What is the key question?

• Is conservative management an acceptable alternative to interventional management for uncomplicated, moderate-to-large primary spontaneous pneumothorax (PSP)?

What is the bottom line?

• Investigators designed an open-label, multicenter, non-inferiority trial whereby they recruited 316 patients 14 to 50 years of age with a first-known, unilateral, moderate-to-large PSP and randomly assigned them to immediate interventional management of the pneumothorax (intervention group) or a conservative observational approach (conservative-management group) and followed them for 12 months. The primary outcome was lung re-expansion within 8 weeks.
• Investigators found that re-expansion within 8 weeks occurred in 129 of 131 patients (98.5%) with interventional management and in 118 of 125 (94.4%) with conservative management (p= 0.02 for non-inferiority).
• It is worth noting that, in the conservative-management group, 25 patients (15.4%) underwent interventions to manage the pneumothorax. In a sensitivity analysis in which all missing data after 56 days were accounted for as treatment failure (with re-expansion in 129 of 138 patients [93.5%] in the intervention group and in 118 of 143 [82.5%] in the conservative-management group), the risk difference changed from -9 % to −11.0% and was considered now to be outside of the pre-specified non-inferiority margin.
• Conservative management resulted in a lower risk of serious adverse events or pneumothorax recurrence than interventional management.
• Limitations of the study included lack of blindness of the outcome assessors that could have introduced some bias. Another major limitation reported by the authors is that the original statistical analysis plan did not specify the window for the 8-week visit nor did it define how missing radiographic data were to be handled for the primary outcome.

Why read on?

• Overall, this trial provides some modest evidence that conservative management of PSP was non-inferior to interventional management, with a lower risk of serious adverse events but the primary outcome was not statistically robust to conservative assumptions about missing data.