Risk Factors and Prediction Model for Spontaneous Pneumothorax Recurrence: A Retrospective Study of 440 Patients

Background Spontaneous pneumothorax (SP) recurrence imposes substantial clinical burdens, yet long-term risk stratification remains poorly defined. Methods This retrospective cohort study analyzed 440 SP patients treated at Longyan First Hospital (2010–2020). Multivariable logistic regression and Cox proportional hazards models identified recurrence predictors, with high-resolution computed tomography quantification of bullae characteristics (diameter/number). Model performance was evaluated using receiver operating characteristic curves. Results The 5-year recurrence rate was 31.82% (140/440), with 75.71% occurring within 1 year. Key independent risk factors: Secondary spontaneous pneumothorax (SSP vs. primary spontaneous pneumothorax[PSP]: adjusted odds ratio [aOR], 2.91; 95% confidence interval [CI], 1.47–5.76); Smoking ≥ 5 pack-years (aOR, 3.28; P  < 0.001); Bullae > 5 cm (adjusted hazard ratio [aHR], 1.77; 95% CI, 0.48–6.56) or > 5 in number (aHR, 7.01; 95% CI, 1.81–27.19); Surgical intervention (video-assisted thoracoscopic surgery [VATS] bullectomy with pleurodesis) reduced recurrence risk by 88% (aOR, 0.13; 95% CI, 0.04–0.42) versus conservative management. The prediction model integrating bullae morphology demonstrated significantly improved discrimination (area under the curve (AUC) = 0.832 vs. 0.797; ΔAUC = + 0.035, P  = 0.001). Conclusions SSP, smoking, and bullae burden (> 5 lesions or > 5 cm diameter) are potent predictors of SP recurrence. VATS bullectomy with pleurodesis significantly mitigates risk. Quantification of bullae morphology enhances prediction accuracy, supporting its integration into risk-stratified clinical protocols to guide surgical decision-making.