Annals of the Royal College of Surgeons of England最新文献

Introduction: Artificial intelligence (AI) chatbots, powered by large language models, are used increasingly for disseminating surgical information, but concerns about accuracy, hallucinations and source reliability persist. This study evaluates the sources of information upon which these systems rely when producing medical information. As these models generate language without true comprehension or reasoning, assessing the credibility and nature of their referenced sources is essential to promote transparency and support evidence-based integration of AI in healthcare.

Methods: Nine AI chatbots (ChatGPT-5, ChatGPT-5 Think, DeepSeek R1, DeepSeek DeepThink, Google Gemini 2.5 Flash, Grok 3, Grok 4, Perplexity Research and Perplexity Search) were queried with six standardised general surgery prompts, both with and without explicit requests for references (n=108 outputs); 1,249 references were extracted and assessed for quantity, authenticity, quality, source category, accessibility, geographic origin and attribution.

Results: Reference provision varied: four chatbots required explicit prompting, whereas others cited consistently. Hallucination rates ranged from 0% (five models) to 34% (Grok 3). Mean quality scores differed significantly, with Perplexity Research achieving the highest score (4.08) and ChatGPT-5 the lowest (2.39), reflecting differences observed in source type. Most references originated from the US or UK. Accessibility was best in Google Gemini (100% open access, clickable citations). Explicit prompting increased reference quantity significantly in six models and quality in one.

Conclusions: AI chatbots exhibit heterogeneous reference integrity, with risks of hallucinations and biases underscoring the need for prompt engineering, model refinements and ongoing evaluation. Our findings suggest ongoing caution is required in surgical contexts to ensure safe, equitable information dissemination.

Introduction: The National Institute for Health and Care Excellence guideline NG12 recommends urgent suspected cancer referrals for adults presenting with dysphagia. Diagnostic yield from upper gastrointestinal endoscopy in this cohort is around 1%. The Cancer Dysphagia Score (CDS) offers a simplified risk stratification tool with a reported sensitivity of 97.8% and negative predictive value (NPV) of 99.5% for upper GI malignancy. This study externally validates the real-world safety and performance of a modified CDS calculated by general practitioners (GPs) to triage dysphagia referrals.

Methods: A retrospective external validation study was conducted across a single NHS trust including adults referred on the suspected cancer pathway for dysphagia between April 2024 and April 2025. A modified CDS compatible with the PRISM electronic referral system was applied by GPs. Data were extracted from the Somerset Cancer Database. Patients were stratified as high (CDS ≥5.5) or low (<5.5) risk. Diagnostic outcomes were confirmed endoscopically and histopathologically. Sensitivity, specificity, positive predictive value (PPV), NPV and relative risk (RR) were calculated.

Results: Of 2,447 referrals, 2,425 met inclusion criteria (1,670 high risk; 755 low risk) and 78 cancers were diagnosed (3.2% detection rate). Cancer prevalence was 4.6% in the high-risk group and 0.3% in the low-risk group. The adjusted CDS achieved sensitivity 97.4%, specificity 32.1%, PPV 4.6%, NPV 99.7% and RR 17.2 (95% CI 4.2-69.8).

Conclusions: GP-applied CDS maintained high sensitivity and NPV, safely identifying low-risk dysphagia patients suitable for routine referral. Incorporating the CDS into suspected cancer referral pathway could optimise endoscopy resource allocation without compromising cancer detection.

Background: Pneumatosis intestinalis (PI) involving the small bowel is often interpreted as a radiological surrogate for ischaemia, prompting urgent surgical intervention. However, non-ischaemic causes-particularly in the postoperative setting-are under-recognised. This case-based systematic review aims to explore and synthesise reported non-ischaemic aetiologies of small bowel PI.

Methods: We present the case of a 74-year-old man who developed small bowel PI following ileostomy reversal, with computed tomography findings initially suggestive of bowel ischaemia. In the absence of clinical or biochemical evidence of sepsis, he was managed conservatively and made a full recovery. A systematic review was conducted using PRISMA methodology across PubMed, Embase and Medline databases. Inclusion criteria focused on reports of small bowel PI in the English language. Data were extracted on clinical context, diagnostic modality, treatment and outcomes.

Results: A total of 24 articles were included, comprising one retrospective cohort and 23 case reports, yielding 37 patients. The most frequent aetiology was small bowel obstruction (59.4%). Rarer causes included a rapidly progressive scleroderma-like disease, congenital lymphangioma and mechanical ventilation. Although 72.9% underwent surgical intervention, 27% were successfully managed conservatively. Only 10.8% had a suspected ischaemic component, and six patients died. Radiological overlap with ischaemic PI was common, yet several patients improved without surgery and only two (5.4%) had confirmed transmural small bowel necrosis.

Conclusions: Not all small bowel pneumatosis indicates ischaemia. Conservative management can be effective in carefully selected patients, guided by clinical status and serial imaging. Early recognition of non-ischaemic PI is crucial to avoid unnecessary laparotomy.

Introduction: The incidence of early-onset colorectal cancer (EoCRC) is rising globally and remains a growing healthcare concern. Despite national guidelines recommending universal referral for genetic testing, adherence remains low in many countries. This study evaluates the clinicopathological features, genetic referral patterns and broader implications of EoCRC in a single Scottish health board, contextualised through a recent literature review.

Methods: A retrospective review was conducted on all patients diagnosed with EoCRC from January 2019 to March 2023 under NHS Ayrshire and Arran. Demographic, clinical, pathological and genetic data were analysed. In addition, a literature review of studies on EoCRC published since 2015 was performed, identifying 19 relevant publications.

Results: Among 998 colorectal cancer cases, 45 (4.5%) were classified as EoCRC. The median age was 45 years, with an equal male-to-female distribution. Most cases were elective presentations, with rectal tumours being the most common (42%). Advanced-stage disease (T3/T4) was seen in 67%, and 31% had metastatic disease. Although microsatellite instability testing was conducted routinely, only 40% of patients aged under 50 years were referred for genetic counselling. Literature review findings reflected similar trends: late-stage presentation, low referral rates for genetic evaluation and mostly sporadic cases.

Conclusions: Our study highlighted suboptimal genetics referral despite national guidance, advanced-stage diagnosis, equal sex distribution and a predominance of rectal tumours. Multidisciplinary collaboration, greater awareness and structured referral pathways are vital to ensure safe, equitable care. Larger multicentre studies are needed to expand on these findings and inform future service planning and research.

Introduction: Congenitally corrected transposition of the great arteries (ccTGA) is a rare disorder. Here, we evaluated the effectiveness of physiological, anatomical and Fontan repair techniques for ccTGA.

Methods: A systematic database search in PubMed, Embase and Cochrane Central Register of Controlled Trials was conducted from inception to 5 June 2023. Random-effects network meta-analysis was performed, comparing mortality, postoperative arrhythmias, reintervention after definitive surgery and transplant-free survival in patients undergoing physiological, anatomical and Fontan repair.

Findings: A total of 1,209 patients were included from nine studies. Fontan repair was associated with a higher rate of transplant-free survival (absolute risk [AR] 0.999 (±0.0101), 95% confidence interval [CI]) and reintervention after definitive surgery (AR 0.098 (±0.067), 95% CI), whereas the difference for mortality and postoperative arrhythmias did not reach statistical significance.

Conclusions: Fontan repair is significantly better than anatomical or physiological repair in terms of transplant-free survival and also had a better rank probability for overall survival and reintervention after definitive surgery.