How to Plan and Run a Real-World Evidence Study, and How to Choose a CRO

AUTHOR
Dragan | Co-Founder & CTO
A real-world evidence (RWE) study answers a question about how a medicine or a medical device performs in routine care. It uses real-world data (RWD): data from everyday practice, such as health records, registries and what patients report themselves, rather than only data from a randomized controlled trial. What RWE and RWD are, and why they matter, is covered in our overview of RWE in clinical trials. This guide is the practical part: how to plan and run an RWE study, and what to ask a CRO before you sign.
An RWE Study in Six Steps
- Start with the decision. Name who will use the evidence, and for what.
- Choose the design. Collect new data, use existing data, or combine both.
- Check the regulatory route before the protocol is final.
- Make the data fit for purpose: relevant to the question and reliable.
- Keep sites and patients in the study, with as little manual work as possible.
- Follow the data while the study runs, not only at the end.
1. Start With the Decision, Not With the Data
Every RWE study supports a decision. Typical ones are a regulatory submission or a label change, a post-authorization safety study, a benefit assessment for pricing and reimbursement, post-market clinical follow-up (PMCF) for a medical device, or a medical affairs question such as how a treatment is used in practice.
The decision sets how strong the evidence must be. A study meant for a regulator should have a protocol, a statistical analysis plan and traceable data from day one. A study for internal planning can be lighter. Write the research question in one sentence, with the population, the treatment or device, the outcome and the time frame. If it does not fit in one sentence, the study is not ready to plan.
2. Choose the Design: New Data, Existing Data, or Both
| Approach | What it means | Fits when |
|---|---|---|
| Prospective non-interventional study | Patients are treated as in routine care; sites and patients record the data during the study. | You need outcomes that existing sources do not record, such as quality of life or other patient-reported outcomes. |
| Study of existing data | Health records, claims data or registries are analyzed after the fact. | The data already exists in sufficient quality, and speed matters. |
| Hybrid study | Existing data plus new collection, for example health records plus ePRO questionnaires. | Existing sources answer part of the question, but not all of it. |
| Pragmatic trial | Patients are randomized, but treated in routine care settings. | You need a randomized comparison under real conditions. Check early which rules apply: randomizing patients goes beyond routine care. |
A prospective study takes longer, but you decide what is collected. A study of existing data is faster, but limited to what was recorded, and that is often less than the question needs: the dose, the reason a treatment was stopped, or how the patient feels.
3. Check the Regulatory Route Before the Protocol Is Final
Whether a study counts as non-interventional decides which rules apply. For studies of medicines in the EU, a non-interventional study is a clinical study that is not a clinical trial (Article 2(2)(4) of Regulation (EU) No 536/2014). Check early whether your design still counts as non-interventional. If it does not, a study of a medicine may be a clinical trial, with its own approval process. Beyond that, check these points early:
- Post-authorization safety studies (PASS): for a non-interventional PASS imposed by the authorities, the marketing authorization holder submits a draft protocol to EMA’s Pharmacovigilance Risk Assessment Committee (PRAC) before the study starts, or, for a study that one member state requested and that runs only in that country, to its national authority (in Germany: Section 63g of the German Medicines Act, AMG). Imposed studies are registered in the HMA-EMA Catalogue of real-world data studies. In Germany, non-interventional PASS that the marketing authorization holder conducts on its own initiative are notified to the competent federal authority (Section 63f AMG). EMA recommends submitting their protocols and study reports in the same way as for imposed studies, but does not require it.
- Germany: studies that collect findings on the use of approved or registered medicines (Anwendungsbeobachtungen) must be notified without delay to the competent federal authority (BfArM or PEI), the National Association of Statutory Health Insurance Physicians (KBV), the National Association of Statutory Health Insurance Funds and the Association of Private Health Insurance (Section 67(6) AMG). The notification includes place, time, objective and observation plan, and, to the KBV and the National Association of Statutory Health Insurance Funds, the participating physicians; for physicians in statutory health insurance care also the payments made to them. Payments must not create an incentive to prescribe or recommend a particular medicine. Non-interventional post-authorization safety studies follow similar rules of their own (Section 63f(3) and (4) AMG): notification to the same three associations, and no incentive to prescribe or recommend a particular medicine.
- Medical devices: the EU MDR sets its own rules for post-market clinical follow-up (PMCF) studies. Check them separately from the rules for medicines.
- Data protection: processing personal data for research needs a legal basis under Article 6 GDPR. Health data is a special category of personal data, so its processing also needs an exception under Article 9 GDPR, such as the patient’s explicit consent or a research provision in Union or national law. In Germany, Section 27 of the German Federal Data Protection Act (BDSG) allows processing for research without consent only if the controller’s interests substantially outweigh the patient’s interest in excluding the processing; the burden of proof lies with the controller.
This is an overview, not legal advice. Have your regulatory affairs team or the CRO confirm the route for each country before the protocol is final.
4. Make the Data Fit for Purpose
Two qualities matter most: is the data relevant to the research question, and is it reliable? The US FDA’s guidance on health records and claims data (final since July 2024) counts accuracy, completeness and traceability as reliability, and the availability of the key study variables and enough representative patients as relevance. In practice, check these points before the study starts:
- Does the source record the exposure, the outcome and the factors you need to adjust for?
- How complete is it, and how will you handle missing data? Write it into the statistical analysis plan.
- Do all sites and countries use the same definitions? Common data models and standard terminologies keep data comparable.
- Can every value be traced back to its source, with an audit trail of who changed what and when?
- Is the system that collects the data validated for its intended use?
5. Keep Sites and Patients in the Study
In most real-world and late-phase studies, data still reaches the eCRF by hand: typed in from paper, PDFs, health records or other clinical documents. That costs sites time, and sites in observational studies are paid less than in interventional ones. The less manual work a study asks for, the more likely sites are to keep recruiting and documenting.
- Data from hospital records: Autonomous Data Capture takes data from EHR integrations, PDF uploads, photos or voice, the AI fills the eCRF fields, and a person at the site reviews them before import.
- Data from patients: ePRO questionnaires on any device, with automated reminders and no app to install.
- Consent: eConsent where the study design and the country allow it.
6. Follow the Data While the Study Runs
An RWE study that is only analyzed at the end finds its problems too late. Track recruitment, questionnaire compliance and missing values per site from the first patient, so you can act while there is still time. Real-time dashboards show recruitment, compliance and selected parameters across visits and sites. Keep these views descriptive, and set out in the statistical analysis plan which interim analyses are planned.
How to Choose a CRO for an RWE Study
Many sponsors run RWE studies with a CRO. These questions help you compare CROs:
- Experience: Which RWE and non-interventional studies has the team run, in which countries and indications? Ask for examples close to your study.
- Regulatory knowledge: Does the team know the rules in your countries, such as the notification of Anwendungsbeobachtungen in Germany?
- Design advice: Which design would they choose for your question, and why? A good partner questions the design, not only the budget.
- System: Which system collects the data? Is it validated, with a complete audit trail, and does it support FDA 21 CFR Part 11 where US submissions are planned?
- Hosting and data protection: Where is the data hosted, and who processes it? For EU studies, ask for EU hosting and a data processing agreement.
- Data from sites: How does data from hospital records reach the study: by hand, through interfaces, or by document upload?
- Patients: Can patients take part from their own device, with reminders and in their language?
- Speed: How long from the final protocol to the first patient?
- Insight during the study: What can you see while the study runs, and how often?
- Deliverables: Which data exports, documentation and reports do you get at the end?
Some sponsors run RWE studies with their own team on a study platform instead, or split the work: the CRO runs the sites, the sponsor keeps the platform. Ask early which model the CRO supports. Not sure which CRO fits your study? Ask us: we work with several CROs and can suggest partners that fit your study.
Common Mistakes
- Choosing the data source before the question is clear.
- Collecting every variable that might be useful. Each extra field costs site time and lowers completeness.
- No plan for missing data and for patients who drop out.
- Underestimating the work for sites, especially when data is typed in from health records.
- Checking the national notification and data protection rules too late.
How Climedo Supports RWE Studies
Climedo runs surveys, direct-to-patient studies with eCOA and ePRO, and non-interventional studies such as post-approval, observational and registry studies on one platform, with EDC, source data verification and site management. CROs can run studies for several sponsors on it, with reusable templates and workflows. Climedo is aligned with ICH GCP (E6), GAMP 5, EU GMP Annex 11 and FDA 21 CFR Part 11. More on the RWE platform for sponsors and CROs.
Planning an RWE study? Book a demo.



