Development and External Validation of Prediction Risk Scores (StrISK and NOFA) to Predict Immediate Surgical Need in Adhesive Small Bowel Obstruction: An Observational Prospective Multicentre Study
Background Introduction
Adhesive Small Bowel Obstruction (ASBO) is one of the leading causes of emergency admissions, accounting for approximately 60% of all small bowel obstruction cases. ASBO is often caused by postoperative intra-abdominal adhesions, with patients presenting symptoms such as abdominal pain, vomiting, and constipation. While most ASBO cases can be managed non-operatively (e.g., intravenous hydration, nasogastric tube decompression, and oral water-soluble contrast agents), some patients may develop intestinal strangulation or experience failure of non-operative treatment, necessitating emergency surgical intervention. However, accurately identifying high-risk patients remains a clinical challenge. Existing prediction models often suffer from issues such as small sample sizes, retrospective study designs, and lack of calibration or external validation, limiting their clinical application.
To address this issue, Panu Räty and his team conducted a multicenter prospective observational study aimed at developing and externally validating two prediction models: one for predicting the risk of strangulation (Strangulation Risk Score, STRISK) and the other for predicting the risk of non-operative treatment failure (Non-Operative Treatment Failure Score, NOFA). The goal of this study was to assist clinicians in making earlier surgical decisions by combining clinical features, laboratory tests, and CT imaging findings, thereby improving short- and long-term patient outcomes.
Source of the Paper
This paper was co-authored by Panu Räty, Akseli Bonsdorff, Helka Parviainen, and others, with the research team hailing from Helsinki University Hospital, Hyvinkää Hospital, and Kanta-Häme Central Hospital in Finland. The paper was published in 2025 in the British Journal of Surgery (BJS) under the title “Development and external validation of prediction risk scores (STRISK and NOFA) to predict immediate surgical need in adhesive small bowel obstruction: an observational prospective multicentre study.”
Research Process
1. Study Design and Participants
The study was conducted in three hospitals in southern Finland, including two university hospitals (Meilahti Hospital and Jorvi Hospital) and one community hospital (Hyvinkää Hospital). The recruitment period spanned from June 2014 to March 2023. Inclusion criteria were patients with CT-confirmed adhesive small bowel obstruction, while exclusion criteria included age under 18, pregnancy, abdominal surgery within the last 30 days, inflammatory bowel disease, intra-abdominal hernia, or peritoneal carcinomatosis. Ultimately, 481 out of 626 patients met the criteria, with 355 patients used for model development and 126 for external validation.
2. Data Collection and Variable Selection
The study collected patient history, clinical symptoms, laboratory test results (e.g., white blood cell count, neutrophil-to-lymphocyte ratio, lactate), and CT imaging features (e.g., closed-loop sign, feces sign, mesenteric edema). CT images were reanalyzed by two expert gastrointestinal radiologists to ensure consistency. Variable selection was based on univariable analysis, literature review, and clinical utility, ultimately identifying six predictor variables: neutrophil-to-lymphocyte ratio, number of previous ASBO episodes, abdominal guarding, mesenteric changes and free abdominal fluid, closed-loop sign, and feces sign.
3. Model Development and Internal Validation
The study used binary logistic regression to develop the STRISK and NOFA models, with internal validation performed using bootstrapping. The STRISK model was designed to predict the risk of strangulation, while the NOFA model predicted the risk of non-operative treatment failure. In the development cohort, the optimism-corrected area under the receiver operating characteristic curve (AUROC) was 0.860 for the STRISK model and 0.751 for the NOFA model, indicating good discriminatory ability.
4. External Validation
The models demonstrated stable discrimination and calibration in the external validation cohort. The AUROC for the STRISK model was 0.907, and for the NOFA model, it was 0.751, further confirming the reliability of the models.
Main Results
- STRISK Model: In the development cohort, 16% of patients experienced strangulation, and the model effectively identified high-risk patients. The most significant predictors were the closed-loop sign (OR=5.68) and mesenteric edema/free abdominal fluid (OR=6.93).
- NOFA Model: In the development cohort, 31% of patients experienced non-operative treatment failure, and the model successfully identified high-risk patients. The most significant predictors were the closed-loop sign (OR=4.23) and mesenteric edema/free abdominal fluid (OR=2.24).
- Clinical Application: The study provided a web-based prediction tool (www.tinyurl.com/strisk) to assist clinicians in assessing surgical needs based on individual patient profiles.
Conclusions and Significance
This study developed and validated two prediction models that effectively identify high-risk patients requiring emergency surgery for adhesive small bowel obstruction. The clinical application of the STRISK and NOFA models is expected to optimize treatment decisions, reduce mortality and complications due to delayed surgery, and shorten hospital stays. Additionally, the study lays the groundwork for future prospective implementation studies to further evaluate the models’ effectiveness in real-world clinical settings.
Research Highlights
- Innovation: The first study to develop and externally validate models for predicting surgical needs in adhesive small bowel obstruction, addressing gaps in existing research.
- Multicenter Design: Conducted across multiple hospitals, enhancing the generalizability and reliability of the results.
- Clinical Value: The provided prediction tool can be directly applied in clinical practice to assist clinicians in making earlier surgical decisions.
- Data Quality: Ensured accuracy and consistency through prospective data collection and expert review.
Other Valuable Information
The study received support from the Finnish Medical Foundation, Orion Research Foundation, and Helsinki University Hospital Research Funds. Authors Panu Räty and Ville Sallinen played significant roles in the research and received funding from multiple research foundations.