Table 1.

Existing registry—historical assessment.

Requirements	Recommendations	Suggested practices
Registry data must demonstrate relevancy and robustness to support regulatory decision-making	Data are relevant  Data are adequate in scope and content  Data are generalizable: Registry reflects high site and patient participation rates compared with total population Data are robust—acceptable for use in one or more of the following  Validated risk prediction  Quality assurance  Performance improvement  Benchmarking  Informing practice guidelines  Post-market surveillance  Generating peer-reviewed publications  Comparative effectiveness research	Evaluate if data generated by an existing registry are adequate for evaluating clinical outcomes or supporting regulatory decision-making Assess whether data and evidence that are generated can address the question at hand (i.e., fit for purpose) Connectivity: Establish whether there are linkages or the ability to link to other existing datasets for additional data not captured directly in the registry Data should be suitable for adequate statistical analysis Data should be interpretable, i.e., evidence derived from analysis of de-identified aggregate data should be sufficient to allow for regulatory decision-making
Registry data must reliably be able to support regulatory decision-making	Design: The registry should be designed to capture reliable data from real-world practice (no protocol-driven treatment)	A standard operating procedure document should exist that defines the processes and procedures for data capture and management The system should have a basic validation package to assure that the software acts as intended
	Patient population: The patient population should be limited to those with specific diseases, conditions, or treatment exposure(s)	The patient population for the registry is associated with a specific disease, condition, family of procedures (e.g., orthopedic surgery), or treatment exposure(s) Inclusion and exclusion criteria should be clearly defined (e.g., total population or population subset)
	Data collection forms: The data collection forms should be standardized	The existing data elements should be fixed and predefined There should be an audit trail for any changes The forms should use standard and uniform data definitions
	Datasets: Data elements should be able to be mapped to industry standards to allow for more direct comparison of data analyses	Documentation should be available that describes the data elements and datasets
	Timing of endpoints/outcomes: The timepoints of each endpoint/outcome in the data collection form should be documented	Evaluate the ability to calculate timing of treatment and treatment outcome (e.g., stroke at discharge or at 30 days post index procedure)
	Timing of data collection: Data collection/entry can occur at any time	The system should be live 24/7 and web-based
	Data completeness and accuracy: Data should be complete, accurate, and attributable	Missing data should be minimized and statistically assessed Assure processes are in place for data collection and entry with documented training The system should allow identification of the data originator (e.g., person[s] performing procedure[s]), data source (e.g., point of care, EHR, procedural record), and data entry person Data logic checks should be included at the time of data entry Processes should be in place to assure accuracy of the data
Registry has assurance of patient protections	Documentation of informed consent or IRB waiver of informed consent is needed for access to the data (e.g., by investigators, patients, regulators) Patient privacy must be assured: Assess for use of de-identified data vs. line-item data	Access to the data needs to be supported by patient informed consent or IRB waiver of informed consent Use a single IRB of record where possible with a broad-use informed consent document Data encryption and security protections should be in place Control/ownership of proprietary data should be addressed If informed consent is waived by the IRB, additional patient protections must be in place to share line-item patient level data (i.e., HIPAA consent or waiver for research)

EHR electronic health record, IRB investigational review board, vs versus