Background: To contextualize adverse event (AE) rates observed in product-exposed populations, expected background rates from similar, drug-naïve populations (i.e., same underlying disease) are critical. Deriving estimates of background incidence rates (IRs)—e.g., from published literature or ad-hoc database studies—can be resource-intensive and challenging to complete within regulatory timelines. For this reason, pre-developing a repository for background IRs of anticipated AEs in diseased populations within a given therapeutic area is expected to expedite AE rate contextualization for safety signal assessments.
Objectives: Establish a repository of background IRs of AEs in populations with various underlying diseases underpinned by shared disease processes.
Methods: The key steps to establishing the repository were as follows: (1) Select underlying diseases of interest. We examined the indications for marketed and pipeline products. We selected a subset of diseases, with preference for those representing indications for multiple products. (2) Select outcomes of interest. We compiled a list of AEs included as potential or identified risks in the summary of product characteristics, risk management plans, and investigator brochures. We selected a subset of AEs, with preference based on potential frequency and severity. (3) Derive incidence rates. We conducted comprehensive literature reviews for each disease-outcome pair. An epidemiologist extracted and summarized IRs. (4) Perform quality control checks. A second epidemiologist reviewed for data entry errors, evidence completeness, and extreme outliers before finalizing the repository entry.
Results: We have piloted this framework in the immunology and inflammation therapeutic area. The general population and six inflammatory diseases of interest were selected: atopic dermatitis, asthma, chronic obstructive pulmonary disease, Crohn’s disease, ulcerative colitis, and hidradenitis suppurativa. Various AEs were selected, including but not limited to all-cause mortality, cardiovascular events, and infections. Comprehensive literature reviews and quality control checks were conducted for 63 disease-outcome pairs. Ranges of IRs were summarized overall and stratified to reflect variation across age groups, sex, and geographic regions as needed. Literature reviews will be updated every 6 months.
Conclusions: During a 6-month period, our team has established a repository for background IRs of multiple diseases that could occur as AEs in various underlying populations of interest. Our framework can be adapted for any therapeutic area to provide epidemiological evidence to support safety signal assessments.
