Clinical Data

Clinical data are information generated from the clinical use of a medical intervention, including data from formal clinical investigations, post market surveillance, published scientific literature, adverse event reports, and routine clinical practice. In the regulatory context, clinical data serve as evidence of a medical device's or drug's safety and performance when used as intended. The exact scope depends on the setting: in EU medical device law (Regulation (EU) 2017/745, the Medical Device Regulation or MDR, Article 2(48)), clinical data means information concerning safety or performance that is generated from the use of a device and comes from clinical investigations of the device, published studies of the device or of a device shown to be equivalent, and post market surveillance, in particular post market clinical follow up.

Clinical data form the evidentiary basis for regulatory decisions about the approval, labeling, and continued market availability of medical products. They are also essential for clinical practice guidelines, health technology assessments, and reimbursement decisions. For researchers, generating high quality clinical data through well designed studies is the primary pathway to demonstrating that an intervention works and is safe. For systematic reviewers, clinical data from multiple studies are synthesized to provide comprehensive assessments of an intervention's benefits and risks. The quality, completeness, and integrity of clinical data directly affect the reliability of all downstream decisions made from that evidence.

Clinical data are generated through several pathways. Clinical investigations (including randomized controlled trials, non randomized studies, and single arm studies) produce prospective data under controlled conditions. These data are often considered the highest quality because the study protocol specifies data collection procedures, outcome definitions, and quality assurance measures.

Post market surveillance generates clinical data from the real world use of approved products. Sources include adverse event reporting systems, patient registries, electronic health records, and post market clinical follow up studies. This data captures the performance of interventions in broader and more diverse populations than those studied in pre market trials.

Published literature provides clinical data that can be compiled through systematic reviews and meta analyses. Literature based clinical evaluations are common in medical device regulation. Under the EU MDR, for example, manufacturers must show through a clinical evaluation that the clinical evidence supports the safety and performance claims for their devices; published data on another device count only if equivalence to that device is demonstrated, and implantable and class III devices generally need a clinical investigation of the device itself.

Clinical data management involves standardized processes for collecting, cleaning, validating, and storing data to ensure their integrity and usability. Good Clinical Practice (GCP), set out in the ICH E6 guideline (currently E6(R3), adopted in January 2025), and the Clinical Data Interchange Standards Consortium (CDISC) standards provide frameworks for data management in clinical research. CDISC formats are a regulatory requirement in some settings: the US FDA requires them for the study data in new drug, abbreviated new drug and certain biologics license applications, for studies that started after 17 December 2016. Data quality dimensions include completeness, accuracy, consistency, timeliness, and traceability.

Clinical data are relevant whenever evidence about the safety, efficacy, or performance of a medical intervention is needed. This includes the development and regulatory approval of new drugs, biologics, and medical devices; post market monitoring and surveillance; clinical guideline development; health technology assessment; and systematic reviews of interventions. The type of clinical data required depends on the regulatory pathway, the stage of product development, and the specific questions being addressed.

A common error is assuming that clinical data from pre market trials fully represent the intervention's performance in routine clinical practice. Pre market trials often enroll selected populations under controlled conditions, and their results may not generalize to broader populations with comorbidities, co medications, and variable adherence patterns. Post market data are essential for capturing the real world performance.

Another mistake is treating all clinical data as equivalent in quality without assessing the methods by which they were generated. Data from well conducted randomized trials carry different evidential weight than data from uncontrolled case series or spontaneous adverse event reports. Systematic reviewers and regulators should assess the quality and risk of bias of each data source. Finally, failing to comply with GCP, including its data integrity requirements, or with required data standards such as CDISC formats can render clinical data unusable for regulatory purposes, even if the underlying study was well designed.

A medical device manufacturer preparing a regulatory submission for a new cardiac implant compiles clinical data from three sources: a prospective multicenter clinical investigation involving 450 patients followed for 24 months, a systematic review of published literature on similar devices (identifying 18 relevant studies), and post market surveillance data from 3,200 patients who received a predecessor device over five years. The clinical investigation shows a device survival rate of 96.5% at two years, and the combined clinical evidence shows an adverse event profile consistent with the intended use, supporting the submission for regulatory approval. The predecessor device data and the literature on similar devices count only as far as the manufacturer can show that they apply to the new implant (under the EU MDR, by demonstrating equivalence), so the investigation of the new device itself carries the most weight.