Thirty studies (n=9,047) were included. Of these, 11 studies (n=2,819; 6 level IIb evidence, 5 level IIIb) investigated the detection of organ injury and 19 (n=6,492; 14 level IIb evidence, 5 level IIIb) investigated the detection of free intraperitoneal fluid.
Organ injury.
Q was estimated as 0.91 (95% CI: 0.76, 1.07) using inverse variance weights and 0.90 (95% CI:0.75, 1.05) using robust regression. The NPV ranged from 0.72 to 0.99.
There was significant heterogeneity for the LRs so a random-effects model was used for pooling. The PLR and NLR were 21.99 (95% CI: 15.08, 32.06) and 0.23 (95% CI: 0.18, 0.28), respectively. The related post-test probabilities were also presented in the review.
Free intraperitoneal fluid.
Q was estimated as 0.89 (95% CI: 0.73, 1.05) using inverse variance weights and 0.89 (95% CI:0.76, 1.03) using robust regression. The NPV ranged from 0.72 to 0.99
There was significant heterogeneity for the LRs so a random-effects model was used for pooling. The PLR and NLR were 36.28 (95% CI: 28.97, 45.44) and 0.24 (95% CI: 0.20, 0.28), respectively. The related post-test probabilities were also presented in the review.
Subgroup analyses.
For studies involving paediatric patients (4 studies, n=501), there was no distinction between recognition of free fluid and organ injury. Q was estimated as 0.88 (95% CI: 0.33, 1.42) using inverse variance weights and 0.87 (95% CI:0.76, 0.99) using robust regression. For ultrasonography in the presence of haemodynamic instability (3 studies, n=155), Q was estimated as 0.89 (95% CI: 0.13, 1.64) using robust regression. For ultrasonography versus different 'gold' standards, higher Q values were computed comparing ultrasonography with diagnostic peritoneal lavage than with CT. These analyses were not reported well but a significant finding appears to be that, compared with CT, the ability of ultrasonography to exclude lacerations remained below clinically acceptable threshold limits (NLR 0.21, 95% CI: 0.16, 0.29).
Linear regression analysis failed to demonstrate any statistically- significant change in the diagnostic accuracy of emergency ultrasound over the preceding 20 years.