Sources
Every source, and what it supports
Two kinds of source sit behind this product. The 29 checks are read from one published episode, with a timestamp on every rule. The legal basis is read from the held frameworks named on the basis page, with the control code for each entry. Nothing else is used, and no line is written from memory and presented as a source.
The recorded source
FDA Watch, The Rise of Real-World Evidence in Regulatory Strategy, Part 3: RWE in Accelerated Approvals. Published 2026-10-20. Read from 19:57 to 39:36. The programme site.
The rule set is read from the published episode, from the point where the panel turns to what the statute requires before approval to the point where it closes on data rights. Each rule carries its timestamp and a short quotation so a reader can check it against the recording.
| Check | What it asks | Timestamp | The sentence it is read from |
|---|---|---|---|
| A1 | The protocol is pre-specified and shared with FDA before the study starts | 29:46 to 30:36 | FDA recommends sharing your protocol and statistical analysis plan before the real-world study is conducted. |
| A2 | The statistical analysis plan is pre-specified, dated and locked before analysis | 30:36 to 31:28 | That is to ensure that the analysis was not cherry-picked to get the best results, but that you really are doing rigorous, prospectively planned, pre-planned science. |
| A3 | The estimand, primary endpoint and time zero are defined in advance | 26:49 to 27:45 | An adequately powered clinically meaningful primary endpoint, a large precise effect, and high quality conduct with minimal missing data. |
| A4 | The analysis populations and eligibility rules are fixed in advance | 30:36 to 31:28 | The analysis was not cherry-picked to get the best results. |
| A5 | Every protocol and SAP version is retained and handed over | 20:55 to 21:53 | Those conditions can be specifically about the protocol, enrollment targets, milestones, and a target completion date. |
| A6 | The confirmatory study is underway, with milestones, not only on paper | 19:57 to 21:53 | Simply just having a protocol on paper, it's not the same thing as having a trial underway. |
| B1 | The data is relevant to the question, the population and current care | 30:36 to 31:28 | The data has to be relevant to the question and reliable. |
| B2 | Data source fit is assessed, with the limits written down | 30:36 to 31:28 | The data must be accurate, complete, and traceable back to the source. |
| B3 | Outcome and exposure definitions are validated, not just coded | 30:36 to 31:28 | That includes validating how outcomes are identified. |
| B4 | Missing data is characterised and handled by a stated method | 30:36 to 31:28 | The data must be accurate, complete, and traceable back to the source. |
| B5 | Every value is traceable from the source to the analysis dataset | 30:36 to 31:28 | Traceable back to the source to allow us to do source data verification or source data validation. |
| B6 | Source data verification is planned, and the records can be retrieved | 36:02 to 36:56 | FDA will go out and inspect these sites to where you collect the data from to make sure that the data that you collect matches up with source documentation. |
| B7 | The data is converted to the standards FDA can read | 30:36 to 31:28 | Converting submitted data to FDA-supported standards like CDISC where necessary. |
| C1 | Sensitivity analyses are pre-specified, each with the assumption it tests | 30:36 to 31:28 | Pre-specifying sensitivity analyses or different ways to look at the data that show your result does not hinge on one assumption alone. |
| C2 | Subgroup analyses are pre-specified, with a multiplicity position | 30:36 to 31:28 | Subgroup analyses, different ways to look at the data. |
| C3 | The result is shown to hold across methods, not to depend on one | 31:28 to 32:20 | The results are consistent when analyzed in different ways. |
| C4 | The plan says in advance what would change the conclusion | 31:28 to 32:20 | The result does not hinge on one assumption alone or one methodological choice alone. |
| D1 | The external control uses the trial's eligibility criteria | 32:20 to 33:15 | A credible external control needs to match the trial eligibility criteria and endpoint definitions. |
| D2 | The endpoint and its ascertainment are the same in both arms | 32:20 to 33:15 | Match the trial eligibility criteria and endpoint definitions. |
| D3 | Confounding is addressed by a pre-specified method, with diagnostics | 32:20 to 33:15 | Make sure the analysis methodology appropriately addresses sources of bias like confounding. |
| D4 | The control is contemporaneous and reflects current care | 32:20 to 33:15 | Contemporaneous data is necessary to make sure you're not comparing against an outdated standard of care. |
| D5 | The sites and countries have a care standard comparable to the reference market | 35:10 to 36:02 | The standard of care has to be similar to what the standard of care is in the U.S. |
| D6 | The single-arm design is justified and the external control is fixed before unblinding | 32:20 to 33:15 | Where a single-arm trial is compared against similar patients drawn from real-world sources. And this is where the bar is highest. |
| E1 | The right to use the data covers the submission, not only the research | 39:36 to 40:34 | Making sure you have the right to the research, you have the right to use the data, not only for the research itself, but also in the submission. |
| E2 | A data use agreement is in place, with its scope written down | 39:36 to 40:34 | All of that is actually usually packaged up in some sort of legal relationship ... accessing with data use agreements. |
| E3 | The source records exist and can be verified, including consent and the protocol | 36:02 to 37:50 | The investigator who did the study no longer had the informed consent forms. They no longer had the protocol for the study. But they did have the patient data. |
| E4 | Privacy and human subject protections are built into the plan | 40:34 to 41:24 | Privacy and human subject protections have to be built into those plans all across the way. |
| E5 | The data partner is obliged to give access, retain records and allow inspection | 39:36 to 40:34 | Do you have the source records? Can they be verified? All of that is actually usually packaged up in some sort of legal relationship. |
| E6 | The traceability chain is papered through every hand | 39:36 to 40:34 | Making sure you have data traceability all the way through the process, understanding where the information originated. |
The held frameworks
Each framework page on the standards site is the source of the control code and title shown here. The one-sentence statement of what each control asks is ours, written for this product. The framework page is the place to check it.
| Market | Framework | Controls read here |
|---|---|---|
| United States | FDA 21 CFR Part 11 | 4 |
| European Union | EU Clinical Trials Regulation (CTR 536/2014) | 3 |
| European Union | GDPR | 3 |
| United Kingdom | UK GDPR (UK General Data Protection Regulation) | 4 |
| Australia | Australia NHMRC National Statement on Ethical Conduct in Human Research | 4 |
| Australia | Privacy Act 1988 (Australia) | 2 |
| International | ICH E6(R3) - Good Clinical Practice | 9 |
The counts are the controls this product reads, not the whole of the instrument. Where a regime is not held, the basis page names it and does not state its requirements.