Real-world implementation of video-assisted thoracic surgery for lung cancer: a decade-long regional propensity score-weighted cohort study

This is the authors' abstract. We don't add key points for this paper.

Background: The perioperative advantages of video-assisted thoracic surgery (VATS) over thoracotomy are established, but the real-world introduction of a minimally invasive program is shaped by patient selection, technical maturation, conversion and secular change in care. We evaluated the decade-long implementation of VATS within a regional thoracic surgery service and estimated its association with early postoperative outcomes.Methods: This is a retrospective cohort study of anatomical lung resections performed over the first decade [2009–2018] of a VATS implementation program at two tertiary hospitals covered by a single thoracic surgery department. The primary exposure was the approach with which the operation was started under an intention-to-treat analysis strategy, and the outcomes were 90-day mortality, a composite of major morbidity and perioperative mortality, and 30-day readmission. Propensity score weighting used preoperative covariates, including calendar year and hospital. Covariate balance before and after weighting was assessed by absolute standardized mean differences. The average treatment effect (ATE), reported as risk difference with 95% confidence interval (CI), was the primary estimand, with average treatment on the treated (ATT) and average treatment effect on the overlap population (ATO) as sensitivity estimands. An exploratory analysis comparing conversion with upfront thoracotomy was carried out under the same prespecification.Results: A total of 964 out of 1,404 anatomical lung resections met the inclusion criteria. The series included 472 operations begun thoracoscopically, of which 97 were converted (20%), versus 492 upfront thoracotomies. Completed VATS rose from 16.7% of anatomical resections in 2009 to 53.7% in 2018, while conversion annual rates fell from 50.0% to 6.5%. Converted procedures had the highest crude rate of every adverse outcome. Under intention to treat, an intended VATS approach was not significantly associated with 90-day mortality (ATE −1.9 percentage points; 95% CI: −4.7 to +0.9; P=0.17), major morbidity and mortality (ATE −4.6 points; 95% CI: −9.4 to +0.2; P=0.06) or readmission (ATE −2.0 points; 95% CI: −5.9 to +2.0; P=0.33); for the composite outcome, the ATT (P=0.04) and ATO (P=0.047) estimands reached significance. In the as-treated analysis, all three outcomes were significant in favor of completed VATS. No statistically significant differences were detected between conversions and upfront thoracotomies in the exploratory analysis.Conclusions: Across a decade-long regional implementation trajectory, adoption of VATS rose and conversion fell. When conversions are analyzed in the arm in which the operation was begun, the association between an intended thoracoscopic approach and better perioperative outcomes is attenuated and no longer significant. These findings do not question the perioperative advantages of thoracoscopic lobectomy, but the benefit attributable to adopting a thoracoscopic strategy at service level is smaller and less certain than that observed among the operations it completes.