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

Data Extraction

One copy of each eligible study was then blinded as to journal source, authors, and authors' affiliations, by using previously published methods.[3,4] Studies were extracted in duplicate, one reviewer working with blinded studies, and the other with unblinded studies.

Data elements extracted included study and patient characteristics and ADR reports. Data elements are listed in Table 1. Special emphasis was given to capturing the relationship of events to drugs, because according to the meta-analysis authors, only patients with events that were definitely or probably related to a drug were used for incidence calculations. In addition, when studies did not use the WHO ADR definition, special attention was given to details of events that were not consistent with the WHO definition, such as mechanism of effect (toxic, idiosyncratic, allergic) and stated cause of events. Therapeutic failures and preventable events, such as ADRs caused by errors in administration, noncompliance, intentional overdose, and drug abuse, are not encompassed by the WHO definition and were to be excluded from the meta-analysis. Serious events were defined as those that require hospitalization, prolong hospitalization, are permanently disabling, or result in death. All preadmission ADRs were, therefore, by definition, serious events. The number of patients with events, as opposed to the number of events, was to be used as the numerator for incidence calculations.

After data consensus by the 2 extractors for each paper, data were entered to an Excel spreadsheet. All computerized data were verified back to the consensus data extraction forms.