O-014ASSESSING TEAM PERFORMANCE IN LUNG RESECTION: TIME SERIES BETTER THAN CROSS-SECTIONAL ANALYSIS

• Published in: Interactive cardiovascular and thoracic surgery 2014-06, Vol.18 (suppl_1), p.S4-S4
• Main Authors: Varela, Gonzalo, Rodríguez, M., Gómez-Hernández, M.T., Jiménez, M.F., Novoa, N., Aranda, J.L.
• Format: Article
• Language: English
• ISSN: 1569-9293; 1569-9285
• DOI: 10.1093/icvts/ivu167.14

Objectives: The performance in lung resection is usually evaluated according to the rates of crude or risk-adjusted rates of adverse events in cross-sectional analysis. The objective of this study is to report the graphical assessment of the team's performance using CUSUM curves constructed on time-series. Methods: The occurrence of the three most prevalent postoperative complications (nosocomial pneumonia, atrial fibrillation, and prolonged air leak) and 30-day mortality was retrospectively reviewed on a series of 1501 lung cancer patients who underwent curative resection. All complications were prospectively recorded according to previously defined criteria. Overall prevalence of each complication was calculated first. Then, predictive models for each complication and mortality were constructed using clinical and functional parameters (age, body mass index, type of resection and spirometry values). The individual probability of each adverse event was calculated and the performance (C-index) of the models was calculated in ROC curves. Individual risk values were used to construct 95% risk-adjusted CUSUM curves. Results: The series consisted in 277 pneumonectomies, 104 bilobectomies, 1075 lobectomies and 45 anatomical segmentectomies. Overall prevalence of each complication was: pneumonia 5.1%, atrial fibrillation 6.9%, air leak 12.1%, 30-day mortality 3.1%. C-indexes of the predictive models were: 0.75 for 30-day mortality; 0.69 for atrial fibrillation; 0.71 for pneumonia and 0.65 for prolonged air leak. Changes in performance can be seen along the period of the study. After an initial bad performance in pneumonia and mortality rates, implemented changes in lung resection process were followed by considerable improvements in quality. Conclusions: The analysis of the team's performance on time-series using CUSUM graphs gives a better idea of the quality of surgery and allows for better monitoring of the effects of implementing improvements in the process of patients' care. Disclosure: No significant relationships.