Fourteen studies (n more than 507,899) including three state, two regional and three level one pre-test post-test studies, and one national, three state and two regional comparisons of different interventions (trauma systems versus non-trauma systems or centres), were included in the review. Six studies (number of patients unclear) including two state, one regional and two level one pre-test post-test studies, and one state comparison of different interventions, were included in the meta-analysis.
The authors did not report the individual quality scores for those studies not included in the meta-analysis. However, quality scores for the six studies included in the meta-analysis ranged from 18 (poor) to 33 (excellent). Overall, one study was rated as poor quality, two were average, two were good, and one was excellent. The graph of quality score versus OR suggested that the higher the study quality the more likely the studies were to approach the line of no effect (i.e. OR=1); the poorest quality study showed the greatest increase in survival rate.
Overall findings (14 studies).
Survival odds were in favour of trauma systems in eight of the fourteen included studies, as compared with normal hospital care. However, the odds of survival were worse for trauma system groups in three comparative studies. There were no significant differences between trauma systems and normal hospital care in the three remaining studies.
Meta-analysis (6 studies).
Two studies showed statistically significant survival ORs in favour of trauma systems, while a further also showed favourable survival ORs but the findings were non significant. Overall, the pooled survival OR was 0.93 (95% CI: 0.87, 1.00) when using a fixed-effect model and 0.88 (95% CI: 0.78, 1.00) when using a random-effects model. Significant statistical heterogeneity was reported (Q=12.286, d.f.=5, p=0.031). The quality-adjusted OR was 0.85 in favour of greater survival with the use of trauma systems.