Abstract
Background:
Within the lung ultrasound (LUS) score, A-lines, discrete B-lines, coalescent B-lines, and lobar consolidation represent scores of 0, 1, 2, and 3, respectively. However, the arbitrary ordinal values of 0 to 3 are not necessarily proportional to their clinical relevance. The objective of this study is to compare the relative strength of association between distinct LUS patterns on relevant clinical outcomes.
Methods:
This is a post hoc analysis of four prospective observational studies in COVID-19 subjects in the emergency department (ED) or in the ICU. Subjects were included if at least 3 zones per hemithorax were examined during ultrasound examination. Each LUS pattern was calibrated for relevant clinical outcomes based on the β coefficients derived from logistic and linear regression analyses, resulting in a rescaled LUS score. All scores were indexed using A-lines as a zero reference and discrete B-lines scaled to 1. Clinical outcomes were computed tomography severity score (CTSS), P/F ratio, ventilator-free days in the first 90 days (VFD-90), and 90-day mortality.
Results:
A total of 418 subjects were included, of which 114 were in the ED and 304 in the ICU. Increase in LUS score was associated with decreased VFD-90 and P/F ratio and increased CTSS and 90-day mortality. The rescaled LUS scores for P/F ratio and CTSS were smaller in magnitude than the original LUS scores (1, 1.24, 1.24 and 1, 1.48, 1.87, respectively), while rescaling for VFD-90 was larger in magnitude (1, 1.38, 2.83). The rescaled LUS score outperformed the original LUS score for every clinical outcome.
Conclusions:
The relative impact of LUS patterns differed depending on the clinical outcome of interest. The original arbitrary ordinal scores of 0 to 3 may not scale properly with any outcome. The LUS score may benefit from rescaling to reflect clinically meaningful outcomes more accurately.
Get full access to this article
View all access options for this article.