International Validation of the Canadian Syncope Risk Score : A Cohort Study

Predicting the Outcomes of Inpatient Cardiac Evaluation for Syncope Using Validated Risk Scores

BACKGROUND

Validated syncope risk scores were aimed to predict a cardiac etiology and are mainly used in the decision of hospital admission. Whether these scores could also predict the outcomes of inpatient cardiac evaluation is unknown and was the subject of our study.

METHODS

This was an observational study including consecutive patients admitted for syncope evaluation. All patients completed prolonged electrocardiogram monitoring and an echocardiography before discharge. The area under the receiver-operating characteristic curve (AUC) was used to evaluate the ability of validated risk scores to predict positive inpatient findings. Subsequently, a multivariate regression was performed to identify independent predictors for positive cardiac evaluation, which were then incorporated into the best predictive risk scores.

RESULTS

Three hundred ninety-seven patients were included, 56 (14%) with a positive inpatient cardiac evaluation. The Osservatorio Epidemiologico sulla Sincope Lazio and Canadian Syncope Risk Score achieved the largest AUC (0.701, 95% confidence interval [CI] 0.63-0.77 and 0.694, 95% CI 0.62-0.77, respectively). Yet, all scores provided relatively high sensitivity with low specificity. Multivariate regression revealed age ≥75 (adjusted odds ratio 3.50, 95% CI 1.5-7.9) and abnormal cardiac auscultation (adjusted odds ratio 4.79, 95% CI 2.5-9.1) to be independent predictors. Incorporating these factors led to a significantly higher prediction ability of the Osservatorio Epidemiologico sulla Sincope Lazio (AUC of 0.787, P < .01) and Canadian Syncope Risk Score (AUC 0.778, P < .01) modified scores.

CONCLUSIONS

Current syncope risk scores provide limited prediction ability for the outcomes of inpatient cardiac syncope work-up. One should specifically consider age > 75 years and either cardiac murmur or irregular heart rate on examination very significant in implying a cardiac etiology for syncope. Although these factors may be obvious, current risk scores can be interpreted in such a fashion that ignores the importance of findings extracted from a good history and physical examination.

Syncope in the Emergency Department: A Practical Approach

Syncope is a common condition encountered in the emergency department (ED), accounting for about 0.6-3% of all ED visits. Despite its high frequency, a widely accepted management strategy for patients with syncope in the ED is still missing. Since syncope can be the presenting condition of many diseases, both severe and benign, most research efforts have focused on strategies to obtain a definitive etiologic diagnosis. Nevertheless, in everyday clinical practice, a definitive diagnosis is rarely reached after the first evaluation. It is thus troublesome to aid clinicians' reasoning by simply focusing on differential diagnoses. With the current review, we would like to propose a management strategy that guides clinicians both in the identification of conditions that warrant immediate treatment and in the management of patients for whom a diagnosis is not immediately reached, differentiating those that can be safely discharged from those that should be admitted to the hospital or monitored before a final decision. We propose the mnemonic acronym RED-SOS: Recognize syncope; Exclude life-threatening conditions; Diagnose; Stratify the risk of adverse events; Observe; decide on the Setting of care. Based on this acronym, in the different sections of the review, we discuss all the elements that clinicians should consider when assessing patients with syncope.

AI in the ED: Assessing the efficacy of GPT models vs. physicians in medical score calculation

BACKGROUND AND AIMS

Artificial Intelligence (AI) models like GPT-3.5 and GPT-4 have shown promise across various domains but remain underexplored in healthcare. Emergency Departments (ED) rely on established scoring systems, such as NIHSS and HEART score, to guide clinical decision-making. This study aims to evaluate the proficiency of GPT-3.5 and GPT-4 against experienced ED physicians in calculating five commonly used medical scores.

METHODS

This retrospective study analyzed data from 150 patients who visited the ED over one week. Both AI models and two human physicians were tasked with calculating scores for NIH Stroke Scale, Canadian Syncope Risk Score, Alvarado Score for Acute Appendicitis, Canadian CT Head Rule, and HEART Score. Cohen's Kappa statistic and AUC values were used to assess inter-rater agreement and predictive performance, respectively.

RESULTS

The highest level of agreement was observed between the human physicians (Kappa = 0.681), while GPT-4 also showed moderate to substantial agreement with them (Kappa values of 0.473 and 0.576). GPT-3.5 had the lowest agreement with human scorers. These results highlight the superior predictive performance of human expertise over the currently available automated systems for this specific medical outcome. Human physicians achieved a higher ROC-AUC on 3 of the 5 scores, but none of the differences were statistically significant.

CONCLUSIONS

While AI models demonstrated some level of concordance with human expertise, they fell short in emulating the complex clinical judgments that physicians make. The study suggests that current AI models may serve as supplementary tools but are not ready to replace human expertise in high-stakes settings like the ED. Further research is needed to explore the capabilities and limitations of AI in emergency medicine.

A Hybrid Model for 30-Day Syncope Prognosis Prediction in the Emergency Department

Syncope is a challenging problem in the emergency department (ED) as the available risk prediction tools have suboptimal predictive performances. Predictive models based on machine learning (ML) are promising tools whose application in the context of syncope remains underexplored. The aim of the present study was to develop and compare the performance of ML-based models in predicting the risk of clinically significant outcomes in patients presenting to the ED for syncope. We enrolled 266 consecutive patients (age 73, IQR 58-83; 52% males) admitted for syncope at three tertiary centers. We collected demographic and clinical information as well as the occurrence of clinically significant outcomes at a 30-day telephone follow-up. We implemented an XGBoost model based on the best-performing candidate predictors. Subsequently, we integrated the XGboost predictors with knowledge-based rules. The obtained hybrid model outperformed the XGboost model (AUC = 0.81 vs. 0.73, p < 0.001) with acceptable calibration. In conclusion, we developed an ML-based model characterized by a commendable capability to predict adverse events within 30 days post-syncope evaluation in the ED. This model relies solely on clinical data routinely collected during a patient's initial syncope evaluation, thus obviating the need for laboratory tests or syncope experienced clinical judgment.

Risk stratification tools for patients with syncope in emergency medical services and emergency departments: a scoping review

BACKGROUND

Patients with a syncope constitute a challenge for risk stratification in (prehospital) emergency care. Professionals in EMS and ED need to differentiate the high-risk from the low-risk syncope patient, with limited time and resources. Clinical decision rules (CDRs) are designed to support professionals in risk stratification and clinical decision-making. Current CDRs seem unable to meet the standards to be used in the chain of emergency care. However, the need for a structured approach for syncope patients remains. We aimed to generate a broad overview of the available risk stratification tools and identify key elements, scoring systems and measurement properties of these tools.

METHODS

We performed a scoping review with a literature search in MEDLINE, CINAHL, Pubmed, Embase, Cochrane and Web of Science from January 2010 to May 2022. Study selection was done by two researchers independently and was supervised by a third researcher. Data extraction was performed through a data extraction form, and data were summarised through descriptive synthesis. A quality assessment of included studies was performed using a generic quality assessment tool for quantitative research and the AMSTAR-2 for systematic reviews.

RESULTS

The literature search identified 5385 unique studies; 38 were included in the review. We discovered 19 risk stratification tools, one of which was established in EMS patient care. One-third of risk stratification tools have been validated. Two main approaches for the application of the tools were identified. Elements of the tools were categorised in history taking, physical examination, electrocardiogram, additional examinations and other variables. Evaluation of measurement properties showed that negative and positive predictive value was used in half of the studies to assess the accuracy of tools.

CONCLUSION

A total of 19 risk stratification tools for syncope patients were identified. They were primarily established in ED patient care; most are not validated properly. Key elements in the risk stratification related to a potential cardiac problem as cause for the syncope. These insights provide directions for the key elements of a risk stratification tool and for a more advanced process to validate risk stratification tools.

Impact of Admission of Patients With Syncope in Non-Teaching Hospitals Versus Teaching Hospitals: A Nationwide Analysis

Syncope is a common condition affecting many individuals, and it remains uncertain whether admission to academic medical centers (AMCs) leads to better outcomes than non-AMCs. This study is aimed to investigate whether there is a difference in mortality, length of stay (LoS), and total hospital charges between patients admitted with syncope to AMCs and non-AMCs. This retrospective cohort study used the National Inpatient Database (NIS) to examine patients aged 18 years and older admitted with a primary diagnosis of syncope to AMCs and non-AMCs from 2016 to 2020. Univariate and multivariate logistic regression analyses were conducted, adjusting for confounders, to assess the primary outcome of all-cause in-hospital mortality and secondary outcomes, including hospital LoS and total cost of admission. Patient characteristics were also described. Of the 451,820 patients who met the inclusion criteria, 69.6% were admitted to AMCs and 30.4% to non-AMCs. Patient age was similar between the two groups (68 years in AMC versus 70 years in non-AMC; p < 0.001), as was sex distribution (52% female in AMC versus 53% in non-AMC; 48% male in AMC versus 47% in non-AMC; p < 0.002). Most patients in both groups were white, while the percentages of black and Hispanic patients were slightly higher in non-AMCs. The study found no difference in all-cause mortality between patients admitted to AMCs and non-AMCs (p = 0.33). However, LoS was marginally longer in AMC patients (2.6 days in AMC versus 2.4 days in the non-AMC group; p < 0.001), and the total cost was higher for AMCs by $3,526 per admission. The estimated total economic burden related to syncope was over 3 billion USD per year. This study suggests that the teaching status of hospitals did not significantly affect the mortality of patients admitted with syncope. However, it may have contributed to marginally longer hospital LoS and higher total hospital charges.

Arrhythmic syncope: From diagnosis to management

Syncope is a concerning symptom that affects a large proportion of patients. It can be related to a heterogeneous group of pathologies ranging from trivial causes to diseases with a high risk of sudden death. However, benign causes are the most frequent, and identifying high-risk patients with potentially severe etiologies is crucial to establish an accurate diagnosis, initiate effective therapy, and alter the prognosis. The term cardiac syncope refers to those episodes where the cause of the cerebral hypoperfusion is directly related to a cardiac disorder, while arrhythmic syncope is cardiac syncope specifically due to rhythm disorders. Indeed, arrhythmias are the most common cause of cardiac syncope. Both bradyarrhythmia and tachyarrhythmia can cause a sudden decrease in cardiac output and produce syncope. In this review, we summarized the main guidelines in the management of patients with syncope of presumed arrhythmic origin. Therefore, we presented a thorough approach to syncope work-up through different tests depending on the clinical characteristics of the patients, risk stratification, and the management of syncope in different scenarios such as structural heart disease and channelopathies.