Recommendation	Rationale
Develop reproducible analytical workflows designed to support the pharmacovigilance signal management use cases, including when appropriate analytics for precomputed results; when possible, evaluate them against accepted benchmarks.	Pharmacovigilance operates under short timelines which require efficient analysis turnaround. Reproducible analytical workflows may generate reliable evidence in the required time frames. Analytics should be accompanied with objective diagnostics to verify that assumptions and limitations inherent to methods are evaluated and determined to be fit‐for‐use for the specific application. N.B.: Reproducible analytical workflows are not ‘one‐size‐fits‐all’, but rather consistent processes that, given a set of analytical design choices, will produce a defined set of analysis results. The analytical design choices, such as target, comparator, outcome, time‐at‐risk, outcome model, represent opportunities for customization within reproducible analytical workflows.
Develop, validate, and share phenotypes for important pharmacovigilance outcomes and covariates, including different levels of specificity and/or severity	Reliable RWE requires validated phenotypes for outcomes and covariates. Timely results require that these preexist for a wide range of medical events, and that there is capacity to rapidly develop and validate new phenotypes when needed.
Develop and maintain mappings between clinical concepts in MedDRA and standard vocabularies for RWD, for important adverse events	Pharmacovigilance signal management processes rely primarily on information encoded in MedDRA. Mappings from clinical concepts in MedDRA to phenotypes in the standard vocabularies for RWD (and vice versa) must be readily available and align on the clinical ideas that they seek to represent to enable rapid transitions between analyses in different types of data as well as evidence synthesis. Assumptions/limitations in developing such mappings should be acknowledged and transparently communicated.
To the extent possible, harmonize common data models (CDMs) for RWD	Efficiencies can be gained if we work with existing communities to evolve current standards rather than independently developing solutions de novo.