*3.2. Traditional Meta-analyses*

A few meta-analyses are available on the topic [28–32]. A traditional meta-analysis of 15 reports of surgical resection found that treatment success was achieved in 84% (95% confidence interval (CI), 78%–89%) of patients, noting substantial heterogeneity among the studies [29].

Two other meta-analyses of MDR-TB patients who had either resection or non-resection surgery found that surgical patients had better outcomes than those who did not [28,30]; however, there was no distinction between the different forms of resection surgery.

In the Marrone's meta-analysis [28], 24 studies identified a significant association between surgery and successful treatment compared to non-surgical interventions (OR 2.24, 95%CI 1.68–2.97). The meta-analysis from 23 single-arm studies demonstrated that, respectively, 92% (95%CI 88.1–95) and 87% (95% CI 83–91) of surgical patients achieved successful short and long-term outcomes. In the sub-group analysis (studies reporting both surgical and non-surgical treatment outcomes) favourable surgical outcomes (treatment success) were associated with increased drug-resistance, i.e., better results for XDR-TB patients than for MDR-TB ones.

Confounding by indication (a form of bias that occurs when the patients most likely to benefit are selected for therapy) was a major limitation in each meta-analysis.

Furthermore, antibiotic regimens were not standardized across studies, meaning that the studies could not account for factors such as the individual drug regimens or the timing of surgery in relation to culture conversion.
