Routine chest X-ray in atraumatic chest pain with normal vital signs: A propensity-matched TriNetX analysis

Authors

Document Type

Journal Article

Publication Date

4-11-2026

Journal

The American journal of emergency medicine

Volume

105

DOI

10.1016/j.ajem.2026.04.012

Keywords

Chest X-ray; Chest pain; Diagnostic yield; Electronic health records; Emergency department; Pneumonia; TriNetX

Abstract

BACKGROUND: Patients presenting to the emergency department (ED) with chest pain frequently undergo a range of diagnostic evaluations aimed at rapidly identifying life-threatening conditions while stratifying overall risk. However, the utility of chest x-rays (CXR) in patients with chest pain requires further investigation, as the diagnostic yield is uncertain. This study examines the prevalence of clinically relevant diagnoses and procedures identified on CXR in a cohort of patients with chest pain who had CXR, versus those with chest pain without CXR. METHODS: A retrospective analysis was conducted using the real-world database TriNetX Research Network, encompassing 20 years of data and 130 million patients. Adult patients presenting to the ED with an ICD-10 code of chest pain were included and stratified based on whether a CXR was performed on the exact same day as the ED visit. Inclusion criteria strictly required documented normal vital signs at the index visit (oxygen saturation ≥ 94%, respiratory rate 12-20, and temperature ≤ 100.4 °F). Patients with trauma, abnormal electrocardiogram, or acute myocardial infarction were excluded. The TriNetX propensity score matching algorithm was used to balance cohorts based on demographic and clinical covariates, including troponin I. The primary outcome was the presence of pneumonia at the index visit. Secondary outcomes included pulmonary edema, pleural effusion, pneumomediastinum, cardiomegaly, and pneumothorax. RESULTS: The query yielded a total of 461,108 eligible patients. After propensity score matching, the final analysis included 204,966 patients (102,483 per group), with a mean age (±SD) of 43.5 (±17.3) years and 56.3% being female. The primary analysis demonstrated a statistically significant but clinically negligible difference in the diagnosis of pneumonia (1.42% vs 0.70%; Risk Difference [RD] 0.72%, p < 0.001). Secondary outcomes, including cardiomegaly (RD 1.16%) and pleural effusion (RD 0.67%), similarly showed statistically significant but minimal risk differences. CONCLUSION: For patients presenting with chest pain and normal vital signs, routine CXR is associated with a statistically significant but clinically negligible increase in the detection of pneumonia. These findings suggest that CXR offers low diagnostic yield in this specific population and support the reduction of routine imaging for patients without specific risk factors such as abnormal vital signs, trauma, or abnormal electrocardiogram findings.

Department

Emergency Medicine

Share

COinS