Influence of Patient and Clinician Gender on Emergency Department HEART Scores: A Secondary Analysis of a Prospective Observational Trial

Bayesian intelligence for medical diagnosis: a pilot study on patient disposition for emergency medicine chest pain

OBJECTIVES

Clinicians can rapidly and accurately diagnose disease, learn from experience, and explain their reasoning. Computational Bayesian medical decision-making might replicate this expertise. This paper assesses a computer system for diagnosing cardiac chest pain in the emergency department (ED) that decides whether to admit or discharge a patient.

METHODS

The system can learn likelihood functions by counting data frequency. The computer compares patient and disease data profiles using likelihood. It calculates a Bayesian probabilistic diagnosis and explains its reasoning. A utility function applies the probabilistic diagnosis to produce a numerical BAYES score for making a medical decision.

RESULTS

We conducted a pilot study to assess BAYES efficacy in ED chest pain patient disposition. Binary BAYES decisions eliminated patient observation. We compared BAYES to the HEART score. On 100 patients, BAYES reduced HEART's false positive rate 18-fold from 58.7 to 3.3 %, and improved ROC AUC accuracy from 0.928 to 1.0.

CONCLUSIONS

The pilot study results were encouraging. The data-driven BAYES score approach could learn from frequency counting, make fast and accurate decisions, and explain its reasoning. The computer replicated these aspects of diagnostic expertise. More research is needed to reproduce and extend these finding to larger diverse patient populations.

Evaluating the performance of the HEART score in a Tanzanian emergency department

OBJECTIVE

The HEART score successfully risk stratifies emergency department (ED) patients with chest pain in high-income settings. However, this tool has not been validated in low-income countries.

METHODS

This is a secondary analysis of a prospective observational study that was conducted in a Tanzanian ED from January 2019 through January 2023. Adult patients with chest pain were consecutively enrolled, and their presenting symptoms and medical history were recorded. Electrocardiograms and point-of-care troponin assays were obtained for all participants. Thirty-day follow-up was conducted, assessing for major adverse cardiac events (MACEs), defined as death, myocardial infarction, or coronary revascularization (coronary artery bypass grafting or percutaneous coronary intervention). HEART scores were calculated for all participants. Likelihood ratios, sensitivity, specificity, and negative predictive values (NPVs) were calculated for each HEART cutoff score to predict 30-day MACEs, and area under the curve (AUC) was calculated from the receiver operating characteristic curve.

RESULTS

Of 927 participants with chest pain, the median (IQR) age was 61 (45.5-74.0) years. Of participants, 216 (23.3%) patients experienced 30-day MACEs, including 163 (17.6%) who died, 48 (5.2%) with myocardial infarction, and 23 (2.5%) with coronary revascularization. The positive likelihood ratio for each cutoff score ranged from 1.023 (95% CI 1.004-1.042; cutoff ≥ 1) to 3.556 (95% CI 1.929-6.555; cutoff ≥ 7). The recommended cutoff of ≥4 to identify patients at high risk of MACEs yielded a sensitivity of 59.4%, specificity of 52.8%, and NPV of 74.7%. The AUC was 0.61.

CONCLUSIONS

Among patients with chest pain in a Tanzanian ED, the HEART score did not perform as well as in high-income settings. Locally validated risk stratification tools are needed for ED patients with chest pain in low-income countries.