Adverse Drug Reactions in Hospitalized Patients: A Critique of a Meta-analysis

Marion Kvasz, MD, MPH; I. Elaine Allen, PhD; Matthew J. Gordon, BA; Eric Y. Ro, BA; Rhonda Estok, RN; Ingram Olkin, PhD; and Susan D. Ross, MD, FRCPC

In This Article

Statistical Methods

We intended to perform a meta-analysis of ADR incidence rates in a manner similar to that of the original meta-analysis. Although details of the statistical methods in the original meta-analysis were sparse, we planned to first calculate, for each source study, the incidence of "all-severities," serious and fatal, ADRs by dividing the total number of hospital patients experiencing at least 1 ADR (according to the WHO definition) by the total number of hospital patients monitored. Then, we planned to use a random effects model, as did the original authors, to combine ADR incidences first within the set of preadmission studies and the set of postadmission studies, and then to combine the 2 sets of studies for a meta-analysis of overall ADR incidence. Finally, a complete replication would include an extrapolation of ADR incidence rates from the meta-analysis to all hospitalized patients in the United States, using hospital admission statistics published in 1994. We also planned an examination of sources of bias, heterogeneity, inferences based on small numbers, and multiple comparison testing.