We appreciate the comments from Drs. Peñalver and Cabrera about our systematic review of rituximab in ITP. Our inclusion criteria for primary reports in the review had to be fairly liberal because most reports were case series or cohort studies, reflecting, perhaps surprisingly, the paucity of methodologically rigorous trials in this field. Nevertheless, we established a priori that care must be taken to avoid redundant publications, because many were published in abstract form only and some were follow-up reports of preliminary studies.

The report by Peñalver and colleagues gave us reason to pause (1). In fact, the authors were kind enough to share their raw data with us for further contemplation. True, their report of 89 patients with ITP treated with rituximab was the largest of any published series, and indeed, patient follow-up was complete. However, their report described the results of a questionnaire sent to 43 different centers in Spain. Our principal reservation about the paper was that we could not be sure that the data contributed by the patients had not been reported elsewhere (that is, their originality). Moreover, survey data are particularly prone to recall and reporting bias, and measures to safeguard against these methodological pitfalls were not described in the paper. As Drs. Peñalver and Cabrera surmise, the inclusion of their report in our systematic review does not significantly change the estimated overall response to rituxumab (62.0% vs. 62.5%), but it does tighten somewhat the 95% CIs (53.0% to 71.0% vs. 52.6% to 72.5%).

The principles of any systematic review require careful scrutiny of primary reports so that the results are comprehensible and free of bias. This is especially true of systematic reviews of observational data (2). We were strict about our inclusion criteria, even if it was at the expense of a report as large as that of Peñalver and colleagues.

We agree that rituximab is a promising therapy for patients with ITP, and its use is becoming widespread in countries where distribution of the drug is unregulated. However, without properly controlled trials, we are putting patients' safety at risk for a benefit that is as of yet undetermined.