Summary of results*
| Median sensitivity (IQR) | Median specificity (IQR)¶ | Pooled sensitivity(95% CI)† | Pooled specificity(95% CI)† ¶ | Likelihood ratio (95% CI)‡ ¶ | ||
|---|---|---|---|---|---|---|
| Positive | Negative | |||||
| IQR, interquartile range. | ||||||
| *For studies that did not use real-time radiological guidance for needle placement. | ||||||
| †Pooled sensitivity and specificity were calculated using a random effects model. | ||||||
| ‡Positive and negative likelihood ratios were calculated from the pooled sensitivity and specificity. | ||||||
| §p value for the comparison between tier 1 and non-tier 1 studies. | ||||||
| ¶Median and pooled specificities for non-tier 1 studies were not calculated because only two studies allowed calculation of specificity (both 100%);33,37 however, both of these studies had fewer than 10 patients without mediastinal metastases. We were thus unable to calculate likelihood ratios for non-tier 1 studies. | ||||||
| Tier 1 studies | 0.36 (0.32–0.38) | 0.98 (0.96–1.00) | 0.39 (0.17 to 0.61) | 0.99 (0.96 to 1.00) | 29.0 (21.9 to 38.4) | 0.62 (0.53 to 0.72) |
| Non-tier 1 studies | 0.82 (0.79–0.84) | – | 0.78 (0.71 to 0.84) | – | – | – |
| p value§ | 0.001 | – | 0.009 | – | – | – |