Abstract
BACKGROUND: Rheumatic heart disease (RHD) affects approximately 40 million people, primarily in low- and middle-income countries. Transthoracic echocardiography is the gold standard for early detection of RHD. Yet, no single echocardiographic finding or combination of findings unequivocally confirms RHD, especially in subclinical stages. As a result, multiple diagnostic guidelines have been proposed. OBJECTIVES: To characterise the echocardiographic RHD findings and criteria included in five of those guidelines; to assess how these diagnostic criteria differ when applied to a concrete population, namely a cohort of high-school students in Sierra Leone; and, as a secondary objective, to provide an initial estimate of the burden of RHD in this country. METHODS: Our study has two parts. First, we performed a qualitative comparison of five major echocardiographic guidelines, mapping every diagnostic criterion into four roles (required, sufficient, combined, or not considered) for each guideline. Second, we conducted a school-based screening in Sierra Leone. A cohort of 604 asymptomatic female students (aged 10–22 years) underwent a two-phase echocardiographic protocol. We quantified regurgitant jet characteristics and morphological signs according to the criteria of the compared guidelines. RESULTS: Guideline discrepancies were substantial, particularly in jet length thresholds and the diagnostic role of morphological features. These differences shifted individual classifications across categories. In our cohort, 13 students (2.2%) had findings beyond normal under at least one guideline, and positive cases ranged from 0.99% to 2.15% depending on the guideline. Using the World Heart Federation (WHF) 2023 criteria, 1.16% (95% CI: 0.47–2.37) were classified as positive. CONCLUSIONS: Variability in echocardiographic criteria influences RHD classification and prevalence estimation. Our empirical study shows that the use of rigid echocardiographic cut-offs may lead to discarding some cases suspicious of RHD, especially in the WHF 2023 screening guideline, because one unfulfilled criterion automatically leads to a negative diagnosis, regardless of the number and relevance of positive findings. We argue that a probabilistic causal model that ponders the numerical measurements and considers correlations among findings would be more appropriate for screening and diagnosis. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12872-026-05758-0.