Study design
This was a prospective study evaluating LUS training and proficiency of physicians performing lung ultrasound in the Patan Hospital Emergency Department. It was conducted from July 2017 through May 2019. This study was approved by the Nepal Health Research Council Ethical Review Board, and written consent was obtained from each physician participant.
Study setting and population
Patan Hospital is a large urban teaching hospital affiliated with Patan Academy of Health Sciences. The ED has an annual patient volume of approximately 48,000 patients. The majority of care in the ED is provided by medical officers, physicians who have completed medical school and are preparing for post-graduate medical education. Patient care is supervised by faculty trained in either general practice or emergency medicine (EM).
All physicians working in Patan Hospital ED were eligible for inclusion in the study. Physician demographics, including medical position (i.e., medical officer, resident, fellow, faculty), and years of practice, were collected. A pre-training assessment of experience and confidence with ultrasound and specifically LUS was obtained.
Physicians performed LUS studies on patients presenting to the ED with dyspnea.
Lung ultrasound training
A total of eight hours of training was provided for ED physicians that consented for participation in the study. The first four hours consisted of a 1-h didactic session on LUS followed by hands-on practice with a human model. This was then followed by four hours of one-on-one proctored scanning in the ED performing LUS on dyspneic patients to ensure learners were comfortable with acquiring images, saving images for submission, and interpreting images according to the BLUE protocol [2].
Study protocol
Ultrasounds were performed using a SonoSite M-Turbo (Fujifilm SonoSite, Inc.) machine and a curvilinear probe. The ultrasound protocol included ten views of the lungs: two anterior, two lateral, and one posterior view on each hemithorax [2].
Following the training, each physician performed scans independently, recorded images in all zones labeled by location, and reported findings in each zone, including an overall interpretation based on the BLUE protocol [2, 16]. Physicians assessed for lung sliding, A-lines, B-lines, consolidations, and/or pleural effusions [2, 17]. A lines were defined as recurrent horizontal echogenic artifacts arising from the pleural line generated by sub-pleural air [2]. B-lines were defined as discrete vertical hyperechoic artifacts arising from the pleural line, extending to the bottom of the ultrasound screen, erasing A lines, and moving with lung sliding [2]. Consolidation was defined as sub-pleural hypoechoic or tissue-like area with B-lines at the far-field border [2].
Physician trainees recorded their interpretation of each zone and overall interpretation on a standardized data collection form immediately following each scan. These examinations were exported from the ultrasound machine and uploaded for review of quality and interpretation by one of two expert sonographers with registered diagnostic medical sonographer certification and > 1000 previously performed ultrasounds. Expert sonographers were blinded to clinical data and physician interpretations of each zone and overall interpretation. After every five scans, physicians received feedback on the quality and interpretation of their ultrasounds. Image quality was graded on a five-point scale (1—Very Poor, 2—Poor, 3—Average, 4—Good, 5—Excellent). Trainees were expected to submit a minimum of five abnormal scans, including both B-lines and consolidation, to ensure proficiency with both normal and abnormal findings [18].
Statistical analysis
Cumulative sum (Cusum) statistical methodology was used to evaluate the number of LUS scans required to reach an adequate level of training [19, 20]. As described in more detail by Russell et al., Cusum analysis uses pre-defined acceptable and unacceptable failure rates and evaluates sequential data to determine when a learner has reached proficiency with a skill [12, 19, 20]. This statistical method assesses procedural competence for learners over time. Two outcomes were considered: whether physicians were at least 70% correct; and whether they had “Yes” for correct interpretation. This predetermined threshold has been used previously in the literature to determine learner competency [12, 21]. Assuming an accepted failure probability was 0.3 and smallest detectable failure probability was 0.7, type 1 error and type 2 error (alpha and beta, respectively) were both set at 0.1.
Interobserver agreement for ultrasound interpretations between the study physician and expert sonographer was calculated using Cohen’s Kappa coefficient. A random 10% of examinations was overread by a second expert sonographer to assess interobserver agreement between experts.