Operator independent continuous ultrasound monitoring of diaphragm excursion predicts successful weaning from mechanical ventilation: a prospective observational study

Abstract Background In mechanically ventilated patients, diaphragm ultrasound can identify diaphragm weakness and predict weaning failure. We evaluated whether a novel operator-independent ultrasound-based medical device allowing continuous monitoring of the diaphragm (CUSdi) could reliably (1) measure diaphragm excursion (EXdi) and peak contraction velocity (PCVdi), (2) predict weaning outcome, and (3) approximate transdiaphragmatic pressure (Pdi). Methods In 49 mechanically ventilated patients, CUSdi was recorded during a 30-min spontaneous breathing trial (SBT), and EXdi and PCVdi were measured. In subgroups of patients, standard ultrasound measurement of EXdi and PCVdi was performed (n = 36), and Pdi derived parameters (peak and pressure time product, n = 30) were measured simultaneously. Results The agreement bias between standard ultrasound and CUSdi for EXdi was 0.1 cm (95% confidence interval -0.7–0.9 cm). The regression of Passing-Bablok indicated a lack of systematic difference between EXdi measured with standard ultrasound and CUSdi, which were positively correlated (Rho = 0.84, p < 0.001). Weaning failure was observed in 54% of patients. One, two and three minutes after the onset of the SBT, EXdi was higher in the weaning success group than in the failure group. Two minutes after the onset of the SBT, an EXdi < 1.1 cm predicted weaning failure with a sensitivity of 0.83, a specificity of 0.68, a positive predictive value of 0.76, and a negative predictive value of 0.24. There was a weak correlation between EXdi and both peak Pdi (r = 0.22, 95% confidence interval 0.15 – 0.28) and pressure time product (r = 0.13, 95% confidence interval 0.06 – 0.20). Similar results were observed with PCVdi. Conclusions Operator-independent continuous diaphragm monitoring quantifies EXdi reliably and can predict weaning failure with an identified cut-off value of 1.1 cm. Trial registration clinicaltrial.gov, NCT04008875 (submitted 12 April 2019, posted 5 July 2019) and NCT03896048 (submitted 27 March 2019, posted 29 March 2019).