Latest / Let's Talk Risk! Podcast / Deep Dive: FDA's RWE Guidance

Transcript

Deep Dive: FDA's RWE Guidance · Let's Talk Risk! Podcast

0:00
SPEAKER_00: Right.
0:28
SPEAKER_00: We're breaking down the December 2025 FDA guidance that,
0:32
SPEAKER_00: frankly, just handed the industry a sledgehammer to smash right through that wall.
0:36
SPEAKER_00: Welcome to the Deep Dive, everyone.
0:38
SPEAKER_01: Thanks for having me.
0:39
SPEAKER_01: And yeah, you're absolutely right.
0:40
SPEAKER_01: This is a massive structural change for anyone listening who works in the medtech
0:45
SPEAKER_01: or life sciences space.
0:46
SPEAKER_00: Yeah, and we are custom tailoring today's Deep Dive specifically for you,
0:49
SPEAKER_00: the senior QA and RA leaders out there.
0:51
SPEAKER_01: Right, because you are the ones who have to actually build this out.
0:55
SPEAKER_01: So we're looking at the finalized guidance.
0:57
SPEAKER_01: It's officially titled Use of Real-World Evidence
1:00
SPEAKER_01: to Support Regulatory Decision-Making for Medical Devices.
1:03
SPEAKER_00: Which officially supersedes the older, much debated 2017 framework, right?
1:08
SPEAKER_01: Yes, exactly.
1:09
SPEAKER_01: What the agency has done here is they've basically reflected on nearly a decade
1:14
SPEAKER_01: of hard lessons from administrative reviews.
1:18
SPEAKER_01: They are fundamentally rewriting how life sciences companies have
1:21
SPEAKER_01: to approach regulatory submissions from here on out.
1:24
SPEAKER_00: I love that.
1:25
SPEAKER_00: What do you have in mind?
1:26
SPEAKER_00: Okay, very topical.
1:42
SPEAKER_00: Right.
1:42
SPEAKER_00: So you have your initial FDA clearance for adults with type 2 diabetes,
1:46
SPEAKER_00: but now you want to expand your label to include teenagers with type 1 diabetes.
1:50
SPEAKER_01: That is a perfect scenario because it perfectly highlights
1:52
SPEAKER_01: the fundamental difference between traditional trials and the actual real world.
1:57
SPEAKER_01: In the regulatory world,
1:58
SPEAKER_01: we differentiate between real world data, RWD, and real world evidence or RWE.
2:04
SPEAKER_00: Right.
2:04
SPEAKER_00: The data versus the evidence.
2:06
SPEAKER_01: Exactly.
2:07
SPEAKER_01: So real world data is just the raw material.
2:09
SPEAKER_01: It's the data routinely collected during patient care.
2:12
SPEAKER_01: We're talking, you know, electronic health records, administrative claims, data from registries.
2:17
SPEAKER_00: And in the case of our hypothetical CGM,
2:19
SPEAKER_00: the actual biometric data literally streaming from the patient's wearable device.
2:23
SPEAKER_01: Yes, that's the RWD.
2:25
SPEAKER_01: The real world evidence, the RWE, is what you actually build from that raw material.
2:30
SPEAKER_01: It's the clinical evidence regarding the CGM's usage or its benefits,
2:34
SPEAKER_01: its risks, and it's derived from a really rigorous analysis of that raw data.
2:38
SPEAKER_00: And just to set the stage before we get into the new rules,
2:41
SPEAKER_00: why is the FDA and the whole industry pushing so hard for RWE right now?
2:45
SPEAKER_00: I mean, traditional clinical trials have been the gold standard forever.
2:48
SPEAKER_01: Well, they have, but traditional highly controlled clinical trials have some severe limitations.
2:53
SPEAKER_01: I mean, they are fantastic for proving basic safety and efficacy in a vacuum.
2:58
SPEAKER_00: Right.
2:58
SPEAKER_00: I always think of traditional trial data like...
3:02
SPEAKER_00: Studying a plant in a highly controlled artificial greenhouse.
3:05
SPEAKER_00: You dictate the soil, the water, the temperature, everything.
3:08
SPEAKER_01: Yes, exactly.
3:08
SPEAKER_00: But real world evidence is like finally getting to see how
3:12
SPEAKER_00: that exact same plant survives out in the actual jungle.
3:16
SPEAKER_01: That is a brilliant analogy.
3:17
SPEAKER_01: The FDA desperately wants to understand those jungle dynamics because think about
3:22
SPEAKER_01: a traditional clinical trial for our AI enabled CGM.
3:26
SPEAKER_01: You enroll a very specific demographic of patients.
3:29
SPEAKER_01: Right.
3:29
SPEAKER_01: They get heavy monitoring.
3:31
SPEAKER_01: They have dedicated study coordinators calling them constantly
3:34
SPEAKER_01: to ensure they calibrate the device properly.
3:38
SPEAKER_00: Which does not reflect reality at all.
3:39
SPEAKER_00: It doesn't reflect a busy,
3:41
SPEAKER_00: I don't know,
3:42
SPEAKER_00: high school student playing soccer who forgets to calibrate their sensor,
3:46
SPEAKER_00: eats erratically, and drops their phone in a puddle.
3:48
SPEAKER_01: Not even a little bit.
3:49
SPEAKER_01: And RWE captures those broader clinical experiences.
3:52
SPEAKER_01: It vastly improves the generalizability of the data.
3:55
SPEAKER_01: You actually get to see how the device performs across highly diverse populations,
3:59
SPEAKER_01: varying comorbidities, unstructured environments.
4:02
SPEAKER_00: It spans the whole life cycle, right?
4:03
SPEAKER_01: Right.
4:03
SPEAKER_01: It allows for continuous long-term tracking across what the FDA calls
4:08
SPEAKER_01: the Total Product Life Cycle or TPLC.
4:11
SPEAKER_00: Plus, there's a massive technological urgency here, especially with our hypothetical AI device.
4:17
SPEAKER_00: Because AI and machine learning algorithms, they aren't static.
4:20
SPEAKER_01: No, they evolve constantly.
4:22
SPEAKER_00: Exactly.
4:23
SPEAKER_00: So a highly controlled greenhouse trial from three years ago tells
4:26
SPEAKER_00: an FDA reviewer absolutely nothing about how that algorithm is behaving out in
4:31
SPEAKER_00: the wild today.
4:32
SPEAKER_01: Spot on.
4:33
SPEAKER_01: And from a strategic perspective for the sponsors listening,
4:37
SPEAKER_01: utilizing RWE significantly reduces your time to market for innovative devices.
4:42
SPEAKER_01: Or like in our example for indication expansions, moving the CGM from adults to teenagers.
4:47
SPEAKER_00: Because you're leveraging data that already exists.
4:49
SPEAKER_01: Exactly.
4:50
SPEAKER_01: Rather than spinning up a multi-year, multi-million dollar prospective trial from scratch.
4:55
SPEAKER_00: Okay, so if the jungle offers these incredible insights
4:59
SPEAKER_00: and it saves sponsors literally millions of dollars,
5:02
SPEAKER_00: why haven't QA and RA teams just been utilizing it heavily all along?
5:05
SPEAKER_01: Well, because the jungle is incredibly messy.
5:08
SPEAKER_00: Fair point.
5:08
SPEAKER_01: Navigating that mess historically presented two massive roadblocks for sponsors.
5:13
SPEAKER_01: First, just the fundamental nature of observational data.
5:17
SPEAKER_01: Right.
5:17
SPEAKER_01: Like a physician in New York coding a glycemic event completely differently than
5:21
SPEAKER_01: a physician in London.
5:30
SPEAKER_01: Yeah, exactly.
5:31
SPEAKER_01: And beyond just missing data,
5:33
SPEAKER_01: observational data inherently introduces confounding factors and bias.
5:37
SPEAKER_01: How so?
5:38
SPEAKER_01: Well, in a randomized controlled trial, you mathematically control for bias.
5:43
SPEAKER_01: But in the real world,
5:44
SPEAKER_01: let's say an endocrinologist only prescribes our advanced AICGM
5:47
SPEAKER_01: to their absolute most brittle highest risk patients.
5:51
SPEAKER_00: Oh, I see where you're going with this.
5:52
SPEAKER_01: Right.
5:53
SPEAKER_01: So if the real-world data shows a higher rate of hypoglycemic events
5:56
SPEAKER_01: for our device compared to a competitor, is our device actually less safe?
6:01
SPEAKER_01: Or were our patients simply sicker to begin with?
6:03
SPEAKER_00: Wow, yeah, that confounding factor totally muddies the water.
6:06
SPEAKER_00: But as I was reviewing the Day 10 2025 guidance,
6:10
SPEAKER_00: data messiness wasn't even the primary bottleneck that stalled the industry, was it?
6:14
SPEAKER_01: No, it wasn't.
6:15
SPEAKER_01: The barrier that truly handcuffed QA and RA leaders was the strict privacy constraints.
6:21
SPEAKER_00: This brings us right back to the impossible paradox we talked about at the beginning.
6:25
SPEAKER_01: Yes.
6:26
SPEAKER_01: This was the fatal flaw in the previous 2017 framework.
6:30
SPEAKER_01: Previous FDA expectations practically mandated that your RWE submissions had
6:34
SPEAKER_01: to include private, confidential information at the individual patient level.
6:39
SPEAKER_00: They wanted the raw patient data.
6:41
SPEAKER_01: Yes.
6:41
SPEAKER_01: The agency wanted to physically inspect the raw participant-level data to verify
6:46
SPEAKER_01: the sponsor's safety claims.
6:48
SPEAKER_00: So QA and RA teams were being told by the agency
6:50
SPEAKER_00: to gather these massive real-world data sets to prove long-term safety.
6:54
SPEAKER_01: Right.
6:55
SPEAKER_00: But when they attempted to hand that macro-level data over,
6:58
SPEAKER_00: they were penalized by completely impossible privacy expectations.
7:01
SPEAKER_01: Exactly.
7:02
SPEAKER_01: Trying to maintain and submit large databases
7:05
SPEAKER_01: while navigating international privacy safeguards,
7:08
SPEAKER_01: think GDPR and HIPAA,
7:09
SPEAKER_01: all to provide individual-level traceability to the FDA,
7:13
SPEAKER_01: it was operationally and financially crippling.
7:15
SPEAKER_00: Just a total gridlock.
7:16
SPEAKER_01: Right.
7:17
SPEAKER_01: Sponsors simply could not maintain the data in a format the FDA would accept
7:21
SPEAKER_01: without exposing themselves to massive legal liability.
7:25
SPEAKER_01: And this regulatory gridlock was literally keeping better devices out of patients' hands.
7:31
SPEAKER_00: So how does the December 2025 guidance actually fix this?
7:36
SPEAKER_01: Oh, wow.
7:36
SPEAKER_01: That is huge.
7:53
SPEAKER_01: It's massive.
7:54
SPEAKER_01: De-identified,
7:55
SPEAKER_01: aggregate,
7:55
SPEAKER_01: and even international data sets are now officially highly viable
7:58
SPEAKER_01: for regulatory use.
8:00
SPEAKER_00: Okay, wait, hold on though.
8:00
SPEAKER_00: Let me play devil's advocate.
8:02
SPEAKER_00: If the FDA can't see the underlying patient data anymore,
8:05
SPEAKER_00: how do they know a sponsor didn't just, I don't know, cherry pick the good outcomes?
8:09
SPEAKER_00: Doesn't removing that raw data requirement make it incredibly easy for a company
8:13
SPEAKER_00: to hide safety issues?
8:14
SPEAKER_01: And that right there is the exact tension the new guidance addresses.
8:17
SPEAKER_01: Because the FDA absolutely did not lower their standards for scientific rigor.
8:21
SPEAKER_01: Instead, they radically shifted the burden of proof.
8:24
SPEAKER_00: Okay, so how does that work?
8:25
SPEAKER_01: Because the FDA can no longer directly inspect that raw participant-level data
8:31
SPEAKER_01: to catch cherry-picking,
8:32
SPEAKER_01: the burden of proving trustworthiness shifts entirely to your documentation as
8:37
SPEAKER_01: the sponsor.
8:38
SPEAKER_01: The agency now evaluates RWE strictly on a case-by-case basis
8:43
SPEAKER_01: through two unyielding pillars, relevance and reliability.
8:47
SPEAKER_00: Relevance and reliability.
8:49
SPEAKER_01: Yes.
8:50
SPEAKER_01: These are the defense mechanisms against the exact kind of fraud you were just talking about.
8:54
SPEAKER_00: So if I'm understanding this right, it feels like the FDA didn't lower the bar.
8:58
SPEAKER_00: They just changed the locks.
9:00
SPEAKER_01: Oh, I like that.
9:01
SPEAKER_00: Like you don't need to hand over the patient's exact name and address anymore,
9:04
SPEAKER_00: but you need an absolutely airtight,
9:06
SPEAKER_00: audit-ready receipt of exactly where that data came from and how it was handled,
9:11
SPEAKER_00: right?
9:11
SPEAKER_01: Change the locks captures the regulatory mindset perfectly.
9:14
SPEAKER_01: So let's look at that first lock, relevance.
9:17
SPEAKER_01: Relevance evaluates whether your data set is actually capable of answering
9:20
SPEAKER_01: the specific regulatory question you are asking.
9:23
SPEAKER_01: The FDA heavily scrutinizes clinical depth here.
9:26
SPEAKER_01: They want to know,
9:27
SPEAKER_01: does the data set actually contain the clinical or physiological measures needed
9:32
SPEAKER_01: to demonstrate safety or effectiveness?
9:35
SPEAKER_00: So applying that to our CGM case study,
9:37
SPEAKER_00: it's not enough to just pull billing codes showing a teenager went to the hospital
9:41
SPEAKER_00: for diabetes management.
9:42
SPEAKER_01: Exactly.
9:42
SPEAKER_01: Billing codes aren't enough.
9:44
SPEAKER_00: You would need the specific continuous glucose values,
9:47
SPEAKER_00: the time and range metrics, the actual A1C lab results to prove the device worked.
9:52
SPEAKER_01: Precisely.
9:52
SPEAKER_01: They also examine temporal alignment.
9:55
SPEAKER_01: Is the data from a relevant time frame that reflects the current standard of care?
9:59
SPEAKER_01: Because medical practices change rapidly.
10:02
SPEAKER_00: Yeah, that makes sense.
10:02
SPEAKER_01: And then they heavily scrutinize population representativeness.
10:06
SPEAKER_01: Does this data actually reflect the target population for your device in the U.S.?
10:11
SPEAKER_00: Ah, so if our CGM company pulls real-world data from a massive registry in Europe.
10:16
SPEAKER_01: Right.
10:16
SPEAKER_01: They must rigorously document the limitations of that European data
10:20
SPEAKER_01: and scientifically justify why European dietary
10:23
SPEAKER_01: and clinical practices legitimately apply to U.S. teenagers.
10:27
SPEAKER_00: Okay, that covers relevance.
10:29
SPEAKER_00: Does the data fit the question?
10:31
SPEAKER_00: So the second pillar,
10:32
SPEAKER_00: reliability, that seems to be where the airtight receipt really comes into play.
10:36
SPEAKER_01: Yes.
10:37
SPEAKER_01: Reliability dictates the mechanics of the data.
10:39
SPEAKER_01: How was it generated?
10:41
SPEAKER_01: How was it curated, processed, and validated?
10:44
SPEAKER_01: This covers data integrity, completeness, consistency.
10:48
SPEAKER_00: So you have to show your work.
10:49
SPEAKER_01: Completely.
10:50
SPEAKER_01: You must provide a very clear audit trail documenting data provenance,
10:54
SPEAKER_01: your collection methods, your quality controls.
10:57
SPEAKER_01: You literally have to explicitly show the statistical methods you use
11:01
SPEAKER_01: to address missing or inconsistent data.
11:04
SPEAKER_00: To ensure you aren't just conveniently ignoring unfavorable outcomes.
11:07
SPEAKER_01: Exactly.
11:08
SPEAKER_01: And the FDA didn't just issue these philosophical pillars and leave it up
11:12
SPEAKER_01: to industry interpretation.
11:13
SPEAKER_00: Right.
11:13
SPEAKER_00: They introduced highly specific procedural updates for how QA
11:17
SPEAKER_00: and RE leaders actually have to package this proof.
11:19
SPEAKER_01: Yes.
11:21
SPEAKER_01: The 2025 guidance mandates standardized submission expectations.
11:24
SPEAKER_01: This is to ensure consistency across all FDA review teams.
11:28
SPEAKER_01: So your regulatory submission cover letters must now explicitly state that
11:32
SPEAKER_01: the application contains RWE.
11:34
SPEAKER_00: It has to be right on the cover.
11:35
SPEAKER_01: Front and center.
11:37
SPEAKER_01: You must specify the study design,
11:38
SPEAKER_01: the primary objective of using the RWE, and exact details of the data sources.
11:44
SPEAKER_01: The name of the registry, the provider, the version, extraction dates every tiny detail.
11:50
SPEAKER_00: And then there's this brand new dedicated report that sponsors have to generate, right?
11:54
SPEAKER_01: Yes, the Relevance and Reliability Report.
11:57
SPEAKER_01: Sponsors submit this dedicated document detailing their entire data quality assessment.
12:02
SPEAKER_01: And to facilitate this,
12:03
SPEAKER_01: the FDA introduced a really vital practical tool,
12:07
SPEAKER_01: the Appendix A Documentation Checklist.
12:10
SPEAKER_00: Let's focus on Appendix A for a second because when I was reading it,
12:13
SPEAKER_00: it essentially operates like a treasure map or like GPS coordinates for
12:16
SPEAKER_00: the FDA reviewer.
12:17
SPEAKER_01: That's a great way to think about it.
12:19
SPEAKER_00: Instead of making an FDA reviewer dig through a 5,000-page PDF trying to find
12:23
SPEAKER_00: where you handle missing data,
12:25
SPEAKER_00: Appendix A literally forces you to provide the exact page,
12:30
SPEAKER_00: paragraph, and line containing the proof of your data's reliability.
12:34
SPEAKER_01: It forces structured transparency.
12:36
SPEAKER_01: You are basically mapping the exact location of your data dictionaries,
12:39
SPEAKER_01: your linkage methodologies, your validation steps.
12:42
SPEAKER_00: And didn't they update the submission software, too?
12:44
SPEAKER_01: They did.
12:45
SPEAKER_01: The FDA updated the electronic submission template and resource forms,
12:49
SPEAKER_01: known as eSTAR,
12:50
SPEAKER_01: in tandem with this guidance
12:51
SPEAKER_01: to natively integrate these RWE reporting recommendations.
12:55
SPEAKER_01: It just standardizes the entire pathway.
12:57
SPEAKER_00: Now there is one more crucial regulatory clarification in the guidance that we need
13:01
SPEAKER_00: to hit regarding Investigational Device Exemptions or IDEs.
13:06
SPEAKER_01: Oh yeah.
13:06
SPEAKER_01: This caused widespread confusion under the old rules.
13:09
SPEAKER_00: Because companies didn't know if they needed an IDE just to look at old data.
13:13
SPEAKER_01: Right.
13:13
SPEAKER_01: So the new guidance clarifies that retrospective analyses
13:16
SPEAKER_01: of real-world data typically do not require an IDE.
13:20
SPEAKER_01: The caveat is, unless the data collection process itself alters standard patient care.
13:25
SPEAKER_00: Okay, so if our CGM company is Just looking back at wearable data and EHR data
13:31
SPEAKER_00: that was already routinely collected during normal daily life
13:34
SPEAKER_00: and standard clinic visits.
13:35
SPEAKER_01: Then they generally do not need an IDE to analyze it.
13:38
SPEAKER_01: The guidance even provides hypothetical examples to really help sponsors navigate
13:42
SPEAKER_01: that boundary of what constitutes altering patient care.
13:44
SPEAKER_00: Which removes a mountain of administrative friction for retrospective studies.
13:48
SPEAKER_01: Absolutely a mountain.
13:49
SPEAKER_00: Okay.
13:50
SPEAKER_00: So we understand the mechanics of the guidance now.
13:53
SPEAKER_00: Let's translate this high-level policy into a Tuesday morning reality for the QA
13:57
SPEAKER_00: and RA leaders listening.
13:59
SPEAKER_00: How do you actually execute this architecture?
14:02
SPEAKER_00: We reviewed some highly strategic lessons from industry experts at Castor and IQVIA
14:08
SPEAKER_00: on operationalizing this.
14:10
SPEAKER_01: Right.
14:10
SPEAKER_01: And Castor's main point is something every leader needs to hear.
14:13
SPEAKER_01: They fundamentally argue
14:15
SPEAKER_01: that medtech organizations must stop treating postmarket surveillance as
14:18
SPEAKER_01: a static compliance exercise.
14:20
SPEAKER_00: Like a box to check.
14:21
SPEAKER_01: Exactly.
14:22
SPEAKER_01: Historically,
14:23
SPEAKER_01: companies would just build a basic cheap registry solely to satisfy
14:26
SPEAKER_01: a postmarket surveillance requirement from the FDA.
14:30
SPEAKER_01: It was viewed as a sunk cost.
14:31
SPEAKER_00: Yes.
14:32
SPEAKER_00: You are burning a potential strategic asset.
14:53
SPEAKER_01: Let's break down what that actually means mechanically.
14:56
SPEAKER_01: We're talking about capturing electronic patient-reported outcomes, or EPRO.
14:58
SPEAKER_01: Right.
15:12
SPEAKER_00: This means our hypothetical CGM company isn't waiting for a teenager to visit
15:16
SPEAKER_00: a clinic every six months to lazily fill out a paper survey.
15:19
SPEAKER_00: They are actively pinging the patient's smartphone app to ask
15:23
SPEAKER_00: about their dietary inputs or symptoms in real time.
15:26
SPEAKER_01: Yes, exactly.
15:27
SPEAKER_01: And by integrating that EPRO data with a pseudonymized infrastructure,
15:31
SPEAKER_01: you are completely protecting the patient's protected health information.
15:34
SPEAKER_01: But crucially, you're still maintaining the investigator-level audit trail the FDA now demands.
15:38
SPEAKER_00: So it's a win-win.
15:39
SPEAKER_01: It is.
15:40
SPEAKER_01: If you build that data asset correctly for your initial post-market compliance,
15:44
SPEAKER_01: you can reuse that exact same data asset three years later
15:47
SPEAKER_01: to support your label expansion to teenagers,
15:50
SPEAKER_01: without starting from scratch with a massive new clinical trial.
15:53
SPEAKER_00: That is incredible.
15:54
SPEAKER_00: Now, moving to IQVIA's lesson,
15:57
SPEAKER_00: they leaned heavily into the mechanics of transparency and advanced privacy.
16:01
SPEAKER_01: Right, because,
16:02
SPEAKER_01: again, the FDA is not inspecting your raw data to look for cherry picking anymore.
16:07
SPEAKER_01: Your documentation carries the entire burden.
16:10
SPEAKER_00: So how do you prove that trustworthiness technically?
16:13
SPEAKER_01: IQVIA emphasizes rigorous validation studies.
16:17
SPEAKER_01: You absolutely must run sensitivity analyses
16:20
SPEAKER_01: to demonstrate how robust your findings are against potential confounding variables.
16:25
SPEAKER_01: And you need crystal clear endpoint definitions.
16:28
SPEAKER_00: Like if you use surrogate endpoints because you can't observe
16:31
SPEAKER_00: the primary endpoint directly in the real world.
16:33
SPEAKER_01: Exactly.
16:33
SPEAKER_01: If you use a surrogate endpoint, your scientific justification for doing so has to be flawless.
16:38
SPEAKER_00: IQVIA also pointed to advanced privacy-preserving techniques like tokenization.
16:43
SPEAKER_00: Now, the concept of linking disparate data sets
16:46
SPEAKER_00: without exposing identity sounds like literal magic to a lot of people.
16:49
SPEAKER_01: It really does.
16:50
SPEAKER_00: How does tokenization actually work mechanically?
16:52
SPEAKER_01: Okay, so tokenization uses cryptographic algorithms to replace
16:57
SPEAKER_01: a patient's personally identifiable information like their name and date of birth
17:01
SPEAKER_01: with a random string of characters.
17:03
SPEAKER_01: That string is known as a token.
17:05
SPEAKER_00: Okay.
17:06
SPEAKER_01: Let's say our CGM company wants to link their wearable app data with
17:10
SPEAKER_01: a massive national insurance claims database.
17:15
SPEAKER_01: They want to see if their device actually reduces emergency room visits for teens.
17:19
SPEAKER_00: But they can't share names.
17:20
SPEAKER_01: Right.
17:21
SPEAKER_01: But because the CGM cloud server and the insurance company use
17:24
SPEAKER_01: the exact same encryption key,
17:27
SPEAKER_01: they mathematically generate the exact same cryptographic token
17:30
SPEAKER_01: for our hypothetical teenager.
17:32
SPEAKER_00: Oh, wow.
17:33
SPEAKER_01: So the sponsor just matches the CGM data to the hospital claims data using that random token.
17:38
SPEAKER_01: It creates a complete longitudinal picture of the patient's journey
17:41
SPEAKER_01: without anyone ever actually knowing the patient's identity.
17:44
SPEAKER_00: That is brilliant.
17:45
SPEAKER_00: It builds the exact comprehensive RWA architecture the FDA wants
17:49
SPEAKER_00: while remaining fully compliant with global privacy laws.
17:52
SPEAKER_01: It's elegant.
17:52
SPEAKER_00: Yeah.
17:53
SPEAKER_01: But, and this is a big bod, Matt,
17:54
SPEAKER_01: a sponsor cannot just drop a massive tokenized data set on an FDA reviewer's desk
17:59
SPEAKER_01: and just hope they accept the methodology.
18:01
SPEAKER_00: Right.
18:02
SPEAKER_00: Which leads to IQVAA's most critical strategic advice.
18:05
SPEAKER_01: Yeah.
18:06
SPEAKER_00: Engage the FDA early.
18:07
SPEAKER_01: Yes.
18:08
SPEAKER_01: Because RWE acceptability is strictly evaluated case by case,
18:12
SPEAKER_01: pre-submission discussions are non-negotiable.
18:15
SPEAKER_00: So if you are listening to this and you are preparing for
18:17
SPEAKER_00: a Q submission meeting next month,
18:20
SPEAKER_00: do not walk into that room with just a vague idea of using tokenized registry data.
18:25
SPEAKER_01: Please don't.
18:25
SPEAKER_00: You need to walk in with your endpoint strategy already mapped out,
18:29
SPEAKER_00: knowing exactly how you are dealing with missing EHR data.
18:32
SPEAKER_01: If you wait for the FDA to tell you how to clean your data,
18:35
SPEAKER_01: you have already lost six months of your timeline.
18:37
SPEAKER_00: At least six months.
18:39
SPEAKER_00: Proactive interaction is the only viable path.
18:42
SPEAKER_00: You use the cue submission process to ask regulators highly specific,
18:45
SPEAKER_00: targeted questions about your study protocol and your relevance
18:49
SPEAKER_00: and reliability assessments.
18:50
SPEAKER_01: Get alignment before you build.
18:52
SPEAKER_00: Exactly.
18:53
SPEAKER_00: You must align with regulators on your data sources,
18:55
SPEAKER_00: your validation plans,
18:56
SPEAKER_00: and evidentiary standards before you invest millions into the architecture.
18:59
SPEAKER_00: All right,
19:00
SPEAKER_00: to sum all of this up,
19:02
SPEAKER_00: The December 2025 FDA guidance on real-world evidence really is
19:07
SPEAKER_00: a profound regulatory evolution.
19:09
SPEAKER_00: It hands the industry unprecedented flexibility by officially removing
19:13
SPEAKER_00: the requirement for individual-level confidential data.
19:17
SPEAKER_00: It fully allows the use of macro-level de-identified data sets.
19:21
SPEAKER_01: But that flexibility demands unyielding standards for scientific rigor.
19:24
SPEAKER_01: The burden of proof now rests entirely on your documentation.
19:28
SPEAKER_01: You have to definitively prove relevance,
19:31
SPEAKER_01: that the data fits the clinical question and reliability,
19:34
SPEAKER_01: that the data was generated and handled impeccably.
19:37
SPEAKER_00: And you have to master the new standardized packaging,
19:39
SPEAKER_00: like the Appendix A checklist,
19:40
SPEAKER_00: the relevance and reliability report, and that native ESTAR integration.
19:44
SPEAKER_01: As a senior key array leader,
19:45
SPEAKER_01: your immediate takeaway should be to audit your current data architecture.
19:49
SPEAKER_01: Treating post-market data as a reusable strategic asset rather than just
19:52
SPEAKER_01: a compliance burden is the absolute key
19:55
SPEAKER_01: to accelerating your future device approvals.
19:57
SPEAKER_00: RWE has clearly transitioned from a nice-to-have supplementary file to a core pillar
20:02
SPEAKER_00: of regulatory strategy across the entire total product lifecycle.
20:06
SPEAKER_00: Well, I want to leave you,
20:06
SPEAKER_00: our listeners,
20:07
SPEAKER_00: with a final,
20:08
SPEAKER_00: somewhat provocative thought
20:09
SPEAKER_00: as you review your current regulatory pipelines this week.
20:13
SPEAKER_00: The FDA is aggressively removing barriers to de-identified real-world data.
20:19
SPEAKER_00: It's a great question.
20:46
SPEAKER_00: Definitely something to think about as you design your next registry.
20:49
SPEAKER_00: Thanks for joining us on this deep dive.