Abstract
Background: The colony-stimulating factors (CSFs) are widely utilized to prevent neutropenic complications in both adults and children, but randomized controlled trials in the pediatric setting have varied considerably in size, methodology and patient population. A systematic review and meta-analysis was conducted to more definitively assess the impact of prophylactic CSFs on the risk of febrile neutropenia (FN) in pediatric oncology patients. The rates of documented infection (DI) and duration of severe neutropenia (ANC <500), hospitalization and antibiotic use were also evaluated.
Methods: MEDLINE and EMBASE were searched and identified references were hand-searched through July 2004 for randomized controlled trials (RCTs) of prophylactic G-CSF or GM-CSF in pediatric oncology patients. Both parallel and crossover studies were included. Of the 625 articles identified, 16 studies were included. Ten studies evaluated children with acute leukemia or high-grade lymphoma, and 6 evaluated those with solid tumors. Study objectives, quality, and outcomes were extracted by 2 independent reviewers. Weighted summary estimates of relative risks (RR) for FN and DI were calculated by the method of Mantel-Haenszel using a fixed effects model. Mean differences in durations (days) were estimated by the method of Cohen using a random effects model due to significant heterogeneity. An economic analysis based on cost minimization was calculated from the results of this analysis and a recent cost study.
Results: Among the 12 reporting RCTs, FN occurred in 68% of 400 controls and 59% of 404 CSF patients. Estimated RR was 0.88 [.81–.97; P=.01] favoring the CSFs for leukemia and high grade lymphoma studies and 0.71 [.51–.97; P=.03] for solid tumor studies. RR estimates were 0.85 [.78–.94; P<.001) for G-CSF and 0.83 [.47–1.49; P=.54] for GM-CSF. In 9 reporting studies, DI occurred in 25% of controls and 20% of CSF patients for an estimated RR of 0.80 [.61–1.06; P=.12]. The mean decrease in duration of neutrophils <500 in 13 reporting studies was 3.5 days [2.2–4.7; P<0.0001). Mean decreases favoring CSF use were also observed for hospital stay of 1.7 days [0.9–2.5] in 13 reporting RCTs (P<.01) and antibiotic use of 2.0 days [0.4–3.6; P=.02] in 10 RCTs. Baseline estimates based on the economic model provide incremental cost savings with CSF of $3,824 per subject and a cost saving FN risk threshold of 24%.
Conclusions: This analysis reveals that prophylactic CSFs significantly decrease the incidence of FN and the durations of severe neutropenia, hospitalization, and antibiotic use in pediatric patients receiving myelosuppressive chemotherapy. Differences in rates of documented infection were not statistically signficant. The economic analysis supports guidelines for CSF prophylaxis above a risk of 20%.
Author notes
Corresponding author
