A non-interventional study (NIS) – also referred to as a non-interventional trial or observational study – is a clinical investigation in which medicinal products or medical devices are used under routine conditions of medical care, without the treatment algorithm being specified by a study protocol. The key difference from an interventional clinical trial is that diagnostic and therapeutic decisions lie exclusively with the treating physician and are made independently of study participation.
Regulatory distinction from a clinical trial
The distinction between interventional and non-interventional studies is of regulatory significance. In the EU, Directive 2001/20/EC, replaced by Regulation 536/2014, defines a clinical trial as interventional if it involves an intervention in patient care. An NIS does not fall under this framework and requires neither authorisation by authorities such as BfArM or PEI nor formal protocol development in accordance with GCP standards. Nevertheless, NIS are not unregulated: they must comply with national requirements (in Germany, notification pursuant to AMG Section 67(6)), undergo ethical review, and are subject to data protection provisions, in particular the GDPR. The sponsor is responsible for ensuring that the non-interventional nature is credibly documented and maintained throughout the study.
Types of non-interventional studies
NIS can be divided into different types according to study design and objective, each with different methodological and regulatory requirements:
- Post-Authorisation Safety Studies (PASS): mandated by authorities or voluntary; used to characterise the risks of authorised medicinal products in routine use.
- Post-Authorisation Efficacy Studies (PAES): assess effectiveness under real-world conditions, often as an obligation following conditional approval.
- Non-interventional studies (AWB): typical in Germany; observe the use and tolerability of an already authorised medicinal product without intervening in therapy.
- Registry and cohort studies: longitudinal observation of patient groups, often over long periods and based on routine data.
Each type has specific requirements for data protection, consent processes, and statistical analysis. Particularly for PASS studies that are mandated by authorities, the protocol, timeline, and submission to PRAC (Pharmacovigilance Risk Assessment Committee) are bindingly defined.
Data sources and methodology
Non-interventional studies use various data sources that differ fundamentally from controlled clinical trials. Commonly used sources include electronic patient records (EPRs), health insurance data, registry data, spontaneous reporting databases, or prospectively collected patient data via eCRF. The methodological quality of an NIS depends heavily on the completeness and consistency of these data sources. Confounding, selection bias, and information bias are typical challenges that must be addressed using appropriate statistical methods such as propensity score matching, instrumental variables, or sensitivity analyses. The study protocol for an NIS should therefore include clear definitions of exposure, outcomes, and confounders in advance.
Importance in the post-market lifecycle
After market authorisation of a medicinal product or medical device, NIS become significantly more important. Real-world evidence from NIS complements the controlled data from Phase III studies and provides insights into safety, effectiveness, and usage patterns in actual patient care. For medical devices under EU MDR 2017/745, post-market clinical follow-up (PMCF) activities are mandatory and are often designed as NIS. For medicinal products, PASS studies are a core component of the risk management plan and ongoing benefit–risk assessment. NIS data can also feed into health technology assessment (HTA) procedures and influence reimbursement decisions by payers.
From a methodological perspective, quality assurance for non-interventional studies places particular demands. As no randomisation takes place, biases due to patient selection or treatment preferences are difficult to control. Prospective NIS with a clearly defined protocol, a pre-registered study design (e.g., in EudraCT or ClinicalTrials.gov), and independent statistical analysis therefore have significantly greater regulatory acceptance than retrospective analyses without prior registration. For sponsors, this means: even if an NIS does not require GCP-mandated documentation, investing in quality assurance is worthwhile—because only methodologically robust data will have an impact in authorisation procedures, PASS assessments, or HTA procedures. Increasingly, sponsors also use real-world data (RWD) platforms that aggregate data from electronic patient records, health insurance claims, or patient registries to conduct NIS more efficiently and cost-effectively. Regulatory acceptance of such secondary data sources is increasing, but it requires a robust data governance concept that transparently describes data protection, data quality, and the representativeness of the source and can be reviewed by independent bodies. This offers sponsors the opportunity to generate cost-efficient evidence—provided that data quality and methodological transparency meet the regulatory requirements for real-world evidence studies, as increasingly specified by the EMA and HTA bodies.
Frequently Asked Questions
Does an NIS require approval by an ethics committee?
In Germany, consultation with an ethics committee is envisaged for most NIS, even though there is no formal approval requirement as there is for clinical trials. An ethics opinion is generally recommended for PASS studies with a regulatory obligation or for studies involving sensitive patient data. GDPR requirements apply without restriction, regardless of the study’s interventional status.
What distinguishes an AWB from a clinical trial?
In a non-interventional study (AWB), the authorised medicinal product is used exclusively in accordance with the approved indication and dosage, and the therapy decision is made solely by the physician. There is no randomisation, no control group, and no standardised protocol in the sense of GCP. By contrast, a clinical trial actively intervenes in the course of treatment—whether through randomisation, blinding, or additional study-related measures.
What reporting obligations apply to NIS in Germany?
Pursuant to AMG Section 67(6), non-interventional studies conducted with authorised medicinal products must be notified to BfArM or PEI (depending on the product type) as well as to the competent state authorities. In addition, serious adverse events must be recorded in accordance with the sponsor’s requirements and reported in line with national pharmacovigilance requirements.