Background and aim Clinical decision rules (CDRs) have sought to identify the few children with chemotherapy-induced febrile neutropenia (FN) really at risk of severe infection to reduce the invasive procedures and costs for those at low risk. Several reports have shown that most rules do not perform well enough to be clinically useful. Our objective was to analyse the derivation methods and validation procedures of these CDRs.

Methods A systematic review using Medline, Ovid, Refdoc, and the Cochrane Library through December 2012 searched for all CDRs predicting the risk of severe infection and/or complications in children with chemotherapy-induced FN. Their methodological quality was analysed by 17 criteria for deriving and validating a CDR identified in the literature. The criteria published by the Evidence Based Medicine Working Group were applied to the published validations of each CDR to assess their level of evidence.

Results The systematic research identified 612 articles and retained 12 that derived CDRs. Overall the CDRs met a median of 65% of the methodological criteria. The criteria met least often were that the rule made clinical sense, or described the course of action, or that the variables and the CDR were reproducible. Only one CDR, developed in South America, met all methodological criteria and provided the highest level of evidence; unfortunately it was not reproducible in Europe.

Conclusion Only one CDR developed for children with FN met all methodological standards and reached the highest level of evidence.