Intraobserver Variability and Reliability of Diaphragm Thickness Measurement on Ultrasonography by Critical Care Physician Among Patients with Sepsis

Background: Diaphragm thickness measurement by ultrasonography (USG) has been shown to be a good surrogate of diaphragmatic functions. However, data regarding reliability of diaphragm thickness measurement by critical care physicians among patients with sepsis is limited. Aims and Objective: This study was designed to assess the variability and reliability of diaphragm thickness measurement on USG by critical care physician in patients with sepsis. Materials and Methods: Study included patients with sepsis who were admitted between March 2015 and December 2016 in a tertiary care center. Three readings of diaphragm thickness were recorded separately during inspiration and expiration using B-mode of USG. Mean [standard deviation (SD)] of variation and intraclass correlation coefficient (ICC) in the measurement were calculated for assessment of variability and reliability, respectively. Results: Four hundred twenty measurements on 70 patients were included in the study. Mean (SD) variation for 1st, 2nd, and 3rd measurement during expiration were 0.233 (0.027) mm, 0.231 (0.029) mm, and 0.233 (0.029) mm, respectively. During inspiration mean (SD) variation for 1st, 2nd, and 3rd measurement were 0.285 (0.033) mm, 0.283 (0.031) mm, and 0.283 (0.033) mm, respectively. The ICC [95% confidence interval (CI)] among 1 vs. 2, 1 vs. 3, and 2 vs. 3 readings taken during expiration were 0.930 (0.889–0.956), 0.919 (0.873–0.949), and 0.940 (0.905–0.963), respectively. During inspiration, ICC (95% CI) among 1 vs. 2, 1 vs. 3, and 2 vs. 3 readings were 0.949 (0.919–0.968), 0.940 (0.905–0.962), and 0.945 (0.914–0.966), respectively. Overall ICC for three readings of diaphragmatic thickness during expiration and inspiration were 0.930 (0.898–0.954); P < 0.001 and 0.945 (0.919–0.963); P < 0.001, respectively. Conclusion: The measurement of diaphragm thickness using USG by critical care physicians among patients with sepsis is reliable with minimal variability.