The study was carried out from December 1st 2018 to February 28th 2019 in the educational rooms of eight Italian University Hospitals (“Magna Graecia” University of Catanzaro, IRCCS San Raffaele Scientific Institute of Milan, University of Udine, “Eastern Piedmont” University of Novara, University of Pisa, University of Parma, University of Catania and Catholic University of the “Sacred Heart” of Rome). The study was approved by the local Ethics Committees and written informed consent was obtained from all participants. The trial was prospectively registered on clinicaltrials.gov (Identifier: NCT03704129; release date 17th October 2018).
We recruited 172 voluntary learners with no experience of ultrasound assessments among medical students or first-year residents. We also designated 14 tutors, two in each centre, with a minimum 2-year experience of DUS in critical care US. Prior to study initiation, all tutors met on the web to standardize the practical training to be administered to the interventional group.
A video tutorial based on the current literature [1, 6,7,8,9] and focusing on key principles of the technique, including acoustic windows and anatomical landmarks featuring diaphragmatic US, was shown to all learners. The video tutorial is available online at https://youtu.be/B7AYP9fElyE.
Afterwards, a questionnaire including 10 multiple-choice questions was administered and considered to be passed when at least 70% of the questions were correctly answered. The questionnaire is enclosed as Additional file 1. Course participants who passed the theoretical test were then randomized to either intervention or control group.
Randomization was achieved with an allocation ratio of 1:1 by means of a computer-generated sequence, operated by an investigator not involved in the trial. Allocation blindness was assured using sequentially numbered sealed opaque envelopes, prepared by the aforementioned investigator. Each envelope contained the allocation of the learners to either control or interventional group, with a unique identifier code. The randomization was based on a centralized phone call system.
Learners randomized to the interventional group had access to the practical training, tutored by an expert evaluator who interactively explained how to perform DUS, before accessing DUS examination. Learners randomized to the control group directly accessed DUS examination, without any practical training by expert tutors. DUS examination was performed on healthy volunteers, not involved in the study protocol. Irrespective of the group of randomizations, learners were asked to independently perform DUS using both acoustic windows. All measurements were performed by learners after images’ acquisition and storage. A local investigator recorded the measurements. A tutor then judged if the images were correctly acquired, and only in such a case, he performed his own measurements on the same acquired images, being blind to the results obtained by the learners. These measurements were also recorded by the local investigator.
Data acquisition and analysis
Diaphragm US was performed by course participants and tutors using one of the following devices: MyLab™30, Esaote, Genova, Italy; MySono U6, Samsung, Seoul, South Korea; EPIQ7 ultrasound system, Philips Healthcare, Bothell, WA, USA.
Sonographic evaluation was conducted on the right hemi-diaphragm, as previously described [1, 6, 9,10,11]. Briefly, DD was ascertained through a 3.5–5 MHz phased array probe, placed immediately below the costal margin in the mid-clavicular line and directed medially, cephalad and dorsally, so that the US beam reached perpendicularly the posterior third of the hemi-diaphragm [1, 6, 9, 11]. The motion of the diaphragm and other anatomical structures along the selected line was displayed in “time-motion” mode (M-mode). DD was measured placing the first caliper at the end of expiration phase, while the second caliper was placed at the apex of inspiration slope [1, 6, 9, 11]. Diaphragm thickness was assessed through a linear 13 MHz probe placed in the 9th–10th intercostal space, closed to the midaxillary line, angled perpendicular to the chest wall, to identify the apposition zone of the diaphragm. Diaphragmatic thickness was the determined in M-mode at end-expiration (Thickexp) and at peak inspiration (Thickinsp) as the distance between the diaphragmatic pleura and the peritoneum [6, 7, 9]. TF was then computed as Thickinsp − Thickexp/Thickexp and expressed in percentage [6, 7, 9].
If a learner failed to correctly display the diaphragm in one of the two acoustic windows, the examination was considered to be negative. Only the measurements by learners who correctly identified both acoustic windows were considered for further analysis. Based on the previous agreement among members of the steering committee, the measurements were considered to be accurate when in the following predetermined ranges: (1) DD ± 2 mm from the value reported by the tutor and (2) TF ± 20% of the assessment recorded by the tutor. If both acoustic windows were correctly identified, and the resulting measurements were included within the predetermined ranges, the participant passed the examination, so that the first study outcome was achieved.
Gaussian data distribution was tested by means of the Kolmogorov–Smirnov test. Data are presented as mean (± standard deviation) or as median [25th–75th interquartile], as indicated. Categorical data were compared through Chi-square test while continuous data with Student t test or Mann–Whitney U test, as appropriate. By means of the Spearman’s rank correlation test, we determined the correlation coefficients (ρ) [95% interval confidence] between measurements (i.e., DD, Thickinsp, Thickexp, and TF) obtained by tutors and learners of each group, and the corresponding p values. To test the statistical significance of the difference, the ρ values separately obtained in the intervention and control groups were then compared and the z and p values were determined . For all comparisons, p values < 0.05 were considered to be significant.