Introduction Exercise training before lung resection for non-small cell lung cancer is believed to decrease postoperative complications (POC) by improving cardiorespiratory fitness. However, this intervention lacks a strong evidence base.

Aim To assess the effectiveness of preoperative exercise training compared with usual care on POC and other secondary outcomes in patients with scheduled lung resection.

Methods A systematic search of randomised trials was conducted by two authors. Meta-analysis was performed, and the effect of exercise training was estimated by risk ratios (RR) and mean differences, with their CIs. Clinical usefulness was estimated according to minimal important difference values (MID).

Results Fourteen studies involving 791 participants were included. Compared with usual care, exercise training reduced overall POC (10 studies, 617 participants, RR 0.58, 95% CI 0.45 to 0.75) and clinically relevant POC (4 studies, 302 participants, Clavien-Dindo score ≥2 RR 0.42, 95% CI 0.25 to 0.69). The estimate of the effect of exercise training on mortality was very imprecise (6 studies, 456 participants, RR 0.66, 95% CI 0.20 to 2.22). The main risks of bias were a lack of participant blinding and selective reporting. Exercise training appeared to improve exercise capacity, pulmonary function and also quality of life and depression, although the clinical usefulness of the changes was unclear. The quality of the evidence was graded for each outcome.

Conclusion Preoperative exercise training leads to a worthwhile reduction in postoperative complications. These estimates were both accurate and large enough to make recommendations for clinical practice.