Impact of simulation training on performance and outcomes of endobronchial ultrasound-guided transbronchial needle aspiration performed by trainees in a tertiary academic hospital

Background: Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) has a high diagnostic yield and low complication rate. Whilst it has been included in international guidelines for the diagnosis and staging of lung cancer, current results are mostly based on EBUS experts performing EBUS-TBNA in centres of excellence. The impact of simulation training on diagnostic yield, complications, scope damage and repair cost in a real-world teaching hospital is unclear. Methods: A review of our hospital EBUS-TBNA registry from August 2008 to December 2016 was performed. A positive diagnosis was defined as a confirmed histological or microbiological diagnosis based on EBUS sampling. Complications were classified as major or minor according to the British Thoracic Society guidelines. In addition, we assessed the cost of repairs for scope damage before and after simulation training was implemented. Using CUSUM analysis, the learning curves of individual trainees and the institution were plotted. Results: There were 608 EBUS-TBNA procedures included in the study. The number of procedures performed by trainees who underwent conventional training was 331 and those who underwent simulation training performed 277 procedures. Diagnostic yield for trainees without simulation training was 88.2% vs. 84.5% for trainees with simulation training (P=0.179). There was no statistical difference in the diagnostic yield between the groups of trainees (OR: 0.781, 95% CI: 0.418-1.460, P=0.438) after adjusting for risk factors. There was an increase in overall complications from 13.6% to 16.6% (OR: 2.247, 95% CI: 1.297-3.891, P=0.004) after introduction of the simulation training, but a trend to decrease in major complications 3.6% to 0.7% (P=0.112). The cost for scope repairs for the trainees without simulation training was SGD 413.88 per procedure vs. SGD 182.79 per procedure for the trainees with simulation training, with the mean difference being SGD 231.09 per procedure (95% CI: 178.40-640.60, P=0.268). CUSUM analysis showed an increasing learning curve for the trainees with simulation training after an initial competency period. Conclusions: There was no statistical difference in diagnostic yield from EBUS-TBNA and cost of scope damage after simulation training was introduced into our training program. Interestingly, there was an increase in minor complications. CUSUM analysis can provide additional information on institutional learning curves. The value of simulation training in EBUS-TBNA remains uncertain.