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


The differences among studies included in the meta-analysis were many and fundamental. These differences included those in event definition, in determination of preventability of events, in methods of identifying and judging drug-event relationships and severity of events, and in patients and hospitals and wards sampled. In addition, we found that the correct numerators, ie, the numbers of patients with definite or probable ADRs, needed for incidence calculations were often unavailable in the source studies. The derived numerators that were used for incidence calculations could lead to results that may overestimate the incidence of ADRs. The meta-analysts stated unequivocally that possible and preventable ADRs were excluded. However, we found in many instances that it was impossible to discern the strength of relationship of event to drug or to determine preventability. Including these events in the numerator clearly leads to an overestimation of the incidence of WHO-defined ADRs. Furthermore, most studies did not report fatalities, and the numbers of events for studies with fatal events were very small. The use of meta-analysis methods without considering problems posed by small numbers may have overstated the fatal ADR estimates.

The answers to the questions originally posed in our review of the ADR meta-analysis are as follows: (1) Were ADRs defined with sufficient precision and consistency across studies to permit pooling of study results? No. (2) Were preventable ADRs reported separately from all adverse events in each study? No. (3) Were the correct numbers (ie, number of patients with probable/definite ADRs) used for computing ADR incidence rates? No. (4) In meta-analyses, were appropriate adjustments made for differences in studies, patients, and outcomes? No.