A pre-check is the step that runs before a human reviewer ever opens the asset. It is not MLR. It does not replace MLR. Its job is narrow and useful: catch the preventable, mechanical failures in promotional content so that the medical, legal, and regulatory reviewers spend their time on judgment, not on counting whether the Important Safety Information is present.
That distinction matters more in 2026 than it did a year ago. On September 9, 2025, the FDA's Office of Prescription Drug Promotion sent 46 untitled letters in a single day, more than double the number it had sent in the previous five years combined, and closed the year with 74 letters to drug and biologic manufacturers, 42 of them aimed at direct-to-consumer television ads. The dominant theme across that wave was not exotic. It was omission and minimization of risk information. Those are exactly the failures a deterministic pre-check is built to catch before a reviewer is even involved.
This article walks through what a pre-check actually does, rule by rule, then draws the line it must never cross.
The pre-check is a triage layer, not a reviewer
Mid and large pharma teams report MLR cycles that stretch 50 to 60 days per content piece under current workflows. A large fraction of that time is consumed by errors that are mechanical: a missing ISI, an efficacy claim with no fair balance nearby, a reference that does not support the sentence it is attached to. A reviewer should never be the first line of defense against those. The pattern the industry has converged on is human-in-the-loop, where assistive automation handles the quantitative checks and qualified reviewers stay accountable for every final approval. The pre-check is the assistive layer. It surfaces issues. It does not decide.
So what does a pre-check actually do when an asset lands? Four things, in order.
Step 1: Extract the claims, from text and image both
A claim is any statement that asserts something about the product: an efficacy claim, a safety claim, a comparative claim, a mechanism claim. The pre-check parses the asset and isolates each claim as a discrete object with a location, because the rules that follow operate per claim, not per page.
Text is the easy half. The hard half is the image. A bar chart that shows a benefit, a callout box, a patient photo with a superimposed result, a visual that implies complete symptom resolution: all of these are claims, and the FDA treats them as such. Its 2025 enforcement letters repeatedly cited misleading efficacy conveyed through visual representations, including exaggerated quality-of-life improvement and implied complete resolution of symptoms. A pre-check that reads only the body copy and ignores the chart is checking half the asset. Claim extraction has to cover both modalities, and it has to record where each claim sits so the next steps can reason about prominence and proximity.
Step 2: Run the deterministic rules
This is the core of the pre-check, and the word that matters is deterministic. Each rule is a fixed, inspectable check with a defined input and a pass or fail output. It does not paraphrase, it does not summarize, it does not have an opinion. It asks a yes or no question and records the answer with its evidence. Four rule families do most of the work.
Fair balance against efficacy claims
Fair balance is the legal requirement that risk information appear with prominence and readability reasonably comparable to benefit information. The standard comes straight from 21 CFR 202.1, which requires a fair balance of benefit and risk and prohibits false or misleading claims, and the FDA's own guidance frames the test in concrete terms of typography, layout, contrast, and proximity. The deterministic rule mirrors that. For every efficacy claim the extractor found, it checks that corresponding risk information is present, is near the claim, and is presented with comparable weight. A benefit on page one and the risk language buried on page four fails the rule, and it fails it the same way every time.
ISI presence and completeness
Important Safety Information is the concise statement of the product's most significant risks, contraindications, warnings, and precautions. Every branded efficacy asset needs it. The rule is binary at its base, ISI present or absent, and then graded: does the ISI carry the contraindications and the boxed warning if one exists, or has a section been dropped. Omission of risk information was the single most cited violation across the 2025 enforcement wave, which is why a present-and-complete ISI check is the highest-value rule a pre-check runs.
On-label indication match
An advertisement may recommend a drug only for the uses in its approved labeling. 21 CFR 202.1 ties promotion to uses for which the drug is generally recognized as safe and effective, supported by adequate and well-controlled investigations. The on-label rule compares each indication or population the asset implies against the approved indication statement in the label. A claim that reaches beyond the indication, names an unapproved population, or implies a use the label does not carry is flagged with the specific label clause it failed against.
Reference substantiation
Every claim that rests on data needs a reference, and the reference has to actually support the claim. The rule links each substantiated claim to its cited source and checks the match. AI-assisted substantiation in this mode links claims to approved references with contextual matching and flags the ones where the link is wrong for manual review. A claim with no reference fails. A claim whose reference says something narrower than the claim fails. The reviewer sees the gap, not a green light.
Step 3: Return a verdict that cites the clause
A pre-check that says "this asset has problems" is nearly useless. A pre-check that says "the efficacy claim in the page-two headline has no risk information within the fold, which fails fair balance, and here is the label section the claim must align to" is something a reviewer can act on in seconds.
So the output is structured per finding: the claim, the rule it touched, the verdict, and the citation. The citation is the load-bearing part. Every flag points back to the controlling source, the specific label clause, the ISI section, the reference passage, so the reviewer is reading evidence, not trusting a score. This is also what makes the pre-check auditable, which leads directly to the guardrails.
What the pre-check must never do
The rules above describe capability. The guardrails describe restraint, and in a regulated setting the restraint is the product. Three lines a pre-check must not cross.
It never writes a new claim
The pre-check evaluates the claims that are already in the asset. It does not draft a better headline, it does not rewrite the ISI, it does not propose a compliant version of a failed claim. The moment a system generates promotional language, that language is itself a claim that must be reviewed, and a tool that both writes and clears its own writing has no independent check. The pre-check reads, scores, and cites. Composing the fix is the human's job, on purpose.
It never auto-publishes
A pass from the pre-check is not an approval. It is a clean triage result that moves the asset to human review faster. Nothing the pre-check does releases content to a channel. Given that over 200 enforcement letters challenged drug promotion in 2025, and that the FDA in two of those letters took an aggressive view of who is responsible for a promotional communication, no automated gate should be the thing that puts a message in front of a patient. A human approves. The system never does.
It backs the reviewer with a Part 11-supporting trail
When the reviewer does sign off, the system's job is to make that sign-off defensible. 21 CFR Part 11 requires secure, computer-generated, time-stamped audit trails capturing who did what and when, and these trails cannot be editable by any user. A pre-check that supports Part 11 records every claim it extracted, every rule it ran, every verdict it returned, and the reviewer's e-signature on the decision, in an immutable trail tied to a unique identity. The distinction in language is deliberate. The tool is Part 11-supporting. It is the customer who validates the system under their own standard operating procedures, runs it inside their quality framework, and owns the compliance posture. A vendor does not get to declare a customer's process compliant. The vendor supplies the controls; the customer validates the use.
A worked example: Varigel
Take a fictional brand, Varigel, approved for one narrow indication, with a known contraindication for patients on a common comorbidity medication. A brand team submits a new digital sales aid. The pre-check runs.
Claim extraction finds eleven claims. Nine are text. Two are visual: a bar chart showing response rate, and a callout reading "lasting relief." On-label match passes for ten of the eleven, but flags one: a subhead implies use in a broader patient population than the approved indication, and the finding cites the exact indication clause in the label it exceeds. Fair balance flags the response-rate chart, an efficacy claim with no risk information on the same screen, failing the prominence-and-proximity test. ISI presence passes, but the completeness grade flags that the contraindication line is absent from the ISI block. Reference substantiation links the "lasting relief" callout to its cited study and flags it: the study reports a defined duration, and "lasting" overstates it.
The verdict is four findings, each with its claim, its rule, and its citation. The pre-check writes nothing. It publishes nothing. It hands the reviewer a triaged asset with the mechanical failures already isolated and pinned to the label, and it logs the whole run, with the reviewer's eventual e-signature, in a Part 11-supporting trail. The reviewer spends the meeting on the one genuinely substantive question, whether that broader population can be supported at all, instead of the forty minutes it would have taken to find the missing contraindication line by hand.
The takeaway
A pre-check earns its place by being narrow. It extracts claims from text and image, runs deterministic rules for fair balance, ISI, on-label match, and reference substantiation, and returns a verdict that cites the controlling clause. It never authors a claim, never auto-publishes, and never declares itself compliant. It backs a human reviewer with a Part 11-supporting audit trail and e-signature, and leaves validation where it belongs, with the customer and their SOPs. Done that way, the pre-check does not weaken review. It gives the reviewer back the time to do the part only a human can.
Sources
- Latham & Watkins, "FDA Begins Crackdown on Direct-to-Consumer Pharmaceutical Advertising," September 2025. lw.com
- Sheppard Mullin, "FDA's Wave of Untitled Letters Signals Stricter Scrutiny for DTC Pharma Ads," October 2025. sheppard.com
- Covington & Burling, "FDA Advertising and Promotion Enforcement Activities: Update," October 2025. cov.com
- King & Spalding, "2025 Year in Review: FDA Drug and Device Advertising and Promotion Enforcement." kslaw.com
- Indegene, "Is the MLR Review Process Slowing Pharma Marketing?" indegene.com
- Indegene, "Future of MLR Review in Pharma: AI and MLR Operations." indegene.com
- Indegene, "AI-Powered MLR Review: Driving Speed, Consistency, and Compliance." indegene.com
- Legal Information Institute, "21 CFR 202.1 Prescription-drug advertisements." law.cornell.edu
- Electronic Code of Federal Regulations, "21 CFR 202.1 Prescription-drug advertisements." ecfr.gov
- SimplerQMS, "FDA 21 CFR Part 11 Audit Trails: Definition, Requirements, and Compliance." simplerqms.com