National Health Service (NHS) Blood and Transplant, Oxford, UK; National Heart and Lung Institute, Imperial College, London, UK.
Division of Epidemiology and Biostatistics, Western University, London, ON, Canada; Department of Medicine, University of Western Ontario, London, ON, Canada.
Lancet Gastroenterol Hepatol. 2017 Sep;2(9):635-643. doi: 10.1016/S2468-1253(17)30150-4. Epub 2017 Jun 23.
Acute lower gastrointestinal bleeding is a common reason for emergency hospital admission, and identification of patients at low risk of harm, who are therefore suitable for outpatient investigation, is a clinical and research priority. We aimed to develop and externally validate a simple risk score to identify patients with lower gastrointestinal bleeding who could safely avoid hospital admission.
We undertook model development with data from the National Comparative Audit of Lower Gastrointestinal Bleeding from 143 hospitals in the UK in 2015. Multivariable logistic regression modelling was used to identify predictors of safe discharge, defined as the absence of rebleeding, blood transfusion, therapeutic intervention, 28 day readmission, or death. The model was converted into a simplified risk scoring system and was externally validated in 288 patients admitted with lower gastrointestinal bleeding (184 safely discharged) from two UK hospitals (Charing Cross Hospital, London, and Hammersmith Hospital, London) that had not contributed data to the development cohort. We calculated C statistics for the new model and did a comparative assessment with six previously developed risk scores.
Of 2336 prospectively identified admissions in the development cohort, 1599 (68%) were safely discharged. Age, sex, previous admission for lower gastrointestinal bleeding, rectal examination findings, heart rate, systolic blood pressure, and haemoglobin concentration strongly discriminated safe discharge in the development cohort (C statistic 0·84, 95% CI 0·82-0·86) and in the validation cohort (0·79, 0·73-0·84). Calibration plots showed the new risk score to have good calibration in the validation cohort. The score was better than the Rockall, Blatchford, Strate, BLEED, AIMS65, and NOBLADS scores in predicting safe discharge. A score of 8 or less predicts a 95% probability of safe discharge.
We developed and validated a novel clinical prediction model with good discriminative performance to identify patients with lower gastrointestinal bleeding who are suitable for safe outpatient management, which has important economic and resource implications.
Bowel Disease Research Foundation and National Health Service Blood and Transplant.
急性下消化道出血是急诊住院的常见原因，确定低风险伤害的患者，因此适合门诊调查，是临床和研究的重点。我们旨在开发和外部验证一个简单的风险评分，以识别可安全避免住院的下消化道出血患者。
我们使用英国 2015 年 143 家医院国家下消化道出血比较审计的数据进行模型开发。多变量逻辑回归模型用于确定安全出院的预测因子，定义为无再出血、输血、治疗干预、28 天再入院或死亡。该模型被转化为简化的风险评分系统，并在英国两家医院（伦敦查令十字医院和伦敦哈默史密斯医院）的 288 名下消化道出血患者（184 例安全出院）中进行外部验证，这些医院没有为开发队列提供数据。我们计算了新模型的 C 统计量，并与之前开发的六个风险评分进行了比较评估。
在开发队列中，2336 例前瞻性入院患者中，1599 例（68%）安全出院。年龄、性别、下消化道出血既往入院、直肠检查结果、心率、收缩压和血红蛋白浓度在开发队列（C 统计量 0.84，95%CI 0.82-0.86）和验证队列（0.79，0.73-0.84）中强烈区分安全出院。校准图显示新的风险评分在验证队列中有良好的校准。该评分在预测安全出院方面优于 Rockall、Blatchford、Strate、BLEED、AIMS65 和 NOBLADS 评分。评分 8 或更低预测安全出院的概率为 95%。
我们开发并验证了一种具有良好判别性能的新型临床预测模型，以识别适合安全门诊管理的下消化道出血患者，这对经济和资源具有重要意义。
肠道疾病研究基金会和英国国家卫生服务血液与移植。
