Background: Evidence supporting similar recurrence rates between video-assisted and open surgery for the treatment of recurrent pneumothorax is questionable, because the number of randomised trials is sparse and they are underpowered to detect any meaningful difference. Our aim was to do a systematic review of randomised and non-randomised studies to compare recurrence rates between the two forms of surgical access.
Methods: We did a systematic literature search for studies on pneumothorax surgery in Medline, Embase, Cochrane Library, trial registers on the internet, and conference abstracts, and identified 29 studies (four randomised and 25 non-randomised) eligible for inclusion. Meta-analysis was done by combining the results of reported recurrence rates in patients undergoing video-assisted thoracoscopic surgery compared with those having open surgery. Both fixed and random effects models were applied to the results pooled for analysis.
Results: In studies that did the same pleurodesis through two different forms of access, the relative risk (RR) of recurrences in patients undergoing video-assisted surgery compared with open surgery was similar between non-randomised and randomised studies (RR 4.880 [95% CI 2.670-8.922] vs 3.951 [0.858-18.193]), yielding an overall RR of 4.731 (2.699-8.291; p<0.0001). There was no evidence to suggest heterogeneity of trial results (p=0.88). The high RR of recurrence for video-assisted surgery remained robust to a random effects model (4.051 [1.996-7.465]; p<0.0001), by including all comparative studies (3.991 [2.584-6.164]; p<0.0001), with only high-quality studies used (4.016 [1.8468.736]; p<0.0001), and on a simulation biased in favour of video-assisted surgery when there were no events in either group (3.559 [2.165-5.852]; p<0.0001).
Interpretation: Both randomised and non-randomised trials are consistent in recurrence of pneumothoraces and show a four-fold increase when a similar pleurodesis procedure is done with a video-assisted approach compared with an open approach.