Thirty five studies (n=2,732) were included in the review. All the studies fulfilled four to five out of 14 QUADAS criteria.
The pooled estimate of sensitivity for endoscopic ultrasound in diagnosing nodal involvement by rectal cancers was 73.2% (95% CI 70.6 to 75.6). The pooled estimate of specificity was 75.8% (95% CI 73.5 to 78.0).
The pooled estimate for positive likelihood ratio was 2.84 (95% CI 2.16 to 3.72), and for the negative likelihood ratio was 0.42 (95% CI 0.33 to 0.52).
The pooled estimate for diagnostic odds ratio was 7.87 (95% CI 5.31 to 11.66).
There was no evidence of statistically significant between-study heterogeneity or publication bias. Estimates were similar for fixed-effect and random-effects models. Pooled estimates of diagnostic accuracy measures did not change significantly over time.