55

(-2.59 to -0.59)

0.002

Neutrophil

6.31(4.03, 10.63)

4.32 (7.53, 2.76)

-1.68

(-2.66 to -0.76)

<0.001

Lymphocyte

1.00 (0.72, 1.37)

1.06 (0.69, 1.50)

0.08

(-0.06 to 0.23)

0.259

CRP

14.79 (2.32, 37.18)

6.54 (2.09, 34.72)

-0.62

(-4.82 to 1.75)

0.504

PCT

0.08 (0.04, 0.56)

0.08 (0.04, 0.26)

-0.01

(-0.04 to 0.02)

0.555

Lactic acid

1.30 (1.00, 1.90)

1.20 (0.90, 1.70)

-0.1

(-0.40 to 0.10)

0.239

D-dimer

1085 (420, 2555)

690 (380, 1642)

-230

(-530 to -30)

0.019

Number (%)

Number (%)

OR

95%CI

P value

Gender?

57 (60)

81 (65.3)

0.796

(0.458 to 1.384)

0.419

Abdominal pain

92 (96.8)

115 (92.7)

2.4

(0.632 to 9.121)

0.186

Distention

62 (65.3)

102 (82.3)

0.405

(0.217 to 0.757)

0.004

Nausea and vomiting

65 (68.4)

93 (75)

0.722

(0.399 to 1.308)

0.282

History of Surgery

74 (77.9)

95 (76.6)

1.076

(0.568 to 2.037)

0.823

Smoking

27 (28.4)

34 (27.4)

1.051

(0.580 to 1.906)

0.87

Tenderness

74 (77.9)

74 (59.7)

2.381

(1.303 to 4.351)

0.004

Rebound

29 (30.5)

19 (15.3)

2.428

(1.261 to 4.676)

0.007

Beak sign

75 (78.9)

62 (50)

3.75

(2.046 to 6.874)

<0.001

>1 beak sign

62 (65.3)

38 (30.6)

4.252

(2.476 to 7.514)

<0.001

Radial distribution

18 (18.9)

8 (6.5)

3.36

(1.392 to 8.113)

0.005

U/C-Shaped loop

8 (8.4)

5 (4.0)

2.189

(0.692 to 6.919)

0.173

Whirl sign

34 (35.8)

11 (8.9)

5.726

(2.711 to 12.095)

<0.001

Small bowel feces sign

21 (22.1)

34 (27.4)

0.751

(0.402 to 1.403)

0.369

Mesenteric haziness

81 (85.3)

82 (66.1)

2.963

(1.504 to 5.840)

0.001

Small bowel wall thickening

20 (21.1)

28 (22.6)

0.914

(0.478 to 1.749)

0.786

Ascites

68 (71.6)

66 (53.2)

2.213

(1.254 to 3.908)

0.006

OL-SBO, open-loop small bowel obstruction; CL-SBO, closed-loop small bowel obstruction; CRP, C-reactive protein; WBC, white blood cell; PCT, Procalcitonin; Alb, albumin; CI, confidence interval.

* percentage of male patients.

shown in Fig. 5. Detailed information on the model is shown in Table 2. The ROC curve is shown in Fig. 6. The Harrell’s C statistic was 0.786 (95%CI, 0.724-0.848), meaning 78.6% of CL-SBO patients can be cor- rectly predicted with this model. When the cut-off value is 0.522, the model has the most Youden index, with 0.632 sensitivity and 0.855 specificity. The Brier score was 0.182, and R2 was 0.325.

    1. Validation

Using 10 repetitions, Harrell’s C statistic in the K-fold cross- validation was 0.766, and the calibration plot is shown in Fig. 7A. Harrell’s C statistic in the bootstrap resampling technique (n = 100) was 0.767, and the calibration plot is shown in Fig. 7B.

Image of Fig. 5

Fig. 5. Nomogram to predict the patients’ probability of developing CL-SBO. Points are assigned for D-dimer, tenderness, more than one beak sign, whirl sign, ascites, and radial disturbance by drawing a line upward called the points line. The sum of these points, plotted on the total points line, corresponds to predictions of CL-SBO. For example, the risk of CL-SBO is estimated at 68% for a patient whose D-dimer is 3000 and who has tenderness, two beak signs, and ascites on CT scan without the whirl sign or radial distribution.

Table 2

The detailed coefficients of the predictive model.

Estimate

Standard Error

z value

Pr (>|z|)

(Intercept)

-2.4533748

0.4270075

-5.746

9.16e-09

D-dimer

0.0002064

0.0001024

2.016

0.043751

Tenderness

0.6601623

0.3540818

1.864

0.062261

More than one beak sign

1.0343065

0.3347645

3.090

0.002004

Whirl sign

1.6458688

0.4311616

3.817

0.000135

Ascites

0.8535280

0.3403688

2.508

0.012153

Radial distribution

0.4632793

0.4913023

0.943

0.345700

External validation data were obtained from the Qingdao Central Hospital. Fifty-seven patients were included in our external validation; 28 of them had OL-SBO confirmed by surgery, and the other 29 had CL-SBO. The characteristics of the external validation cohort are shown in Table 3. The Harrell’s C statistic of external validation was 0.784 (95%CI, 0.664-0.905), Brier score was 0.190, and R2 was 0.315. A calibration plot of the external validation is shown in Fig. 7C.

    1. Diagnostic performance of CT findings

The diagnostic performance of CT findings is shown in Table 4. None of the variables had a high sensitivity or specificity. Radial distribution, U/C-shaped loop, and whirl sign had high specificity (93.5%, 96.0%, and 91.1%, respectively) but low sensitivity (18.9%, 8.4%, and 35.8%, re- spectively). Beak sign and mesenteric haziness had high sensitivity (78.9% and 85.3%, respectively) but low specificity (50% and 33.9%, re- spectively). Among all the CT findings, the whirl sign had the best PPV (75.6%), and beak sign had the highest NPV (75.6%).

  1. Discussion

CL-SBO is an important type of SBO. In CL-SBO, the bowel is obstructed at two points, and the portion of bowel between the two transition zones is considered a closed loop, which cannot decompress into the upstream bowel loops. As a result, the intestinal contents are more likely to compress the intestinal wall and nearby blood vessels. Moreover, because the closed-loop bowel is obstructed at two or more points, it is easier for the bowel to be compressed by a band or hernial orifice. Alternatively, twisting of the mesentery may impair circulation. Therefore, in clinical practice, CL-SBO is often accompanied by bowel is- chemia. In a retrospective study in the Netherlands [14], bowel ischemia or necrosis occurred in 81% of patients with CL-SBO. In our analysis, bowel ischemia was confirmed in 57.9% of patients with CL-SBO com- pared with 18.5% in the OL-SBO group. Diagnosing CL-SBO in time can

Image of Fig. 6

Fig. 6. The predictive model’s receiver operating characteristic (ROC) curve.

reduce the possibility of adverse consequences such as bowel resection (sometimes short bowel syndrome) and sepsis.

In some studies [6,16], CL-SBO was defined as having two adjacent beak signs, a C/U-shaped bowel, or radial distribution. In our analysis, if the diagnostic criterion for CL-SBO was meeting one of three CT find- ings, it would have had a sensitivity of 0.705 (95%CI, 0.602-0.752) and a specificity of 0.653 (95%CI, 0.562-0.735). We believe that this is inaccu- rate. In the study by Makar et al., [11] average reader sensitivity and specificity for CL-SBO were 53% and 83%, respectively, and they did not find any specific CT findings associated with CL-SBO. Our analysis found no CT signs with high sensitivity and specificity (Table 3). In an- other study from the Cerrahpafla Medical Faculty [15], only five of

Image of Fig. 7

Fig. 7. The calibration plot of K-fold cross-validation (a), bootstrap resampling technique (b), and external validation (c). The apparent and bias-corrected lines correspond to der- ivation and validation data, respectively.

Table 3

Characteristics of the external validation cohort.

CL-SBO (n = 29)

OL-SBO (n = 28)

mean +- SD

mean +- SD

Age

55.93 +- 15.047

58.57 +- 12.054

Alb

37.77 +- 9.036

35.25 +- 7.857

Longest diameter

38.08 +- 6.380

37.72 +- 9.076

Median (P25, P75)

Median (P25, P75)

WBC

7.68 (6.08, 11.70)

8.21 (5.69, 12.40)

Neutrophil

5.98 (3.68, 9.96)

6.13 (4.11, 10.80)

Lymphocyte

1.08 (0.84, 1.53)

0.94 (0.68, 1.49)

CRP

9.00 (3.69, 73.86)

21.57 (0.25, 142.73)

PCT

0.08 (0.02, 3.40)

0.25 (0.12, 0.67)

Lactic acid

1.30 (0.98, 2.30)

0.80 (0.55, 1.05)

D-dimer

1350 (590, 3420)

660 (435, 1770)

*

Number (%)

15 (51.7)

Number (%)

16 (57.1)

Abdominal pain

29 (100)

25 (89.3)

Distention

19 (65.5)

19 (67.9)

Nausea and vomiting

19 (65.5)

22 (78.6)

History of surgery

15 (51.7)

18 (64.3)

Smoking

3 (10.3)

8 (28.6)

Tenderness

25 (86.2)

24 (85.7)

Rebound

9 (31.0)

3 (10.7)

Beak sign

25 (86.2)

17 (60.7)

>1 beak sign

22 (75.9)

7 (25.0)

Radial distribution

7 (24.1)

1 (3.6)

U/C-Shaped loop

7 (24.1)

3 (10.7)

Gender

Whirl sign 7 (24.1) 1 (3.6)

Small bowel feces sign 8 (27.6) 11 (39.3)

Mesenteric haziness 25 (86.2) 20 (71.4)

Small bowel wall thickening 10 (34.5) 9 (32.1)

Ascites 25 (86.2) 17 (60.7)

OL-SBO, open-loop small bowel obstruction; CL-SBO, closed-loop small bowel obstruction; CRP, C-reactive protein; WBC, white blood cell; PCT, Procalcitonin; Alb, albumin.

* Percentage of male patients.

3000 cases of Internal hernia were detected using CT. Therefore, diag- nosing CL-SBO early remains challenging.

We constructed a nomogram diagnostic model based on HIS, which found that D-dimer, tenderness, CT findings–more than one beak sign, radial distribution, and whirl sign–and ascites were predictive factors for CL-SBO, with good diagnostic performance (area under the curve, 0.786; 95%CI, 0.724-0.848). The Diagnostic ability of each CT finding was also evaluated. We then performed a subgroup analysis in the CL- SBO group between patients with ischemic bowel and those with viable bowel to identify the reason for the differences between the CL-SBO and OL-SBO groups.

We found that tenderness was a predictive factor for CL-SBO. Few studies have reported an association between CL-SBO and tenderness. Based on our subgroup analysis, we found more tenderness in patients with ischemic CL-SBO than in those with viable bowel (85.7% vs. 68.5%, p < 0.05). Therefore, we believe that the differences between CL-SBO and OL-SBO arise from the larger proportion of patients with ischemia in the CL-SBO group. Peritoneal irritation signs, such as

tenderness, rebound tenderness, and guarding, are more common in patients with strangulated bowels (Odds ratio, 13) [17]. A retrospective study [18] also found that bowel ischemia was associated with peritonitis (36% vs. 1%).

D-dimer was another important variable in our diagnostic model. As a Fibrin degradation product, D-dimer is a valuable marker in the diag- nosis of venous thromboembolism and disseminated intravascular co- agulation [19]. It is also an important serological marker to diagnose bowel ischemia, with a sensitivity of 0.96 (95%CI, 0.89-0.99) and a spec- ificity of 0.40 (95%CI, 0.33-0.47) [20]. Therefore, we believe that D- dimer is another variable that is associated with the proportion of pa- tients with ischemia. This may be related to the microthrombi resulting from strangulation of the mesentery. However, a prospective study [21] found that D-dimer levels were not significantly different between pa- tients with internal hernia and those with other types of SBO after Roux-en-Y gastric bypass surgery. We believe that this was because in their patient cohort, no bowel ischemia was found in either group.

CT is the preferred technique for diagnosing CL-SBO. Therefore, most of our predictive factors in nomogram(4 of 6) were CT findings. A beak sign is observed when the bowel is occluded with adhesive bands or her- nia orifices. A CT scan can detect more than one beak sign if there are two or more occluded points. More than one beak sign is an important CT find- ing to diagnose an internal hernia [22,23]. In our analysis, more than one beak sign was detected in 65.3% of CT images (sensitivity, 65.3%; 95%CI, 54.7-74.5%) of patients with CL-SBO. This CT finding also had a median specificity of 69.4% (95%CI, 60.3-77.1%). We believe that this is because some images of adhesive bands may have been misidentified as beak signs. Radial distribution on a string appearance, also known as balloons, is another significant CT finding. If the closed loop is perpendicular to the CT scan level, a radial distribution will be formed, with the embedded mesentery pointing toward the adhesion zone or opening of the internal hernia [24]. This CT sign is difficult to determine if the closed loop is not exactly vertical to the scan level. Thus, it had high specificity (93.5%, 95% CI, 87.2-96.9%) but low sensitivity (18.9%, 95%CI, 11.9-28.6%).

A whirl sign appears when afferent and efferent bowel loops and vessels rotate around a fixed point of obstruction. It is often regarded as a sign of small bowel volvulus. Some investigators have suggested that this appearance can also be observed after small bowel surgery due to the disruption of the normal anatomical relationship between vessels and intestinal collaterals [25]. A recent paper [26] reporting on 1493 patients with swirling signs found that the swirling signs detected on CT were not specific to small bowel closed-collar obstruction. This study showed a low sensitivity and high specificity for the whirl sign, which is almost identical to what we found (a sensitivity of 35.8% and a specificity of 91.1%). Ascites is usually considered an independent pre- dictive factor for bowel ischemia [14,18]. However, in our analysis, there were no significant differences between the ischemia and no-Ischemia groups. The first reason for this may be the fact that some patients with SBO had nutritional disorders with hypoalbuminemia, which causes fluid transfer from the blood to the abdominal cavity. The second reason could be that the patients had liver disease or were undergoing Peritoneal dialysis [27].

Table 4

The diagnostic performance of computed tomography findings.

Sensitivity (95% CI)

Specificity (95% CI)

PPV (95% CI)

NPV (95% CI)

Beak sign

78.9% (69.1-86.4)

50.0% (40.9-59.1)

54.7% (46.0-63.2)

75.6% (64.7-84.1)

>1 beak sign

65.3% (54.7-74.5)

69.4% (60.3-77.1)

62.0% (51.7-71.4)

72.3% (63.2-79.9)

Radial distribution

18.9% (11.9-28.6)

93.5% (87.2-96.9)

69.2% (48.1-84.9)

59.9% (52.6-66.8)

U/C-Shaped Loop

8.4% (4.0-16.4)

96.0% (90.4-98.5)

61.5% (32.3-84.9)

57.8% (50.7-64.5)

Whirl sign

35.8% (26.4-46.3)

91.1% (84.3-95.3)

75.6% (60.1-86.6)

65.0% (57.3-71.9)

Small bowel feces sign

22.1% (14.5-32.0)

72.6% (63.7-80.0)

38.2% (25.7-52.3)

54.9% (46.9-62.6)

Mesenteric haziness

85.3% (76.2-91.4)

33.9% (25.8-43.0)

49.7% (41.8-57.6)

75.0% (61.4-85.2)

Small bowel wall thickening

21.1% (13.6-30.9)

77.4% (68.9-84.2)

41.7% (27.9-56.7)

56.1% (48.4-63.6)

Ascites

71.6% (61.3-80.1)

46.8% (37.8-55.9)

50.7% (42.0-59.4)

68.2% (57.1-77.7)

CI, confidence interval; PPV, positive predictive value; NPV, negative predictive value.

A diagnostic model of CL-SBO using a machine learning technique has been built [28]. Our study differs from previous studies mainly because of three factors: (1) we excluded patients with MBO because we believed there were other factors that should be considered, such as the stage of the tumor and metastasis; (2) we did not include contrast-enhanced CT scan findings because this test is usually not the first choice in our emergency room for patients suspected to have SBO due to its higher cost and contraindications such as renal failure; and (3) we used univar- iate analysis and backward stepwise regression, rather than machine learning. With fewer variables in our predictive model containing a larger sample size, we had better discriminative ability.

Our study has some limitations. First, it was a retrospective study. Although almost all our patients underwent all the tests within three days before surgery, it still differs from an actual operation. Second, some variables (CRP, PCT, and lactic acid) had missing values that were replaced with multiple imputations. Nine patients who only un- derwent abdominal CT (pelvic CT scan was not accessible) were in- cluded in our analysis. This may have caused the data we analyzed to differ from the true results. However, we do not think it will have a significant effect on our final results. Further more, only patients who underwent surgeries can be included in our cohort, which can lead to potential selective bias because SBO patients who underwent surgery are usually the ones with more Serious conditions. Fourth, our sample size was small. The predictive model is more accurate if more cases are included. Further external validation is also needed in the future.

  1. Conclusion

A diagnostic model for CL-SBO was created, and tenderness, D-dimer, more than one beak sign, whirl sign, radial distribution, and ascites were the predictive variables in this model. Urgent surgery should be considered in patients with a high risk for developing CL-SBO.

Ethical approval

This article describes a retrospective study and all patients’ treat- ments were not influenced by our study. The collection of patients’ in- formation was approved by the ethnic institution of the affiliated hospital of Qingdao university.

Declaration of Competing Interest

None.

Acknowledgments

The authors would like to thank Dongfeng Zhang from the Depart- ment of Epidemiology and Health Statistics, Public Health College, Qing- dao University for his professional help with statistical methods.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2022.10.022.

References

  1. Diaz Jr JJ, Bokhari F, Mowery NT, Acosta JA, Block EF, Bromberg WJ, et al. Guidelines for management of small bowel obstruction. J Trauma. 2008;64:1651-64. https:// doi.org/10.1097/TA.0b013e31816f709e.
  2. Hwang JY, Lee JK, Lee JE, Baek SY. Value of multidetector CT in decision making re- garding surgery in patients with Small-bowel obstruction due to adhesion. Eur Radiol. 2009;19:2425-31. https://doi.org/10.1007/s00330-009-1424-4.
  3. Ray NF, Denton WG, Thamer M, Henderson SC, Perry S. Abdominal adhesiolysis: in- patient care and expenditures in the United States in 1994. J Am Coll Surg. 1998; 186:1-9. https://doi.org/10.1016/s1072-7515(97)00127-0.
  4. Sikirica V, Bapat B, Candrilli SD, Davis KL, Wilson M, Johns A. The inpatient burden of abdominal and gynecological adhesiolysis in the US. BMC Surg. 2011;11:13. https:// doi.org/10.1186/1471-2482-11-13.
  5. Ten Broek RPG, Krielen P, Di Saverio S, Coccolini F, Biffl WL, Ansaloni L, et al. Bologna guidelines for diagnosis and management of adhesive small bowel obstruction (ASBO): 2017 update of the Evidence-based guidelines from the world society of emergency surgery ASBO working group. World J Emerg Surg. 2018;13:24. https:// doi.org/10.1186/s13017-018-0185-2.
  6. Millet I, Boutot D, Faget C, Pages-Bouic E, Molinari N, Zins M, et al. Assessment of strangulation in adhesive small bowel obstruction on the basis of combined CT find- ings: implications for clinical care. Radiology. 2017;285:798-808. https://doi.org/10. 1148/radiol.2017162352.
  7. Geffroy Y, Boulay-Coletta I, Julles MC, Nakache S, Taourel P, Zins M. Increased unenhanced bowel-wall attenuation at multidetector CT is highly specific of ische- mia complicating small-bowel obstruction. Radiology. 2014;270:159-67. https:// doi.org/10.1148/radiol.13122654.
  8. Sheedy SP, Earnest 4th F, Fletcher JG, Fidler JL, Hoskin TL. CT of small-bowel ischemia associated with obstruction in emergency department patients: diagnostic perfor- mance evaluation. Radiology. 2006;241:729-36. https://doi.org/10.1148/radiol. 2413050965.
  9. Nakashima K, Ishimaru H, Fujimoto T, Mizowaki T, Mitarai K, Nakashima K, et al. Di- agnostic performance of CT findings for bowel ischemia and necrosis in closed-loop small-bowel obstruction. Abdom Imaging. 2015;40:1097-103. https://doi.org/10. 1007/s00261-014-0335-2.
  10. Li Z, Zhang L, Liu X, Yuan F, Song B. diagnostic utility of CT for small bowel obstruc- tion: systematic review and meta-analysis. PLoS One. 2019;14:e0226740. https:// doi.org/10.1371/journal.pone.0226740.
  11. Makar RA, Bashir MR, Haystead CM, Iseman C, Mayes N, Hebert S, et al. Diagnostic performance of MDCT in identifying closed loop small bowel obstruction. Abdom Radiol (NY). 2016;41:1253-60. https://doi.org/10.1007/s00261-016-0656-4.
  12. Collins GS, Reitsma JB, Altman DG, Moons KG. Transparent reporting of a multivari- able prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ. 2015;350:g7594. https://doi.org/10.1136/bmj.g7594.
  13. Balachandran VP, Gonen M, Smith JJ, DeMatteo RP. Nomograms in oncology: more than meets the eye. Lancet Oncol. 2015;16:e173-80. https://doi.org/10.1016/ S1470-2045(14)71116-7.
  14. de Kok BM, Toneman MK, Oei S, Westerterp M, van Acker G, van der Pool A, et al. Correlation of CT findings with intra-operative outcome in closed-loop small bowel obstruction (CL-SBO). Eur J Radiol. 2021;142:109844. https://doi.org/10. 1016/j.ejrad.2021.109844.
  15. Selcuk D, Kantarci F, Ogut G, Korman U. radiological evaluation of internal abdomi- nal hernias. Turk J Gastroenterol. 2005;16:57-64.
  16. Zielinski MD, Eiken PW, Heller SF, Lohse CM, Huebner M, Sarr MG. Prospective, ob- servational validation of a multivariate small-bowel obstruction model to predict the need for operative intervention. J Am Coll Surg. 2011;212:1068-76. https:// doi.org/10.1016/j.jamcollsurg.2011.02.023.
  17. Huang X, Fang G, Lin J, Xu K, Shi H, Zhuang L. A prediction model for recognizing strangulated small bowel obstruction. Gastroenterol Res Pract. 2018:7164648. https://doi.org/10.1155/2018/7164648.
  18. Zielinski MD, Eiken PW, Bannon MP, Heller SF, Lohse CM, Huebner M. Small bowel obstruction-who needs an operation? A multivariate prediction model. World J Surg. 2010;134:910-9. https://doi.org/10.1007/s00268-010-0479-3.
  19. Weitz JI, Fredenburgh JC, Eikelboom JW. A test in context: D-dimer. J Am Coll Cardiol. 2017;70:2411-20. https://doi.org/10.1016/j.jacc.2017.09.024.
  20. Derikx JP, Schellekens DH, Acosta S. Serological markers for human intestinal ische- mia: a systematic review. Best Pract Res Clin Gastroenterol. 2017;31:69-74. https:// doi.org/10.1016/j.bpg.2017.01.004.
  21. Zaigham H, Olsson Regner A, Ekelund M, Regner S. The role of citrulline, Intestinal fatty acid-binding protein, and D-dimer as potential biomarkers in the diagnosis of internal herniation after roux-en-Y gastric bypass. Surg Obes Relat Dis. 2021;17: 1704-12. https://doi.org/10.1016/j.soard.2021.05.028.
  22. Zhang F, Qiao Y, Zhang H. Multidetector computed tomography findings of adhesive internal hernias. J Comput Assist Tomogr. 2020;44:131-7. https://doi.org/10.1097/ RCT.0000000000000955.
  23. Yen CH, Chen JD, Tui CM, Chou YH, Lee CH, Chang CY. Internal hernia: computed to- mography diagnosis and differentiation from adhesive small bowel obstruction. J Chin Med Assoc. 2005;68:21-8. https://doi.org/10.1016/S1726-4901(09)70127-3.
  24. Gore RM, Silvers RI, Thakrar KH, Wenzke DR, Mehta UK, Newmark GM, et al. Bowel obstruction. Radiol Clin N Am. 2015;53:1225-40. https://doi.org/10.1016/j.rcl.2015. 06.008.
  25. Matsumoto S, Mori H, Okino Y, Tomonari K, Yamada Y, Kiyosue H. Computed tomo- graphic imaging of abdominal volvulus: pictorial essay. Can Assoc Radiol J. 2004;55: 297-303.
  26. Gollub MJ, Yoon S, Smith LM, Moskowitz CS. Does the CT whirl sign really predict small bowel volvulus?: experience in an oncologic population. J Comput Assist Tomogr. 2006;30:25-32. https://doi.org/10.1097/01.rct.0000191680.28344.f0.
  27. Kim J, Lee Y, Yoon JH, Lee HJ, Lim YJ, Yi J. Non-strangulated adhesive small bowel ob- struction: CT findings predicting outcome of conservative treatment. Eur Radiol. 2021;31:1597-607. https://doi.org/10.1007/s00330-020-07406-3.
  28. Goyal R, Mui LW, Riyahi S, Prince MR, Lee HK. Machine learning based prediction model for closed-loop small bowel obstruction using computed tomography and clinical findings. J Comput Assist Tomogr. 2022;46:169-74. https://doi.org/10. 1097/RCT.0000000000001269.