Determinants of image quality in transthoracic echocardiography: a retrospective cohort study

Objectives

To identify patient factors associated with poor transthoracic echocardiographic image quality and to evaluate whether a simple pre-test triage model could improve imaging efficiency.

Design

Retrospective cohort study with derivation and independent validation cohorts, together with a model-based cost-effectiveness analysis.

Setting

Single large UK tertiary centre using routinely collected data from 2010 to 2020.

Participants

70,597 adult transthoracic echocardiograms, divided into a derivation cohort (n = 40,000) and an independent validation cohort (n = 30,597).

Main outcome measures

Poor image quality (limited or non-diagnostic vs good or adequate), model discrimination, sensitivity, specificity and comparative imaging costs.

Results

Of 70,597 studies, 24,213 (34.3%) were poor quality, including 8582 (12.2%) non-diagnostic and 15,631 (22.1%) limited studies. Lung disease was the strongest predictor (OR 2.04, 95% CI 1.59 to 2.61), followed by suspected heart failure, inpatient status, arrhythmia, prior cardiac surgery and permanent pacemaker (all p < 0.01). Validation performance was modest (AUC 0.58; sensitivity 67.3%; specificity 45.7%). In a model-guided simulation using 2024/25 NHS tariffs, total imaging costs were lower than with standard care (£11.85 million vs £12.16 million), yielding an estimated saving of £317,331.

Conclusions

Several routinely available clinical factors influence transthoracic echocardiographic image quality. Although individual-level prediction is modest, pre-test triage may help direct higher-risk patients to contrast echocardiography or alternative imaging, reducing repeat testing and improving efficiency.