(234) Features of Case Reports That Support Putative Causal Relationships Between Medicinal Products and Suspected Adverse Drug Reactions. Preliminary Results from a Scoping Review

Publication Number: 1279

Background: Bradford Hill’s viewpoints, as he called them, also called guidelines, can be used to help determine whether an adverse event can form the basis of a pharmacovigilance signal. Beside a few applications of the viewpoints, little has been done to study which features of reports most often support such assessments.

Objectives: To characterise the features of reports of adverse events used to support signals of suspected adverse drug reactions (ADRs).

Methods: We retrieved electronic records (PubMed, EMBASE, Web of Science, PsycINFO) and grey literature records that described findings as signals of ADRs or signals of disproportionate reporting, without time or language restrictions; when necessary, we contacted regulatory agencies and authors to obtain other records or clarifications. We included previously undocumented signals and excluded records that did not explicitly describe findings as signals. We also charted the features of reports of suspected ADRs that authors advanced as supportive of signals and when possible coded them to mirror the Bradford Hill viewpoints, omitting biological plausibility and strength of association. One author performed title/abstract screening, eligibility assessment, and data charting; a second author independently cross-validated the findings. We analysed the data descriptively.

Results: We screened the titles/abstracts of 9525 electronic records and the full texts of 1509. We also reviewed the full texts of 2250 entries from websites/cited references/original authors, and included 1728 in the review. In all, we screened 11,775 unique records and included 2132. Of those, 925 concerned clinical reviews of reports of ADRs (either alone or with other types of evidence); 134 of these mentioned at least one feature and concerned 228 distinct signals: 89 presented one feature, 79 two, 48 three, and 12 > three. We recorded 440 instances of relevant features; the most frequent was positive dechallenge/positive rechallenge (217 occurrences), followed by temporality (131) and exclusion of competing causes/only one suspect drug in a report (53). Other signals depended on available information to ascertain the suspected ADRs (14), the report’s consistency (12), and biological gradient (6).

Conclusions: In this scoping review positive dechallenge/rechallenge, temporality, and exclusion of competing causes were the most frequent factors supporting signals.