Comparison of Ultrasound-Derived Muscle Thickness With Computed Tomography Muscle Cross-Sectional Area on Admission to the Intensive Care Unit: A Pilot Cross-Sectional Study

Introduction The development of bedside methods to assess muscularity is an essential critical care nutrition research priority. We aimed to compare ultrasound-derived muscle thickness at 5 landmarks with computed tomography (CT) muscle area at intensive care unit (ICU) admission. Secondary aims were to (1) combine muscle thicknesses and baseline covariates to evaluate correlation with CT muscle area and (2) assess the ability of the best-performing ultrasound model to identify patients with low CT muscle area. Methods Adult patients who underwent CT scanning at the third lumbar area <72 hours after ICU admission were prospectively recruited. Muscle thickness was measured at mid-upper arm, forearm, abdomen, and thighs. Low CT muscle area was determined using published cutoffs. Pearson correlation compared ultrasound-derived muscle thickness and CT muscle area. Linear regression was used to develop ultrasound prediction models. Bland-Altman analyses compared ultrasound-predicted and CT-measured muscle area. Results Fifty ICU patients were enrolled, aged 52 +/- 20 years. Ultrasound-derived muscle thickness at each landmark correlated with CT muscle area (P < .001). The sum of muscle thickness at mid-upper arm and bilateral thighs, including age, sex, and the Charlson Comorbidity Index, improved the correlation with CT muscle area (r = 0.85; P < .001). Mean difference between ultrasound-predicted and CT-measured muscle area was -2 cm(2) (95% limits of agreement, -40 cm(2) to +36 cm(2)). The best-performing ultrasound model demonstrated good ability to identify 14 patients with low CT muscle area (area under curve = 0.79). Conclusion Ultrasound shows potential for assessing muscularity at ICU admission (Clinicaltrials.gov NCT03019913).

Automated summary

Ultrasound-derived muscle thickness at various landmarks was found to correlate with CT muscle area at ICU admission. Combining muscle thickness at multiple landmarks improved the correlation with CT muscle area. The best-performing ultrasound model demonstrated good ability to identify patients with low CT muscle area.