Short-term and long-term outcomes for video-assisted thoracoscopic surgery and open lobectomy: Results from statistical analysis adjusting for treatment selection bias

Abstract

BACKGROUND: Lack of equipoise limits the feasibility of a randomised study to compare short-term and long-term outcomes between video-assisted thoracoscopic surgery (VATS) and open lobectomy. CALGB observational data from a multi-institutional setting is available for the comparison. Because of the potential bias associated with treatment selection, standard statistical methods can yield biased estimates for treatment effect. We aim to investigate several statistical approaches, including multivariable regression modelling and propensity-score methods, which can be used to adjust for selection bias in the estimation of treatment effect. METHODS: 519 patients with stage I and II non-small-cell lung cancer, whose tumours had been collected as part of CALGB lung cancer tissue bank, were eligible to participate. Multivariable regression modelling and propensity score based methods (eg, matching and inverse probability weighting) were used to compare VATS and open lobectomy in terms of the length of post-operative hospital stay and overall survival. Propensity score for treatment selection was estimated via logistic model with age, race, sex, performance status, comorbidities, histology, tumour stage, and size as covariates. FINDINGS: Consistent with the primary findings from Nwogu et al. (2015), all statistical analyses indicate that VATS is associated with significantly shorter length of post-operative hospital stay. Univariate analysis shows VATS has longer overall survival than open lobectomy, but the survival benefit disappears in the multivariable regression and all propensity-score-based methods investigated. INTERPRETATION: Several statistical methods yield consistent results. This multi-institutional study supports the assertion that VATS lobectomy results in shorter hospital length of stay compared with open lobectomy for early stage non-small-cell lung cancer. Survival was similar between the two groups.