Eleven studies (n=344 patients) were included in the review. There was one randomised controlled trial (n=28); the other trials were controlled pre-test/post-test designs. Mean quality score was 24.65 (standard deviation 1.29, range 23 to 27 out of 32).
Statistically significant differences in favour of comprehensive EIBI were reported for full scale IQ (WMD 11.98, 95% CI 6.73 to 17.23; 10 studies), non-verbal IQ (WMD 11.09, 95% CI 1.88 to 20.30; five studies), expressive language (WMD 15.21, 95% CI 8.43 to 21.99; five studies), receptive language (WMD 13.94, 95% CI 10.14 to 17.75; five studies), adaptive behaviour (WMD 5.92, 95% CI 2.72 to 9.13; seven studies). Subscales of adaptive behaviour were communication (WMD 10.44, 95% CI 5.03 to 15.85; seven studies), daily living skills (WMD 5.49, 95% CI 3.36 to 7.61; seven studies) and socialisation (WMD 4.96, 95% CI 0.18 to 9.75; seven studies). The sole RCT (n=28) showed statistically significant differences in favour of EIBI for full scale IQ and non-verbal IQ. Results for all other outcome measures were not statistically significant.
Statistical heterogeneity was reported for IQ, communication and daily living skills domains of the Vineland Adaptive Behaviour Scale and expressive language.