The Future of Acute Abdomen Management in Emergency Care

The debate between traditional methods and modern technology in gastrointestinal emergency medicine can be seen in two recent publications. One, a review article by American surgeons Selwyn O. Rogers and Orlando C. Kirton, published inThe New England Journal of Medicine, advocated for the surgeon’s pivotal role in abdominal emergencies, supported by imaging, just like in the early 20th century. The other, published in the Canadian Association of Radiologists Journal, anticipated the imminent arrival of artificial intelligence (AI) in managing acute abdomen and minimizing human bias.
Acute abdominal pain accounts for 4%-10% of emergency department visits, with hospitalization admission rates ranging from 18% to 42%. Today, the first-line physicians — emergency doctors or advanced practice professionals (nurses or physician assistants) — are often the first to evaluate a patient with acute abdominal pain. 
These clinicians may administer analgesics, order laboratory tests, and request diagnostic imaging. On the basis of the results, they decide when to call for a surgical consultation. Despite advancements in imaging techniques, diagnostic errors and delays in surgical intervention for abdominal emergencies persist. This article summarizes the key points from these two North American reviews, with a synthesis of the findings.
The myth that analgesia should not be prescribed for patients with abdominal pain in the emergency room because it masks the diagnosis has long been debunked. Studies and meta-analyses conducted between 2000 and 2015 highlighted the superiority of opioid treatment over placebo for pain relief, without increasing diagnostic errors in properly medicated patients. These studies also showed no worsening of outcomes and improved comfort for patients awaiting test results and surgical evaluation. The rapid administration of protocolized analgesic treatments, such as paracetamol, nefopam, tramadol, morphine hydrochloride, and NSAIDs (in the absence of sepsis), is essential for quality care. A patient who is not in pain is more easily assessable and transportable for a CT scan or ultrasonography.
During abdominal pain management, it’s important for the physician and emergency nurses to repeat clinical exams, including hemodynamic monitoring and pain scale assessments. The progression of some conditions may be slow or, conversely, may worsen rapidly. Rogers and Kirton strongly recommended administering modest doses of analgesics for patients with acute abdominal pain.
With the decline of plain abdominal radiography, contrast-enhanced abdominopelvic CT has become the primary imaging technique for accurately identifying intra-abdominal inflammation, perforation, or ischemia. It has a high predictive value for specific abdominal diagnoses. In one study, it increased diagnostic certainty, reduced hospital admissions by 23.8%, and led to faster surgical interventions. However, contrast-enhanced CT is harmful in cases of renal failure, and noncontrast abdominal and pelvic CTs are about 30% less accurate. Early vascular imaging in contrast-enhanced exams is particularly useful for evaluating acute abdomen and identifying secondary diagnoses.
Rapid interpretation of cross-sectional images is vital so that surgical consultation can begin promptly. A wait of more than 2 hours for a final interpretation by a radiologist is associated with an increased risk for systemic complications and death. Teleradiology has become an indispensable tool for some emergency departments, offering 24/7 access to specialized radiologists, particularly in digestive imaging, optimizing resources for smaller hospitals. However, it also has drawbacks, such as a lack of direct communication with the radiologist, degraded image quality depending on network congestion, and diluted legal responsibility.
Teleradiology is a valuable tool for improving access to rapid and specialized diagnostics in emergency rooms, but it must be well supervised to avoid technical, communication, and legal problems. The number of CT scans performed on patients presenting with new, nontraumatic acute abdominal pain has steadily increased over the past few decades. However, abdominal imaging may not reveal all intra-abdominal diseases, and in one study, 20% of exams were deemed unnecessary by third-party radiologists. Yet, CT has become a legal safeguard, reassuring all involved physicians when the results are normal and aiding decisions regarding hospitalization for observation.
CT scanning offers distinct benefits for an aging population, as older people may present unreliable medical histories, vague abdominal symptoms, altered cognitive states, multiple comorbidities, challenging physical exams, and diminished physiologic reserves. Identifying pneumoperitoneum, inflammatory changes, or intestinal ischemia during a CT angiogram can facilitate rapid laparotomy, preventing progression to severe systemic disease or death in this vulnerable population.
Ultrasonography and MRI can be used as alternatives to CT, which uses x-rays. However, they have their own disadvantages, including variable sensitivity and operator dependency in the case of ultrasonography, and variable availability for MRI and particularly biliary MRI.
Ultrasonography is the initial imaging modality of choice for pregnant patients suspected of having appendicitis or other pelvic or gynecologic conditions associated with acute abdominal pain (eg, tubo-ovarian abscess, degenerative ovarian cyst, or ruptured ectopic pregnancy).
When performed by a suitably trained emergency physician, ultrasonography remains a valuable diagnostic and monitoring tool when rapid access to contrast-enhanced CT is unavailable. It can assist in the triage of a wide range of conditions manifesting as acute abdominal pain (eg, cholecystitis and biliary duct dilation, appendicitis, gastrointestinal perforation, acute pancreatitis, colitis, intestinal obstruction, deep abscesses, and aneurysms).
Every effort should be made to avoid delays in surgical consultations due to the risk for complications and death. A thorough medical history and a complete physical examination, including a systematic rectal exam, are essential components of evaluating abdominal pain. Clinical decisions should not be clouded by disagreements in the interpretation of radiologic results.
In cases of acute abdomen, studies have shown that consultation with in-house surgeons reduces mortality in patients with life-threatening conditions requiring emergency surgery. Unfortunately, unlike in France, 24/7 availability of consulting surgeons is not a reality in many rural or underserved areas in the United States, which explains the advocacy of American surgeons on this sensitive issue.
In challenging cases, cognitive biases related to training, previous experiences, personal beliefs, and clinicians’ expectations can lead to the selection of inappropriate datasets that may guide subsequent reasoning. Surgeons are not immune to these biases, particularly in older patients with multiple conditions and elevated anesthetic risk. There are three main types of biases: 
Attribution bias is the tendency to ascribe symptoms to unrelated diagnostic tests on the basis of one’s preconceptions. 
Confirmation bias is the tendency to seek out data that corroborate one’s existing mental model of the patient’s condition and recall information in a way that confirms or reinforces previous beliefs or values. This can result in the selective mobilization of medical support for a favored hypothesis. Diagnostic momentum may lead to a presumptive prehospital diagnosis being carried into the emergency room, where clinicians seek evidence to confirm the diagnosis without considering other differential diagnoses, such as abdominal presentation of myocardial infarction.
Anchoring bias involves the tendency to rely too heavily on the first piece of information mentioned in the patient’s record, which then becomes a reference point for subsequent judgments.
The influences of these cognitive biases is most pronounced in high-stakes scenarios where rapid decision-making is required, such as clinical instability and/or abdominal complications with life-threatening potential.
If symptoms do not align with the suspected diagnosis, the clinician must reassess and consider alternative diagnoses and treatments. This is where the on-call gastroenterologist comes in. While typically an endoscopist, their role shifts to that of a consultant to validate the medical approach when the surgeon does not intervene immediately. A classic example is acute ileitis.
The diagnosis of acute ileitis is straightforward in clinical practice, but determining the cause can be more challenging. In two thirds of cases, acute ileitis resolves spontaneously without specific treatment and does not necessitate routine hospitalization. However, if symptoms persist for 2-4 weeks after the initial episode, an ileocolonoscopy should be performed to investigate the possibility of Crohn’s disease.
Conversely, the potential pitfall of acute jejunitis is the ambiguous nature of the radiologic diagnosis, which may obscure early ischemia in an older patient with vascular disease. Mesenteric infarction, or “acute mesenteric ischemia,” is a largely underrecognized condition that affects more than 10,000 people in France each year, with a rising incidence and a mortality rate of 70%. 
The main challenge is the identification of early forms and the implementation of prompt management in specialized centers, analogous to the approach used in stroke centers for cerebrovascular accidents or coronary care units for myocardial infarctions. Urgent treatment relies on protective medical measures for the intestines, interventional radiologic revascularization, and, if necessary, surgical resection of necrotic bowel.
This care requires a multidisciplinary approach focused on bowel viability and coordinated by interventional radiologists and specialized gastroenterologists. A specialized medical structure for treating intestinal vascular emergencies, particularly acute mesenteric ischemia, called SURVI (for Structure d’URgences Vasculaires Intestinales) is being used at the Beaujon Hospital, Assistance Publique–Hôpitaux de Paris. It reported a significant improvement in both survival and intestinal prognosis, with a mortality rate of around 20% and an intestinal resection rate of 30%. Establishing similar centers would be desirable in other French university hospitals, given the high frequency of acute mesenteric ischemia and the limitations of regional hospital centers.
Consequently, Rogers and Kirton’s review recommended that clinical examination (ideally conducted by a surgeon) should be the primary method of diagnosis in cases of acute abdomen. This probably reflects the organizational challenges faced by small rural American hospitals, where patients may experience inferior outcomes due to the absence of on-site surgical expertise.
In French regional hospitals, the current trend for nonimmediate surgical acute abdomen cases is to integrate “gastroenterology beds” within digestive surgery departments or in general medicine or geriatric units. Gastroenterologists are then asked to act as specialized consultants, provided they have the technical resources, such as echo-endoscopy and endoscopic surgery. Otherwise, the patient may be transferred to a tertiary center, although availability can be problematic. 
Beyond the popular use of ChatGPT-4 by tech-savvy doctors, clinical applications of AI in acute abdominal imaging are rapidly evolving and could help diagnose various conditions (pneumoperitoneum, cholecystitis, appendicitis, obstructions, renal pathology, and aortic conditions). AI algorithms, including deep learning, convolutional neural networks, and large language models, have demonstrated high accuracy in identifying and characterizing both traumatic and nontraumatic acute abdominal pathologies, often with sensitivity and specificity exceeding 90%.
Recent studies have suggested that radiologists assisted by AI can work more efficiently, even in emergency situations, where AI can streamline workflow and reduce reading time in certain applications. Beyond image interpretation, AI can assist emergency radiologists with tasks such as setting up study protocols and prioritizing worklists. Wang and colleagues observed that the workflow interval for CT, defined as the time from the initial exam request to the delivery of the first report, accounted for 29% of the patient’s total average emergency department stay. Of this, radiology processing time — from completing CT to providing the first report — was estimated at 32%.
There are already many commercially available AI algorithms for radiology, approved by Health Canada and the US Food and Drug Administration, or registered with the French National Agency for Medicines and Health Products Safety with a CE (for Conformité Européenne) mark (signifying conformation to the essential requirements set by European Union regulations for health, safety, and environmental protection). These algorithms are used for tasks such as lesion detection, organ segmentation, and result classification.
Nevertheless, only a limited number of software programs are directly linked to tasks involving acute abdominopelvic imaging. This may be attributed to the heterogeneity of images, which vary by the data used to train these algorithms. The continuous evolution of model architectures (radiomics, 3D reconstructions) and the training techniques of some skeptical radiologists will probably facilitate the development of more diverse and clinically validated applications for abdominal imaging, endorsed by the French National Authority for Health and professional societies.
Hybrid clinical decision-making models that combine imaging data, laboratory results, and symptoms can also help provide a comprehensive assessment by integrating multiple data sources. For example, Reismann and colleagues developed a diagnostic algorithm to predict complicated appendicitis on the basis of a combination of ultrasonography, biology, and clinical data.
Other clinical decision algorithms for ovarian torsion, intestinal obstruction, and abdominal trauma have demonstrated excellent diagnostic and prognostic performance. However, they are often monofocused and require more detailed real-world evaluations before they can be considered to replace the expertise of a seasoned radiologist.
Additional studies have shown that AI can reduce human error by providing a secondary review of scans, particularly in high-stress emergency settings. These integrated models can offer individualized data beyond the current computational capabilities of emergency departments. 
Emerging research is exploring how natural language processing models, which allow computers to understand human language, can assist physicians with diagnosis and decision-making. The DALVIA software, developed in a collaboration between Docaposte and the French generative AI start-up Mistral AI, has recently been commercialized and is expected to soon help extract “the essence” of complex patient records.
The future application of this AI to emergency department software (which contains the 18 mandatory elements of the emergency department discharge summary) should be closely monitored. Several experiments are also underway in Amiens University Hospital, France, with the aim of optimizing resources (eg, downstream beds), in Niort and Rennes to more precisely analyze AI algorithms in medical imaging, and in Besançon for medical emergency predictions.
Despite these advances, several challenges limit the broader adoption of AI in clinical practice for acute abdominal imaging. There is considerable variation in the size and quality of anonymized datasets used for model development. This has an impact on the evaluation of the performance of computer models in real-world scenarios.
The development of robust and generalizable models requires the availability of large, high-quality, multi-institutional, annotated public datasets.
Additionally, seamless integration into hospital IT systems is necessary for AI tools to function optimally in real time. Finally, AI should serve as an assistive tool and not as a substitute for the expertise of radiologists and emergency physicians, as unusual or complex cases still require human judgment and supervision.
In conclusion, the application of AI to acute abdominal imaging is an emerging field of innovation with significant potential to transform the diagnosis and care of patients admitted to emergency departments. While there are ongoing challenges, including ethics, responsibility, security, regulation, and data reliability, the combined efforts of the medical and technology sectors are moving toward a future where AI will not only enhance image interpretation but also speed up clinical decision-making and patient triage in stressful environments.
This story was adapted from JIM using several editorial tools, including AI, as part of the process. Human editors reviewed this content