Acute internal medicine physicians' clinical intuition based on acute care telephone referral: A prospective study.

INTRODUCTION: In the Netherlands, most emergency department (ED) patients are referred by a general practitioner (GP) or a hospital specialist. Early risk stratification during telephone referral could allow the physician to assess the severity of the patients' illness in the prehospital setting. We aim to assess the discriminatory value of the acute internal medicine (AIM) physicians' clinical intuition based on telephone referral of ED patients to predict short-term adverse outcomes, and to investigate on which information their predictions are based. METHODS: In this prospective study, we included adult ED patients who were referred for internal medicine by a GP or a hospital specialist. Primary outcomes were hospital admission and triage category according to the Manchester Triage System (MTS). Secondary outcome was 31-day mortality. The discriminatory performance of the clinical intuition was assessed using an area under the receiver operating characteristics curve (AUC). To identify which information is important to predict adverse outcomes, we performed univariate regression analysis. Agreement between predicted and observed MTS triage category was assessed using intraclass and Spearman's correlation. RESULTS: We included 333 patients, of whom 172 (51.7%) were referred by a GP, 146 (43.8%) by a hospital specialist, and 12 (3.6%) by another health professional. The AIM physician's clinical intuition showed good discriminatory performance regarding hospital admission (AUC 0.72, 95% CI: 0.66-0.78) and 31-day mortality (AUC 0.73, 95% CI: 0.64-0.81). Univariate regression analysis showed that age ≥65 years and a sense of alarm were significant predictors. The predicted and observed triage category were similar in 45.2%, but in 92.5% the prediction did not deviate by more than one category. Intraclass and Spearman's correlation showed fair agreement between predicted and observed triage category (ICC 0.48, Spearman's 0.29). CONCLUSION: Clinical intuition based on relevant information during a telephone referral can be used to accurately predict short-term outcomes, allowing for early risk stratification in the prehospital setting and managing ED patient flow more effectively.

A comparison of machine learning models versus clinical evaluation for mortality prediction in patients with sepsis.

INTRODUCTION: Patients with sepsis who present to an emergency department (ED) have highly variable underlying disease severity, and can be categorized from low to high risk. Development of a risk stratification tool for these patients is important for appropriate triage and early treatment. The aim of this study was to develop machine learning models predicting 31-day mortality in patients presenting to the ED with sepsis and to compare these to internal medicine physicians and clinical risk scores. METHODS: A single-center, retrospective cohort study was conducted amongst 1,344 emergency department patients fulfilling sepsis criteria. Laboratory and clinical data that was available in the first two hours of presentation from these patients were randomly partitioned into a development (n = 1,244) and validation dataset (n = 100). Machine learning models were trained and evaluated on the development dataset and compared to internal medicine physicians and risk scores in the independent validation dataset. The primary outcome was 31-day mortality. RESULTS: A number of 1,344 patients were included of whom 174 (13.0%) died. Machine learning models trained with laboratory or a combination of laboratory + clinical data achieved an area-under-the ROC curve of 0.82 (95% CI: 0.80-0.84) and 0.84 (95% CI: 0.81-0.87) for predicting 31-day mortality, respectively. In the validation set, models outperformed internal medicine physicians and clinical risk scores in sensitivity (92% vs. 72% vs. 78%;p<0.001,all comparisons) while retaining comparable specificity (78% vs. 74% vs. 72%;p>0.02). The model had higher diagnostic accuracy with an area-under-the-ROC curve of 0.85 (95%CI: 0.78-0.92) compared to abbMEDS (0.63,0.54-0.73), mREMS (0.63,0.54-0.72) and internal medicine physicians (0.74,0.65-0.82). CONCLUSION: Machine learning models outperformed internal medicine physicians and clinical risk scores in predicting 31-day mortality. These models are a promising tool to aid in risk stratification of patients presenting to the ED with sepsis.

Risk stratification by abbMEDS and CURB-65 in relation to treatment and clinical disposition of the septic patient at the emergency department: a cohort study.

BACKGROUND: Sepsis leads to high mortality, therefore risk stratification is important. The abbMEDS (abbreviated Mortality Emergency Department Sepsis) score assesses sepsis severity and predicts mortality. In community-acquired pneumonia, the CURB-65 (Confusion, Urea, Respiration, Blood pressure, Age) also provides support in clinical decisions regarding antibiotic treatment and clinical disposition. We investigated the predictive value and feasibility of the abbMEDS and CURB-65 in sepsis patients at the ED and the relationship between the scores and antibiotic treatment and clinical disposition (i.e. admission and type of ward). METHODS: In this retrospective cohort study, we included 725 sepsis patients at the ED. We investigated the value in predicting 28-day mortality and feasibility of both scores. We calibrated the abbMEDS. We further assessed the relationship between the three risk categories per score and antibiotic treatment (i.e. oral and intravenous narrow or broad-spectrum) and clinical disposition. RESULTS: Both abbMEDS and CURB-65 were good predictors of 28-day mortality (13.0%) (AUC 0.77 [95% CI 0.72 - 0.83] and 0.73 [95% CI 0.67 - 0.78], respectively) and feasible (complete score 92.7 and 93.9%, respectively). In the high risk category of the abbMEDS, all patients were admitted and treated with intravenous broad-spectrum antibiotics. In the high risk category of the CURB-65, 2.5% were not admitted and 4.4% received no antibiotics. CONCLUSION: Both abbMEDS and CURB-65 are good predictors of 28-day mortality in septic ED patients. The abbMEDS is well calibrated and matches current clinical decisions concerning antibiotic treatment and clinical disposition, while this is less so for the CURB-65. In the future, use of the abbMEDS at the ED may improve sepsis care when its value as a decision support tool can be confirmed.