Eleven RCTs were included in the review (n=5,214). Samples sizes ranged from 78 to 1,432 participants. Follow up ranged from six to 29 months. The mean Jadad score was four, indicating high quality. In general, the reporting of data was poor; differences between groups were small, and were below the minimally clinically important difference specified for each measure.
Six trials (n=3,784) using the SF-36 measure (including three with complete data on all domains) provided sufficient data for inclusion in the meta-analysis.
Statistically significant changes in favour of targeting high haemoglobin levels were noted in terms of the impact of treatment on: physical function, WMD 2.91 (95% CI: 1.29, 4.53); general health, WMD 2.71 (95% CI: 1.26, 4.15); vitality, WMD 3.17 (95% CI: 1.89, 4.44); and mental health, WMD 0.44 (95% CI: 0.06, 0.83).
No significant differences were found for the domains of physical role, pain, social function and emotional role.
Significant heterogeneity was reported in analyses for: physical role (I2 = 68.8%, p=0.02); social function (I2 = 72.6%, p=0.01); and mental health (I2 = 70.5%, p=0.02). Smaller differences between study groups were noted in sensitivity analyses where all domains of the SF-36 were included, or where no difference was assumed. These analyses favoured targeting the low/intermediate target level in more domains.
The authors reported that none of the results were clinically meaningful.