A meta-analysis of the impact of bronchial stump coverage on the risk of bronchopleural fistula after pneumonectomy.

The occurrence of bronchopleural fistula (BPF) after pneumonectomy is associated with high morbidity and mortality. The incidence of BPF in historical patients not subjected to bronchial stump coverage (BSC) was between 6 and 12% after pneumonectomy for lung cancer surgery or benign disease. BSC is considered an important prophylactic measure against BPF and is widely used, but its efficacy remains unknown. Our aim was to systematically review the literature, in order to quantify BPF risk in patients receiving or not receiving BSC with any tissue after pneumonectomy. We performed a systematic review in PubMed, for papers published between 1999 and 2012, analysing series of patients treated with pneumonectomy and including both patients receiving coverage and patients not receiving coverage. Both randomized and non-randomized series were eligible. Proportion of failures (i.e. BPF) was analyzed separately in the two groups (patients receiving BSC and patients not receiving BSC). For each study and for the overall series, 95% confidence interval (CI) (without continuity correction) of the observed proportion was calculated. Overall, 21 series were eligible, with 3879 patients (1774 receiving BSC and 2105 not receiving coverage). The decision to perform or not the BSC was randomized only in one small trial, limited to diabetic patients, showing a significant reduction of BPF in favour of coverage. In the 20 remaining studies, baseline risk of BPF in the group of patients receiving BSC and in the group of patients who did not receive coverage was different. In patients receiving coverage, the proportion of BPF was 6.3% (95% CI: 5.3-7.5%). In patients not receiving coverage, the proportion of BPF was 4.0% (95% CI: 3.2-4.9%). In recently published series, the vast majority of patients considered at high risk for BPF received BSC. This common practice hinders an unbiased estimate of the efficacy of BSC in reducing BPF risk. Results of this meta-analysis show that, despite a clear negative selection, the incidence of BPF in patients considered at high risk and receiving coverage was only slightly higher compared with patients considered at low risk and not covered. A randomized trial would help answer the question.