Sixteen RCTs (934 patients), 6 with a crossover design and 10 with parallel design, were included. The study size ranged from 12 to 287 patients.
The randomisation process was described in detail in 2 studies. Two RCTs were double blinded. All studies clearly stated inclusion and exclusion criteria, accounted for withdrawals and missing data, and used an intention-to-treat analysis.
There was a statistically significant lower incidence of FN with CSF prophylaxis compared with control (12 trials; OR 0.59, 95% CI: 0.43, 0.81, p=0.001). There was no heterogeneity (I-squared 0%). This benefit was evident both for leukaemia and high-grade lymphoma (OR 0.62, 95% CI: 0.43, 0.90, p=0.012) and for solid tumours (OR 0.51, 95% CI: 0.28, 0.94, p=0.029). There was no statistically significant difference in effect estimates between cancer types, study design, or type of CSF. CSF prophylaxis did not reduce the incidence of DI compared with the control (9 trials; OR 0.75, 95% CI: 0.52, 1.08, p=0.12). There was moderate statistical heterogeneity (I-squared 41%). The duration of neutropenia was decreased by prophylactic CSF (13 trials) by a mean of 3.40 days (95% CI: 1.85, 4.96, p<0.0001). Compared with controls, there was a decrease with CSF of 1.7 days (95% CI: 0.9, 2.5, p<0.001) in the mean length of hospitalisation and as well as a decrease of 2.0 days (95% CI: 0.35, 3.6, p=0.017) of antibiotic use (10 trials). Eight studies did not report on side-effects. Four studies reported no side-effects. Visual examination of the funnel plot suggested no publication bias.