Ten RCTs (n=1,688) were included in the review. Sample sizes were small (range 28 to 427). Attrition ranged from 11% to 50%. Allocation concealment was reported adequate in six trials. Intention to treat was reported in two trials. Sensitivity analyses were conducted using two trials (n=433).
Neurodevelopment outcomes: Using BSID-II, MDI was significantly higher in infants fed with supplemented formula (WMD 3.44, 95% CI 0.56 to 6.31, p=0.02; five trials, n=879). Using BSID-I, PDI was significantly decreased in infants fed with supplemented formula (WMD -7.99, 95% CI -14 to -1.99, p=0.009; two trials, n=87). There were no other statistically significant differences between the study groups.
Sensitivity analyses revealed no differences in MDI or PDI, although results were restricted by small sample size. Moderate heterogeneity was reported (X2=14.5 to 20.72, p<0.02).
Diseases associated with prematurity: There were no statistically significant differences in all analyses of infants fed with supplemented formula compared with control for sepsis, necrotising enterocolitis, retinopathy of prematurity, intraventricular haemorrhage or bronchopulmonary dysplasia. Sensitivity analysis did not alter this result.