1,720,972 research outputs found
PTH-32 development of a novel electronic referral grading & triage system
No full textIntroduction: prior to Covid-19, demand for secondary care appointments continued to rise year on year suggesting unsustainable future post-pandemic demand. Now is thus the right time to invest in triage and clinical pathway innovation.Methods: anew fully-integrated digital triage system was built at our institution allowing for document upload and electronic triage. Data pertaining to referral time, triage decision, outpatient appointments and direct-to-test was extracted from the backend to plot empirical cumulative distribution functions, interquartile ranges and allow statistical comparison using the Kruskal-Wallis’ test.Results: we analysed the first 704 luminal Gastroenterology referrals through the new triage system with the following sub-specialty classifications: Iron deficiency anaemia (IDA) – 200, Upper gastrointestinal symptoms (UGI) – 152, Inflammatory bowel disease (IBD) – 116, Irritable bowel syndrome (IBS/Functional) – 95, Lower gastrointestinal symptoms/change in bowel habit alone (LGI/CIBH) – 59, Coeliac – 27, Surgical – 25, Complex Functional – 12, Intestinal failure (IF/Nutrition) – 12, Hepatology – 4. 664 (95%) of referrals were accepted with 179 (27%) being sent direct to test. Of these only 42 (23.5%) had a subsequent clinic appointment booked, vs 436 (90%) for those not going direct to test. In addition, sending patients direct to test increased the proportion of subsequent routine clinic appointments from 55% to 70%. Median timelag from referral to grading was four days with grading taking a single day and appointments occurring 17 days later on average. Direct-to-test was most common amongst patients in the UGI (52.6%) and IBD (50%) sub-cohorts. This was significantly different vs other groups at the (p<0.05) level. [PTH-32 Figure 1 Subspecialty Referrals vs Direct-To-Test Numbers not included].Conclusions: using a system as described here substantially improves data capture and efficiency. Direct to test reduces both need for clinic appointments and the urgency of subsequent appointments. IBD and UGI are the subspecialties most likely to benefit from direct to test approaches. IDA could be another suitable specialty and the plan is to address this in the future
P58 anonymous electronic IBD patient service feedback
No full textIntroduction: collecting structured patient feedback is challenging, particularly during the pandemic with many virtual appointments. Our electronic IBD-patient feedback covers outpatient (OP), endoscopy and flare-line experiences.Methods: IBD patients provide anonymous feedback at the time-of-service contact. GATHER, a survey platform hosted by our institution, collects anonymous information via QR codes (scan QR codes for surveys), electronic links or handheld tablet. Demographics, disease characteristics and medication were noted in all 3 surveys. The OP survey collated clinic type/modality and feedback on individual health care professionals based on an adapted Royal College of Physicians questionnaire as well as preferences for future appointments. Endoscopy surveys gathered information on referral pathway, endoscopy type, treatment advice, length of wait and pre-test information. Flare line surveys allowed individual feedback on IBD nurses, assessed response time and outcomes. Patients’ attitudes regarding use of our online portal My Medical Record (MyMR) were explored. All surveys allowed sign up for MyMR. Patients could leave individual comments.Results: since September 2021, 425 patients responded. Figure 1 outlines the findings of the surveys. [P58 Figure 1 not included].Conclusion: electronic surveys are well accepted by our IBD patients and provides useful demographic data. It gives patients the option to inform the service of their preferences for future appointments and allows clinicians to get personal patient feedback for appraisals. Furthermore, it provides feedback on new services such as direct access endoscopy service and the acceptability of patient directed online healthcare (MyMR). Patient-centred feedback enables the user to help shape their future local IBD service
O69 outcome of direct access IBD physician delivered endoscopy for general practice referrals with suspected IBD
No full textIntroduction: patients with suspected IBD referred by primary care (GP) are traditionally seen in gastroenterology outpatient clinics followed by endoscopic investigations. This 2 phase model leads to delay in diagnosis and treatment, increasing pressure on gastroenterology outpatient services while still requiring endoscopic intervention. Our novel pilot project compared outcomes between direct-access IBD physician-delivered endoscopy versus the traditional clinic model for patients with suspected IBD.Method: a prospective cohort of consecutive patients referred by GP with suspected IBD were triaged either direct to IBD endoscopy (n=50) or to outpatient IBD clinic followed by IBD endoscopy (n=50) at the discretion of 10 gastroenterology consultants grading GP referrals. Data on demographics, faecal calprotectin, C-reactive protein, endoscopy outcomes, treatment, and follow up was collected. (Group A = direct to IBD endoscopy and Group B = IBD endoscopy via IBD clinic).Results: both groups were age and gender-matched. Group A had a higher mean calprotectin (1363 ug/g vs 302 ug/g) and a higher C-reactive protein (10.6 mg/l vs 4.5 mg/l). In Group A only 38% had a full colonoscopy versus 86% in Group B. Definitive diagnosis and treatment at time of IBD endoscopy took 27 days in Group A versus 212 days in Group B. Treatment with immunomodulators and biologics was similar in both groups but mesalazine and steroid use was higher in Group A due to more severe disease and higher rate of ulcerative colitis, table 1 shows the diagnostic breakdowns from both groups following endoscopy. The IBD pick up was significantly higher in Group A with 70% vs 42%. Endoscopy DNA rate was twice as high in Group B (n=6). The direct to IBD endoscopy pathway resulted in 50 less initial IBD consultant clinics (100% reduction) with a follow-up shift from IBD consultant to IBD nurse clinics. [O69 Table 1 Diagnostic breakdown not included].Conclusion: triaging patients referred with suspected IBD directly to IBD physician delivered endoscopy resulted in more than a 26-week reduction in time to diagnosis and treatment while saving 100% of initial IBD consultant clinics. IBD pick up was high at 70% in direct to IBD endoscopy group, identifying a higher-need IBD population. Triaging GP referrals with suspected IBD direct to IBD endoscopy delivers rapid assessment and treatmen
Robust comparative evaluation of 15 natural language processing algorithms to positively identify patients with inflammatory bowel disease from secondary care records
No full textObjective: natural language processing (NLP) can identify cohorts of patients with inflammatory bowel disease (IBD) from free text. However, limited sharing of code, models, and data sets continues to hinder progress. The aim of this study was to evaluate multiple open-source NLP models for identifying IBD cohorts, reporting on document-to-patient-level classification, while exploring explainability, generalisability, fairness and cost.Methods: 15 algorithms were assessed, covering all types of NLP spanning over 50 years of NLP development. Rule-based (regular expressions, spaCy with negation), and vector-based (bag-of-words (BoW), term frequency inverse document frequency (TF IDF), word-2-vector), to transformers: (two sentence-based sBERT models, three bidirectional encoder representations from transformers (BERT) models (distilBERT, BioclinicalBERT, RoBERTa), and five large language models (LLMs): (Mistral-Instruct-v0.3-7B, M42-Health/Llama-v3-8B, Deepseek-R1-Distill-Qwen-v2.5-32B, Qwen-v3-32B, and Deepseek-R1-Distill-Llama-v3-70B). Models were comparatively evaluated based on full confusion matrices, time/environmental costs, fairness, and explainability.Results: a total of 9311 labelled documents were evaluated. The fine-tuned DistilBERT_IBD model achieved the best performance overall (micro F1: 93.54%), followed by sBERT-Base (micro F1: 93.05%); however, specificity was an issue for both: (67.80-64.41%) respectively. LLMs performed well, given that they had never seen the training data (micro F1: 86.47-92.20%), but were comparatively slow (18-300 hours) and expensive. Bias was a significant issue for all effective model types.Conclusion: NLP has undergone significant advancements over the last 50 years. LLMs appear likely to solve the problem of re-identifying patients with IBD from clinical free text sources in the future. Once cost, performance and bias issues are addressed, they and their successors are likely to become the primary method of data retrieval for clinical data warehousing.</p
Identification of cohorts with inflammatory bowel disease amidst fragmented clinical databases via machine learning
No full textPurpose: inflammatory bowel disease (IBD) cohort identification typically relies primarily on read/billing codes, which may miss some patients. However, a complete picture cannot typically be obtained due to database fragmentation/missingness. This study used novel cohort retrieval methods to identify the total IBD cohort from a large university teaching hospital with a specialist intestinal failure unit.Methods: between 2007 and 2023, 11 clinical databases (ICD10 codes, OPCS4 codes, clinician-entry IBD registry, IBD patient portal, prescriptions, biochemistry, flare line calls, clinic appointments, endoscopy, histopathology, and clinic letters) were identified as having the potential to help identify local patients with IBD. The 11 databases were statistically compared, and a penalized logistic regression (LR) classifier was robustly trained and validated.Results: the gold-standard validation cohort comprised 2800 patients: 2092(75%) with IBD and 708(25%) without. All the databases contained unique patients that were not covered by the Casemix ICD-10 database. The penalizsed LR model (AUROC:0.85-Validation) confidently identified 8,159 patients with IBD (threshold: 0.496). By combining the likely true-positive predictions from the LR model with likely true-positive IBD clinic letters, a final estimate of 13,048 patients with IBD was obtained. ICD-10 codes combined with medication data identified only 8,048 patients, suggesting that present recapture methods missed 38.3% of the local cohort.Conclusion: diagnostic billing codes and medication data alone cannot accurately identify complete cohorts of individuals with IBD in secondary care. A multimodal cross-database model can partially compensate for this deficit. However, to improve this situation in the future, more robust natural language processing (NLP)-based identification mechanisms will be required.</p
Systematic review of natural language processing applied to gastroenterology & hepatology: the current state of the art
No full textObjective:This review assesses the progress of NLP in gastroenterology to date, grades the robustness of the methodology, exposes the field to a new generation of authors, and highlights opportunities for future research.Design:Seven scholarly databases (ACM Digital Library, Arxiv, Embase, IEEE Explore, Pubmed, Scopus and Google Scholar) were searched for studies published 2015–2023 meeting inclusion criteria. Studies lacking a description of appropriate validation or NLP methods were excluded, as were studies unavailable in English, focused on non-gastrointestinal diseases and duplicates. Two independent reviewers extracted study information, clinical/algorithm details, and relevant outcome data. Methodological quality and bias risks were appraised using a checklist of quality indicators for NLP studies.Results:Fifty-three studies were identified utilising NLP in Endoscopy, Inflammatory Bowel Disease, Gastrointestinal Bleeding, Liver and Pancreatic Disease. Colonoscopy was the focus of 21(38.9%) studies, 13(24.1%) focused on liver disease, 7(13.0%) inflammatory bowel disease, 4(7.4%) on gastroscopy, 4(7.4%) on pancreatic disease and 2(3.7%) studies focused on endoscopic sedation/ERCP and gastrointestinal bleeding respectively. Only 30(56.6%) of studies reported any patient demographics, and only 13(24.5%) scored as low risk of validation bias. 35(66%) studies mentioned generalisability but only 5(9.4%) mentioned explainability or shared code/models.Conclusion:NLP can unlock substantial clinical information from free-text notes stored in EPRs and is already being used, particularly to interpret colonoscopy and radiology reports. However, the models we have so far lack transparency, leading to duplication, bias, and doubts about generalisability. Therefore, greater clinical engagement, collaboration, and open sharing of appropriate datasets and code are needed
Early real-world effectiveness of ustekinumab for Crohn's disease
No full textObjective: To understand the effectiveness of ustekinumab in treating Crohn's disease (CD) in a UK real-world setting. Design: Retrospective cohort study using prospectively maintained clinical records. Setting: Single UK inflammatory bowel disease centre. Patients: Adult patients with an established diagnosis of CD prescribed ustekinumab outside of clinical trials at University Hospital Southampton (UHS). Interventions: Ustekinumab, a monoclonal antibody to the shared p40 subunit of interleukin (IL) 12 and IL-23 as part of routine clinical care. Main outcome measures: Effectiveness as measured by an improvement in physician's global assessment, drug persistence and improvement in biomarkers (C-reactive protein (CRP), albumin and calprotectin). Results: 84 patients were included, 72 had a postinduction review and 49 had 1-year data. At postinduction clinical review, clinical response occurred in 53% of patients and clinical remission occurred in 8%. For patients on ustekinumab at 1 year, clinical response occurred in 71% and remission in 14%. Adverse events included four patients with infections requiring admission, one drug-related rash, five CD surgeries and two CD exacerbations. Conclusions: Ustekinumab was well tolerated in a complex UK CD population and demonstrated benefit to patients in terms of clinical response and improvement of biomarkers and with some patients attaining clinical remission. No unexpected safety signals were seen.</p
PTH-36 identification & service evaluation of a primary sclerosing cholangitis cohort using natural language processing
No full textIntroduction: primary sclerosing cholangitis (PSC) is a rare and difficult to treat condition. PSC is strongly associated with malignancy, therefore screening and surveillance are paramount. PSC however does not have a unique UK ICD-10 diagnostic code, hence reliable patient cohort identification and thorough service evaluation is challenging. We used natural language processing (NLP) to identify the PSC patient cohort at University Hospital Southampton (UHS) and audited associated outcomes against recently updated British Society of Gastroenterology (BSG) management guidelines.Method: records of all patients with PSC at our institution between 2008-2020 were identified using our NLP methodology. We used fuzzy matching to analyse clinical records, and tokenized and lemmatized key paragraphs to identify key diagnostic patterns and exclude diagnostically uncertain or exclusive sentences. Anonymised discharge summaries, clinic letters, radiology reports, endoscopy records and histology were extracted and digitally trawled to identify the cohort characteristics.Results: we identified 125 patients with PSC followed-up at UHS. 39.2% (49) of these patients were missed in a parallel criterion-based review of case notes.We calculated an age-standardised point prevalence of 12.52 cases per 100,000 patients, 124% higher than typically cited UK figures. Service evaluation revealed high rates of clinic follow-up however lower than recommended rates of screening with colonoscopy and imaging (see Table 1). Introduction of a combined PSC/IBD clinic as a targeted service delivery intervention is addressing this shortfall with significant impact after 1 year. [PTH-36 Table 1 not included].Conclusions: PSC cohorts are difficult to identify due to a lack of a UK clinical code. An NLP based methodology proved highly effective at identifying all cases within our institution, with a 64.5% increase compared to conventional methods. This allowed rapid patient cohort identification and conversion of unstructured data to clinically useful structured data and could be reproduced at other institutions and for other diseases
P318 electronic referral grading system that puts patients and clinicians first
No full textIntroduction: the NHS Long Term Plan estimates demand for gastroenterology outpatient appointments (OPA) will continue to increase with the Royal College of Physicians deeming up to 20% of new referrals unnecessary. Bespoke electronic referral grading provides direct access to all community and hospital electronic patient records allowing fully informed immediate decision making. This can avoid unnecessary OPAs by redirecting/rejecting and sending appropriate patients direct-to-test while getting the right patient to the right clinician through sub-speciality tagging.Methods Outcomes: proportion/number of first OPA appointments saved, time to first appointment and subspecialty focus. Pre-grading period: Sep 2018–Mar 2020. Post-grading period: Aug 2020–Feb 2022. The separate 2WW IDA pathway was excluded. Cardiology/rheumatology were used as controls. Wait in days at the 50th percentile was calculated. One-tailed Mann Whitney U test calculated statistical significance at the p<0.05 level. Number of appointments saved per year and resultant financial implications were estimated.Results: GI received 3,768 consultant-graded referrals (Sep 2018–Mar 2020) and 3,908 (Aug 2020–Feb 2022).Referral sub-specialty groupings:1) Inflammatory Bowel Disease–765(19.6%)1. Irritable Bowel Syndrome–755(19.3%)2. Upper GI Diseases–753(19.3%)3. Non–IBD Colorectal–426(10.9%)4. Non–2WW Iron Deficiency Anaemia–345(8.8%)5. Unexplained Weight Loss–165(4.2%)6. Surgical/Hepatology–159(4.1%)7. Coeliac–139(3.6%)8. Endoscopy–126(3.2%)9. Complex Functional–106(2.7%)10. Not Tagged–88(2.3%)11. Intestinal Failure–81(2.1%)Documented referral rejection rates increased from 1% to 19% (n=745/3908) in the second period. 456 (11.6%) of patients were diverted directly to endoscopy, cumulatively saving 30.7% (n=1201) first OPA appointments equivalent to 150 new patient clinics. Total savings = £101,731/year in first GI OPA alone given the current block contract structure. Time to first appointment reduced by 58% but no improvement was seen in comparator specialties: [P318 Table 1Clinic wait times compared @ the 50th percentile not included]. Conclusions: use of a consultant-led electronic grading system had dramatic effects on the quality of data collected and significantly reduced first OPA waits at the trust. The reasons for this were triage direct-to-test and proactive rejection/redirection of referrals using sub-speciality tagging to get the right patient to the right clinician
- …
