Validation of GPT-4 for clinical event classification: A comparative analysis with ICD codes and human reviewers.

BACKGROUND AND AIM: Effective clinical event classification is essential for clinical research and quality improvement. The validation of artificial intelligence (AI) models like Generative Pre-trained Transformer 4 (GPT-4) for this task and comparison with conventional methods remains unexplored. METHODS: We evaluated the performance of the GPT-4 model for classifying gastrointestinal (GI) bleeding episodes from 200 medical discharge summaries and compared the results with human review and an International Classification of Diseases (ICD) code-based system. The analysis included accuracy, sensitivity, and specificity evaluation, using ground truth determined by physician reviewers. RESULTS: GPT-4 exhibited an accuracy of 94.4% in identifying GI bleeding occurrences, outperforming ICD codes (accuracy 63.5%, P < 0.001). GPT-4's accuracy was either slightly lower or statistically similar to individual human reviewers (Reviewer 1: 98.5%, P < 0.001; Reviewer 2: 90.8%, P = 0.170). For location classification, GPT-4 achieved accuracies of 81.7% and 83.5% for confirmed and probable GI bleeding locations, respectively, with figures that were either slightly lower or comparable with those of human reviewers. GPT-4 was highly efficient, analyzing the dataset in 12.7 min at a cost of 21.2 USD, whereas human reviewers required 8-9 h each. CONCLUSION: Our study indicates GPT-4 offers a reliable, cost-efficient, and faster alternative to current clinical event classification methods, outperforming the conventional ICD coding system and performing comparably to individual expert human reviewers. Its implementation could facilitate more accurate and granular clinical research and quality audits. Future research should explore scalability, prompt and model tuning, and ethical implications of high-performance AI models in clinical data processing.

CLINICAL UTILITY OF RISK SCORES IN VARICEAL BLEEDING.

BACKGROUND: Variceal bleeding remains important cause of upper gastrointestinal bleed. Various risk scores are used in risk stratification for non-variceal bleed. Their utility in variceal bleeding patients is not clear. This study aims to compare probability of these scores in predicting various outcomes in same population. OBJECTIVE: This study aims to compare probability of these scores in predicting various outcomes in same population. To study characteristics and validate AIMS65, Rockall, Glasgow Blatchford score(GBS), Progetto Nazionale Emorragia Digestiva (PNED) score in variceal Upper Gastrointestinal Bleed (UGIB) patients for predicting various outcomes in our population. METHODS: Three hundred subjects with UGIB were screened prospectively. Of these 141 patients with variceal bleeding were assessed with clinical, blood investigations and endoscopy and risk scores were calculated and compared to non-variceal cases. All cases were followed up for 30 days for mortality, rebleeding, requirement of blood transfusion and need of radiological or surgical intervention. RESULTS: Variceal bleeding (141) was more common than non variceal (134) and 25 had negative endoscopy. In variceal group, cirrhosis (85%) was most common etiology. Distribution of age and sex were similar in both groups. Presence of coffee coloured vomitus (P=0.002), painless bleed (P=0.001), edema (P=0.001), ascites (P=0.001), hemoglobin <7.5 gms (P<0.001), pH<7.35 (P<0.001), serum bicarbonate level <17.6 mmol/L (P<0.001), serum albumin<2.75 gms% (P<0.001), platelet count <1.2 lacs/µL (P<0.001), high INR 1.35 (P<0.001), BUN >25mmol/L (P<0.001), and ASA status (P<0.001), high lactate >2.85 mmol/L (P=0.001) were significant. However, no factor was found significant on multivariate analysis. Rockall was found to be significant in predicting mortality and rebleed. AIMS65 was also significant in predicting mortality. GBS was significant in predicting blood transfusion and need of intervention. PNED score was significant in all events except mortality. CONCLUSION: All four scores had lower predictive potential in predicting events in variceal bleed. However, AIMS65 & Rockall score were significant in predicting mortality, while GBS in predicting need of transfusion and intervention. PNED score was significant in all events except mortality.

Clinical coding data algorithm to categorize type of gastrointestinal bleeding as a primary reason for massive transfusion: results from the Australian and New Zealand Massive Transfusion Registry.

BACKGROUND: Management of major gastrointestinal bleeding (GIB) may require massive transfusion (MT), but limited data are available. Upper and lower GIB have different aetiologies, prognosis, bleeding patterns and outcomes. Better understanding of current transfusion management and outcomes in these patients is important. We sought to define and validate an algorithm based on clinical coding data to distinguish critical upper and lower GIB using data from the Australian and New Zealand Massive Transfusion Registry (ANZ-MTR). STUDY DESIGN AND METHODS: Australian and New Zealand Massive Transfusion Registry hospital-source data on adult patients receiving a MT (defined as ≥5 red cell units within 4 h) for any bleeding context were used. An algorithm allocating ICD-10-AM codes into 'probable' or 'possible' causes of GIB was developed and applied to the ANZ-MTR. Source medical records of 69 randomly selected cases were independently reviewed to validate the algorithm. RESULTS: Of 5482 MT cases available from 25 hospitals, 716 (13%) were identified as GIB with 538/716 (75%) categorized 'probable' and 178/716 'possible' GIB. Upper and lower GIB causes of MT were identified for 455/538 (85%) and 76/538 (14%) 'probable' cases, respectively; 7/538 (1·3%) cases had both upper and lower GIB. Allocation by the algorithm into a 'probable' GIB category had a 95·7% (CI: 90-100%) positive predictive value when validated against source medical records. CONCLUSION: An algorithm based on ICD-10-AM codes can be used to accurately categorize patients with luminal GIB as the primary reason for MT, enabling further study of this critically unwell and resource-intensive cohort of patients.

Utilidade clínica dos escores de risco no sangramento varicoso / Clinical utility of risk scores in variceal bleeding

ABSTRACT BACKGROUND: Variceal bleeding remains important cause of upper gastrointestinal bleed. Various risk scores are used in risk stratification for non-variceal bleed. Their utility in variceal bleeding patients is not clear. This study aims to compare probability of these scores in predicting various outcomes in same population. OBJECTIVE: This study aims to compare probability of these scores in predicting various outcomes in same population. To study characteristics and validate AIMS65, Rockall, Glasgow Blatchford score(GBS), Progetto Nazionale Emorragia Digestiva (PNED) score in variceal Upper Gastrointestinal Bleed (UGIB) patients for predicting various outcomes in our population. METHODS: Three hundred subjects with UGIB were screened prospectively. Of these 141 patients with variceal bleeding were assessed with clinical, blood investigations and endoscopy and risk scores were calculated and compared to non-variceal cases. All cases were followed up for 30 days for mortality, rebleeding, requirement of blood transfusion and need of radiological or surgical intervention. RESULTS: Variceal bleeding (141) was more common than non variceal (134) and 25 had negative endoscopy. In variceal group, cirrhosis (85%) was most common etiology. Distribution of age and sex were similar in both groups. Presence of coffee coloured vomitus (P=0.002), painless bleed (P=0.001), edema (P=0.001), ascites (P=0.001), hemoglobin <7.5 gms (P<0.001), pH<7.35 (P<0.001), serum bicarbonate level <17.6 mmol/L (P<0.001), serum albumin<2.75 gms% (P<0.001), platelet count <1.2 lacs/µL (P<0.001), high INR 1.35 (P<0.001), BUN >25mmol/L (P<0.001), and ASA status (P<0.001), high lactate >2.85 mmol/L (P=0.001) were significant. However, no factor was found significant on multivariate analysis. Rockall was found to be significant in predicting mortality and rebleed. AIMS65 was also significant in predicting mortality. GBS was significant in predicting blood transfusion and need of intervention. PNED score was significant in all events except mortality. CONCLUSION: All four scores had lower predictive potential in predicting events in variceal bleed. However, AIMS65 & Rockall score were significant in predicting mortality, while GBS in predicting need of transfusion and intervention. PNED score was significant in all events except mortality.

RESUMO CONTEXTO: O sangramento varicoso permanece como importante causa de sangramento gastrointestinal superior. Vários escores são utilizados na estratificação do risco para sangramento não varicoso. Sua utilidade em pacientes de sangramento varicoso não é clara. OBJETIVO: Este estudo tem como objetivo comparar a probabilidade desses escores em prever vários desfechos na mesma população. Estudar característica e validar o AIMS65, o Rockall, a Pontuação de Glasgow Blatchford (GBS), o escore Progetto Nazionale Emorragia Digestiva (PNED), na pontuação em hemorragia gastrointestinal varicosa superior (UGIB) em pacientes para prever vários resultados em nossa população. MÉTODOS: Um total de 300 indivíduos com UGIB foram rastreados prospectivamente. Destes, 141 pacientes com sangramento varicoso foram submetidos à avaliação clínica, hematológica e endoscopia tendo seus escores de risco calculados e comparados aos casos não-varicosos. Todos os casos foram acompanhados por 30 dias para mortalidade, necessidade de transfusão sanguínea por ressangramento ou de necessidade de intervenção radiológica ou cirúrgica. RESULTADOS: O sangramento varicoso (141) foi mais comum do que não varicoso (134) e em 25 teve endoscopia negativa. No grupo varicoso, a cirrose foi a etiologia mais comum (85%). A distribuição da idade e do sexo foi semelhante em ambos os grupos. Presença de vômito colorido em borra de café (P=0,002), sangramento indolor (P=0,001), edema (P=0,001), ascite (P=0,001), hemoglobina <7,5 GMS (P<0,001), pH <7,35 (P<0,001), nível de bicarbonato sérico <17,6 mmol/L (P<0,001), albumina sérica <2,75 GMS% (P<0,001), contagem plaquetária <1,2 Lacs/μL (P<0,001), INR elevada 1,35 (P<0,001), Bun >25 mmol/L (P<0,001) e estado ASA (P<0,001), lactato elevado >2,85 mmol/L (P=0,001) foram significativos. Entretanto, nenhum fator foi encontrado como significativo na análise multivariada. Rockall foi significativo em prever a mortalidade e ressangrar. O AIMS65 também foi significante na predição da mortalidade. O GBS foi significativo na predição de transfusão sanguínea e necessidade de intervenção. O escore de PNED foi significante em todos os eventos, exceto mortalidade. CONCLUSÃO: Todos os quatro escores apresentaram menor potencial preditivo na predição de eventos em sangramento varicoso. Entretanto, o AIMS65 e o escore de Rockall foram significantes na predição da mortalidade, enquanto o GBS na predição da necessidade de transfusão e intervenção. O escore de PNED foi significante em todos os eventos, exceto mortalidade.

Joint recommendations on management of anaemia in patients with gastrointestinal bleeding in Hong Kong.

The demand for blood products continues to grow in an unsustainable manner in Hong Kong. While anaemia associated with gastrointestinal bleeding (GIB) is the leading indication for transfusion, there is no local recommendation regarding best practices for transfusion. We aimed to provide evidence-based recommendations regarding management of anaemia in patients with acute and chronic GIB. We reviewed all original papers, meta-analyses, systematic reviews, or guidelines that were available in PubMed. For acute GIB, a restrictive transfusion strategy, targeting a haemoglobin threshold of 7 to 8 g/dL, should be adopted because overtransfusion is associated with significantly higher all-cause mortality and re-bleeding. A liberal transfusion strategy should only be considered in patients with co-existing symptomatic coronary artery disease, targeting a haemoglobin threshold of 9 to 10 g/dL. When acute GIB settles, patients should be prescribed iron supplements if iron deficiency is present. For chronic GIB, iron stores should be replenished aggressively via iron supplementation before consideration of blood transfusion, except in patients with symptoms of severe anaemia. Oral iron replacement is the preferred first-line therapy, while intravenous iron is indicated for patients with inflammatory bowel disease, poor response or poor tolerability to oral iron, and in whom a rapid correction of iron deficit is preferred. Intravenous iron is underutilised and the risk of anaphylactic reaction to current preparations is extremely low. These recommendations are provided to local clinicians to facilitate judicious and appropriate use of red cell products and iron replacement therapy in patients with GIB.

Comparison of AIMS65, Glasgow-Blatchford and Rockall scoring approaches in predicting the risk of in-hospital death among emergency hospitalized patients with upper gastrointestinal bleeding: a retrospective observational study in Nanjing, China.

BACKGROUND: This study aims to compare the performance of AIMS65, Glasgow-Blatchford (GBS) and Rockall scores (RS) in predicting the death risk among emergency-hospitalized patients with upper gastrointestinal bleeding (UGIB) in regional China. METHODS: A retrospective study was implemented between January 2014 and December 2015. Eligible participants were those who were hospitalized with UGIB. The outcome variable was in-hospital death, while explanatory variables were AIMS65, GBS and RS scores. Odds ratios (OR) and 95% confidence interval (CI) were estimated to assess the association of AIMS65, GBS and RS with death risk using multivariate logistic regression models. The areas under the receiver operating characteristics curve (AUC) of three scoring systems were computed to compare their predictive power. RESULTS: Among 799 UGIB participants, 674 were non-variceal bleeding (NVUGIB) and 125 variceal bleeding (VUGIB) patients. AIMS65 (OR = 14.72, 95% CI = 6.48, 33.43) and RS (OR = 1.60, 95% CI = 1.20, 2.13) were positively associated with the risk of in-hospital death. Moreover, AIMS65 (AUC = 0.91, 95% CI = 0.84, 0.98) performed the best in predicting in-hospital death, followed by RS (AUC = 0.79, 95% CI = 0.72, 0.86) and GBS (AUC = 0.71, 95% CI = 0.59, 0.83) among overall UGIB participants. AIMS65 was also the best indicator to predict in-hospital death among either NVUGIB participants (AUC = 0.89, 95% CI = 0.80, 0.98) or VUGIB participants (AUC = 0.94, 95% CI = 0.89, 1.00). CONCLUSIONS: AIMS65, GBS and RS scoring approaches were all acceptable for predicting in-hospital death among UGIB patients irrespective of the subtype of UGIB in China. The AIMS65 might be the most powerful predictor.

Outcome pre-specification requires sufficient detail to guard against outcome switching in clinical trials: a case study.

BACKGROUND: Pre-specification of outcomes is an important tool to guard against outcome switching in clinical trials. However, if the outcome is not sufficiently clearly defined, then different definitions could be applied and analysed, with only the most favourable result reported. METHODS: In order to assess the impact that differing outcome definitions could have on treatment effect estimates, we re-analysed data from TRIGGER, a cluster randomised trial comparing two red blood cell transfusion strategies for patients with acute upper gastrointestinal bleeding. We varied several aspects of the definition of further bleeding: (1) the criteria for what constitutes a further bleeding episode; (2) how further bleeding is assessed; and (3) the time-point at which further bleeding is measured. RESULTS: There were marked discrepancies in the estimated odds ratios (OR) (range 0.23-0.94) and corresponding P values (range < 0.001-0.89) between different outcome definitions. At the extremes, differing outcome definitions led to markedly different conclusions; one definition led to very little evidence of a treatment effect (OR = 0.94, 95% confidence interval [CI] = 0.37-2.40, P = 0.89), while another led to very strong evidence of a treatment effect (OR = 0.23, 95% CI = 0.11-0.50, P < 0.001). CONCLUSIONS: Outcomes should be pre-specified in sufficient detail to avoid differing definitions being analysed and only the most favourable result being reported. TRIAL REGISTRATION: Clinical Trials.gov, NCT02105532 . Registered on 7 April 2014.

Asia-Pacific working group consensus on non-variceal upper gastrointestinal bleeding: an update 2018.

Non-variceal upper gastrointestinal bleeding remains an important emergency condition, leading to significant morbidity and mortality. As endoscopic therapy is the 'gold standard' of management, treatment of these patients can be considered in three stages: pre-endoscopic treatment, endoscopic haemostasis and post-endoscopic management. Since publication of the Asia-Pacific consensus on non-variceal upper gastrointestinal bleeding (NVUGIB) 7 years ago, there have been significant advancements in the clinical management of patients in all three stages. These include pre-endoscopy risk stratification scores, blood and platelet transfusion, use of proton pump inhibitors; during endoscopy new haemostasis techniques (haemostatic powder spray and over-the-scope clips); and post-endoscopy management by second-look endoscopy and medication strategies. Emerging techniques, including capsule endoscopy and Doppler endoscopic probe in assessing adequacy of endoscopic therapy, and the pre-emptive use of angiographic embolisation, are attracting new attention. An emerging problem is the increasing use of dual antiplatelet agents and direct oral anticoagulants in patients with cardiac and cerebrovascular diseases. Guidelines on the discontinuation and then resumption of these agents in patients presenting with NVUGIB are very much needed. The Asia-Pacific Working Group examined recent evidence and recommends practical management guidelines in this updated consensus statement.

The CDD System in Computed Tomographic Diagnosis of Diverticular Disease.

Purpose This overview sums up the Classification of Diverticular Disease (CDD) with regard to its application in computed tomographic diagnosis and briefly recapitulates its targeted advantages over preliminary systems. Primarily, application of the CDD in computed tomography diagnostics is described. Differences with respect to the categories of the older systems are pointed out on the level of each CDD type using imaging examples. Materials and Methods The presented images are derived from our institute according to the S2k criteria. Literature was researched on PubMed. Results The CDD constitutes an improvement compared to older systems for categorizing the stages of diverticular disease. It provides more discriminatory power on the descriptive-morphological level and defines as well as differentiates more courses of the disease. Furthermore, the categories translate more directly into state-of-the-art decision-making concerning hospitalization and therapy. Conclusion The CDD should be applied routinely in the computed tomographic diagnosis of diverticular disease. Typical imaging patterns are presented. Key points · The CDD is superior to its predecessors. It better stratifies categories of diverticular disease by morphology, course and modern options for treatment of the disease.. · Computed tomography is the dominant imaging modality. Different stages show typical imaging patterns.. · Non-abscessed phlegmonous peridiverticulitis is now interpreted as an uncomplicated course.. · Minimal paracolic air does not constitute a full-fledged perforation in terms of a pneumoperitoneum (CDD type 2c).. Citation Format · Pustelnik D, Elsholtz FH, Bojarski C et al. The CDD System in Computed Tomographic Diagnosis of Diverticular Disease. Fortschr Röntgenstr 2017; 189: 740 - 747.

ACR Appropriateness Criteria® Nonvariceal Upper Gastrointestinal Bleeding.

Upper gastrointestinal bleeding (UGIB) remains a significant cause of morbidity and mortality with mortality rates as high as 14%. This document addresses the indications for imaging UGIB that is nonvariceal and unrelated to portal hypertension. The four variants are derived with respect to upper endoscopy. For the first three, it is presumed that upper endoscopy has been performed, with three potential initial outcomes: endoscopy reveals arterial bleeding source, endoscopy confirms UGIB without a clear source, and negative endoscopy. The fourth variant, "postsurgical and traumatic causes of UGIB; endoscopy contraindicated" is considered separately because upper endoscopy is not performed. When endoscopy identifies the presence and location of bleeding but bleeding cannot be controlled endoscopically, catheter-based arteriography with treatment is an appropriate next study. CT angiography (CTA) is comparable with angiography as a diagnostic next step. If endoscopy demonstrates a bleed but the endoscopist cannot identify the bleeding source, angiography or CTA can be typically performed and both are considered appropriate. In the event of an obscure UGIB, angiography and CTA have been shown to be equivalent in identifying the bleeding source; CT enterography may be an alternative to CTA to find an intermittent bleeding source. In the postoperative or traumatic setting when endoscopy is contraindicated, primary angiography, CTA, and CT with intravenous contrast are considered appropriate. The American College of Radiology Appropriateness Criteria are evidence-based guidelines for specific clinical conditions that are reviewed annually by a multidisciplinary expert panel. The guideline development and revision include an extensive analysis of current medical literature from peer reviewed journals and the application of well-established methodologies (RAND/UCLA Appropriateness Method and Grading of Recommendations Assessment, Development, and Evaluation or GRADE) to rate the appropriateness of imaging and treatment procedures for specific clinical scenarios. In those instances where evidence is lacking or equivocal, expert opinion may supplement the available evidence to recommend imaging or treatment.

[Clinical epidemiological characteristics and change trend of upper gastrointestinal bleeding over the past 15 years].

OBJECTIVE: To investigate the clinical epidemiology change trend of upper gastrointestinal bleeding (UGIB) over the past 15 years. METHODS: Consecutive patients who was diagnosed as continuous UGIB in the endoscopy center of The First Affiliated Hospital of Sun-Yat University during the period from 1 January 1997 to 31 December 1998 and the period from 1 January 2012 to 31 December 2013 were enrolled in this study. Their gender, age, etiology, ulcer classification, endoscopic treatment and hospitalization mortality were compared between two periods. RESULTS: In periods from 1997 to 1998 and 2012 to 2013, the detection rate of UGIB was 9.99%(928/9 287) and 4.49%(1 092/24 318)(χ2=360.089, P=0.000); the percentage of male patients was 73.28%(680/928) and 72.44% (791/1 092) (χ2=0.179, P=0.672), and the onset age was (47.3±16.4) years and (51.4±18.2) years (t=9.214, P=0.002) respectively. From 1997 to 1998, the first etiology of UGIB was peptic ulcer bleeding, accounting for 65.2%(605/928)[duodenal ulcer 47.8%(444/928), gastric ulcer 8.3%(77/928), stomal ulcer 2.3%(21/928), compound ulcer 6.8%(63/928)],the second was cancer bleeding(7.0%,65/928), and the third was esophageal and gastric varices bleeding (6.4%,59/928). From 2012 to 2013, peptic ulcer still was the first cause of UGIB, but the ratio obviously decreased to 52.7%(575/1092)(χ2=32.467, P=0.000)[duodenal ulcer 31.9%(348/1092), gastric ulcer 9.4%(103/1092), stomal ulcer 2.8%(30/1092), compound ulcer 8.6%(94/1092)]. The decreased ratio of duodenal ulcer bleeding was the main reason (χ2=53.724, P=0.000). Esophageal and gastric varices bleeding became the second cause (15.1%,165/1 092, χ2=38.976, P=0.000), and cancer was the third cause (9.2%,101/1 092, χ2=3.352, P=0.067). The largest increasing amplitude of the onset age was peptic ulcer bleeding [(46.2±16.7) years vs. (51.9±18.9) years, t=-5.548, P=0.000), and the greatest contribution to the amplitude was duodenal ulcer bleeding [(43.4±15.9) years vs. (48.4±19.4) years, t=-3.935, P=0.000], while the onset age of esophageal and gastric varices bleeding [(49.8±14.1) years vs. (48.8±13.9) years, t=0.458, P=0.648] and cancer [(58.4±13.4) years vs. (58.9±16.7) years, t=-0.196, P=0.845] did not change significantly. Compared with the period from 1997 to 1998, the detection rate of high risk peptic ulcer rebleeding (Forrest stage I(a, I(b, II(a and II(b) increased (χ2=39.958, P=0.000) in the period from 2012 to 2013. From 1997 to 1998, 54 patients underwent endoscopic treatment, and the achievement ratio of hemostasis was 79.6% (43/54). From 2012 to 2013, 261 patients underwent endoscopic treatment and the achievement ratio of hemostasis was 96.9%(253/261), which was significantly higher (χ2=23.287, P=0.000). Compared to the period from 1997 to 1998, more patients with variceal bleeding or non-variceal bleeding received endoscopic treatment in time (39.0% vs. 70.3%, χ2=51.930, P=0.000; 3.6% vs. 15.6%, χ2=62.292, P=0.000, respectively), and higher ratio of patients staging Forrest stage I(a to II(b also received endoscopic treatment in the period from 2012 to 2013 [27.4%(26/95) vs. 68.5%(111/162), χ2=40.739, P=0.000]. More qualified endoscopic hemostatic techniques were used, containing thermocoagulation (0 vs. 15.2%, χ2=79.518, P=0.000), hemostatic clip (0 vs. 55.9%, χ2=20.879, P=0.000), hemostatic clip combined with thermocoagulation (4.3% vs. 16.4%, χ2=5.154, P=0.023), while less single injection was used (87.1% vs. 6.2%, χ2=10.420, P=0.001), and single spraying for hemostasis was completely abandoned in the period from 2012 to 2013. The ratio of inpatients undergoing reoperation decreased obviously in the period from 2012 to 2013 [9.3%(86/928) vs. 6.0%(65/1092), χ2=7.970, P=0.005], while no significant difference was found in mortality during hospitalization between two periods. CONCLUSION: Compared with the period from 1997 to1998, the mean onset age of UGIB increased, and the ratio of peptic ulcer bleeding decreased due to the reduction of duodenal ulcer bleeding, the detection rate of high risk peptic ulcer rebleeding increased, the cure rate of endoscopic treatment for UGIB increased, more reasonable and immediate hemostatic methods were used, but overall mortality did not change obviously in the period from 2012 to 2013.

Evaluation of clinical coding data to determine causes of critical bleeding in patients receiving massive transfusion: a bi-national, multicentre, cross-sectional study.

OBJECTIVES: To evaluate the use of routinely collected data to determine the cause(s) of critical bleeding in patients who receive massive transfusion (MT). BACKGROUND: Routinely collected data are increasingly being used to describe and evaluate transfusion practice. MATERIALS/METHODS: Chart reviews were undertaken on 10 randomly selected MT patients at 48 hospitals across Australia and New Zealand to determine the cause(s) of critical bleeding. Diagnosis-related group (DRG) and International Classification of Diseases (ICD) codes were extracted separately and used to assign each patient a cause of critical bleeding. These were compared against chart review using percentage agreement and kappa statistics. RESULTS: A total of 427 MT patients were included with complete ICD and DRG data for 427 (100%) and 396 (93%), respectively. Good overall agreement was found between chart review and ICD codes (78·3%; κ = 0·74, 95% CI 0·70-0·79) and only fair overall agreement with DRG (51%; κ = 0·45, 95% CI 0·40-0·50). Both ICD and DRG were sensitive and accurate for classifying obstetric haemorrhage patients (98% sensitivity and κ > 0·94). However, compared with the ICD algorithm, DRGs were less sensitive and accurate in classifying bleeding as a result of gastrointestinal haemorrhage (74% vs 8%; κ = 0·75 vs 0·1), trauma (92% vs 62%; κ = 0·78 vs 0·67), cardiac (80% vs 57%; κ = 0·79 vs 0·60) and vascular surgery (64% vs 56%; κ = 0·69 vs 0·65). CONCLUSION: Algorithms using ICD codes can determine the cause of critical bleeding in patients requiring MT with good to excellent agreement with clinical history. DRG are less suitable to determine critical bleeding causes.

Evaluation of scoring models for identifying the need for therapeutic intervention of upper gastrointestinal bleeding: A new prediction score model for Japanese patients.

BACKGROUND AND AIM: Multiple scoring systems have been developed to predict outcomes in patients with upper gastrointestinal bleeding. We determined how well these and a newly established scoring model predict the need for therapeutic intervention, excluding transfusion, in Japanese patients with upper gastrointestinal bleeding. METHODS: We reviewed data from 212 consecutive patients with upper gastrointestinal bleeding. Patients requiring endoscopic intervention, operation, or interventional radiology were allocated to the therapeutic intervention group. Firstly, we compared areas under the curve for the Glasgow-Blatchford, Clinical Rockall, and AIMS65 scores. Secondly, the scores and factors likely associated with upper gastrointestinal bleeding were analyzed with a logistic regression analysis to form a new scoring model. Thirdly, the new model and the existing model were investigated to evaluate their usefulness. RESULTS: Therapeutic intervention was required in 109 patients (51.4%). The Glasgow-Blatchford score was superior to both the Clinical Rockall and AIMS65 scores for predicting therapeutic intervention need (area under the curve, 0.75 [95% confidence interval, 0.69-0.81] vs 0.53 [0.46-0.61] and 0.52 [0.44-0.60], respectively). Multivariate logistic regression analysis retained seven significant predictors in the model: systolic blood pressure <100 mmHg, syncope, hematemesis, hemoglobin <10 g/dL, blood urea nitrogen ≥22.4 mg/dL, estimated glomerular filtration rate ≤ 60 mL/min per 1.73 m2 , and antiplatelet medication. Based on these variables, we established a new scoring model with superior discrimination to those of existing scoring systems (area under the curve, 0.85 [0.80-0.90]). CONCLUSION: We developed a superior scoring model for identifying therapeutic intervention need in Japanese patients with upper gastrointestinal bleeding.

Predictive Parameters of Symptomatic Hematochezia Following 5-Fraction Gantry-Based SABR in Prostate Cancer.

PURPOSE: This study identified predictors of high-grade late hematochezia (HH) following 5-fraction gantry-based stereotactic ablative radiation therapy (SABR). METHODS AND MATERIALS: Hematochezia data for 258 patients who received 35 to 40 Gy SABR in 5-fractions as part of sequential phase 2 prospective trials was retrieved. Grade 2 or higher late rectal bleeding was labeled HH. Hematochezia needing steroid suppositories, 4% formalin, or 1 to 2 sessions of argon plasma coagulation (APC) was labeled grade 2. More than 2 sessions of APC, blood transfusion, or a course of hyperbaric oxygen was grade 3 and development of visceral fistula, grade 4. Various dosimetric and clinical factors were analyzed using univariate and multivariate analyses. Receiver operating characteristic (ROC) curve analysis and recursive partitioning analysis were used to determine clinically valid cut-off points and identify risk groups, respectively. RESULTS: HH was observed in 19.4%, grade ≥3 toxicity in 3.1%. Median follow-up was 29.7 months (interquartile range [IQR]: 20.6-61.7) Median time to develop HH was 11.7 months (IQR: 9.0-15.2) from the start of radiation. At 2 years, cumulative HH was 4.9%, 27.2%, and 42.1% in patients who received 35 Gy to prostate (4-mm planning target volume [PTV] margin), 40 Gy to prostate (5-mm PTV margin), and 40 Gy to prostate/seminal vesicles (5-mm PTV margin), respectively (P<.0001). In the ROC analysis, volume of rectum receiving radiation dose of 38 Gy (V38) was a strong predictor of HH with an area under the curve of 0.65. In multivariate analysis, rectal V38 (≥2.0 cm(3); odds ratio [OR]: 4.7); use of anticoagulants in the follow-up period (OR: 6.5) and presence of hemorrhoids (OR: 2.7) were the strongest predictors. Recursive partitioning analysis showed rectal V38 < 2.0 cm(3), and use of anticoagulants or rectal V38 ≥ 2.0 cm(3) plus 1 other risk factor resulted in an HH risk of >30%. CONCLUSIONS: Rectal V38 and 2 clinical factors were strong predictors of HH following 5-fraction SABR. Planning constraints should keep rectal V38 below 2.0 cm(3).

Risk assessment scores for patients with upper gastrointestinal bleeding and their use in clinical practice.

Upper gastrointestinal bleeding (UGIB) is a common cause for emergency admission to hospital representing a significant clinical as well as economic burden. UGIB encompasses a wide range of severities from life-threatening exsanguination to minor bleeding that may not require hospital admission. Patients with UGIB are often initially assessed and managed by junior doctors and non-gastroenterologists. Several risk scores have been created for the assessment of these patients, some requiring endoscopic data for calculation and others that are calculable from clinical data alone. A key question in clinical practice is how to accurately identify patients with UGIB at high risk of adverse outcome. Patients considered high risk are more likely to experience adverse outcomes and will require urgent intervention. In contrast, those patients with UGIB who are considered to be low risk could potentially be managed on an outpatient basis. The Glasgow Blatchford Score (GBS) appears best at identifying patients at low risk of requiring intervention or death and therefore may be best for use in clinical practice, allowing outpatient management in low risk cases. There has been some debate as to the optimal GBS cut-off score for safely identifying this low-risk group. Many guidelines suggest that patients with a GBS of zero can be safely managed as outpatients, but more recent studies have suggested that this threshold could potentially be safely increased to ≤1. Most other patients require inpatient endoscopy within 24 h and the full Rockall score remains important for risk assessment following endoscopy, particularly as it includes the endoscopic diagnosis. A minority of patients will require emergency endoscopy following resuscitation, but at present there is no evidence that risk scores can accurately identify this very high-risk group. Studies have shown the latest risk assessment score, the AIMS65, looks promising in the prediction of mortality. However, to date there is no data on the use of the AIMS65 in identifying low risk patients for possible outpatient management.

Acute upper gastrointestinal bleeding in childhood: development of the Sheffield scoring system to predict need for endoscopic therapy.

BACKGROUND/AIMS: Upper gastrointestinal bleeding (UGIB) is a rare and potentially life-threatening condition in childhood. In adults with UGIB, validated scoring systems exist, but these are not applicable to children. The aim of this study was to construct a clinical scoring system to accurately predict the need for endoscopic haemostatic intervention. METHODS: A retrospective data collection occurred during a 3-year period at a tertiary children's hospital. A total of 69 patients who had had endoscopic assessment were divided into group 1 (no intervention required) and group 2 (intervention required). A wide range of clinical parameters were collated including preexisting conditions, melaena, haematemesis and degree, transfusion requirement, parameters of hypovolaemia, presenting haemoglobin (Hb), Hb drop during 24 hours, platelet count, coagulation indices, liver function tests, and urea/electrolytes. RESULTS: Parameters that reached statistical significance for endoscopic intervention (group 1 vs group 2) were the presence of significant preexisting condition, melaena, large haematemesis, heart rate (HR) >20 mean HR for age, prolonged capillary refill time (CRT), Hb drop of >20 g/L, need for fluid bolus, need for blood transfusion (Hb < 80 g/L), and need for other blood products. Using these parameters, a number of scoring models were tested, and the most predictive resulted in a scoring system constructed with a total of 24 and a cutoff for intervention of 8. According to this design, there were 4 false-negatives in the interventional group with 3 false-positives in the noninterventional group. This resulted in a positive predictive value (PPV) of 91.18% (95% confidence interval [CI] 76.3-98.04), negative predictive value (NPV) of 88.57% (95% CI 73.24-96.73), sensitivity of 88.7% (95% CI 73.24-96.73), and specificity of 91.18% (95% CI 76.3-98.04). CONCLUSIONS: In our study population, we were able to formulate a scoring system with reasonable PPV and NPV to predict the need for endoscopic intervention in acute UGIB in children. Prospective evaluation is now required.

Low interrater reliability in grading of rectal bleeding using National Cancer Institute Common Toxicity Criteria and Radiation Therapy Oncology Group Toxicity scales: a survey of radiation oncologists.

PURPOSE: To measure concordance among genitourinary radiation oncologists in using the National Cancer Institute Common Toxicity Criteria (NCI CTC) and Radiation Therapy Oncology Group (RTOG) grading scales to grade rectal bleeding. METHODS AND MATERIALS: From June 2013 to January 2014, a Web-based survey was sent to 250 American and Canadian academic radiation oncologists who treat prostate cancer. Participants were provided 4 case vignettes in which patients received radiation therapy and developed rectal bleeding and were asked for management plans and to rate the bleeding according to NCI CTC v.4 and RTOG late toxicity grading (scales provided). In 2 cases, participants were also asked whether they would send the patient for colonoscopy. A multilevel, random intercept modeling approach was used to assess sources of variation (case, respondent) in toxicity grading to calculate the intraclass correlation coefficient (ICC). Agreement on a dichotomous grading scale (low grades 1-2 vs high grades 3-4) was also assessed, using the κ statistic for multiple respondents. RESULTS: Seventy-two radiation oncologists (28%) completed the survey. Forty-seven (65%) reported having either written or been principal investigator on a study using these scales. Agreement between respondents was moderate (ICC 0.52, 95% confidence interval [CI] 0.47-0.58) when using NCI CTC and fair using the RTOG scale (ICC 0.28, 95% CI 0.20-0.40). Respondents who chose an invasive management were more likely to select a higher toxicity grade (P<.0001). Using the dichotomous scale, we observed moderate agreement (κ = 0.42, 95% CI 0.40-0.44) with the NCI CTC scale, but only slight agreement with the RTOG scale (κ = 0.19, 95% CI 0.17-0.21). CONCLUSION: Low interrater reliability was observed among radiation oncologists grading rectal bleeding using 2 common scales. Clearer definitions of late rectal bleeding toxicity should be constructed to reduce this variability and avoid ambiguity in both reporting and interpretation.

Validation study in four health-care databases: upper gastrointestinal bleeding misclassification affects precision but not magnitude of drug-related upper gastrointestinal bleeding risk.

OBJECTIVE: To evaluate the accuracy of disease codes and free text in identifying upper gastrointestinal bleeding (UGIB) from electronic health-care records (EHRs). STUDY DESIGN AND SETTING: We conducted a validation study in four European electronic health-care record (EHR) databases such as Integrated Primary Care Information (IPCI), Health Search/CSD Patient Database (HSD), ARS, and Aarhus, in which we identified UGIB cases using free text or disease codes: (1) International Classification of Disease (ICD)-9 (HSD, ARS); (2) ICD-10 (Aarhus); and (3) International Classification of Primary Care (ICPC) (IPCI). From each database, we randomly selected and manually reviewed 200 cases to calculate positive predictive values (PPVs). We employed different case definitions to assess the effect of outcome misclassification on estimation of risk of drug-related UGIB. RESULTS: PPV was 22% [95% confidence interval (CI): 16, 28] and 21% (95% CI: 16, 28) in IPCI for free text and ICPC codes, respectively. PPV was 91% (95% CI: 86, 95) for ICD-9 codes and 47% (95% CI: 35, 59) for free text in HSD. PPV for ICD-9 codes in ARS was 72% (95% CI: 65, 78) and 77% (95% CI: 69, 83) for ICD-10 codes (Aarhus). More specific definitions did not have significant impact on risk estimation of drug-related UGIB, except for wider CIs. CONCLUSIONS: ICD-9-CM and ICD-10 disease codes have good PPV in identifying UGIB from EHR; less granular terminology (ICPC) may require additional strategies. Use of more specific UGIB definitions affects precision, but not magnitude, of risk estimates.