ABSTRACT
INTRODUCTION: Paediatric NAFLD is an increasing global health concern, which can be effectively managed with early detection. Screening, using accurate, affordable, and accessible tests is recommended, however, there is currently no consensus on the most appropriate tests. Although ultrasound techniques are widely used, their performance against reference tests have not been fully assessed. METHODS: A literature search of related databases for peer-reviewed original articles published from January 2010-March 2024 was conducted. Appropriate tools were used to systematise and document the search results and selected studies were quality assessed and critically appraised. Extracted data was subjected to thematic analysis and narrative synthesis. RESULTS: Eighteen articles met the inclusion criteria. B-mode and Quantitative ultrasound techniques were compared against MR spectroscopy, MRI-PDFF and Liver biopsy. CONCLUSION: Liver echogenicity and Steato-scores were the B-mode methods used. The former was less effective, with a maximum reported sensitivity of 70%. The latter reached up to 100% sensitivity, and >80% specificity. Ultrasound performed better with moderate-severe steatosis. There was not enough evidence to support steatosis grading, possibly due to small sample sizes and lack of established cut-off values. QUS (Quantitative Ultrasound)) methods including Continuous Attenuation Parameter (CAP), Attenuation Coefficient (AC), Ultrasound derived fat fraction (UDFF), Tissue Scatter Imaging (TSI) Hepato-Renal Index (HRI), Heterogeneity Index (HIA), Computer Assisted Ultrasound (CAUS) and Picture Archiving and Communication System (PACS-based Image analysis performed better than B-mode methods. Although QUS demonstrated excellent performance, with sensitivity and specificity of up to 100%, this will require further verification before implementation in practice. PRACTICE IMPLICATIONS: Ultrasound techniques can effectively be used for paediatric NAFLD screening, especially in higher-risk subjects. The steato-scores method is currently recommendable for this, with excellent potential for the use of QUS, after cut-off values and validation requirements have been addressed.
