Real-world data for clinical trials: what it is, and how CROs use it
Real-world data is information about patients’ health and care collected during routine care, not inside a trial. It can tell you how many patients actually fit a protocol, where they are treated and who treats them. The question for most CROs and sponsors is not whether to use it, but whether to license raw data or get the answer.
- Real-world data is information collected during routine care, not inside a trial.
- Real-world evidence is what you learn from it: counts, locations, investigators.
- In trial planning it answers three questions: how many patients fit, where they are treated, and who treats them.
- It comes from health records, claims, pharmacy and lab data, and registries.
- Patient-level data counts real patients; aggregate data gives you totals and averages.
- Four kinds of provider offer it. They differ in who does the analysis and what you get back.
- You can license the data and analyze it yourself, or get the finished answer. Most mid-size teams want the answer.
The basicsWhat real-world data is
The raw record of routine care.
Information about patients’ health and care collected outside a trial: health records, billing records, pharmacy and lab data, and registries.
What the data tells you.
The answer you get by analyzing it: how many patients meet a protocol, where they are treated, and which investigators treat them.
Regulators draw the same line: real-world data is what is collected during routine care, and real-world evidence is the clinical insight that comes from analyzing it. For trial planning, the evidence is what you need; the data is how you get there.
The sourcesThe main types of real-world data
Real-world data comes from the records that routine care already produces. Each source answers a different part of a trial planning question.
| Source | What it records | What it helps answer |
|---|---|---|
| Electronic health records (EHR) | Diagnoses, procedures, test results and clinical notes from care visits | Whether a patient meets clinical criteria, such as a diagnosis or a lab value |
| Claims (billing) data | What was billed for visits, procedures and prescriptions, across providers | A patient’s treatment history over time, including care at other institutions |
| Pharmacy data | Prescriptions written and filled | Prior and current medicines, which many exclusion criteria depend on |
| Lab data | Test results and their dates | The thresholds written into inclusion criteria |
| Disease registries | Patients with one condition, followed by a disease group or a center | Natural history and small populations, often in rare disease |
No single source holds everything. Patient-level data that brings these records together for the same de-identified patient is what makes it possible to check a protocol’s criteria one by one.
In practiceHow CROs and sponsors use it
- Protocol feasibility. How many patients actually meet the inclusion and exclusion criteria, and which criteria remove the most. See clinical trial feasibility.
- Site selection. Which institutions treat those patients today, so the site list rests on counts rather than reputation. See site selection on real patient data.
- Investigator selection. Which investigators treat the population and have enrolled trials like this one.
- The bid defense. Evidence a sponsor can check, in the meeting where the award is decided. See bid defense.
- Rare disease. Where a handful of qualifying patients are scattered across the country. See rare disease site selection.
The detailPatient-level or aggregate?
Not all real-world data answers the same question. Aggregate data gives totals and averages: how many people have a diagnosis in a region. Patient-level data follows each de-identified patient across the institutions that treat them, so you can count how many actually meet a protocol, not just how many have the condition. For a feasibility or site decision, that difference is usually the whole answer. More in what patient-level data is.
The choiceLicense the data, or get the answer
There are two ways to put real-world data to work on a trial. Both use real-world data. They differ in who does the work, and in what lands on your desk.
| A data license | Finished answers | |
|---|---|---|
| What you receive | Access to data, and often a tool to query it | Counted patients, a ranked site list and the reasoning, for your protocol |
| Who does the analysis | Your own analysts | The provider’s team |
| What you need in-house | A data team that knows the dataset | Someone to read the result and use it |
| Time to a first answer | Depends on your team’s queue | In days |
| What a sponsor sees | Whatever your team builds from it | Slides in your own template, every line explained |
| Best fit | Large organizations with data science teams | Mid-size CROs and sponsors that need the answer, not the infrastructure |
Neither is wrong. A large organization with a data science group may want the raw data. A mid-size CRO answering an RFP next week usually wants the answer, and the reasoning to defend it.
The marketWho provides real-world data for clinical trials
A search for a real-world data vendor turns up four kinds of provider. All of them start from real-world data. They differ in what you get back and who does the analysis.
| Provider | What you get | Who does the analysis | Best fit |
|---|---|---|---|
| Data licensors | Access to datasets, often with a query tool | Your analysts | Organizations with a data science team |
| Analytics platforms | Software that runs on top of the data | Your team, in the provider’s tool | Teams that want to run their own queries |
| In-house feasibility | Surveys, trial databases and the team’s own experience | Your feasibility team | Common indications with plenty of trial history |
| Finished-answer partners | Counted patients, ranked sites and named investigators for each protocol | The partner’s team | Mid-size CROs and sponsors with an RFP, bid defense or protocol deadline |
Quantum Research Cloud is a finished-answer partner. Some teams combine types: a license for steady internal work, and finished answers when a bid or a protocol decision is due.
The limitsWhat real-world data cannot tell you
Real-world data is strong evidence, not a complete picture. A good provider states the gaps plainly.
- Not every criterion is recorded. Routine care does not capture everything a protocol asks for, such as some performance scores or a patient’s willingness to join a trial.
- Coverage differs by country. Data outside the US is often thinner, so ask what each country in your plan can support.
- Patients are not the same as capacity. The data shows which institutions treat qualifying patients, not whether a site has the staff, equipment or start-up speed to run your trial.
- Recent months can be incomplete. Records take time to arrive, so the newest data is usually the least complete.
That is why every number in a good answer comes with its reasoning, so a sponsor can see what it rests on.
Before you chooseWhat to ask any provider
- Is the data patient-level, or totals and averages?
- Does it follow patients across every institution they visit, or only some?
- Can you count patients against our protocol’s criteria, not just the diagnosis?
- Do we get data to analyze, or the finished answer?
- Can every number be explained to a sponsor?
The full list is in how to choose a real-world data partner.
The termsReal-world data terms, defined
- Real-world data (RWD). Information about patients’ health and care collected during routine care rather than in a trial. Also written real world data.
- Real-world evidence (RWE). The findings that come from analyzing real-world data.
- Electronic health record (EHR). The record a hospital or clinic keeps of a patient’s care.
- Claims data. Billing records of the care a patient received, from every provider that billed for it.
- Patient-level data. Data kept as one de-identified record per patient, so patients can be counted against a protocol’s criteria.
- Aggregate data. Totals and averages, such as how many people in a region have a diagnosis.
- Longitudinal data. Data that follows the same patient over time, so treatment history can be checked.
- De-identified data. Data with names and other identifiers removed, so no one can be recognized.
- Qualifying patients. Patients who meet a protocol’s inclusion and exclusion criteria, not just its diagnosis.
Real-world data, answered
What is real-world data in clinical trials?
Real-world data is information about patients’ health and care that is collected during routine care rather than inside a clinical trial, such as health records, billing records, pharmacy and lab data, and registries. In trial planning it is used to see how many patients actually fit a protocol, where they are treated and which investigators treat them.
What is the difference between real-world data and real-world evidence?
Real-world data is the raw information collected during routine care. Real-world evidence is what you learn by analyzing it, for example how many patients meet a protocol’s criteria or which sites treat them. Data is the input; evidence is the answer.
How is real-world data used for clinical trial site selection?
It shows which institutions actually treat patients who meet a protocol’s criteria, and which investigators have run trials like it. A site list built that way rests on counted patients rather than reputation or questionnaire answers.
Is real-world data de-identified?
Yes. For trial planning, patient-level real-world data is de-identified: each record carries an anonymized ID instead of a name, so care can be followed across institutions without revealing who anyone is. Planning works with counts, never with individual identities.
Do we need our own data team to use real-world data?
Not if you buy answers instead of a data license. A license gives your analysts raw data to work through. With finished answers, the analysis is done for you and you receive the counts, the ranked sites and the reasoning, ready to use.
What are examples of real-world data?
The most common examples are electronic health records, insurance claims, pharmacy records, lab results and disease registries. Each is collected during routine care rather than inside a clinical trial, and each answers a different part of trial planning, from whether a patient meets a lab threshold to which treatments they have already had.
What types of real-world data vendors are there?
There are four main types: data licensors that provide datasets for your analysts, analytics platforms that provide software on top of the data, in-house feasibility teams working from surveys and trial databases, and finished-answer partners such as Quantum Research Cloud, whose team delivers counted patients, ranked sites and named investigators for each protocol.
What are the limitations of real-world data?
Routine care does not record every protocol criterion, coverage differs by country, the most recent months are usually incomplete, and the data shows which institutions treat qualifying patients rather than whether a site has the capacity to run a trial. A good provider states these gaps and explains every number.
How is Quantum Research Cloud different from a real-world data vendor?
Most real-world data vendors license data and tools that your own team has to analyze. Quantum Research Cloud delivers the finished result for each protocol: counted qualifying patients, a ranked and explained site shortlist and the investigators behind it, built by our team in days and in your own template.