Real-world evidence, explained: what a study of medical records can show that a trial can't, and the other way around
Scroll any health feed this week and you'll hit the phrase "real-world data." A drugmaker presents it at a liver conference. A clinic posts a chart review. Someone in the comments says a trial "doesn't reflect real life." So which is it: the gold standard's messy cousin, or the thing trials miss? FDA has written a fair amount on this, and its answer is more interesting than either camp's.
Facts are as of November 7, 2026. Sources, with retrieval dates, are at the bottom.
Before you read on
- FDA defines real-world data as health information collected routinely, such as electronic health records, insurance claims and registries, and real-world evidence as what you learn by analyzing it. [V]
- In a randomized trial, a protocol assigns the treatment. In an observational study, doctors and patients chose it in ordinary care, and researchers analyze what happened. [V]
- FDA has used real-world data for decades to watch for safety problems after approval, and on a more limited basis to support effectiveness. [V]
- FDA's 2018 framework says randomized trials have the stronger scientific footing for showing a drug works, because randomization balances the groups. [V]
The rules today
Congress put real-world evidence into drug law with the 21st Century Cures Act, signed December 13, 2016. It added section 505F to the Food, Drug, and Cosmetic Act, which told FDA to build a program for judging when real-world evidence can help support a new use for an already-approved drug, or meet a study required after approval. [V] FDA published its framework for that program in December 2018. [V] In August 2023 it finalized guidance on how sponsors should run and document observational studies they plan to submit, replacing a December 2021 draft. [V]
That guidance draws the key line in plain terms. An interventional study, or clinical trial, assigns people to treatments according to a protocol. The classic version is the randomized controlled trial, where some participants get the drug and others get a comparator or placebo. [V] A non-interventional study, also called observational, looks at patients who received a marketed drug during routine care without anyone assigning it. [V]
The 2018 framework adds a second split. A retrospective observational study picks its population and treatments from data that already existed before the study began. A prospective one identifies its population at the start and collects data going forward. [V]
What randomization buys you
Here's the core of it. In ordinary care, a doctor's judgment decides who gets which drug, and those decisions aren't random. Sicker patients, healthier patients, people with better insurance or more follow-up visits can end up clustered in one group. FDA's framework says that kind of assignment makes cause and effect hard to pin down, and that randomization exists to prevent it by building groups that are balanced on the risk factors that matter. [V]
“there is a stronger scientific justification for deriving evidence of a drug effect from randomized controlled trials”
FDA doesn't stop there. The same framework notes that some observational studies have reached the same conclusions as randomized trials, and that others got effects that couldn't be reproduced or that differed in size or even direction. [V] Statistical tools like propensity weighting help, but they can only balance what's recorded. [U] That's our summary of the general method, not a quote from FDA.
What real-world data does well
None of this makes real-world evidence second-rate. It answers questions trials struggle with.
Safety at scale. Trials enroll hundreds or a few thousand people for a set period. Once a drug is on the market, millions may take it. FDA's program page says the agency has a long history of using real-world data to monitor approved drugs' safety. [V]
Real patients. FDA's framework describes traditional trials as running on research infrastructure separate from routine care, often with restrictive entry rules. [V] Records and registries can include the older, sicker and more complicated patients a trial's entry rules may leave out. [U]
Ideas for the next trial. The framework lists generating hypotheses for randomized trials, checking whether a trial is feasible, and finding patients for rare-disease studies among the ways real-world data helps drug development. [V]
Sometimes, approval. On July 16, 2021, FDA approved a new use for the transplant drug Prograf (tacrolimus) in lung transplant patients, based on an observational study. [V] It used the national Scientific Registry of Transplant Recipients, which covers every U.S. lung transplant, plus federal death records. In that study, FDA said outcomes improved dramatically compared with the well-documented course of transplants with little or no immunosuppression. [V] Randomized trials of Prograf in other organ transplants served as confirmatory evidence. [V] FDA keeps a running list of drug approvals that used real-world evidence. One entry, from November 2024, is a labeling change for the acetylcysteine product Acetadote that leaned on a retrospective chart-review study, in a case where FDA's summary says randomized trials were considered unethical and impractical. [V]
Where this shows up in peptide and GLP-1 news
You'll see both kinds of evidence this month. Novo Nordisk's November 6 liver-conference results for semaglutide were two retrospective studies of medical records, and Novo's own release spelled out the limits. [V] We laid out those numbers and limits in our brief on Novo's real-world MASH data. The randomized ESSENCE trial is what supported the drug's MASH approval. Our semaglutide and the liver explainer covers that trial.
In wellness peptides, the published human record is thinner. Of the human studies on BPC-157 that a 2026 review counted, one is a retrospective chart review from a single clinic, and none had a placebo group. [V] Our BPC-157 human studies page goes through each one. That's an early stage of evidence, the usual first step for any compound, and it's exactly where randomized trials come next.
Testimonials and before-and-after posts are a different thing again. FDA's definition of real-world evidence is the analysis of routinely collected data. [V] A single story isn't that analysis. [U] That's our reading of the definition.
Three common mix-ups
"Real-world evidence is just anecdotes." FDA means structured analysis of records, claims and registries, sometimes covering tens of thousands of patients. [V]
"Observational studies can't count for anything at FDA." They've supported safety decisions for decades, and in some cases new approvals. [V]
"A big real-world study beats a small trial." Size doesn't fix how people got into each group. That's the job randomization does. [V]
How to size up a real-world headline
Four questions cover most of it. Who chose the treatment, a protocol or ordinary care? What's being compared: two drugs, or people who started a drug against people who didn't? What was measured: a blood test, a scan, or a hard outcome like hospitalization? And how long was follow-up? A study that answers all four clearly, and lists its own limits, is doing its job. For trials, our guide to reading a ClinicalTrials.gov listing covers the matching fields, and the evidence ladder shows where each study type sits.
Peptide Policy Watch
Real-world evidence, explained: what a study of medical records can show that a trial can't, and the other way around
Printable checklist
Questions worth bringing to your clinician
Points from this story to talk through with a clinician or pharmacist who knows your history.
- Was the research behind this treatment a randomized trial, an observational study, or both?
- Were the people studied similar to me in age, health and other conditions?
- Did the studies measure the outcome I care about, or a lab marker?
- Is there safety data from people using it outside trials?
Print view Select it, then use your browser’s Print command to print only this box.
Why both kinds keep mattering
Most medicines that start human testing never reach approval, and our piece on why drugs fail in trials shows where they drop out. The trials that get a drug approved are the start of its evidence record. The records, registries and claims that build up afterward fill in the rest, sometimes for decades. Expect to see more of both in the GLP-1 and peptide stories this year.
Sources (retrieved November 7, 2026)
- FDA, "Real-World Evidence" (program page and definitions) [V] — https://www.fda.gov/science-research/science-and-research-special-topics/real-world-evidence
- FDA, "Framework for FDA's Real-World Evidence Program," December 2018 (PDF) [V] — https://www.fda.gov/media/120060/download
- FDA, "Considerations for the Use of Real-World Data and Real-World Evidence To Support Regulatory Decision-Making for Drug and Biological Products," final guidance, August 2023 [V] — https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-real-world-data-and-real-world-evidence-support-regulatory-decision-making-drug
- Federal Register, notice of availability of the August 2023 final guidance (finalizes the December 9, 2021 draft), August 31, 2023 [V] — https://www.federalregister.gov/documents/2023/08/31/2023-18841/considerations-for-the-use-of-real-world-data-and-real-world-evidence-to-support-regulatory
- FDA, "FDA approves new use of transplant drug based on real-world evidence," July 16, 2021 [V] — https://www.fda.gov/drugs/news-events-human-drugs/fda-approves-new-use-transplant-drug-based-real-world-evidence
- FDA, "CDER Drug Approvals that Used Real-World Evidence," content current as of June 3, 2026 [V] — https://www.fda.gov/science-research/real-world-evidence/cder-drug-approvals-used-real-world-evidence
- Novo Nordisk, AASLD 2026 real-world MASH release, PR Newswire, November 6, 2026 [V] — https://www.prnewswire.com/news-releases/novo-real-world-studies-link-wegovy-2-4-mg-to-reductions-in-liver-enzymes-weight-and-blood-sugar-in-adults-with-mash-at-the-liver-meeting-aasld-2026--302932360.html
Cite or link to this page
Peptide Policy Watch staff. “Real-world evidence, explained: what a study of medical records can show that a trial can't, and the other way around.” Peptide Policy Watch. Published November 7, 2026. https://peptidepolicywatch.com/2026/11/07/real-world-evidence-vs-clinical-trials-observational-studies/
Spot a mistake? See how we source and our corrections policy.
Image: AI-generated. Image credits